--- title: "When Consumers Prefer Point Versus Range Estimates of Product Performance" authors: "Amin Shiri, Felipe M. Affonso, Keith Wilcox" journal: "Journal of Marketing Research" year: 2026 doi: "10.1177/00222437261490921" citation: "Shiri, Amin, Felipe M. Affonso, and Keith Wilcox (2026), \"When Consumers Prefer Point Versus Range Estimates of Product Performance,\" Journal of Marketing Research, forthcoming. https://doi.org/10.1177/00222437261490921" version: "Accepted manuscript (JMR-24-0926.R3, accepted 2026-09-04), before copyediting" --- > **Disclaimer:** This is a machine-readable conversion of the accepted manuscript (after peer review, before copyediting) of the published paper, for use with AI tools. It may contain conversion errors in formatting, tables, or equations, and the published version may differ in wording after copyediting. Always verify against the [published version](https://doi.org/10.1177/00222437261490921). Figures and stimulus images are replaced by written descriptions (alt text plus a long description); the images themselves are not included. ```bibtex @article{shiri2026point, author = {Shiri, Amin and Affonso, Felipe M. and Wilcox, Keith}, title = {When Consumers Prefer Point Versus Range Estimates of Product Performance}, journal = {Journal of Marketing Research}, year = {2026}, note = {Forthcoming}, doi = {10.1177/00222437261490921} } ``` **When Consumers Prefer Point Versus Range Estimates of Product Performance** Amin Shiri Felipe M. Affonso Keith Wilcox Amin Shiri (amin.shiri@uga.edu) is Assistant Professor of Marketing, University of Georgia, USA. Felipe M. Affonso (felipe.affonso@okstate.edu) is Assistant Professor of Marketing, Spears School of Business, Oklahoma State University, USA. Keith Wilcox is Professor of Marketing and the Macy's Foundation Professor at the Mays Business School, Texas A&M University, Wehner 220S, College Station, TX, Phone: 979-845-3028, kwilcox@mays.tamu.edu. Please address correspondence to Amin Shiri. Supplementary materials are included in the Web Appendix accompanying the online version of this article. Acknowledgments: The authors thank the review team for their constructive guidance across the review process. During the preparation of this manuscript, the authors used generative AI tools (Claude Code 2.1.220 using Claude Fable 5) to assist with verification of statistical reporting against raw data. All analyses, results, and conclusions were produced and verified by the authors, who take full responsibility for the content of this article. No references were generated with AI tools. Data Availability Statement: The data, code, and materials that support the findings of this article are available in the Journal of Marketing Research Dataverse at https://doi.org/10.7910/DVN/JIZHV9. Declarations: The authors declare no conflicts of interest. All studies were conducted in accordance with the authors' institutional review board policies, and participants provided informed consent. **When Consumers Prefer Point Versus Range Estimates of Product Performance** **Abstract** Product performance claims play an important role in shaping consumers' purchase decisions. Through six preregistered studies, including one field study, the authors demonstrate that consumers' preferences for point versus range performance estimates in common comparative purchase settings depend on whether their attention is directed toward the content of the claim or the manner of its presentation. At baseline, consumers focus on what information is communicated and prefer point over range estimates because point estimates convey greater certainty about the expected performance outcome. When consumers' persuasion knowledge is activated, however, their attention shifts toward how information is communicated. In this mode, the format of the claim, rather than the stated performance figures, becomes relatively more informative: range estimates signal brand transparency by acknowledging the natural variability of real-world performance, enhancing perceived trustworthiness and reversing consumer preference. These effects are attenuated when products are evaluated in isolation rather than compared side by side. Collectively, these findings inform both theory and practice by revealing when and why consumers prefer point versus range estimates of product performance. *Keywords*: uncertainty, point and range estimates, product performance, persuasion knowledge Product performance claims play an important role in shaping consumers' purchase decisions. Whether it is the travel distance of an electric vehicle or the battery life of a device, consumers often depend on these claims to make purchase decisions (Nelson 1974; Palmeira et al. 2020). Brands communicate performance in different ways, most commonly through either point estimates (e.g., "5 hr energy boost") or range estimates (e.g., "4--6 hr energy boost"). This variation appears not only across brands but also within the same brand and even the same retailer. For instance, on Amazon, candles with point estimates of burn time (e.g., "30 hours of burn time") are often displayed alongside candles with range estimates (e.g., "28--32 hours of burn time"). Similarly, BMW recently promoted its electric vehicles by reporting a point estimate for one model's battery range (e.g., "296 miles") and a range estimate for another (e.g., "305--311 miles") in the same advertisement (see Web Appendix H for this and additional examples). Although consumers are routinely exposed to both point and range estimates, prior research has not directly examined how these formats influence consumers' preferences. The present research addresses this gap by demonstrating that consumers' preference for point versus range estimates depends on their attentional focus when evaluating the claims. At baseline, consumers focus on what information is communicated and prefer point over range estimates because point estimates convey greater certainty about the expected performance outcome. When persuasion knowledge is activated, however, they shift attention toward how information is communicated and approach claims with skepticism. This shift leads them to prefer range estimates, which are perceived as more trustworthy acknowledgments of natural performance variability. Our investigation focuses on contexts in which consumers compare products that share similar attribute values, with the format of a single performance claim (point versus range) serving as the key distinguishing feature between options. This research makes several contributions. First, we extend work on persuasion knowledge. Activating persuasion knowledge shifts consumers into an agentic, inference-driven stance that can result in skepticism and less favorable evaluations. However, persuasion knowledge can enhance evaluation when brands utilize credible persuasion tactics (Isaac and Grayson 2017). We build on this insight by demonstrating that persuasion knowledge can also lead to more favorable evaluations when brands employ range estimates, even though consumers perceive point and range estimates to be equally credible. In doing so, we establish perceived trustworthiness as the mechanism driving this effect. Second, we contribute to research on uncertainty aversion, which is a well-established phenomenon describing people's tendency to avoid ambiguous or uncertain options (e.g., Gneezy, List, and Wu 2006; Newman and Mochon 2012). Because range estimates inherently convey uncertainty, our finding that consumers avoid range estimates in many situations is consistent with uncertainty aversion. However, by showing that consumers prefer range estimates when persuasion knowledge is activated, we demonstrate a novel condition under which consumers may become uncertainty seeking. These findings advance understanding of the contextual factors that shape consumer preferences for uncertainty. Finally, our research offers actionable implications for managers. Currently, marketers appear to communicate performance information in an ad hoc manner, without systematic consideration of how point versus range estimates influence consumer response. Our findings suggest that marketers can be more strategic. We show that point estimates are generally more effective, but range estimates become advantageous when persuasion knowledge is activated, such as when products appear in sponsored listings on platforms such as Amazon or Google Shopping. Importantly, our findings suggest that in situations where persuasion knowledge is activated, marketers must calibrate ranges carefully, as overly wide ranges diminish their effectiveness. **Theoretical Background** ***Range Versus Point Estimates in Marketing*** Marketers frequently use numerical information to inform and persuade consumers, who in turn rely on the information to anticipate product performance or benefits (Nelson 1974; Van Osselaer and Janiszewski 2012). In these contexts, numerical product attributes serve as signals of potential outcomes. Such information is pervasive across marketing domains, including pricing, promotions, product specifications, and financial forecasts (Janiszewski and Uy 2008). Importantly, the format in which numbers are presented, whether as points or ranges, shapes consumers' interpretation and responses (Thomas and Morwitz 2009; Zhang and Schwarz 2013). Previous research has explored how consumers respond to numerical information presented in point versus range formats across a variety of contexts. In pricing, price points (e.g., "\$49.99") tend to generate higher purchase intentions than price ranges (e.g., "\$45--\$55"), as consumers often anchor on the upper bound of the range when evaluating value (Janiszewski and Lichtenstein 1999). Similar effects appear in promotions, where point discounts increase deal attractiveness relative to range discounts (Biswas and Burton 1993; Biswas and Burton 1994). The use of point versus range formats has also been investigated in social exchanges such as negotiations (Ames and Mason 2015). We extend this line of research to product performance claims. A distinctive feature of such claims is that the outcomes they describe depend on the complex interplay between product capabilities and usage conditions (Bolton and Drew 1991; Zeithaml, Berry, and Parasuraman 1993). For example, an electric vehicle's battery range fluctuates with driving speed, terrain, and weather conditions (Jägerbrand and Sjöbergh 2016), while a laptop's battery life varies with screen brightness, processing demands, and usage patterns (Yan, Tan, and Fu 2019). Accordingly, performance claims are estimates of how products are likely to perform. Consumers themselves recognize this variability in performance attributes from their everyday experience with products. Of course, not all product attributes are perceived as variable. Specifications such as storage capacity, screen size, or weight are fixed properties that do not change with usage, and for these attributes, providing a range estimate might signal ignorance or deception, not transparency. Friestad and Wright (1994) refer to this accumulated understanding as "topic knowledge," one of three knowledge structures that consumers bring to persuasion encounters. Topic knowledge encompasses consumers' beliefs about the product domain, including the recognition that real-world performance varies across usage conditions. To validate the premise that consumers draw this distinction, we conducted a pilot in which 200 Prolific participants (M~age~ = 42.2, SD = 13.7; 51.5% female) rated how much they believed each of six product attributes could vary depending on usage conditions (1 = not at all, 7 = very much). A t-test showed that the perceived variability of performance attributes (M = 5.65, SD = 0.91) was significantly higher than that of non-performance attributes (M = 2.76, SD = 1.59; t(199) = 22.55, *p* < .001, d = 1.59). See Web Appendix G for full details. Given that consumers perceive real-world performance as inherently variable, the way a marketer chooses to present a performance claim, whether as a point or a range, could itself become informative. Yet how consumers respond when performance claims are presented in these two formats remains largely unexplored. In the next section, we examine how consumers may react to performance claims communicated in point versus range formats. ***Attentional Focus on Product Claims*** Previous research suggests that when consumers evaluate product claims, they can adopt different modes of attention that shape their responses to the information (Wickens 2021). One common mode involves attending to *what* information is communicated. In this mode, attention centers on product attributes, expected outcomes, and measurable indicators of performance (Anderson 1981; Bettman, Luce, and Payne 1998). The goal is to use the available information to form expectations about how the product will function in practice. This process reflects an effort to link product claims directly to real-world outcomes. Within this mode, performance claims are interpreted in ways that shape consumers' beliefs about expected outcomes. Point estimates (e.g., "5 hours of battery life") are viewed as deterministic predictions about a future outcome, whereas range estimates (e.g., "4-6 hours of battery life") are understood as bounded sets of possible outcomes (Lynch and Srull 1982; Yaniv and Foster 1997; Teigen and Jørgensen 2005). Therefore, when consumers are in this mode, they attend to the substantive information being conveyed, and their responses depend on how they interpret that information. Research also finds that when consumers evaluate product claims, they can focus on *how* information is communicated. This mode typically emerges when persuasion knowledge is activated (Campbell 1995; Campbell and Kirmani 2000; Kirmani and Zhu 2007; Eisend and Tarrahi 2022). Persuasion knowledge refers to consumers' awareness of marketers' tactics and motives, as well as their judgments about the appropriateness of those tactics (Friestad and Wright 1994). Once activated, it increases the relative salience of the manner of presentation over the substantive content of the claim. As a result, consumers become relatively less concerned with linking product claims to real-world outcomes and relatively more attentive to what the format of the claim signals about the marketer. Therefore, when persuasion knowledge is activated, consumers use performance claims as cues to evaluate the marketer. A point estimate, for instance, may be interpreted as an attempt to project confidence or to oversimplify performance in a way that risks appearing overreaching. By contrast, a range estimate may be taken as a signal of transparency because it acknowledges the natural variability of product performance across usage conditions (Mohan, Buell, and John 2020). Ultimately, the meaning consumers derive from performance claims in this mode reflects judgments about the marketer's tactics rather than expectations about the performance outcome. Which of these two interpretations prevails is adjudicated by consumers' topic knowledge. A point estimate could, in principle, signal confidence and expertise. Because topic knowledge holds that real-world performance varies with usage conditions, however, the precision a point estimate promises exceeds what consumers regard as deliverable, so a point estimate cannot earn the competence premium the confidence interpretation anticipates and instead risks an overreaching reading. A range that spans the variability consumers already expect matches this reference standard and earns the transparency credit instead. We state the resolution comparatively: relative to a range, a point estimate is less able to earn credit for transparency once evaluation is directed at the marketer's conduct. Importantly, the magnitude of a claim and the precision of its format are distinct. A point estimate at the category norm is plausible in magnitude and is taken at face value at baseline; its precision becomes motive-relevant only once persuasion knowledge directs evaluation toward the marketer's conduct, which is why plausible point claims do not activate persuasion knowledge on their own. The same reference standard implies the limit we test in Study 4: a range that exceeds the variability consumers consider realistic no longer matches topic knowledge and loses the transparency credit. These two modes of attention are conceptually distinct. Although they typically co-occur, one tends to predominate (Campbell and Kirmani 2000; Friestad and Wright 1994; Isaac and Grayson 2017). In most situations (i.e., at baseline), consumers focus primarily on what is communicated (Bettman et al. 1998; Lynch and Srull 1982). However, when persuasion knowledge is activated, attention shifts toward how information is communicated. To provide direct evidence for our proposed attention-shifting mechanism, we conducted a pilot study testing whether consumers focus on what is communicated at baseline and how information is communicated when persuasion knowledge is activated. Two-hundred Prolific participants evaluated two otherwise identical water filter pitchers differing only in filter lifespan (Brand A: "40 gallons" vs. Brand B: "35-45 gallons"). Persuasion knowledge was manipulated using instructions adapted from Isaac and Grayson (2017), which directed participants to consider brands' underlying motives, while control participants evaluated the information as they normally would. Attentional focus was measured with three 7-point items (α = .87): "I found myself concentrating more on\...," "When considering the information, my focus was on\...," and "While reading I focused on\..." (1 = the substance of the claims being made, 7 = how it was being communicated/presentation style). See Web Appendix G for full details. Results showed that participants in the persuasion knowledge condition paid significantly more attention to how claims were presented (M = 4.28, SD = 1.30) than those in the control condition (M = 2.78, SD = 1.42; F(1, 198) = 60.84, *p* < .001, ηp² *=* .235). The mean in the persuasion knowledge condition was modestly above the scale midpoint (t(99) = 2.15, *p* = .034, *d* = .22), whereas the mean in the control condition was well below it (t(99) = -8.60, *p* < .001, *d* = -.86). These results suggest that activating persuasion knowledge increases the relative salience of how information is communicated rather than fully redirecting attention away from the substance of the claim. This relative shift is sufficient for our theorizing, because it implies that format-based inferences about the marketer become more diagnostic under activation even as consumers continue to process the content of the claim. In the next section, we develop our predictions about when consumers will prefer point or range estimates of product performance. **Consumer Preference for Point Versus Range Estimates of Performance** We propose that consumers' responses to point versus range estimates depend on whether they focus on what or how information is communicated. We predict that at baseline, when consumers focus on what is communicated, they will prefer point estimates to range estimates. By contrast, when persuasion knowledge is activated and consumers focus on how information is communicated, they will prefer range estimates to point estimates. In the sections below, we elaborate on each pathway of our conceptual model presented in Figure 1. **Figure 1:** Conceptual Model of Preference for Point Versus Range Estimates **Alt text:** Figure 1. Two-panel conceptual model: at baseline a range estimate raises perceived uncertainty and lowers purchase likelihood; with persuasion knowledge active a range estimate raises perceived trustworthiness and raises purchase likelihood. [Two-panel diagram. Each panel is a left-to-right chain of three rounded boxes joined by arrows, with a sign above each arrow. Panel A: Baseline. The box "Range (vs. Point) Performance Attribute" points to "Perceived Uncertainty" with a plus sign, and "Perceived Uncertainty" points to "Purchase Likelihood" with a minus sign: a range estimate increases perceived uncertainty, which decreases purchase likelihood. Panel B: Persuasion Knowledge Active. The box "Range (vs. Point) Performance Attribute" points to "Perceived Trustworthiness" with a plus sign, and "Perceived Trustworthiness" points to "Purchase Likelihood" with a plus sign: a range estimate increases perceived trustworthiness, which increases purchase likelihood.] *Note: Figure described by AI (Claude Fable 5.1) from the figure source file (Figure 1.tex, TikZ) and the manuscript image. Values cross-checked against the paper text and the source file. The figure image is not included in this machine-readable version; see the published paper for the visual.* ***Baseline (What is Communicated)*** At baseline, when consumers focus on what information is communicated, range estimates tend to create uncertainty about performance outcomes. Unlike point estimates that offer a deterministic prediction about performance levels, ranges communicate bounded possibilities, making it difficult to form deterministic expectations about product utility (Lynch and Srull 1982; Yaniv and Foster 1997; Teigen and Jørgensen 2005). For example, a consumer seeing "removes 65--75% of stains" faces uncertainty about whether the product will consistently perform at the higher end or frequently fall toward the lower end of the stated range. Moreover, research on range processing shows that consumers often struggle with distributional information, sometimes fixating on endpoints without considering the full spectrum of possibilities, which can amplify uncertainty (Janiszewski and Lichtenstein 1999). By contrast, point estimates reduce perceived uncertainty by providing a single anchor for expectations (Zhang and Schwarz 2013). Even when consumers recognize that point estimates may represent idealized conditions, the cognitive advantage of having a deterministic reference point typically outweighs concerns about realism (Janiszewski and Lichtenstein 1999). A significant amount of research finds that consumers display an aversion to uncertainty (Gneezy et al. 2006; Newman and Mochon 2012). The uncertainty effect, for example, shows that people value an uncertain prospect (e.g., a 50% chance at winning a \$100 gift card and a 50% chance at winning a \$50 gift card) less than its worst possible outcome (e.g., a *certain* \$50 gift card) (Gneezy et al. 2006). The uncertainty effect is robust across various choice contexts (Newman and Mochon 2012; Trautmann and van de Kuilen 2015). Consumer avoidance of uncertainty can lead them to prefer less uncertain options even when more uncertain alternatives might offer superior benefits (see Camerer and Weber 1992). Therefore, at baseline, we predict that consumers will prefer a product whose performance is presented using a point estimate to one whose performance is presented using a range estimate, because consumers feel more certain about the performance of the product that uses a point estimate. Next, we consider consumers' response to point (vs. range) estimates when persuasion knowledge is activated. ***Persuasion Knowledge Activation (How it is Communicated)*** When persuasion knowledge is salient, and consumers focus on how information is communicated, they approach product claims with greater skepticism (Escalas 2007; Nielsen and Escalas 2010). In these situations, they are more likely to analyze information objectively to draw their own conclusion about the information presented (Obermiller, Spangenberg, and MacLachlan 2005). As a result, consumers not only become more suspicious of promotional appeals (Campbell 1995; Kirmani and Zhu 2007) but also direct their attention toward assessing whether the information presented accurately reflects reality (Escalas 2007; Obermiller and Spangenberg 1998; Obermiller et al. 2005). This shift in attention changes how consumers interpret performance claims. When persuasion knowledge is active, consumers' topic knowledge (Friestad and Wright 1994) about real-world performance variability becomes a salient standard against which claims are evaluated. Consumers know from experience that an electric vehicle's range depends on driving conditions, that battery life varies with usage patterns, and that cleaning effectiveness differs across applications. Because this experiential understanding is now at the forefront, consumers are more likely to view range estimates as realistic reflections of product performance and, therefore, as signals of transparency (Mohan et al. 2020). This transparency about product performance variability should enhance brand trustworthiness (Erdem and Swait 2004; see also Khamitov et al. 2024), which will have a positive effect on purchase intentions. Therefore, when persuasion knowledge is activated, we predict that consumers will prefer a product with a range estimate of product performance to one that uses a point estimate. This preference will occur because consumers perceive the brand that uses a range estimate to be more trustworthy. Importantly, we do not expect point versus range estimates of product performance to influence the non-focal perception in either condition. The accessibility-diagnosticity framework is fundamentally a theory of judgment construction. Perceptions such as outcome uncertainty and brand trustworthiness are not fixed evaluations that consumers continuously maintain, but are instead constructed at the time of judgment from the information that is both accessible and perceived to be diagnostic for the specific inference being made (Feldman and Lynch 1988; Kardes, Posavac, and Cronley 2004). Diagnosticity therefore determines the extent to which a cue contributes to constructing a particular perception. Cues perceived to be highly diagnostic receive greater weight in constructing that perception, whereas cues perceived to be less diagnostic receive little or no weight. In the present research, the range-versus-point format cue is accessible in both conditions, but its perceived diagnosticity changes depending on consumers' attentional focus. Under baseline conditions, consumers view the format as highly diagnostic of outcome uncertainty. Consequently, the format cue receives substantial weight when constructing uncertainty judgments but little weight when constructing brand trustworthiness judgments, leading range estimates to increase perceived uncertainty while leaving trustworthiness unchanged. When persuasion knowledge is activated, however, consumers instead view the same format cue as diagnostic of the firm's transparency and honesty. As a result, the cue receives greater weight when constructing brand trustworthiness judgments and less weight when constructing uncertainty judgments, increasing trustworthiness without affecting perceived uncertainty. Thus, our argument concerns first-stage moderation in the judgment formation process (Edwards and Lambert 2007; Muller, Judd, and Yzerbyt 2005). Activating persuasion knowledge changes the diagnosticity of the format cue for different inferences, thereby changing the weight the cue receives when constructing perceptions of uncertainty versus trustworthiness. The observed differences in the measured mediators therefore reflect differences in how the cue contributes to constructing those perceptions, rather than differences in the weight assigned to uncertainty or trustworthiness after those perceptions have already been formed. Two features of this account define its scope. First, the format acquires motive-relevant meaning only upon activation, which follows from the change-of-meaning principle of the persuasion knowledge model (Friestad and Wright 1994): once persuasion knowledge is active, the choice of a point or a range ceases to be ordinary product information and becomes conduct relevant to evaluating the marketer's tactics. Second, the baseline prediction is scoped to claims that fall within the plausible bounds implied by topic knowledge, as ours do. Claim extremity is itself a bottom-up activator of persuasion knowledge, because sufficiently salient self-serving tactics elicit inferences of manipulative intent without any external prime (Campbell 1995; Kirmani and Zhu 2007) and extreme claims lose credibility when they exceed what consumers know to be plausible (Goldberg and Hartwick 1990); a blatantly exaggerated point claim is therefore not a baseline stimulus, and we would expect it to damage trustworthiness for that reason. Conversely, doubting a marketer's motives can lower the certainty of the conclusions a consumer draws, but such discounting applies to the marketer's claims as a class rather than selectively by format (Darke and Ritchie 2007; Obermiller and Spangenberg 1998), so it can shift overall certainty without producing a format-based difference. Accordingly, we state all of our predictions as relative shifts. ***Joint Evaluation as a Foundational Condition*** A foundational condition for our theorizing is whether consumers evaluate products jointly or separately, as evaluation mode determines the extent to which format differences are perceptually accessible. We propose that joint evaluation enhances consumers' ability to discern the distinction between point and range estimates. Under joint evaluation, this distinction becomes highly salient because consumers focus on features that vary across options while discounting shared attributes (Dhar and Sherman 1996). When one product presents a point estimate and another a range, format emerges as the key differentiating feature and captures attention. In contrast, in separate evaluation, where only a single option is assessed, the concept of unique versus shared features becomes less applicable (Krüger, Mata, and Ihmels 2014). Without direct comparison, format differences are less noticeable, and consumers may focus instead on other product aspects such as absolute performance levels, brand reputation, or other readily assessable features. Moreover, product performance claims are inherently difficult to evaluate in isolation because consumers often lack clear benchmarks for what constitutes good performance in most categories (Van Osselaer and Janiszewski 2012). Under these conditions, the format of the estimate, whether point or range, lacks the comparative contrast needed to become diagnostic for either uncertainty or trustworthiness inferences. Therefore, we expect consumers' preference for point (vs. range) estimates, both at baseline and when persuasion knowledge is activated, to emerge only under joint evaluation, where the format difference is salient. In separate evaluation, consumers cannot readily discern the format distinction, which attenuates the effect of point versus range estimates. As a result, the influence of format on downstream judgments should be minimal or absent when products are evaluated in isolation. **Empirical Overview** We evaluate our theory across six preregistered studies, including one field study. Studies 1a and 1b demonstrate that consumers prefer point versus range estimates in baseline conditions and their preference reverses when persuasion knowledge is activated. Study 2 demonstrates that consumers' preference for point estimates at baseline is mediated by uncertainty aversion and their preference for range estimates when persuasion knowledge is activated is mediated by perceived trustworthiness. Study 3 investigates evaluation mode as a foundational condition. Study 4 demonstrates that wide ranges attenuate the benefits observed under persuasion knowledge activation. Finally, Study 5 provides a field test of our core predictions in a real marketplace setting, demonstrating external validity for our laboratory findings. Data and code for all studies are available in the Journal of Marketing Research Dataverse (see the data availability statement on the title page). The Web Appendix provides full materials and additional information for all studies. **Study 1a: Testing When Consumers Prefer Point or Range Estimates** Study 1a () examines how consumers respond to point versus range performance estimates when persuasion knowledge is or is not activated. Based on our theoretical framework, we predicted that consumers at baseline would prefer products with point estimates over those with range estimates. However, when persuasion knowledge is activated, consumers would instead prefer products with range estimates over those with point estimates. To test these predictions, we designed a study using a sponsored product listings context, which is a common feature of e-commerce platforms where products gain prominence through paid placement. Thus, we manipulated persuasion knowledge using naturalistic marketplace cues. When consumers encounter products marked as "sponsored," they become aware that the placement is paid advertising, which should shift their focus toward evaluating how the information is presented, such as scrutinizing the format, tone, and structure of a message (Eisend and Tarrahi 2022). In contrast, encountering products in organic search results represents a more common (i.e., baseline) shopping experience, where the focus is on the content of the message (e.g., product attributes, performance, price). Web Appendix A provides full materials. ***Method*** *Participants and Design.* Two hundred participants (M~age~ = 38.91 years, SD = 11.79; 57.0% female) from Prolific completed the study in exchange for a payment. No participants failed the attention check at the beginning of the survey. The study employed a two-cell (condition: persuasion knowledge vs. control) between-subjects design. As preregistered, the sample size was predetermined by a target of 100 participants per cell. *Procedure.* First, we told all participants that they would participate in a market research study about shopping preferences. In the persuasion knowledge condition, participants evaluated two competing wireless earbuds that appeared as sponsored listings in a simulated search result. Both products featured identical specifications including price, physical attributes, and customer ratings (4.8/5), differing only in their stated battery life. One brand specified a 10-hour battery duration, while the other advertised a battery range of 8-12 hours. Participants in the control condition received the same product scenario without any mention of sponsored listings. Next, participants saw two earbuds, side by side, with brand name (A vs. B) and the advertised battery life (10 hours vs. 8 to 12 hours). Those in the sponsored condition saw the earbuds with a "Sponsored" tag above both earbuds, whereas those in the control condition did not see this information. We conducted a separate pretest (N = 198) to validate our persuasion knowledge manipulation. As expected, participants who saw the "Sponsored" tag reported higher persuasion knowledge activation (see Web Appendix A for more details) As our dependent measure, we asked participants to indicate their product preference on a 7-point bipolar scale (1 = "more likely Brand A"; 7 = "more likely Brand B"). Importantly, we counterbalanced whether the brand advertised with point (10 hours) or range (8 to 12 hours) estimates was Brand A (on the left) or Brand B (on the right), as well as the product image associated with Brand A or B. We coded the dependent variable such that lower (higher) scores represent preference for the product with a point (range) estimate regardless of counterbalancing. There were no main effects or interactions involving specific pairings of brand name (A vs. B) and product image with estimate type (point vs. range) (all Fs ≤ 1.01, ps ≥ .316). Thus, we collapsed across these counterbalancing factors for the subsequent analyses. Finally, we examined whether consumer preferences were driven by different performance expectations. Given that consumers often fixate on a range's endpoints (Janiszewski and Lichtenstein 1999), one could argue that participants at baseline fixate on the lower bound, which makes the range less attractive. In contrast, activating persuasion knowledge could prompt a shift in fixation to the upper bound, making the range more attractive (Fan, Li, and Jiang 2019). We therefore measured open-ended performance expectations for brands A and B to directly test whether participants indeed anchored at the low or high end of the estimate rather than its midpoint. Finally, participants provided demographics. ***Results*** *Product Preference.* An ANOVA with product preference as the dependent variable revealed a main effect of condition (F(1, 198) = 18.99, *p* < .001, ηp² *=* .087). Participants were more likely to purchase the product advertised with a range (vs. point) estimate in the persuasion knowledge condition (M = 4.44, SD = 2.18) compared to those in the control condition (M = 3.11, SD = 2.14). We also compared participants' product preference against the neutral midpoint 4 for each condition. Participants in the persuasion knowledge condition preferred the product advertised with a range (vs. point) estimate more than the neutral midpoint (t(99) = 2.02, *p* = .046, d = .20). Participants in the control condition preferred the product advertised with a point (vs. range) estimate more than the neutral midpoint (t(99) = -4.16, *p* < .001, d = -.42). *Alternative Explanations.* A paired samples t-test revealed no difference in expected performance between the product advertised with a point estimate (M = 9.74, SD = .79) and the product advertised with a range estimate (M = 9.55, SD = 1.61; t(183)[^1] = 1.71, *p* = .090, d = .13). Furthermore, a mixed ANOVA revealed no significant interaction between condition and estimate type (point vs. range) on expected performance (F(1, 182) = 2.92, *p* = .089, ηp² = .016), indicating that activating persuasion knowledge (vs. control) did not alter how participants formed expectations for the range versus the point estimate. ***Discussion*** Study 1a provides initial support for our framework, demonstrating that consumers prefer point estimates at baseline, but this preference reverses in favor of range estimates when a common marketplace cue (i.e., a "Sponsored" label) activates their persuasion knowledge. Importantly, this preference shift cannot be attributed to changes in performance expectations. Study 1b seeks to conceptually replicate and extend these results. **Study 1b: Testing Point vs. Range Estimates with an Explicit Competitor** Study 1b () was designed to replicate and extend our initial findings in a context where a focal product is evaluated against an explicit and consistent competitor. Whereas Study 1a involved a direct preference choice between two unbranded alternatives, Study 1b presents participants with a focal brand and asks for their purchase likelihood after comparing its performance claim to that of a competitor whose performance is always communicated with a point estimate. At baseline, we expected preference for the focal brand to be higher when performance is presented as a point (vs. range). However, when persuasion knowledge is activated, we expected preference for the focal brand to be higher when performance was presented as a range (vs. point). ***Method*** *Participants and Design.* Four hundred participants (M~age~ = 38.73 years, SD = 12.90; 56.3% female) from Prolific completed the study in exchange for a payment. No participants failed the attention check at the beginning of the survey. The study employed a 2 (condition: persuasion knowledge vs. control) × 2 (focal brand: point vs. range performance estimate) between-subjects design. As preregistered, the sample size was predetermined by a target of 100 participants per cell. *Procedure*. We first told all participants they would be participating in a study about shopping decisions on Amazon. We then manipulated persuasion knowledge using a procedure adapted from Isaac and Grayson (2017). In the persuasion knowledge condition, we instructed participants to think about the marketers' intentions, noting that "some brands lay out details exactly as they are, whereas others emphasize specific aspects and downplay others to shape shoppers' impressions". In the control condition, we simply told participants to "consider the product information as you normally would when shopping online." Next, we presented all participants with a scenario in which they were searching on Amazon for a new water filter pitcher and had narrowed their search to a brand called VIO. We informed them that VIO met their requirements for price, design, and other features, and had a high customer rating (4.8 out of 5). We then showed participants a simulated product listing that displayed the VIO pitcher alongside a competitor pitcher labeled "Others." We held the performance of the competitor constant, always presenting its filter lifespan as a point estimate ("40 Gallons"). We manipulated the performance of the focal brand, VIO. In the point estimate condition, we described its lifespan as "40 Gallons", while in the range estimate condition, we described its lifespan as "36-44 Gallons". We counterbalanced the on-screen position (left vs. right) of the VIO and "Others" (competitors) pitchers to control for order effects. There were no main effects or interactions involving focal brand position (all Fs ≤ 1.65, ps ≥ .200). As a design feature of the simulated Amazon listing, the VIO product displayed "2025 BPA-Free Plastic" as a material attribute, while the competitor displayed "2024 BPA-Free Plastic." Because this difference is constant across all conditions, it can only affect overall preference for the focal brand rather than the observed interaction. As our key dependent measure, we asked participants to indicate their purchase likelihood for the focal brand on a 7-point scale (1 = "Not likely at all"; 7 = "Very likely"), by having them respond to the question: "Given other brands' filter life (40 gallons as shown in the image above), how likely would you be to purchase the VIO water pitcher that can filter [40 Gallons / 36-44 Gallons] before it needs replacement?". We then collected secondary measures. First, to investigate the alternative explanation that our effects are an artifact of shifting endpoint fixation (Janiszewski and Lichtenstein 1999; Fan, Li, and Jiang 2019) or differences in performance expectations, we asked: "How many gallons of water do you expect the VIO water pitcher to have before it needs a replacement?" with a slider scale from 0 to 100. This measure allows us to test whether activating persuasion knowledge simply shifts participants' anchoring from the range's lower bound to its upper bound. Second, as a manipulation check for our performance attribute manipulation, we asked participants to rate the precision of the VIO filter lifespan estimate on two 7-point scales ("How precise was VIO's filter lifespan estimate?" and "How specific was the information about the VIO's expected filter lifespan?") (r = .812, *p* < .001). Finally, we collected demographics. Web Appendix B provides full materials. ***Results*** *Manipulation Check.* A 2 (condition) × 2 (focal brand) ANOVA revealed a main effect of focal brand (F(1, 396) = 147.12, *p* < .001, ηp² = .271), such that participants perceived the focal brand with the range estimate as less precise (M = 4.40, SD = 1.55) than the one with the point estimate (M = 5.93, SD = .90). There was no main effect of condition (F(1, 396) = 1.81, *p* = .180). Although the interaction was significant (F(1, 396) = 3.90, *p* = .049, ηp² = .010), it occurred because the manipulation was stronger in the control condition (M~range~ = 4.19, SD~range~ = 1.65; M~point~ = 5.97, SD~point~ = .91; F(1, 396) = 98.98, *p* < .001, ηp² = .200) than in the persuasion knowledge condition (M~range~ = 4.61, SD~range~ = 1.42; M~point~ = 5.89, SD~point~ = .90; F(1, 396) = 51.81, *p* < .001, ηp² = .116). Importantly, this interaction does not confound the main analysis (see Discussion; see also Web Appendix C for supplementary analyses). *Product Preference.* As expected, a 2 (condition) × 2 (focal brand) ANOVA revealed an interaction (F(1, 396) = 17.68, *p* < .001, ηp² = .043). There were no main effects of condition or focal brand performance estimate (all Fs < 1). Participants in the control condition were more likely to purchase the water pitcher when it was advertised with a point estimate (M = 5.59, SD = 1.26) rather than a range estimate (M = 5.04, SD = 1.39; F(1, 396) = 9.34, *p* = .002, ηp² = .023). Participants in the persuasion knowledge condition preferred the product when it was advertised with a range estimate (M = 5.61, SD = 1.17) rather than a point estimate (M = 5.10, SD = 1.22; F(1, 396) = 8.36, *p* = .004, ηp² = .021). These results are consistent with our predictions. *Alternative Explanations.* A 2 (condition) × 2 (focal brand) ANOVA on performance expectations revealed the main effects and the interaction were not significant (all Fs < 1). Thus, it is unlikely that the observed results occurred due to changes in performance expectations or differential anchoring on lower vs. upper endpoints depending on the condition. ***Discussion*** The results of Study 1b conceptually replicate and extend our initial findings. Importantly, this effect cannot be attributed to a shift in performance expectations or endpoint anchoring. Taken together, Studies 1a and 1b establish this behavioral pattern as a reliable phenomenon across different contexts and operationalizations. Further, while persuasion knowledge activation slightly attenuated the precision gap between the range and point formats (perhaps because consumers scrutinized the claims more closely), participants in the persuasion knowledge condition preferred the range estimate despite perceiving it as significantly less precise than the point estimate, which makes our hypothesis testing more conservative. See Web Appendix C for an empirical test ruling out differences in perceived precision. **Study 2: Uncertainty Aversion and Brand Trustworthiness** Study 2 () builds on the findings of Studies 1a and 1b by examining the mechanisms underlying the effect of point (vs. range) estimates at baseline and when persuasion knowledge is activated. We predict that in baseline contexts, the preference for point estimates is driven by an aversion to the uncertainty. In contrast, we expect that when persuasion knowledge is active, the preference for range estimates is driven by the enhanced brand trustworthiness. ***Method*** *Participants and Design.* Four hundred participants (M~age~ = 39.57 years, SD = 13.53; 54.5% female) from Prolific completed the study in exchange for a payment. This was the final sample size after removing one participant who failed an attention check at the beginning of the survey. The study employed a 2 (condition: persuasion knowledge vs. control) × 2 (focal brand: advertised with point vs. range performance estimate) between-subjects design. As preregistered, the sample size was predetermined by a target of 100 participants per cell. *Procedure.* The procedure used the same paradigm as that of Study 1b, but used earbuds as the focal product category. We first told all participants they would be participating in a study to understand how people make shopping decisions on Amazon. We then manipulated persuasion knowledge using a procedure adapted from Isaac and Grayson (2017) used in Study 1b. Next, we presented all participants with a scenario in which they were searching for earbuds and had narrowed their search to a brand called VIO, which had a high customer rating (4.6 out of 5). We then showed participants a simulated product listing displaying the VIO earbuds alongside a competitor labeled "Others". We held the competitor's performance constant, always presenting its battery life as a point estimate ("5 hours"). We manipulated the performance of the focal brand, VIO. In the point estimate condition, we described its battery life as "5 hours"; in the range estimate condition, we described its battery life as "4-6 hours". We counterbalanced the on-screen position (left vs. right) of the VIO and "Others" earbuds. There were no main effects or interactions involving focal brand position (all Fs < 1). As our primary dependent variable, we asked participants to indicate their purchase likelihood for the focal brand on a 7-point scale (1 = "not likely at all," and 7 = "very likely"). We then measured two key mediators in a randomized order. To measure perceived uncertainty, we asked participants to rate their agreement with three statements: "I am doubtful about VIO's actual performance," "I am unsure about VIO's actual performance," and "I am uncertain about VIO's actual performance" on a 7-point scale (1 = "Strongly disagree"; 7 = "Strongly agree") (α = .96). To measure brand trustworthiness, we asked participants to rate their agreement with three different statements: "VIO is a trustworthy brand," "VIO is a believable brand," and "VIO is an honest brand" on the same 7-point scale (α = .93). We also collected several additional measures. For exploratory purposes, we measured processing ease, "how easy or difficult was it to understand the performance information provided for this product?" (1 = "very easy", 7 = "very difficult") and the mental effort with the question, "how much mental effort did it take you to evaluate this product's performance based on the information given?" (1 = "very little mental effort", 7 = "a great deal of mental effort"). As a manipulation check, we asked participants, "how precise was the battery life estimate provided?" (1 = "not at all precise", 7 = "very precise") and "How specific was the information about the brand's expected battery life?" (1 = "not at all specific", 7 = "very specific") (r = .838, *p* < .001). Finally, participants provided demographics. ***Results*** *Manipulation Check.* A 2 (condition) × 2 (focal brand) ANOVA revealed a main effect of focal brand (F(1, 396) = 115.52, *p* < .001, ηp² = .226), such that participants perceived the focal brand with the range estimate as less precise (M = 4.42, SD = 1.62) than the one with the point estimate (M = 5.88, SD = 1.08). There was no main effect of condition (F(1, 396) = 1.06, *p* = .305). Although the interaction was significant (F(1, 396) = 9.95, *p* = .002, ηp² = .025), it occurred because the manipulation was stronger in the control condition (M~range~ = 4.13, SD~range~ = 1.70; M~point~ = 6.03, SD~point~ = 1.03; F(1, 396) = 96.64, *p* < .001, ηp² = .196) than in the persuasion knowledge condition (M~range~ = 4.70, SD~range~ = 1.49; M~point~ = 5.74, SD~point~ = 1.13; F(1, 396) = 28.83, *p* < .001, ηp² = .068). However, the difference in perceived precision had the same pattern and was significant in each condition. We empirically demonstrate that this interaction does not explain the preference reversal (see Web Appendix C). *Product Preference.* As expected, a 2 (condition) × 2 (focal brand) ANOVA revealed an interaction (F(1, 396) = 20.92, *p* < .001, ηp² = .050). There were no main effects of condition or focal brand performance estimate (all Fs < 1). Participants in the control condition were more likely to purchase the earbuds when it was advertised with a point estimate (M = 5.34, SD = 1.51) rather than a range estimate (M = 4.63, SD = 1.57; F(1, 396) = 10.92, *p* = .001, ηp² = .027). Participants in the persuasion knowledge condition preferred the product when it was advertised with a range estimate (M = 5.45, SD = 1.42) rather than a point estimate (M = 4.77, SD = 1.57; F(1, 396) = 10.02, *p* = .002, ηp² = .025). *Perceived Uncertainty.* A 2 (condition) × 2 (focal brand) ANOVA revealed an interaction (F(1, 396) = 6.47, *p* = .011, ηp² = .016). There were no significant main effects of condition (*F*(1, 396) = 1.81, *p* = .179) or focal brand performance estimate (*F*(1, 396) = 1.76, *p* = .186). As expected, participants in the control condition perceived the earbuds advertised with a range estimate as more uncertain (M = 3.92, SD = 1.55) than the one with a point estimate (M = 3.32, SD = 1.54; F(1, 396) = 7.48, *p* = .007, ηp² = .019). However, this difference was not significant in the persuasion knowledge condition (*M*~range~ = 3.31, *SD*~range~ = 1.52; *M*~point~ = 3.50, *SD*~point~ = 1.63; F(1, 396) = .742, *p* = .390). *Perceived Trustworthiness.* A 2 (condition) × 2 (focal brand) ANOVA revealed an interaction (F(1, 396) = 7.49, *p* = .006, ηp² = .019). There were no main effects of condition (F(1, 396) = 3.80, *p* = .052) or focal brand performance estimate (F(1, 396) = 1.04, *p* = .309). In the control condition, there was no difference in perceived trustworthiness between the earbuds advertised with range (M = 4.91, SD = 1.24) or point estimates (M = 5.10, SD = 1.22; F(1, 396) = 1.47, *p* = .225). As predicted, participants in the persuasion knowledge condition perceived the earbuds advertised with a range estimate as more trustworthy (M = 5.44, SD = .89) than the one with a point estimate (M = 5.01, SD = 1.19; F(1, 396) = 7.05, *p* = .008, ηp² = .017). *Cross-Condition Comparisons.* Table 1 reports the means and standard deviations for all four cells. Viewed across conditions, both interactions arise from responses to the range estimate changing when persuasion knowledge is activated, whereas responses to the point estimate remain statistically stable on both perceptions. For the range estimate, perceived uncertainty fell from 3.92 at baseline to 3.31 under persuasion knowledge (F(1, 396) = 7.56, *p* = .006) and perceived trustworthiness rose from 4.91 to 5.44 (F(1, 396) = 10.98, *p* = .001). For the point estimate, neither perception differed significantly across conditions (uncertainty: 3.32 vs. 3.50, F(1, 396) = 0.72, *p* = .398; trustworthiness: 5.10 vs. 5.01, F(1, 396) = 0.31, *p* = .579). This pattern is consistent with the change of meaning we propose. Once the format is recruited for inferences about the marketer rather than for expectations about the outcome, a range no longer signals ambiguity about the outcome, so its baseline uncertainty penalty disappears and the two formats converge on a similar level of uncertainty, while the range alone earns the transparency credit. *Table 1: Means and Standard Deviations for Study 2 by Condition and Performance Estimate Format* | | Control: Point | Control: Range | Persuasion Knowledge: Point | Persuasion Knowledge: Range | |---|---|---|---|---| | Perceived uncertainty | 3.32 (1.54) | 3.92 (1.55) | 3.50 (1.63) | 3.31 (1.52) | | Perceived trustworthiness | 5.10 (1.22) | 4.91 (1.24) | 5.01 (1.19) | 5.44 (0.89) | | Purchase likelihood | 5.34 (1.51) | 4.63 (1.57) | 4.77 (1.57) | 5.45 (1.42) | *Notes:* n = 100 per cell. Standard deviations appear in parentheses. All measures were assessed on seven-point scales. *Mediation Analysis.* We conducted a moderated mediation analysis using PROCESS model 8 (Hayes 2022; 5,000 samples) with focal brand (1 = range estimate, 0 = point estimate) as the independent variable, condition (1 = persuasion knowledge, 0 = control) as the moderator, product preference as the dependent variable, and (a) perceived uncertainty and (b) perceived trustworthiness as parallel mediators. The index of moderated mediation was significant both for perceived uncertainty (β = .17, 95% CI: [.0306 to .3819]) and perceived trustworthiness (β = .38, 95% CI: [.0983 to .7041]). The indirect effect through perceived uncertainty was significant in the control condition (β = -.13, 95% CI: [-.2731 to -.0286]), but not in the persuasion knowledge condition (β = .04, 95% CI: [-.0564 to .1592]). These findings suggest that the increased preference for the product with point (vs. range) estimates at the baseline level is driven by uncertainty aversion associated with the range estimate. The indirect effect through perceived trustworthiness was not significant in the control condition (β = -.12, 95% CI: [-.3465 to .0811]), but significant in the persuasion knowledge condition (β = .26, 95% CI: [.0807 to .4693]). See Figure 2. These findings suggest that the increased preference for the product with range (vs. point) estimates when persuasion knowledge is active is driven by perceived trustworthiness of the range estimate. Web Appendix C provides further detail on Study 2 results, including analysis to rule out several alternative explanations. *Locating the Moderation.* Our theorizing implies that activating persuasion knowledge changes the weight the format cue receives when the two perceptions are constructed, rather than the weight consumers place on those perceptions once they have been formed. The two possibilities leave different signatures in this design. If activation changed the weight assigned to already-formed perceptions, format would shift both perceptions in both conditions and the mediator-to-preference paths would differ across conditions. If instead activation changes which perception the format cue is recruited to construct, format should shift each perception primarily in its focal condition, with no corresponding change in the mediator-to-preference paths. The data show the second signature. The format × condition interactions on the mediators are significant for both perceived uncertainty and perceived trustworthiness, with no significant main effect of format on either. Re-estimating the model with condition additionally moderating both mediator-to-preference paths (the specification of Model 59 in Hayes 2022, estimated with ordinary least squares and 5,000 percentile bootstrap samples) yields no evidence that either path is moderated (joint test of the two product terms: F(2, 392) = 0.52, *p* = .594; individual interaction ps ≥ .575). Both perceptions predicted preference to a similar degree in each condition (see Figure 2), and the conditional indirect effects from this less constrained model replicate the pattern reported above. The nonsignificant interaction terms indicate an absence of evidence for path moderation rather than a demonstration of equivalence; the direct effect of format remains moderated by condition (b = .83, *p* = .001), so we scope this conclusion to the two measured perceptions, whose paths to preference are correlational (Pieters 2017; Spencer, Zanna, and Fong 2005). Web Appendix C reports the full analysis and its qualifications. **Figure 2:** Parallel Mediation Results for Baseline and Persuasion Knowledge Conditions **Panel A: Baseline Condition** **Alt text:** Figure 2, Panel A (baseline condition). Parallel mediation diagram: the range versus point manipulation raises perceived uncertainty (a1 = .60, p = .006), which lowers purchase likelihood (b1 = -.25, p < .001); the trustworthiness path is not significant (a2 = -.20, p = .260; b2 = .62, p < .001); direct effect c' = -.44 (p = .010). [Parallel mediation diagram for the baseline condition of Study 2. Four rounded boxes: "Range (vs. Point) Performance Attribute" at the left, "Perceived Uncertainty" at the top center, "Perceived Trustworthiness" at the bottom center, and "Purchase Likelihood" at the right. The upper path, drawn in solid black, runs from the performance attribute to perceived uncertainty (a1 = .60, p = .006) and from perceived uncertainty to purchase likelihood (b1 = -.25, p < .001). The lower path, drawn in light gray dashed lines, runs from the performance attribute to perceived trustworthiness (a2 = -.20, p = .260) and from perceived trustworthiness to purchase likelihood (b2 = .62, p < .001). A dashed horizontal line from the performance attribute to purchase likelihood carries the direct effect, c' = -.44 (p = .010). Solid black marks the path that carries the effect in this condition (uncertainty); gray marks the path that does not (trustworthiness).] *Note: Figure described by AI (Claude Fable 5.1) from the figure source file (Figure 2 panel A.tex, TikZ) and the manuscript image. Values cross-checked against the paper text and the source file. The figure image is not included in this machine-readable version; see the published paper for the visual.* **Panel B: Persuasion Knowledge Active** **Alt text:** Figure 2, Panel B (persuasion knowledge active). Parallel mediation diagram: the range versus point manipulation raises perceived trustworthiness (a2 = .43, p = .004), which raises purchase likelihood (b2 = .57, p < .001); the uncertainty path is not significant (a1 = -.19, p = .395; b1 = -.20, p = .004); direct effect c' = .40 (p = .031). [Parallel mediation diagram for the persuasion knowledge condition of Study 2, same layout as Panel A. Four rounded boxes: "Range (vs. Point) Performance Attribute" at the left, "Perceived Uncertainty" at the top center, "Perceived Trustworthiness" at the bottom center, and "Purchase Likelihood" at the right. The upper path, drawn in light gray dashed lines, runs from the performance attribute to perceived uncertainty (a1 = -.19, p = .395) and from perceived uncertainty to purchase likelihood (b1 = -.20, p = .004). The lower path, drawn in solid black, runs from the performance attribute to perceived trustworthiness (a2 = .43, p = .004) and from perceived trustworthiness to purchase likelihood (b2 = .57, p < .001). A solid horizontal line from the performance attribute to purchase likelihood carries the direct effect, c' = .40 (p = .031). Solid black marks the path that carries the effect in this condition (trustworthiness); gray marks the path that does not (uncertainty).] *Note: Figure described by AI (Claude Fable 5.1) from the figure source file (Figure 2 panel B.tex, TikZ) and the manuscript image. Values cross-checked against the paper text and the source file. The figure image is not included in this machine-readable version; see the published paper for the visual.* Note: Path coefficients are estimated within each condition; the simple effects reported in the text use the pooled 2 × 2 ANOVA error term, so figure and text p values for the same contrasts differ slightly. ***Discussion*** Study 2 provides direct evidence for the psychological processes underlying the preference reversal established in our previous studies. The results demonstrate that at baseline, the preference for point estimates is mediated by greater perceived uncertainty associated with ranges. In contrast, in the persuasion knowledge condition, the preference for range estimates is mediated by the greater perceived trustworthiness they signal. By demonstrating these distinct mediation paths within a single experiment these findings clarify the separate psychological drivers behind consumers' context-dependent preferences for point versus range estimates. **Study 3: The Role of Evaluation Mode** Study 3 () builds on our findings by demonstrating that the effect of points versus range estimates only emerges in joint evaluation, and not in separate evaluation. We propose that when a product with a point estimate is evaluated next to one with a range estimate, the format of the claim becomes a unique, distinguishing feature. This heightened salience makes the uncertainty of the range (in the baseline condition) or the transparency of the range (in the persuasion knowledge condition) a focal point of the decision. In contrast, in separate evaluation, without direct comparison, format differences become less noticeable, which should attenuate the effect of format on judgment. We therefore predict that the preference for point estimates at baseline and the preference for range estimates when persuasion knowledge is active will occur in joint evaluation but will be attenuated when options are evaluated separately. ***Method*** *Participants and Design.* Eight hundred participants (M~age~ = 44.70 years, SD = 14.05; 51.9% female) from Prolific completed the study in exchange for a payment. No participants failed the attention check at the beginning of the survey. The study employed a 2 (condition: persuasion knowledge vs. control) × 2 (focal brand: advertised with point vs. range performance estimate) × 2 (evaluation mode: isolated vs. comparative) between-subjects design. As preregistered, the sample size was predetermined by a target of 100 participants per cell. *Procedure.* We first told all participants they would be participating in a study about shopping decisions. We then manipulated persuasion knowledge using a procedure adapted from prior research (Clarkson, Janiszewski, and Cinelli 2013; Tormala and DeSensi 2008). Specifically, participants indicated their level of agreement with several statements on a 7-point scale (1 = definitely disagree, 7 = definitely agree). In the persuasion knowledge condition, the statements were designed to make persuasion knowledge salient and included: "We cannot depend on companies' advertised product claims," and "Advertised product performance rarely matches the actual performance," among others. In the baseline condition, participants indicated their agreement to everyday statements, such as: "This summer felt more humid than last year's" and "Some people are born to be artists". Next, we presented all participants with a scenario in which they were searching for a new electric scooter and had narrowed their search to a brand called iScooter, which had a high customer rating (4.6 out of 5). We manipulated evaluation mode and the focal brand's performance attribute. In the comparative evaluation condition, we showed participants the iScooter alongside a generic competitor labeled "Others". We held the competitor's performance constant at "24 miles" per charge. In the isolated evaluation condition, we showed participants only the iScooter product listing. The focal brand's performance was presented as either a point estimate ("24 miles") or a range estimate ("22-26 miles"). As our key dependent variable, we asked all participants to indicate their purchase likelihood for the focal brand on a 7-point scale (1 = "Very Unlikely"; 7 = "Very Likely"), with the question: "How likely would you be to purchase the iScooter electric scooter with a travel distance of [24 miles / 22-26 miles] per full charge?". Finally, participants provided demographics. Web Appendix D provides full materials. ***Results*** *Product Preference.* A 2 (condition) × 2 (focal brand) × 2 (evaluation mode) ANOVA revealed a three-way interaction (F(1, 792) = 5.22, *p* = .023, ηp² = .007). In the joint evaluation mode condition, there was a two-way interaction between condition and focal brand (F(1, 792) = 15.97, *p* < .001, ηp² = .020). As expected, participants in the control condition were more likely to purchase the electric scooter when it was advertised with a point estimate (M = 5.56, SD = 1.40) rather than a range estimate (M = 5.01, SD = 1.53; F(1, 792) = 8.22, *p* = .004, ηp² = .010). Participants in the persuasion knowledge condition preferred the product when it was advertised with a range estimate (M = 5.50, SD = 1.16) rather than a point estimate (M = 4.96, SD = 1.45; F(1, 792) = 7.76, *p* = .005, ηp² = .010). Conversely, in the isolated evaluation mode condition, the interaction between condition and focal brand was not significant (F(1, 792) = .576, *p* = .448). That is, the purchase likelihood in the control condition was similar regardless of whether the electric scooter was advertised with a point (M = 5.07, SD = 1.36) or range estimate (M = 5.00, SD = 1.39; F(1, 792) = .130, *p* = .718). Similarly, the purchase likelihood in the persuasion knowledge condition was similar in the point (M = 4.69, SD = 1.42) and range conditions (M = 4.83, SD = 1.23; F(1, 792) = .506, *p* = .477). These results show that the effects of presentation format primarily emerge in joint evaluation. See Figure 3. **Figure 3:** Purchase Likelihood as a Function of Format, Condition, and Evaluation Mode **Alt text:** Figure 3. Two-panel bar chart of purchase likelihood (1 to 7) in Study 3. Joint evaluation: point beats range at baseline (5.56 vs. 5.01, p = .004) and range beats point under persuasion knowledge (5.50 vs. 4.96, p = .005). Separate evaluation: no significant differences (5.07 vs. 5.00; 4.69 vs. 4.83). [Two side-by-side bar charts sharing a vertical axis labeled "Purchase Likelihood" with tick marks 1 through 7. In each panel the horizontal axis has two groups, "Baseline" and "Persuasion Knowledge", and each group holds two bars: a white bar for the point estimate and a dark gray bar for the range estimate (legend at the bottom: "Point estimate", "Range estimate"). Each bar carries a thin error bar of plus or minus one standard error and its mean printed above it. Panel A, "Joint Evaluation": Baseline point 5.56, Baseline range 5.01, with a bracket over the pair labeled p = .004; Persuasion Knowledge point 4.96, Persuasion Knowledge range 5.50, with a bracket labeled p = .005. The preference reverses: the white bar is taller at baseline and the gray bar is taller under persuasion knowledge. Panel B, "Separate Evaluation": Baseline point 5.07, Baseline range 5.00, with a gray bracket labeled ns; Persuasion Knowledge point 4.69, Persuasion Knowledge range 4.83, with a gray bracket labeled ns. In Panel B the pairs of bars are nearly equal in height.] *Note: Figure described by AI (Claude Fable 5.1) from the figure source file (Figure 3.R, ggplot2, from Study 3 Clean Data.sav) and the manuscript image. Values cross-checked against the paper text and the source file. The figure image is not included in this machine-readable version; see the published paper for the visual.* Note: Error bars represent ±1 standard error. In the separate-evaluation conditions, the point-versus-range differences were not significant (ps = .718 and .477). ***Discussion*** Study 3 demonstrates that evaluation mode is a key moderator for the preference for point versus range estimates of product performance. The results replicated our core findings in the joint evaluation condition. Participants in the baseline condition preferred the point estimate, while those in the persuasion knowledge condition preferred the range estimate. However, both of these effects were attenuated when products were evaluated in isolation. One could argue that a joint evaluation context by itself could activate persuasion knowledge, as it makes it salient that one brand is offering a range estimate, while the other is offering a point, potentially increasing suspicion. To address this possibility, we reran Study 3 with 401 participants using the persuasion knowledge activation measure (α = .82) used in Study 1a's manipulation validation. A 2 (condition) × 2 (focal brand) × 2 (evaluation mode) ANOVA revealed a significant main effect of condition (F(1, 393) = 8.47; *p* = .004, ηp² = .021), such that participants in the persuasion knowledge condition reported greater levels of persuasion knowledge activation (M = 5.13, SD = 1.09) than those in the control condition (M = 4.79, SD = 1.18). No other main effects or interactions were significant. This suggests that mode of evaluation does not affect persuasion knowledge activation. **Study 4: Attenuation for Wide Ranges** Study 4 () tests a boundary condition for our finding that consumers prefer range estimates when persuasion knowledge is activated. We propose that consumers prefer range estimates in this condition because such estimates signal brand trustworthiness by acknowledging the natural variability of product performance. However, this signal should be diagnostic only when the range is perceived as a realistic reflection of performance variability. Research on interval judgments suggests that as ranges become wider, they progressively lose their diagnostic value and informativeness (Yaniv and Foster 1997). A moderately wide range is consistent with consumers' experiential knowledge that performance fluctuates within reasonable bounds, and therefore serves as a credible signal of transparency. By contrast, an excessively wide range exceeds the bounds of what consumers consider realistic variability, weakening the transparency inference. In other words, wide ranges become less diagnostic of trustworthiness. Therefore, when persuasion knowledge is activated, we do not expect consumers to prefer range estimates to point estimates when the performance range is excessively wide. ***Method*** *Participants and Design.* Four hundred forty-four respondents from Amazon Mechanical Turk completed the study in exchange for a payment. The study employed a four-cell (condition: persuasion knowledge vs. control) × (range width: moderate vs. wide) between-subjects design. Consistent with our preregistration, we removed 30 participants who failed the attention check, resulting in a final sample of 414 respondents (M~age~ = 40.57 years, SD = 12.36; 44.0% female). The sample size was predetermined by a target of 100 participants per cell. *Procedure.* First, participants in all four conditions read a scenario involving purchasing an electric vehicle. We manipulated the persuasion knowledge vs. control conditions using the same procedure from Study 1b (adapted from Isaac and Grayson 2017). Next, we told all participants to imagine they were searching on Autotrader, a popular car sales website, to purchase an electric vehicle. We told participants they had narrowed down their search to two electric vehicles from different manufacturers. The two electric vehicles were identical in price, safety measures, performance, technology, and rating from Consumer Reports (4.7 out of 5). We then presented participants with two electric vehicles on the same page for two different brands (Brand A and Brand B). We manipulated range width by presenting the advertised travel distance per full battery charge, a performance attribute, with a point estimate (Brand A; certain) or with a range estimate (Brand B; uncertain). We manipulated range width level (moderate vs. wide) on the ranges provided for Brand B. Specifically, participants in the moderate range condition saw Brand A with 260 miles, and Brand B with 240 to 280 miles. Participants in the wide range condition saw Brand A with 260 miles, and Brand B with 210 to 310 miles. Finally, as our dependent measure, we asked participants to indicate their brand preference: "If you were to buy one of these electric vehicles, which one would you be more likely to purchase?" measured on a 7-point bipolar scale (1 = "more likely Brand A"; 7 = "more likely Brand B"). Higher scores indicate a stronger relative preference for the electric vehicle with range (vs. point) performance claims. Next, participants answered our attention check measure, which asked the advertised travel distance of Brand A's electric car. Finally, participants provided demographics. Web Appendix E provides the full materials for Study 4, including results for a separate pre-test for the range width manipulation. ***Results*** *Product Preference.* As expected, a 2 (condition) × 2 (range width) ANOVA revealed a main effect of condition (F(1, 410) = 24.69, *p* < .001, ηp² = .057), such that participants were more likely to purchase the electric vehicle advertised with a range (vs. point) in the persuasion knowledge condition (M = 4.26, SD = 2.05) compared to those in the control condition (M = 3.34, SD = 1.81). The effect of range width was marginally significant (F(1, 410) = 3.45, *p* = .064, ηp² = .008), suggesting that participants were less likely to purchase the electric vehicle with the range (vs. point) in wide range width condition (M = 3.63, SD = 1.99) compared to those in the moderate range width condition (M = 3.94, SD = 1.97). These main effects were qualified by the predicted condition × range width interaction effect (*F*(1, 410) = 5.76, *p* = .017, ηp² = .014). As predicted, when the range width of the performance claim was moderate, participants were more likely to purchase the electric vehicle with the range (vs. point) performance claim in the persuasion knowledge condition (M = 4.68, SD = 1.96) compared to those in the control condition (M = 3.29, SD = 1.74; F(1, 410) = 26.34, *p* < .001, ηp² = .06). In contrast, when the range width of the performance claim was wide, this difference was smaller and only marginally significant (persuasion knowledge: M = 3.88, SD = 2.07; control: M = 3.39, SD = 1.89; F(1, 410) = 3.40, *p* = .066, ηp² = .008). ***Discussion*** Study 4 reveals an important boundary condition for the positive effects of range estimates under persuasion knowledge activation. The results show that persuasion knowledge increased preference for range estimates when the range width was moderate. When ranges become excessively wide, the increase in preference for range estimates under persuasion knowledge activation is attenuated. This attenuation occurred even for electric vehicles, a high-involvement product category, suggesting that range width is a crucial factor across diverse purchase contexts, not just low-cost, frequently purchased goods. **Study 5: Large-Scale Field Study Using Meta Ads** To test our theory in a naturalistic setting, Study 5 () was conducted using social media advertising where two key marketplace phenomena converge. First, on platforms like Facebook and Instagram, all advertisements are explicitly labeled as "Sponsored," a feature that naturally activates consumers' persuasion knowledge (Study 1a). Second, brands frequently employ direct comparative formats in their digital advertising, often using "Us vs. Them" or "Our Brand vs. Other Brands" messaging to highlight performance advantages. This comparative approach is particularly common in product categories where performance metrics are central to consumer choice, from battery life in electronics to efficiency ratings in appliances to burn time in candles. These characteristics create an ideal context to test how consumers respond to range versus point performance claims. Building on Study 1a's finding that sponsored content activates persuasion knowledge, we conducted a large-scale field study using Facebook's advertising platform. We created two versions of a comparative advertisement: one where the focal brand uses a range estimate while the competitor uses a point estimate, and another where these formats are reversed. If our theorizing is correct, the sponsored nature of these ads should lead consumers to respond more favorably when the focal brand communicates performance through ranges rather than points in comparative advertising. ***Method*** *Design and Stimuli.* The study employed a two-condition (focal brand performance attribute: point vs. range) between-subjects design manipulating performance claim format in sponsored comparative advertising. Following recent research using Facebook Ads (Affonso and Janiszewski 2023; Castelo, Bos, and Lehmann 2019) we established a fictitious candle brand named Ember and Leaf Co. to eliminate any prior brand familiarity effects. We designed two comparative advertisements with identical layouts split into contrasting sections (labeled "US" vs "THEM"). Each section displayed visually identical handmade candles in amber glass jars against complementary orange and cream backgrounds. Both sections maintained consistent product attributes (handmade, more concentrated), varying only in burn time claims, which was the performance attribute. In the range performance attribute condition, our focal brand advertised "Burns for 45-55 hours" while the competitor claimed "Burns for 50 hours." In the point performance attribute condition, we reversed these claims, with the focal brand stating "Burns for 50 hours" and the competitor "Burns for 45-55 hours." The exact midpoint (50 hours) remained constant across conditions to control for absolute performance expectations. Web Appendix F provides additional materials for Study 5. *Procedure.* Using Meta Ads Manager's A/B test feature, we conducted a multiple-ad study optimized for link clicks. A/B testing in digital advertising allows marketers to compare two versions of an advertisement by simultaneously showing different variations to separate groups of users, measuring which version performs better on key metrics like engagement or clicks. The campaign targeted U.S. adults aged 18-65+ with demonstrated interest in or like pages related to "Candle (home furnishings)", reaching a potential audience of approximately 32.3 million users. We conducted the experiment during a five-day period in December 2024. We set the daily budget as \$35 per advertisement. It is important to note that while Meta's A/B testing tool randomly assigns users to be eligible for different ads, the platform's algorithm ultimately determines which eligible users see each ad through a process called "divergent delivery" (Braun et al. 2024). This means that users who see different ads may systematically differ on unobservable characteristics. Therefore, the results should be interpreted as correlational rather than causal. Both ads featured a "Shop Now" call-to-action button. When people clicked the ad, they were directed to a landing page that provided a short debriefing statement explaining the advertisement was designed for research purposes, an approach which has been used in prior research (Castelo et al. 2019). The dependent variable was advertising click-through rates (CTRs), which is based on the number of clicks and impressions. CTRs are a key determinant of advertising effectiveness and are associated with real product interest. ***Results*** The campaign generated 32,780 impressions across both conditions, resulting in 429 total link clicks. We performed a chi-square test with click (0 = no, 1 = yes) as the dependent variable, and performance attribute format (focal brand performance claim format: point vs. range) as the independent variable. As predicted, the ad featuring the range estimate generated a higher CTR (242 link clicks / 16,269 impressions = 1.49%) than the ad featuring the point estimate (187 link clicks / 16,511 impressions = 1.13%; χ*^2^*(1) = 7.99, *p* = .005, φ = .016, OR = 1.32). The results were similar for the unique-CTR metric, with the ad featuring the range estimate generating a higher unique-CTR (230 unique link clicks / 13,243 people reached = 1.74%) than the ad featuring a point estimate (175 unique link clicks / 12,603 people reached = 1.39%; χ^2^(1) = 5.08, *p* = .024, φ = .014, OR = 1.26). ***Discussion*** Using a real-life social media environment with actual users, Study 5 provides field evidence that range performance estimates can enhance advertising effectiveness in competitive contexts where persuasion knowledge is naturally activated. The superior performance of range estimates in sponsored comparative advertising is particularly noteworthy given that this format diverges from the industry norm of using point estimates when comparing products. These results extend our laboratory findings to a real-world digital advertising context where consumers make consequential engagement decisions in their natural social media environment, highlighting the practical value of matching performance claim format to the sponsored comparative advertising context. **General Discussion** Through six preregistered studies, including one large-scale field study, we demonstrate that consumer preferences for point versus range performance estimates depend on an attention-shifting mechanism. At baseline, consumers typically focus on what information a performance claim communicates, leading them to prefer the certainty of point estimates due to uncertainty aversion (Gneezy et al. 2006). However, when consumers' persuasion knowledge is activated, their attention shifts toward how the information is communicated, which becomes relatively more salient than the substance of the claim (Campbell and Kirmani 2000). This shift leads them to prefer range estimates, which are perceived as more trustworthy acknowledgments of natural performance variability (Mohan et al. 2020). Study 2 provides direct evidence for these distinct psychological pathways, measuring both perceived uncertainty and trustworthiness simultaneously and demonstrating that their influence depends on the consumer's attentional focus. At baseline, the preference for point estimates is mediated by a reduction in perceived uncertainty. In contrast, when persuasion knowledge is activated, the preference for range estimates is mediated by an increase in perceived brand trustworthiness. These effects emerge only in joint evaluation, where the format difference between point and range estimates becomes a salient distinguishing feature (Hsee et al. 1999). In separate evaluation, the format distinction is less noticeable and the effects are attenuated. We also identify range width as a boundary condition: the benefit of range estimates when persuasion knowledge is activated is attenuated when ranges become excessively wide, as this diminishes the claim's diagnostic value and the positive signal of trustworthiness (Yaniv and Foster 1997). ***Theoretical Implications*** Our research advances theory by providing an integrated framework that explains when and why the same uncertainty can be problematic or beneficial in consumer decision-making. This integration moves beyond documenting separate effects to identifying the underlying psychological mechanism, attention shifting, that determines when uncertainty aversion (Gneezy et al. 2006) versus trustworthiness concerns (Erdem and Swait 2004) dominate. We extend persuasion knowledge theory (Friestad and Wright 1994) by demonstrating that persuasion knowledge activation alters the psychological lens through which consumers interpret identical information. While prior research has shown that persuasion knowledge can enhance evaluations of tactics already perceived as credible (Isaac and Grayson 2017), we demonstrate that persuasion knowledge can create credibility distinctions between formats that consumers initially view as equally credible. When persuasion knowledge shifts consumer attention from "what is promised" toward "how it is communicated", range estimates gain a transparency advantage by acknowledging performance variability that point estimates appear to mask. Furthermore, this finding reveals that activating persuasion knowledge can enhance perceptions of brand trustworthiness when brands present range estimates that align with consumers' knowledge of performance variability. Our findings contribute to uncertainty aversion research by identifying how consumers' interpretation of uncertainty's source alters their responses to it. While uncertainty aversion research suggests stable preference for known outcomes, we show that the same uncertainty can enhance trustworthiness when consumers interpret it as reflecting natural performance variation rather than marketer intentions (Kirmani and Rao 2000). This finding demonstrates that uncertainty aversion is not a fixed response but depends on the processing lens consumers adopt, helping explain why prior research has found variation in uncertainty preferences across contexts (Trautmann and van de Kuilen 2015). We advance evaluability theory (Hsee et al. 1999) by demonstrating that joint evaluation is a necessary condition for format differences to influence consumer judgment. When products are compared side by side, the distinction between point and range estimates becomes a salient, distinguishing feature that activates either uncertainty concerns or transparency inferences depending on consumers' attentional focus. In separate evaluation, format differences fade into the background of consumer decision-making, and their influence on downstream judgments is attenuated. Finally, our research provides a unifying framework for numerical format effects in performance contexts. Because product performance naturally varies with real-world usage conditions (Bolton and Drew 1991; Zeithaml et al. 1993), performance claims are inherently estimates. This means that the choice of format (point versus range) simultaneously communicates information about the product's likely capabilities and the marketer's transparency, a dual signaling function that shapes consumer responses in ways that prior numerical cognition research has not addressed (Thomas and Morwitz 2009; Zhang and Schwarz 2013). ***Implications for Marketers and Businesses*** Our findings provide guidance for communicating product performance information based on context and consumer processing mode. The effectiveness of point versus range estimates depends on where and how products are encountered, enabling marketers to optimize their communication strategies rather than choosing formats unsystematically. When products are encountered in contexts that naturally activate persuasion knowledge, such as sponsored content on digital platforms or comparative advertising (Campbell 1995; Isaac and Grayson 2017), marketers should consider employing range estimates. Our field study demonstrates that range estimates generate higher engagement rates than point estimates in sponsored comparative advertising, despite diverging from industry norms of using deterministic point estimates when comparing products. This finding is particularly relevant given the prevalence of paid digital advertising. Sponsored designations typically activate persuasion knowledge (Campbell and Kirmani 2000), which can work in marketers' favor when paired with range estimates that consumers view as more credible. For products displayed in comparative contexts, such as retail shelves or e-commerce category pages, the choice of format depends on whether persuasion knowledge is likely to be activated (Hsee et al. 1999). When products are compared side-by-side without explicit persuasion cues, point estimates tend to be more effective because uncertainty becomes salient as a distinguishing feature. However, when comparative contexts include persuasion-activating elements, range estimates become more effective by signaling appropriate transparency about performance variability. Importantly, the range advantage under persuasion knowledge also depends on competitive context. If all firms in a category converged on range communication, the format would no longer stand apart as distinctively transparent, and the trustworthiness benefit could attenuate. Early adopters of range estimates in persuasion-rich environments are therefore best positioned to benefit from this format. Marketers need to carefully calibrate range width when using that format to communicate their product performance. Moderate ranges that reflect realistic performance variation enhance brand trustworthiness and, when persuasion knowledge is active, increase preference for the brand. For excessively wide ranges this preference benefit is attenuated rather than reversed, consistent with such ranges exceeding what consumers consider realistic variability and thereby becoming less diagnostic of transparency. This suggests that marketing and product development teams should collaborate to understand natural performance variation and communicate it appropriately, avoiding misleading deterministic claims about performance. How, then, should a marketer choose a range? Our theorizing implies that a communicated range is evaluated against the reference distribution supplied by consumers' topic knowledge: it must be wide enough to acknowledge the variability consumers know exists, yet narrow enough to remain within what they consider realistic, which is the accuracy-informativeness trade-off documented for interval judgments (Yaniv and Foster 1995, 1997). This standard suggests a two-step calibration. First, ground candidate endpoints in observed performance distributions across representative usage conditions, data that often already exist as engineering or certification byproducts, and communicate the central mass of typical outcomes rather than extreme-condition tails. Second, because the operative standard is consumers' perceived variability rather than engineered variability, pretest the minimum and maximum performance consumers consider plausible in the category and verify that the advertised range falls within those bounds. For concreteness, in Study 4 a range spanning roughly ±8% of the point value preserved the range advantage under persuasion knowledge whereas a range spanning roughly ±19% attenuated it, while ranges near ±20% of the point value preserved the advantage in the categories of Studies 1a and 2. The threshold is therefore category-specific rather than a fixed percentage, and identifying the functional form of this realistic-variability threshold is a promising direction for future research. ***Limitations and Future Research*** While our effects emerged across diverse product categories and price ranges, boundary conditions merit exploration. The benefits of range estimates under persuasion knowledge activation may be less evident for products where outcomes are publicly observable or for services requiring extensive information gathering. Future research should examine how product characteristics and usage contexts influence the dual signaling function we identify. Understanding how prior experience with performance variability moderates these effects represents another feasible research direction. Consumers' actual experience with performance variation in a category might influence their response to different claim formats independent of persuasion knowledge activation. Our experimental approach holds all product attributes constant except format, providing a clean test of the diagnosticity mechanism while constraining what can be inferred about richer decision environments. In most real-world settings, consumers compare products that differ on multiple dimensions, and format may interact with these other features in ways that amplify or attenuate its diagnostic value. Extending the framework to such contexts would also make it possible to investigate how consumers evaluate competing range estimates of different widths rather than contrasting a range against a point, and how they trade off expected performance levels against the variability implied by a range when the two are not held constant. These extensions would clarify whether the diagnosticity of format depends on the point-range contrast itself or generalizes to subtler variation in how performance information is communicated. **References** Affonso, Felipe M. and Chris Janiszewski (2023), "Marketing by Design: The Influence of Perceptual Structure on Brand Performance," *Journal of Marketing*, 87 (5), 736--54. Ames, Daniel R. and Malia F. Mason (2015), "Tandem Anchoring: Informational and Politeness Effects of Range Offers in Social Exchange," *Journal of Personality and Social Psychology*, 108 (2), 254--74. Anderson, Norman H. (1981), *Foundations of Information Integration Theory*, New York: Academic Press. Bettman, James R., Mary Frances Luce, and John W. Payne (1998), "Constructive Consumer Choice Processes," *Journal of Consumer Research*, 25 (December), 187--217. Biswas, Abhijit and Scot Burton (1993), "Consumer Perceptions of Tensile Price Claims in Advertisements: An Assessment of Claim Types Across Different Discount Levels," *Journal of the Academy of Marketing Science*, 21 (3), 217--29. Biswas, Abhijit and Scot Burton (1994), "An Experimental Assessment of Effects Associated with Alternative Tensile Price Claims," *Journal of Business Research*, 29 (1), 65--73. Bolton, Ruth N. and James H. Drew (1991), "A Multistage Model of Customers' Assessments of Service Quality and Value," *Journal of Consumer Research*, 17 (4), 375--84. Braun, Michael, Bart De Langhe, Stefano Puntoni, and Eric M. Schwartz (2024), "Leveraging Digital Advertising Platforms for Consumer Research," *Journal of Consumer Research*, 51 (1), 119--28. Camerer, Colin F. and Martin Weber (1992), "Recent Developments in Modeling Preferences: Uncertainty and Ambiguity," *Journal of Risk and Uncertainty*, 5 (4), 325--70. Campbell, Margaret C. (1995), "When Attention-Getting Advertising Tactics Elicit Consumer Inferences of Manipulative Intent: The Importance of Balancing Benefits and Investments," *Journal of Consumer Psychology*, 4 (3), 225--54. Campbell, Margaret C. and Amna Kirmani (2000), "Consumers' Use of Persuasion Knowledge: The Effects of Accessibility and Cognitive Capacity on Perceptions of an Influence Agent," *Journal of Consumer Research*, 27 (1), 69--83. Castelo, Noah, Maarten W. Bos, and Donald R. Lehmann (2019), "Task-Dependent Algorithm Aversion," *Journal of Marketing Research*, 56 (5), 809--25. Clarkson, Joshua J., Chris Janiszewski, and Melissa D. Cinelli (2013), "The Desire for Consumption Knowledge," *Journal of Consumer Research*, 39 (6), 1313--29. Darke, Peter R. and Robin J. B. Ritchie (2007), "The Defensive Consumer: Advertising Deception, Defensive Processing, and Distrust," *Journal of Marketing Research*, 44 (1), 114--27. Dhar, Ravi and Steven J. Sherman (1996), "The Effect of Common and Unique Features in Consumer Choice," *Journal of Consumer Research*, 23 (3), 193--203. Edwards, Jeffrey R. and Lisa Schurer Lambert (2007), "Methods for Integrating Moderation and Mediation: A General Analytical Framework Using Moderated Path Analysis," *Psychological Methods*, 12 (1), 1--22. Eisend, Martin and Farid Tarrahi (2022), "Persuasion Knowledge in the Marketplace: A Meta-Analysis," *Journal of Consumer Psychology*, 32 (1), 3--22. Erdem, Tülin and Joffre Swait (2004), "Brand Credibility, Brand Consideration, and Choice," *Journal of Consumer Research*, 31 (1), 191--98. Escalas, Jennifer Edson (2007), "Self-Referencing and Persuasion: Narrative Transportation Versus Analytical Elaboration," *Journal of Consumer Research*, 33 (4), 421--29. Fan, Linying, Xueni Li, and Yuwei Jiang (2019), "Room for Opportunity: Resource Scarcity Increases Attractiveness of Range Marketing Offers," *Journal of Consumer Research*, 46 (1), 82--98. Feldman, Jack M. and John G. Lynch Jr. (1988), "Self-Generated Validity and Other Effects of Measurement on Belief, Attitude, Intention, and Behavior," *Journal of Applied Psychology*, 73 (3), 421--35. Friestad, Marian and Peter Wright (1994), "The Persuasion Knowledge Model: How People Cope with Persuasion Attempts," *Journal of Consumer Research*, 21 (1), 1--31. Gneezy, Uri, John A. List, and George Wu (2006), "The Uncertainty Effect: When a Risky Prospect Is Valued Less than Its Worst Possible Outcome," *Quarterly Journal of Economics*, 121 (4), 1283--309. Goldberg, Marvin E. and Jon Hartwick (1990), "The Effects of Advertiser Reputation and Extremity of Advertising Claim on Advertising Effectiveness," *Journal of Consumer Research*, 17 (2), 172--79. Hayes, Andrew F. (2022), *Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach*, 3rd ed., New York, NY: The Guilford Press. Hsee, Christopher K., George F. Loewenstein, Sally Blount, and Max H. Bazerman (1999), "Preference Reversals Between Joint and Separate Evaluations of Options: A Review and Theoretical Analysis," *Psychological Bulletin*, 125 (5), 576--90. Isaac, Mathew S. and Kent Grayson (2017), "Beyond Skepticism: Can Accessing Persuasion Knowledge Bolster Credibility?" *Journal of Consumer Research*, 43 (6), 895--912. Jägerbrand, Annika K. and Jonas Sjöbergh (2016), "Effects of Weather Conditions, Light Conditions, and Road Lighting on Vehicle Speed," *SpringerPlus*, 5 (1), 505. Janiszewski, Chris and Donald R. Lichtenstein (1999), "A Range Theory Account of Price Perception," *Journal of Consumer Research*, 25 (4), 353--68. Janiszewski, Chris and Dan Uy (2008), "Precision of the Anchor Influences the Amount of Adjustment," *Psychological Science*, 19 (2), 121--27. Kardes, Frank R., Steven S. Posavac, and Maria L. Cronley (2004), "Consumer Inference: A Review of Processes, Bases, and Judgment Contexts," *Journal of Consumer Psychology*, 14 (3), 230--56. Khamitov, Mansur, Koushyar Rajavi, Der-Wei Huang, and Yuly Hong (2024), "Consumer Trust: Meta-Analysis of 50 Years of Empirical Research," *Journal of Consumer Research*, 51 (1), 7--18. Kirmani, Amna and Akshay R. Rao (2000), "No Pain, No Gain: A Critical Review of the Literature on Signaling Unobservable Product Quality," *Journal of Marketing*, 64 (2), 66--79. Kirmani, Amna and Rui Zhu (2007), "Vigilant Against Manipulation: The Effect of Regulatory Focus on the Use of Persuasion Knowledge," *Journal of Marketing Research*, 44 (4), 688--701. Krüger, Tobias, André Mata, and Max Ihmels (2014), "The Presenter's Paradox Revisited: An Evaluation Mode Account," *Journal of Consumer Research*, 41 (4), 1127--36. Lynch Jr., John G. and Thomas K. Srull (1982), "Memory and Attentional Factors in Consumer Choice: Concepts and Research Methods," *Journal of Consumer Research*, 9 (1), 18--37. Mohan, Bhavya, Ryan W. Buell, and Leslie K. John (2020), "Lifting the Veil: The Benefits of Cost Transparency," *Marketing Science*, 39 (6), 1105--21. Muller, Dominique, Charles M. Judd, and Vincent Y. Yzerbyt (2005), "When Moderation Is Mediated and Mediation Is Moderated," *Journal of Personality and Social Psychology*, 89 (6), 852--63. Nelson, Phillip (1974), "Advertising as Information," *Journal of Political Economy*, 82 (4), 729--54. Newman, George E. and Daniel Mochon (2012), "Why Are Lotteries Valued Less? Multiple Tests of a Direct Risk Aversion Mechanism," *Judgment and Decision Making*, 7 (1), 19--24. Nielsen, Jesper H. and Jennifer Edson Escalas (2010), "Easier Is Not Always Better: The Moderating Role of Processing Type on Preference Fluency," *Journal of Consumer Psychology*, 20 (3), 295--305. Obermiller, Carl and Eric R. Spangenberg (1998), "Development of a Scale to Measure Consumer Skepticism Toward Advertising," *Journal of Consumer Psychology*, 7 (2), 159--86. Obermiller, Carl, Eric R. Spangenberg, and Douglas L. MacLachlan (2005), "Ad Skepticism: The Consequences of Disbelief," *Journal of Advertising*, 34 (3), 7--17. Palmeira, Mauricio, Eduardo Andrade, Shahin Sharifi, Wen Mao, and Jorge Jacob (2020), "The Influence of Arbitrary Breakpoints on Judgments of Maximum Output," *Journal of Consumer Psychology*, 30 (2), 260--76. Pieters, Rik (2017), "Meaningful Mediation Analysis: Plausible Causal Inference and Informative Communication," *Journal of Consumer Research*, 44 (3), 692--716. Spencer, Steven J., Mark P. Zanna, and Geoffrey T. Fong (2005), "Establishing a Causal Chain: Why Experiments Are Often More Effective Than Mediational Analyses in Examining Psychological Processes," *Journal of Personality and Social Psychology*, 89 (6), 845--51. Teigen, Karl Halvor and Magne Jørgensen (2005), "When 90% Confidence Intervals Are 50% Certain: On the Credibility of Credible Intervals," *Applied Cognitive Psychology*, 19 (4), 455--75. Thomas, Manoj and Vicki G. Morwitz (2009), "The Ease-of-Computation Effect: The Interplay of Metacognitive Experiences and Naive Theories in Judgments of Price Differences," *Journal of Marketing Research*, 46 (1), 81--91. Tormala, Zakary L. and Victoria L. DeSensi (2008), "The Perceived Informational Basis of Attitudes: Implications for Subjective Ambivalence," *Personality and Social Psychology Bulletin*, 34 (2), 275--87. Trautmann, Stefan T. and Gijs van de Kuilen (2015), "Ambiguity Attitudes," in Gideon Keren and George Wu (eds.), *The Wiley Blackwell Handbook of Judgment and Decision Making*, Chichester, UK: John Wiley & Sons, 89--116. Van Osselaer, Stijn M. J. and Chris Janiszewski (2012), "A Goal-Based Model of Product Evaluation and Choice," *Journal of Consumer Research*, 39 (2), 260--92. Wickens, Christopher (2021), "Attention: Theory, Principles, Models and Applications," *International Journal of Human--Computer Interaction*, 37 (5), 403--17. Yan, Kaige, Jingweijia Tan, and Xin Fu (2019), "Improving Energy Efficiency of Mobile Devices by Characterizing and Exploring User Behaviors," *Journal of Systems Architecture*, 98 (September), 126--34. Yaniv, Ilan and Dean P. Foster (1995), "Graininess of Judgment Under Uncertainty: An Accuracy-Informativeness Trade-Off," *Journal of Experimental Psychology: General*, 124 (4), 424--32. Yaniv, Ilan and Dean P. Foster (1997), "Precision and Accuracy of Judgmental Estimation," *Journal of Behavioral Decision Making*, 10 (1), 21--32. Zeithaml, Valarie A., Leonard L. Berry, and A. Parasuraman (1993), "The Nature and Determinants of Customer Expectations of Service," *Journal of the Academy of Marketing Science*, 21 (1), 1--12. Zhang, Y. Charles and Norbert Schwarz (2013), "The Power of Precise Numbers: A Conversational Logic Analysis," *Journal of Experimental Social Psychology*, 49 (5), 944--46. [^1]: The question was open-ended and did not force a specific response format, resulting in some participants providing ranges of expectations instead of single numbers. We removed those participants from the alternative explanations analyses, which influenced the number of degrees of freedom. --- # Supplementary Materials (Web Appendix) When Consumers Prefer Point Versus Range Estimates of Product Performance Web Appendix Amin Shiri (University of Georgia, amin.shiri@uga.edu), Felipe M. Affonso (Oklahoma State University, felipe.affonso@okstate.edu), and Keith Wilcox (Texas A&M University, ) These materials have been supplied by the authors to aid in the understanding of their paper. The AMA is sharing these materials at the request of the authors. Web Appendix A: Additional Methodological Details for Study 1a (page 2) Web Appendix B: Additional Methodological Details for Study 1b (page 6) Web Appendix C: Additional Methodological Details for Study 2 (page 10) Web Appendix D: Additional Methodological Details for Study 3 (page 20) Web Appendix E: Additional Methodological Details for Study 4 (page 27) Web Appendix F: Additional Methodological Details for Study 5 (page 31) Web Appendix G: Additional Details for Pilot Studies (page 33) Web Appendix H: Examples of Point Versus Range Estimates in the Marketplace (page 37) Web Appendix References (page 40) **Web Appendix A: Additional Methodological Details for Study 1a** **Materials** **Consent form** **[condition: control;** (order of brands counterbalanced)**]\** Imagine you're shopping online for a new pair of earbuds and you come across two options at the top of the search results. Both earbuds are from different brands, but they are identical in price, weight, sound quality, and color. Each product also has an average review of 4.8 out of 5. However, the two earbuds **differ **in **the advertised battery life**. - Brand A claims its earbuds have a **battery life of 10 hours [ranging from 8 to 12 hours]**. - Brand B claims its earbuds have a **battery life ranging from 8 to 12 hours [10 hours]**. **[condition: sponsored;** (order of brands counterbalanced)**]\** Imagine you're shopping online for a new pair of earbuds and you come across two options at the top of the search results. **You notice that these two products are labeled as sponsored, meaning the brand owners have paid to ensure their products appear more prominently in your search results.** Both earbuds are from different brands, but they are identical in price, weight, sound quality, and color. Each product also has an average review of 4.8 out of 5. However, the two earbuds **differ **in **the advertised battery life**. - Brand A claims its earbuds have a **battery life of 10 hours [ranging from 8 to 12 hours]**. - Brand B claims its earbuds have a **battery life ranging from 8 to 12 hours [10 hours]**. **[condition: control;** (order of brands counterbalanced)**]** **Alt text:** Study 1a stimulus, control condition: two black wireless earbud sets side by side; Brand A "Battery Lasts 10 Hours", Brand B "Battery Lasts 8 to 12 Hours". [A two-column stimulus. The top row shows product photos: on the left, a pair of black wireless earbuds standing in an open black charging case; on the right, a different pair of black wireless earbuds standing in an open dark gray charging case. Below the photos a dark header row spanning both columns reads "Advertised Battery Life". The next row reads "Brand A" (left) and "Brand B" (right). The bottom row reads "Battery Lasts 10 Hours" under Brand A and "Battery Lasts 8 to 12 Hours" under Brand B. There is no sponsored label.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* **[condition: sponsored;** (order of brands counterbalanced)**]** **Alt text:** Study 1a stimulus, sponsored condition: the same two earbud sets with a red "Sponsored" tag above each column; Brand A 10 hours, Brand B 8 to 12 hours. [The same two-column stimulus as the control condition (two black wireless earbud sets in open charging cases, the header "Advertised Battery Life", "Brand A" with "Battery Lasts 10 Hours" and "Brand B" with "Battery Lasts 8 to 12 Hours"), with one addition: a red rectangular tag reading "Sponsored" in white text sits above each product column.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* **[condition: all conditions]** If you are to buy one of these earbuds, which one are you more likely to buy? **(1 = more likely Brand A; 7 = more likely brand B)** **[condition: all conditions]** How many hours of battery life do you expect to get from\... (please enter a number) Brand A: Brand B: What is your age? What is your gender? - Male - Female - Other **Pre-test study to test manipulation effectiveness** *Manipulation Check -- Persuasion Knowledge Manipulation.* We conducted a pretest (N = 198) to validate our persuasion knowledge manipulation. We randomly assigned participants to the same two conditions and showed them the same two earbuds. However, rather than the dependent measure, we asked participants their level of persuasion knowledge activation based on the extent to which they agreed with the following statements: "The main goal of how these products are presented is to sell earbuds rather than inform consumers," "The way these products are presented seems to involve strategic marketing techniques," "The brands are trying to influence my purchase decision through how they present their products," "I can tell these brands are using marketing tactics to make their offerings more appealing" (1 = "disagree," and 7 = "agree"; α = .84). These measures assess recognition of persuasive intent and perceived manipulative intent, consistent with Campbell (1995) and Campbell and Kirmani (2000). As expected, participants in the persuasion knowledge condition reported higher persuasion knowledge activation (M = 5.38, SD = 1.17) than participants in the control condition (*M* = 4.96, *SD* = 1.09; F(1, 196) = 6.70, *p* = .010; ηp² *=* .033). **Web Appendix B: Additional Methodological Details for Study 1b** **Materials** **Consent form** We are conducting a study to understand how people make shopping decisions on Amazon. Your task is to help us learn more about how you shop and make choices among suggested items. **[condition: persuasion knowledge salient]** You will be reading a scenario involving purchasing a product on Amazon. Please keep this in mind while you complete the entire survey. While reading, think about the intentions and motives behind the way each brand presents its information. Some brands lay out details exactly as they are, whereas others emphasize specific aspects and downplay others to shape shoppers' impressions. **[condition: control]** You will be reading a scenario involving purchasing a product on Amazon. Please keep this in mind while you complete the entire survey. While reading the scenario, consider the product information as you normally would when shopping online---for example, the features that matter to you and whether the product seems to meet everyday needs. **[condition: focal product with point performance]** **Alt text:** Screenshot strip of the Amazon website header with the search box reading "water pitcher with filter for tap water". [A thin screenshot strip of the Amazon.com header: the Amazon logo, "Delivering to Metropolis 62960, Update location", a search box containing "water pitcher with filter for tap water" with the orange magnifier button, "EN", "Hello, sign in, Account & Lists", "Returns & Orders", and a cart showing 0. Below it the dark navigation bar: All, Medical Care, Best Sellers, Amazon Basics, Prime, Today's Deals, New Releases, Music, Customer Service, Amazon Home, Registry, Books, Pharmacy, Gift Cards, Fashion, Smart Home, Sell. The strip sets the scene of an Amazon search; it carries no product information.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* Imagine you are searching on Amazon to buy a new water filter pitcher. You have narrowed down your search to a water filter pitcher called VIO. The VIO water pitcher meets your requirement in terms of material, price, color, design and features an overall rating of +4.8 (out of 5). While you are browsing VIO's product listing images, you will notice that VIO is making the following claims: - VIO can filter **40 Gallons** before needing replacement. - Others can filter **40 Gallons** before needing replacement. **[condition: focal product with range performance]** **Alt text:** Screenshot strip of the Amazon website header with the search box reading "water pitcher with filter for tap water". [A thin screenshot strip of the Amazon.com header: the Amazon logo, "Delivering to Metropolis 62960, Update location", a search box containing "water pitcher with filter for tap water" with the orange magnifier button, "EN", "Hello, sign in, Account & Lists", "Returns & Orders", and a cart showing 0. Below it the dark navigation bar: All, Medical Care, Best Sellers, Amazon Basics, Prime, Today's Deals, New Releases, Music, Customer Service, Amazon Home, Registry, Books, Pharmacy, Gift Cards, Fashion, Smart Home, Sell. The strip sets the scene of an Amazon search; it carries no product information.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* Imagine you are searching on Amazon to buy a new water filter pitcher. You have narrowed down your search to a water filter pitcher called VIO. The VIO water pitcher meets your requirement in terms of material, price, color, design and features an overall rating of +4.8 (out of 5). While you are browsing VIO's product listing images, you will notice that VIO is making the following claims: - VIO can filter **36-44 Gallons** before needing replacement. - Others can filter **40 Gallons** before needing replacement. **[condition: focal product with point performance; product shape counterbalanced]** **Alt text:** Study 1b stimulus, point condition: mock Amazon listing "VIO: Water Filter Pitcher, Filters 40 Gallons of Water"; VIO and Others both show Filter Lifespan 40 Gallons. [A mock Amazon product page. At the top, the Amazon header with the search box reading "water pitcher with filter for tap water", the Electronics sub-menu, and a breadcrumb (Electronics, Headphones, Earbuds & Accessories, Headphones & Earbuds, Earbud Headphones). The main image is a two-column comparison graphic: a blue panel headed "VIO" with a white water filter pitcher and a filter cartridge, and a light blue panel headed "Others" with a different white pitcher and cartridge. Under each pitcher three specification lines separated by rules. VIO: "Filtration: Reduces Chlorine", "Material: 2025 BPA-Free Plastic", "Filter Lifespan: 40 Gallons". Others: "Filtration: Reduces Chlorine", "Material: 2024 BPA-Free Plastic", "Filter Lifespan: 40 Gallons". To the right, the product title "VIO: Water Filter Pitcher" with the bold line "Filters 40 Gallons of Water", a "Visit VIO store" link, and a rating of 4.6 with four and a half orange stars.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* **[condition: focal product with range performance; product shape counterbalanced]** **Alt text:** Study 1b stimulus, range condition: mock Amazon listing "VIO: Water Filter Pitcher, Filters 36 to 44 Gallons of Water"; VIO shows Filter Lifespan 36-44 Gallons, Others 40 Gallons. [The same mock Amazon product page as the point condition, with the focal claim changed to a range. The blue "VIO" panel reads "Filtration: Reduces Chlorine", "Material: 2025 BPA-Free Plastic", "Filter Lifespan: 36-44 Gallons"; the light blue "Others" panel reads "Filtration: Reduces Chlorine", "Material: 2024 BPA-Free Plastic", "Filter Lifespan: 40 Gallons". The product title reads "VIO: Water Filter Pitcher" with the bold line "Filters 36–44 Gallons of Water", a "Visit VIO store" link, and a 4.6 star rating.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* **{product images repeated}** If you were in this scenario\... Given other brands' filter life (40 gallons as shown in the image above) How likely would you be to purchase the VIO water pitcher that can filter **[40 Gallons / 36-44 Gallons] **before it needs replacement? **(1 = not likely at all; 7 = very likely)** How many gallons of water (filter lifespan) do you expect VIO water pitcher to actually have before it needs a replacement? **(slider scale 0 to 100)** How precise was VIO's filter lifespan estimate? (**1 = not at all precise; 7 = very precise)** How specific was the information about the VIO's expected filter lifespan? (**1 = not at all specific; 7 = very specific)** What is your sex? - Male - Female - Prefer Not to Answer What's your age? **Web Appendix C: Additional Methodological Details for Study 2** **Materials** **Consent form** We are conducting a study to understand how people make shopping decisions on Amazon. Your task is to help us learn more about how you shop and make choices among suggested items. **[condition: persuasion knowledge salient]** You will be reading a scenario involving purchasing a product on Amazon. Please keep this in mind while you complete the entire survey. While reading, think about the intentions and motives behind the way each brand presents its information. Some brands lay out details exactly as they are, whereas others emphasize specific aspects and downplay others to shape shoppers' impressions. **[condition: control]** You will be reading a scenario involving purchasing a product on Amazon. Please keep this in mind while you complete the entire survey. While reading the scenario, consider the product information as you normally would when shopping online---for example, the features that matter to you and whether the product seems to meet everyday needs. When ready, click on the button below to proceed. **[condition: focal product with point performance]** **Alt text:** Screenshot strip of the Amazon website header with the search box reading "earbuds" in the Electronics department. [A thin screenshot strip of the Amazon.com header: the Amazon logo, "Hello, Select your address", a department selector set to "Electronics" next to a search box containing "earbuds", the orange magnifier button, a flag icon, "Hello, Sign in, Account & Lists", "Returns & Orders", and a cart showing 0. Below it the dark navigation bar (All, Best Sellers, Prime, Customer Service, New Releases, Pharmacy, Today's Deals, Books, Fashion, Toys & Games, Registry, Kindle Books, Gift Cards, Amazon Home, Sell, Computers, Automotive, Coupons, "Get your school supplies, fast") and a lighter Electronics sub-bar (All Electronics, Deals, Best Sellers, TV & Video, Audio & Home Theater, Computers, Camera & Photo, Wearable Technology, Car Electronics & GPS, Portable Audio, Cell Phones, Office Electronics, Musical Instruments, New Arrivals, Trade-In). The strip sets the scene; it carries no product information.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* Imagine you are searching on Amazon to buy a new pair of earbuds. You have narrowed down your search to an earbud called VIO. The VIO earbuds meet your requirement in terms of material, price, weight, sound quality, color, and a rating of +4.6 (out of 5). While you are browsing VIO's product listing images, you will notice that VIO is making the following claims: - VIO's battery will last **5 hours**. - Other earbuds battery will last **5 hours.** **[condition: focal product with range performance]** **Alt text:** Screenshot strip of the Amazon website header with the search box reading "earbuds" in the Electronics department. [A thin screenshot strip of the Amazon.com header: the Amazon logo, "Hello, Select your address", a department selector set to "Electronics" next to a search box containing "earbuds", the orange magnifier button, a flag icon, "Hello, Sign in, Account & Lists", "Returns & Orders", and a cart showing 0. Below it the dark navigation bar (All, Best Sellers, Prime, Customer Service, New Releases, Pharmacy, Today's Deals, Books, Fashion, Toys & Games, Registry, Kindle Books, Gift Cards, Amazon Home, Sell, Computers, Automotive, Coupons, "Get your school supplies, fast") and a lighter Electronics sub-bar (All Electronics, Deals, Best Sellers, TV & Video, Audio & Home Theater, Computers, Camera & Photo, Wearable Technology, Car Electronics & GPS, Portable Audio, Cell Phones, Office Electronics, Musical Instruments, New Arrivals, Trade-In). The strip sets the scene; it carries no product information.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* Imagine you are searching on Amazon to buy a new pair of earbuds. You have narrowed down your search to an earbud called VIO. The VIO earbuds meet your requirement in terms of material, price, weight, sound quality, color, and a rating of +4.6 (out of 5). While you are browsing VIO's product listing images, you will notice that VIO is making the following claims: - VIO's battery will last **4-6 hours**. - Other earbuds battery will last **5 hours.** when ready click next to see the VIO's listing page **{text repeated}** **[condition: focal product with point performance; product placement order counterbalanced]** **Alt text:** Study 2 stimulus, point condition: mock Amazon listing "VIO: Wireless Earbuds with 5 HOURS of BATTERY LIFE"; VIO and Others both claim 5 hours battery life. [A mock Amazon product page. At the top, the Amazon header with the search box reading "wireless earbuds", the Electronics sub-menu, and a breadcrumb (Electronics, Headphones, Earbuds & Accessories, Headphones & Earbuds, Earbud Headphones). The main image is a two-column comparison graphic: a blue panel headed "VIO" showing black earbuds in an open case, and a light blue panel headed "Others" showing a different pair of black earbuds in an open case. Under each, three lines separated by rules. VIO: "Bluetooth 5.0", "Earbud controls", "5 hours battery life" (bold). Others: "Bluetooth 4.0", "Earbud controls", "5 hours battery life" (bold). To the right, the title "VIO: Wireless Earbuds with" and the bold line "5 HOURS of BATTERY LIFE.", a "Visit VIO store" link and a 4.6 star rating. Under the image: "Roll over image to zoom in".] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* **[condition: focal product with range performance; product placement order counterbalanced]** **Alt text:** Study 2 stimulus, range condition: mock Amazon listing "VIO: Wireless Earbuds with 4-6 HOURS of BATTERY LIFE"; VIO claims 4-6 hours, Others 5 hours. [The same mock Amazon product page as the point condition, with the focal claim changed to a range. The blue "VIO" panel reads "Bluetooth 5.0", "Earbud controls", "4-6 hours battery life" (bold); the light blue "Others" panel reads "Bluetooth 4.0", "Earbud controls", "5 hours battery life" (bold). The title reads "VIO: Wireless Earbuds with" and the bold line "4-6 HOURS of BATTERY LIFE.", with the "Visit VIO store" link and the 4.6 star rating. Under the image: "Roll over image to zoom in".] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* **{images repeated}** If you were in this scenario\... Given the average battery life of other earbuds (**5 hours** as shown in the image above): How likely would you be to purchase the VIO buds with **[5 / 4-6]** hours battery life? **(1 = not likely at all; 7 = very likely)** **{images repeated}** Given the average battery life of other earbuds (5** hours** as shown in the image above), indicate your agreement with each statement regarding VIO **(1 = strongly disagree; 4 = neither agree nor disagree; 7 = strongly agree)**: - I am **doubtful** about VIO's actual performance - I am **unsure** about VIO's actual performance - I am **uncertain** about VIO's actual performance **{images repeated}** Given the average battery life of other earbuds (5** hours** as shown in the image above), indicate your agreement with each statement regarding VIO **(1 = strongly disagree; 4 = neither agree nor disagree; 7 = strongly agree)**: - VIO is a **trustworthy** brand - VIO is a **believable** brand - VIO is a **honest** brand The following statements are about the way VIO's battery performance is described. How easy or difficult was it to understand the performance information provided for this product? **(1 = very easy; 7 = very difficult)** How much mental effort did it take you to evaluate this product's performance based on the information given? **(1 = very little mental effort; 7 = a great deal of mental effort)** How precise was the battery life estimate provided? (**1 = not at all precise; 7 = very precise)** How specific was the information about the brand's expected battery life? (**1 = not at all specific; 7 = very specific)** What is your sex? - Male - Female - Prefer Not to Answer What's your age? **Separate analyses to rule out alternative explanations** *Alternative Explanations.* A 2 (condition) × 2 (focal brand) ANOVA on processing ease did not reveal an interaction (F(1, 396) = 1.51, *p* = .220). There was a main effect of condition (F(1, 396) = 4.35, *p* = .038, ηp² = .011) such that participants in the control condition felt more difficulty understanding the performance information (M = 2.97, SD = 1.93) than those in the persuasion knowledge condition (M = 2.58, SD = 1.84). Further, there was a main effect of focal brand (F(1, 396) = 6.86, *p* = .009, ηp² = .017) such that participants in the range estimate condition felt more difficulty understanding the performance information (M = 3.02, SD = 1.95) than those in the point estimate condition (M = 2.53, SD = 1.80). A 2 (condition) × 2 (focal brand) ANOVA on mental effort did not reveal a significant interaction (F(1, 395) = 3.23, *p* = .073) or main effects (Fs < 1). Given the main effects of processing ease, we conducted another mediation analysis using it as an additional parallel mediator in addition to the previous specifications. However, the index of moderated mediation for processing ease was not significant (β = -.03, 95% CI: [-.1054 to .0172]), with nonsignificant indirect effects (e.g., including zero) both in the control and persuasion knowledge conditions. Thus, it is unlikely that the observed results occurred due to changes in processing ease/difficulty. **Supplementary Analysis: Ruling out Perceived Precision as an Explanation for the Preference Reversal** In Studies 1b and 2, we observed an interaction between Condition (PK vs. Control) and Format (Point vs. Range) on the manipulation check measure, Perceived Precision. This occurred because PK activation slightly attenuated the difference in precision between the formats. To empirically ensure this perceptual shift does not explain the preference reversal observed on the main DV, we conducted a variety of additional analyses. *Additional Analyses -- Study 1b.* First, one could argue that the point estimate is perceived as less precise in the persuasion knowledge condition than in the baseline condition, increasing the preference for the range (vs. point) estimates. To address this concern, we compared the perceived precision of the point estimate across the baseline and persuasion knowledge conditions, and simple main effects analyses showed no differences between the baseline (M~point~ = 5.97, SD~point~ = .91) and persuasion knowledge (M~point~ = 5.89, SD~point~ = .90; F(1, 396) = .198, *p* = .656). Still, this explanation would not follow from our main results because (a) the point estimate was nonetheless perceived as more precise than the range estimate in both conditions, and (b) participants in the persuasion knowledge preferred the option perceived as less precise (i.e., the option with a range estimate), (c) the correlation between our dependent variable (product preference) and manipulation check (perceived precision) is positive (r = .270, *p* < .001). Second, one could argue that the range estimate is perceived as more precise in the persuasion knowledge condition than in the baseline condition, increasing the preference for range (vs. point) conditions. While the simple main effects showed that the range was perceived as more precise in the persuasion knowledge (M~range~ = 4.61, SD~range~ = 1.42) than in the baseline (M~range~ = 4.19, SD~range~ = 1.65; F(1, 396) = 5.54, *p* = .019) condition, this explanation is inconsistent with the fact that (a) the range estimate was nonetheless perceived as less precise than the point estimate in both conditions, (b) participants in the persuasion knowledge preferred the option perceived as less precise (i.e., the option with a range estimate), and (c) the correlation between product preference and perceived precision is positive. Thus, an increase in the perceived precision of the range estimate makes our hypothesis testing more conservative. *Additional Analyses -- Study 2.* Third, one might wonder if precision mediates our observed effects. To address this, we conducted a moderated mediation analysis using PROCESS model 8 (Hayes 2022; 5,000 samples) with focal brand (1 = range estimate, 0 = point estimate) as the independent variable, condition (1 = persuasion knowledge, 0 = control) as the moderator, product preference as the dependent variable, and (a) perceived uncertainty, (b) perceived trustworthiness, and (c) perceived precision as parallel mediators. The index of moderated mediation for perceived precision was not significant (β = .016, 95% CI: [-.0750 to .1180]). Furthermore, the moderated mediation indexes for perceived uncertainty (β = .175, 95% CI: [.0270 to .3676]) and perceived trustworthiness (β = .376, 95% CI: [.1000 to .7199]) continued to be significant, with indirect effects consistent with the patterns observed in Study 2. Collectively, these results suggest that it is unlikely that shifts in precision explain our effects, and the mediating roles of perceived uncertainty and perceived trustworthiness are more parsimonious explanations. **Supplementary Analysis: Examining Serial Mediation Between Trustworthiness and Uncertainty** Perceived trustworthiness and perceived uncertainty are negatively correlated (r = -.54, *p* < .001), raising the possibility that trustworthiness serves as an antecedent to uncertainty reduction rather than operating as a distinct parallel pathway. To examine this, we estimated a moderated serial mediation model (PROCESS Model 85; Hayes 2022; 5,000 bootstrap samples) with format (1 = range, 0 = point) as the independent variable, condition (1 = persuasion knowledge, 0 = control) as the moderator, perceived trustworthiness as the first mediator, perceived uncertainty as the second mediator, and purchase likelihood as the dependent variable. Three conditional indirect effects were estimated. The indirect effect through trustworthiness alone was not significant at baseline (β = -.12, 95% CI: [-.3487 to .0855]) but was significant when PK was activated (β = .26, 95% CI: [.0826 to .4599]). The indirect effect through uncertainty alone was significant at baseline (β = -.10, 95% CI: [-.2272 to -.0130]) but not when PK was activated (β = -.03, 95% CI: [-.1229 to .0530]). The serial indirect effect (Trustworthiness → Uncertainty) was not significant at baseline (β = -.03, 95% CI: [-.0971 to .0248]) but was significant when PK was activated (β = .07, 95% CI: [.0190 to .1416]). Several features of these results support treating the parallel model as the more appropriate framework for our theorizing. First, the parallel pathway through trustworthiness alone (β = .26) is nearly four times larger than the serial pathway (β = .07) in the PK condition, indicating that the majority of the trustworthiness-related indirect effect does not pass through uncertainty. Second, the correlation between trustworthiness and uncertainty is similar across conditions (r = -.51 at baseline vs. r = -.57 under PK; Fisher's z = 0.87, p = .385) and across all four cells (range: r = -.48 to -.58), consistent with a general association between the constructs rather than a process specific to the format manipulation or PK activation. Third, the interaction between format and condition on perceived uncertainty is not significant after controlling for trustworthiness (β = -.34, p = .208), indicating that format does not differentially affect uncertainty across conditions beyond its shared variance with trustworthiness. Collectively, these results suggest that the serial pathway reflects the natural covariation between trustworthiness and uncertainty rather than a distinct sequential mechanism, and the parallel mediation model provides a more parsimonious account of the data. **Supplementary Analysis: Testing Whether Persuasion Knowledge Moderates the Mediator-to-Preference Paths** The parallel moderated mediation reported in the main text (PROCESS Model 8) constrains the paths from the two mediators to purchase likelihood to be equal across conditions. Our theorizing implies that activating persuasion knowledge changes the weight the format cue receives when the two perceptions are constructed, rather than the weight consumers place on those perceptions once formed (Edwards and Lambert 2007; Muller, Judd, and Yzerbyt 2005). To test the alternative, we re-estimated the model allowing condition to moderate the paths from both mediators to purchase likelihood as well as the format-to-mediator paths (the specification of Model 59 in Hayes 2022), with format (1 = range, 0 = point) as the independent variable and condition (1 = persuasion knowledge, 0 = control) as the moderator. Paths were estimated with ordinary least squares and indirect effects with 5,000 percentile bootstrap samples; the analysis code is available with the data. The format-to-mediator (first-stage) interactions mirrored the ANOVA results: format × condition was significant for perceived uncertainty (b = -.79, SE = .31, t(396) = -2.54, *p* = .011) and perceived trustworthiness (b = .63, SE = .23, t(396) = 2.74, *p* = .006). In the outcome model, neither mediator-to-preference path showed evidence of moderation by condition (uncertainty × condition: b = .05, SE = .09, t(392) = 0.56, *p* = .575, 95% CI: [-.13 to .23]; trustworthiness × condition: b = -.05, SE = .13, t(392) = -0.41, *p* = .680, 95% CI: [-.31 to .20]), and a joint test of the two product terms, computed from the equivalent ordinary least squares estimation, was not significant (F(2, 392) = 0.52, *p* = .594). The conditional paths were similar across conditions (uncertainty: b = -.25 at baseline vs. -.20 under persuasion knowledge; trustworthiness: b = .62 vs. .57; see Figure 2 in the main text). The conditional indirect effects replicated the pattern from the parallel model: the indirect effect through uncertainty was significant at baseline (-.15, 95% CI: [-.33 to -.03]) but not under persuasion knowledge (.04, 95% CI: [-.06 to .15]), and the indirect effect through trustworthiness was significant under persuasion knowledge (.25, 95% CI: [.07 to .47]) but not at baseline (-.13, 95% CI: [-.37 to .09]). The indices of moderated mediation remained significant for both mediators (uncertainty: .19, 95% CI: [.02 to .39]; trustworthiness: .37, 95% CI: [.09 to .70]). Three qualifications apply. First, the nonsignificant interaction terms indicate an absence of evidence for path moderation rather than equivalence, since the confidence intervals do not exclude moderate changes; the positive evidence for our account is the pair of significant first-stage interactions, which the alternative predicts should be absent. Second, the direct effect of format remained moderated by condition (b = .83, SE = .25, t(392) = 3.37, *p* = .001, 95% CI: [.35 to 1.32]), so this conclusion is scoped to the two measured perceptions. Third, because the mediators were measured rather than manipulated, the mediator-to-preference paths are correlational (Pieters 2017; Spencer, Zanna, and Fong 2005). The pattern is therefore consistent with attentional focus changing which perception the format cue is recruited to construct, not how much either is relied upon. **Web Appendix D: Additional Methodological Details for Study 3** In the manuscript, the comparative conditions are the joint evaluation mode and the isolated conditions are the separate evaluation mode. **Materials** **Consent form** You passed the attention check. Thank you for being an attentive participant! First study: On the next page, you will read a number of statements. Your task is to indicate the extent to which you agree with each of the statements. When you are ready to begin please click the arrow button. **[condition: control; one question per page with block display order randomized]** This summer felt more humid than last year's. We can rely on our intuition to solve problems. Some people are born to be artists. We can learn new skills by repeating a task many times. **(1 = definitely disagree; 7 = definitely agree)** **[condition: persuasion knowledge salient; one question per page with block display order randomized]** We cannot depend on companies' advertised product claims. Advertised product performance rarely matches the actual performance. Companies are not a reliable source of information about the quality and performance of their products. In general, companies do not present a true picture of their products' performance. **(1 = definitely disagree; 7 = definitely agree)** Click next to continue with the next study. **[condition: comparative with focal brand point performance]** **Alt text:** Simplified Amazon header strip with the search box reading "electric scooter". [A thin dark-blue header strip in the style of Amazon.com: the Amazon logo at the left, a white search box containing "electric scooter" with an orange "Go" button, and the links "Your Account", "Help", and "Cart" at the right. The strip sets the scene; it carries no product information.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* Imagine you are searching on Amazon to buy a new electric scooter. You have narrowed down your search to a few electric scooters from several different US-based manufacturers. You have further narrowed down your search to visit a product page by a brand called iScooter. The iScooter electric scooter meets your requirement in terms of material, price, color, design and features an overall rating of +4.6 (out of 5). While you are browsing iScooter's product listing images, you notice that iScooter has a comparison table suggesting: - iScooter can travel **24 miles per full charge. ** - Others are also **24 miles per full charge. ** **[condition: comparative with focal brand range performance]** **Alt text:** Simplified Amazon header strip with the search box reading "electric scooter". [A thin dark-blue header strip in the style of Amazon.com: the Amazon logo at the left, a white search box containing "electric scooter" with an orange "Go" button, and the links "Your Account", "Help", and "Cart" at the right. The strip sets the scene; it carries no product information.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* Imagine you are searching on Amazon to buy a new electric scooter. You have narrowed down your search to a few electric scooters from several different US-based manufacturers. You have further narrowed down your search to visit a product page by a brand called iScooter. The iScooter electric scooter meets your requirement in terms of material, price, color, design and features an overall rating of +4.6 (out of 5). While you are browsing iScooter's product listing images, you notice that iScooter has a comparison table suggesting: - iScooter can travel **22-26 miles per full charge. ** - Others are **24 miles per full charge. ** **[condition: isolated with focal brand point performance]** **Alt text:** Simplified Amazon header strip with the search box reading "electric scooter". [A thin dark-blue header strip in the style of Amazon.com: the Amazon logo at the left, a white search box containing "electric scooter" with an orange "Go" button, and the links "Your Account", "Help", and "Cart" at the right. The strip sets the scene; it carries no product information.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* Imagine you are searching on Amazon to buy a new electric scooter. You have narrowed down your search to a few electric scooters from several different US-based manufacturers. You have further narrowed down your search to visit a product page by a brand called iScooter. The iScooter electric scooter meets your requirement in terms of material, price, color, design and features an overall rating of +4.6 (out of 5). While you are browsing iScooter's product listing images, you notice that iScooter has a table suggesting: - iScooter can travel **24 miles per full charge. ** **[condition: isolated with focal brand range performance]** **Alt text:** Simplified Amazon header strip with the search box reading "electric scooter". [A thin dark-blue header strip in the style of Amazon.com: the Amazon logo at the left, a white search box containing "electric scooter" with an orange "Go" button, and the links "Your Account", "Help", and "Cart" at the right. The strip sets the scene; it carries no product information.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* Imagine you are searching on Amazon to buy a new electric scooter. You have narrowed down your search to a few electric scooters from several different US-based manufacturers. You have further narrowed down your search to visit a product page by a brand called iScooter. The iScooter electric scooter meets your requirement in terms of material, price, color, design and features an overall rating of +4.6 (out of 5). While you are browsing iScooter's product listing images, you notice that iScooter has a table suggesting: - iScooter can travel **22-26 miles per full charge. ** when ready click next to see the iScooter's listing page on amazon. **[condition: comparative; focal product with point performance]** **Alt text:** Study 3 stimulus, comparative evaluation, point condition: mock Amazon listing "iScooter Electric Scooter Dual Motors, Travel Distance: 24 Miles per Full Charge"; iScooter and Others both show 24 miles. [A mock Amazon product page under the simplified header (search box "electric scooter"). The main image is a two-column comparison graphic: a light gray panel headed "iScooter" with a photo of a black electric kick scooter with orange wheel accents, and a darker gray panel headed "Others" with a photo of a different black electric kick scooter. A white box under each photo reads "Travel Distance: 24 miles" (iScooter) and "Travel Distance 24 miles" (Others). To the right, the title "iScooter Electric Scooter Dual Motors" with the bold line "Travel Distance: 24 Miles per Full Charge", a "Visit iScooter store" link and a 4.6 star rating.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* **[condition: comparative; focal product with range performance]** **Alt text:** Study 3 stimulus, comparative evaluation, range condition: mock Amazon listing "Travel Distance: 22 to 26 Miles per Full Charge"; iScooter shows 22-26 miles, Others 24 miles. [The same two-column mock Amazon page as the comparative point condition, with the focal claim changed to a range. The white box under the iScooter photo reads "Travel Distance: 22-26 miles"; the box under the Others photo reads "Travel Distance: 24 miles". The title reads "iScooter Electric Scooter Dual Motors" with the bold line "Travel Distance: 22—26 Miles per Full Charge", the "Visit iScooter store" link and the 4.6 star rating.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* **[condition: isolated; focal product with point performance]** **Alt text:** Study 3 stimulus, isolated evaluation, point condition: mock Amazon listing for the iScooter alone, "Travel Distance: 24 Miles per Full Charge". [A mock Amazon product page under the simplified header (search box "electric scooter"). The main image is a single light gray panel headed "iScooter" with a photo of a black electric kick scooter with orange wheel accents and a white box reading "Travel Distance: 24 miles". There is no "Others" panel. To the right, the title "iScooter Electric Scooter Dual Motors" with the bold line "Travel Distance: 24 Miles per Full Charge", a "Visit iScooter store" link and a 4.6 star rating.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* **[condition: isolated; focal product with range performance]** **Alt text:** Study 3 stimulus, isolated evaluation, range condition: mock Amazon listing for the iScooter alone, "Travel Distance: 22 to 26 Miles per Full Charge". [The same single-panel mock Amazon page as the isolated point condition, with the claim changed to a range: the white box under the scooter photo reads "Travel Distance: 22-26 miles" and the title carries the bold line "Travel Distance: 22—26 Miles per Full Charge", with the "Visit iScooter store" link and the 4.6 star rating.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* **{images repeated}** **[condition: all]** If you were in this scenario\... **[condition: comparative]** Given the travel distance per full charge of other electric scooters as shown in the image above): **[condition: all]** How likely would you be to purchase the iScooter electric scooter with a travel distance of **[24 miles / 22-26 miles] **per full charge? **(1 = very unlikely; 7 = very likely)** What is your sex? - Male - Female - Prefer Not to Answer What's your age? **Web Appendix E: Additional Methodological Details for Study 4** **Materials** **Consent Form** **[condition: persuasion knowledge salient]** You will be reading a scenario involving purchasing an electric vehicle. When reading the scenario, you should imagine yourself as a shopper considering purchasing a new electric vehicle. While reading the scenario, please take a minute to think about the advertised product information. Think about the **intentions and motives** behind the way each brand advertised information about their product. **Some brands try to be as realistic and honest** as possible when advertising information about their products. **Other brands exaggerate and are dishonest** when advertising information about their products. **[condition: control]** You will be reading a scenario involving purchasing an electric vehicle. When reading the scenario, you should imagine yourself as a shopper considering purchasing a new electric vehicle. While reading the scenario, **please take a minute to think about considerations you have when purchasing a new vehicle.** **[condition: moderate range width]** **Alt text:** Screenshot strip of the Autotrader website header and its "Search Cars for Sale" banner. [A screenshot strip of the Autotrader.com header on a dark blue band: the Autotrader logo, then the menu items Used Cars, New Cars, Sell My Car, Instant Cash Offer, Car Research & Tools, Find Local Dealers, and at the right a price-tag icon, an account icon and "Sign In". Below it, over a dim mountain-road photo, the banner "Search Cars for Sale" with three controls (an "Any Make" drop-down, an "Any Model" drop-down, a "ZIP Code" field) and an orange "Search" button, plus the links "Search by Budget", "Browse by Style", and "Advanced Search". The strip sets the scene; it carries no product information.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* Imagine you are searching on Autotrader to purchase an electronic vehicle. You have narrowed down your search to two electronic vehicles from two different manufacturers. Further, the two electronic vehicles are identical in terms of price, safety measures, performance, and technology. Both of these vehicles have a rating of +4.7 (out of 5) from Consumer Reports. The two vehicles differ on their **mileage after a full battery charge**: - Brand A claims that its electric car battery will last **260 miles**. - Brand B claims that its electric car battery will last **240 to 280 miles**. On the next page, you will see the cars. **[condition: wide range width]** **Alt text:** Screenshot strip of the Autotrader website header and its "Search Cars for Sale" banner. [A screenshot strip of the Autotrader.com header on a dark blue band: the Autotrader logo, then the menu items Used Cars, New Cars, Sell My Car, Instant Cash Offer, Car Research & Tools, Find Local Dealers, and at the right a price-tag icon, an account icon and "Sign In". Below it, over a dim mountain-road photo, the banner "Search Cars for Sale" with three controls (an "Any Make" drop-down, an "Any Model" drop-down, a "ZIP Code" field) and an orange "Search" button, plus the links "Search by Budget", "Browse by Style", and "Advanced Search". The strip sets the scene; it carries no product information.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* Imagine you are searching on Autotrader to purchase an electronic vehicle. You have narrowed down your search to two electronic vehicles from two different manufacturers. Further, the two electronic vehicles are identical in terms of price, safety measures, performance, and technology. Both of these vehicles have a rating of +4.7 (out of 5) from Consumer Reports. The two vehicles differ on their **mileage after a full battery charge**: - Brand A claims that its electric car battery will last **260 miles**. - Brand B claims that its electric car battery will last **210 to 310 miles**. On the next page, you will see the cars. **[moderate range width]** **Alt text:** Study 4 stimulus, moderate range width: two blurred black electric sedans, Brand A "260 Miles" and Brand B "240 to 280 Miles" of advertised travel distance per full charge. [Two product photos side by side, each a black sedan seen from the front three-quarter view and deliberately blurred, labeled "Brand A" (left) and "Brand B" (right). Below them a three-row table: a green header row reading "Advertised Travel Distance Per Full Battery Charge"; a yellow row reading "Brand A" and "Brand B"; and a value row reading "260 Miles" under Brand A and "240 to 280 Miles" under Brand B.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* **[wide range width]** **Alt text:** Study 4 stimulus, wide range width: the same two blurred sedans, Brand A "260 Miles" and Brand B "210 to 310 Miles". [The same layout as the moderate range width stimulus (two blurred black sedans labeled Brand A and Brand B over a table headed "Advertised Travel Distance Per Full Battery Charge"), with the Brand B value widened: the value row reads "260 Miles" under Brand A and "210 to 310 Miles" under Brand B.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* **(The pictures are blurred for copyright reasons)** If you are to buy one of these cars, which one are you more likely to buy? **(1 = more likely Brand A; 7 = more likely brand B)** Attention check: what was the advertised travel distance of Brand A's electric car? (200 miles, 220 miles, 240 miles, 260 miles, 280 miles) -- choice question. What is your sex? - Male - Female - Other - Prefer Not to Answer What's your age? \_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_ **Pre-test study to test the range manipulation** *Pretest -- Range Level Manipulation.* We conducted a pretest (*N* = 205) to validate our range level manipulation. Participants were randomly assigned to the same conditions used in Study 4. We then asked participants to evaluate the range width of the option with range performance claims on a scale ranging from 1 = "very narrow," to 4 = "neither narrow nor wide," to 7 = "very wide." As predicted, the 210-310 range was perceived as wider (*M* = 4.46 *SD* = 1.23) than the 240-280 range (*M* = 3.12, SD = 1.20; *F*(1, 203) = 62.33, *p* < .001). **Web Appendix F: Additional Methodological Details for Study 5** **Facebook Advertisements** **Alt text:** Study 5 Facebook advertisement, focal brand with a range claim: an "Ember & Leaf Candle Co." sponsored post whose image compares "US" (Burn for 45-55 hours) with "THEM" (Burn for 50 hours). [A Facebook sponsored post shown on a phone screen. Header: a round logo, "Ember & Leaf Candle Co.", "Sponsored". Post text: "Looking for candles that create a cozy, inviting atmosphere? Our handmade candles are crafted with care, offering rich, concentrated fragrances and a burn time designed to last. Perfect for adding warmth to any moment". The ad image is split in two: an orange left half headed "US" with a lit amber candle jar, and a cream right half headed "THEM" with a lit candle jar, a black circular "VS" badge between them. Under each candle, three lines: "Handmade", "More concentrated", and the burn-time claim, "Burn for 45-55 hours" on the US side and "Burn for 50 hours" on the THEM side. Below the image a link card reads "www.emberleaf.com", "Handmade Candles That Last", with a "Shop now" button, then reaction icons with the count 6 and "2 comments", and the Like, Comment, Share bar.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* **Alt text:** Study 5 Facebook advertisement, focal brand with a point claim: the same "Ember & Leaf Candle Co." post with "US" (Burn for 50 hours) versus "THEM" (Burn for 45-55 hours). [The same Facebook sponsored post as the range version (same logo, post text, orange "US" half and cream "THEM" half with the "VS" badge, "Handmade" and "More concentrated" lines, link card "www.emberleaf.com, Handmade Candles That Last" with a "Shop now" button), with the burn-time claims swapped: "Burn for 50 hours" on the US side and "Burn for 45-55 hours" on the THEM side. The reaction row shows the count 4 and "1 comment".] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* **Meta Ads Manager Screenshot** Note: Reach values are per-ad audiences as reported by Meta Ads Manager; the deduplicated campaign-level reach was 25,750. All reported tests use the per-ad values. **Alt text:** Meta Ads Manager screenshot for Study 5: two ad sets, "Focal Brand: Point" (187 link clicks, 16,511 impressions, reach 12,603, CTR 1.13%) and "Focal Brand: Range" (242 link clicks, 16,269 impressions, reach 13,243, CTR 1.49%). [A screenshot of the Meta Ads Manager "Ad sets" view, date range "This month: Dec 1, 2024 – Dec 31, 2024", with the account identifiers covered by red boxes. Tabs show "Campaigns, 2 selected", "Ad sets, 2 selected", "Ads for 2 Ad sets". The results table has two rows, both switched on. Row "Focal Brand: Point": Results 187 (Link clicks), Link clicks 187, Impressions 16,511, Unique link clicks 175, Reach 12,603, CTR (link click-through) 1.13%, Unique CTR 1.39%. Row "Focal Brand: Range": Results 242 (Link clicks), Link clicks 242, Impressions 16,269, Unique link clicks 230, Reach 13,243, CTR 1.49%, Unique CTR 1.74%. Totals row ("Results from 2 ad sets"): 429 link clicks, 429 link clicks, 32,780 impressions, 405 unique link clicks, 25,750 Accounts Center accounts reached, CTR 1.31% per impressions, unique CTR 1.57% per Accounts Center accounts.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* Note: portions of the screenshots that identify the advertising account are obscured. **Landing Page** **Alt text:** Study 5 landing page: an "Ember & Leaf Co." page that tells the visitor the advertisement was fictitious and part of a research study. [A plain web page. At the top center, "Ember & Leaf Co." in large orange letters over a short orange rule. Below, three lines of gray text each reading "Candles that last!". At the bottom, a debriefing paragraph: "Hello. The advertising you have just clicked is a fictitious advertising presented to you by research purposes. Ember & Leaf Co. is not an existing organization. Rest assured that none of your personal data have been collected. The only information we have is the aggregated number of people who clicked or not the ad. If you want more information about the study, please do not hesitate to contact [HIDDEN FOR PEER-REVIEW] Thank you!" (the contact details are covered by a box in the review copy).] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* We note that this approach to the landing page has been used in research using Facebook Ads (Castelo, Bos, and Lehmann 2019; Mookerjee, Cornil, and Hoegg 2021). **Web Appendix G: Additional Details for Pilot Studies** **Materials for Pilot Study 1: Attribute Variability** PRODUCT FEATURE EVALUATION STUDY Thank you for participating. In this survey, we will ask your opinions about several product features. There are no right or wrong answers. **[the following six items were presented in randomized order]** Consider the following product feature: **Electric scooter range** (the maximum distance an electric scooter can travel on a single charge). How much do you think this feature can vary depending on how the product is used or the conditions the product is used in? (1 = not at all, 4 = moderately, 7 = very much) Consider the following product feature: **Laptop battery life** (how long a laptop battery lasts between charges). How much do you think this feature can vary depending on how the product is used or the conditions the product is used in? (1 = not at all, 4 = moderately, 7 = very much) Consider the following product feature: **Stain removal effectiveness** (how well a laundry detergent removes stains from clothing). How much do you think this feature can vary depending on how the product is used or the conditions the product is used in? (1 = not at all, 4 = moderately, 7 = very much) Consider the following product feature: **Hard drive storage capacity** (the total amount of data a hard drive can store, e.g., 1 TB). How much do you think this feature can vary depending on how the product is used or the conditions the product is used in? (1 = not at all, 4 = moderately, 7 = very much) Consider the following product feature: **Laptop screen size** (the diagonal measurement of a laptop display, e.g., 15.6 inches). How much do you think this feature can vary depending on how the product is used or the conditions the product is used in? (1 = not at all, 4 = moderately, 7 = very much) Consider the following product feature: **Smartphone weight** (the weight of a smartphone as listed on its specifications, e.g., 6.1 oz). How much do you think this feature can vary depending on how the product is used or the conditions the product is used in? (1 = not at all, 4 = moderately, 7 = very much) What is your age? (Dropdown list) What is your gender? - Female - Male - Other / prefer not to specify --------------------------------end of survey-------------------------------- **Materials for Pilot Study 2: Attentional Focus** In today's study, you will be asked to respond to a variety of different questions. It's important that you pay close attention and read all directions carefully. To indicate that you're able to read and understand directions please select the word below that describes an animal. Rock Car Stove Guitar Shoe Painting Dog Clock **Alt text:** Pilot Study 2 study badge: a round dark blue emblem reading "Understanding Shopping Preferences on amazon". [A round dark navy badge with a light blue drop shadow. Inside, in white capitals arched around a small shopping-cart icon, "UNDERSTANDING", then in large bold letters "SHOPPING PREFERENCES", a small "ON", and the Amazon wordmark with its orange smile arrow at the bottom.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* **[condition: all conditions]** We are conducting a study to understand how people make shopping decisions on Amazon. Your task is to help us learn more about how you shop and make choices among suggested items. **[condition: persuasion knowledge]** You will be reading a scenario involving purchasing a product on Amazon. Please keep this in mind while you complete the entire survey. While reading, think about the **intentions and motives** behind the way each brand presents its information. Some brands lay out details exactly as they are, whereas others emphasize specific aspects and downplay others to shape shoppers' impressions. **[condition: control]** You will be reading a scenario involving purchasing a product on Amazon. Please keep this in mind while you complete the entire survey. While reading the scenario, consider the product information as you normally would when shopping online, for example, the features that matter to you and whether the product seems to meet everyday needs. **Alt text:** Screenshot strip of the Amazon website header with the search box reading "water pitcher with filter for tap water". [A thin screenshot strip of the Amazon.com header: the Amazon logo, "Delivering to Metropolis 62960, Update location", a search box containing "water pitcher with filter for tap water" with the orange magnifier button, "EN", "Hello, sign in, Account & Lists", "Returns & Orders", and a cart showing 0. Below it the dark navigation bar: All, Medical Care, Best Sellers, Amazon Basics, Prime, Today's Deals, New Releases, Music, Customer Service, Amazon Home, Registry, Books, Pharmacy, Gift Cards, Fashion, Smart Home, Sell. The strip sets the scene of an Amazon search; it carries no product information.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* Imagine you are searching on Amazon to buy a new water filter pitcher. You have narrowed down your search to two pitchers from two different manufacturers. Both pitchers are identical in terms of material, price, capacity, and design. Both of these products have a rating of +4.9 (out of 5). However, the two pitchers differ in their advertised filter lifespan: **Brand A** claims that its pitcher's filter will last for 40 gallons. **Brand B** claims that its pitcher's filter will last between 35 to 45 gallons. Carefully look at the products below. **Alt text:** Pilot Study 2 stimulus: two water filter pitchers, Brand A "Filter lasts for 40 gallons" and Brand B "Filter lasts for 35 - 45 gallons". [Two rounded white cards side by side. The left card is headed "Brand A" and shows a white water filter pitcher with a filter cartridge standing beside it, with the red caption "Filter lasts for 40 gallons". The right card is headed "Brand B" and shows a different white pitcher with a cartridge, with the red caption "Filter lasts for 35 - 45 gallons".] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* When you were reading the product information... I found myself concentrating more on... (1 = The substance of the claims being made; 7 = The communication techniques employed) When considering the information, my focus was on... (1 = What was being communicated (the facts/details); 7 = How it was being communicated (the style/delivery)) While reading I focused on... (1 = Substance of the information; 7 = Presentation style) What was the product you saw in this survey? (open ended) What is your gender identity? - Male - Female - Other - Prefer Not to Answer What's your age? (dropdown list) --------------------------------end of survey-------------------------------- **Results** The attention-focus index was reliable (α = .87). Participants in the persuasion knowledge condition reported greater relative focus on how information was communicated (M = 4.28, SD = 1.30) than those in the control condition (M = 2.78, SD = 1.42; F(1, 198) = 60.84, p < .001, ηp² = .235). The persuasion knowledge mean was modestly above the scale midpoint (t(99) = 2.15, p = .034, d = .22), whereas the control mean was well below it (t(99) = -8.60, p < .001, d = -.86), indicating that persuasion knowledge activation increases the relative salience of how information is communicated rather than fully redirecting attention away from the substance of the claim. **Web Appendix H: Examples of Point Versus Range Estimates in the Marketplace** Web Appendix Figure H1. Two BMW iX models, side-by-side, with travel range displayed as point vs. range estimates, on BMW's official website. **Alt text:** Web Appendix Figure H1: BMW's website showing two iX models side by side, the iX xDrive50 with "Range from 305 to 311 miles" and the iX M60 with "Range of 296 miles". [A screenshot of a BMW web page on a black background headed "BMW iX All-Electric SUV" with the giant letters "THE iX". Two SUVs face the viewer: on the left a silver "iX xDrive50" (marked ELECTRIC) with "Starting MSRP $87,250" and "Range from 305 to 311 miles"; on the right a blue-gray "iX M60" (marked M ELECTRIC) with "Starting MSRP $111,500" and "Range of 296 miles". Under each, a "See Full Specifications" link and a white "Start Building" button. The two models sit in the same advertisement, one with a range estimate and one with a point estimate.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* Web Appendix Figure H2. Example of two electric scooters that appear side-by-side when "electric scooter" is searched on Amazon.com. **Alt text:** Web Appendix Figure H2: two Amazon search results for electric scooters, one advertising "17-21 Miles Range" and the other "18Miles Travel Range". [Two Amazon search-result tiles side by side. Left: a gray and black folding scooter with a phone app shown, titled "NAVIC T7 Electric Scooter, 17-21 Miles Range, 19 Mph Folding Commute Electric...", five stars with 22 ratings, "50+ bought in past month", price $259.99 with the Prime badge. Right: a black scooter with a phone app shown, titled "iScooter Electric Scooter, 18Miles Travel Range, 15.6 Mph Top Speed, 350W Scooter Electric...", four and a half stars with 57 ratings, price $249.99 with the Prime badge.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* Web Appendix Figure H3. Example of two portable fire pits that appear side-by-side when "portable campfire" is searched on Amazon.com. **Alt text:** Web Appendix Figure H3: two Amazon search results for portable fire pits, one advertising "3-5 Hours of Burn Time" and the other "3 Hours of Warmth and Burn Time". [Two Amazon search-result tiles side by side. Left, tagged "Popular Brand Pick": a small tin fire pit with a flame and its box, titled "City Bonfires Portable Fire Pit - Mini Fire Pit, Table Top Firepit, No Wood, No Embers, 3-5 Hours of Burn Time - Food Safe for Smores - Lightweight,...", four and a half stars with 1,425 ratings, "50+ bought in past month", price $39.95 (list $41.95), Prime Two-Day free delivery, "Small Business", "More Buying Choices $37.36 (3 used & new offers)". Right, tagged "Previously viewed": a tin fire pit with a tall flame, titled "Radiate Portable Campfire: The Go-Anywhere Outdoor Fire Pit - Portable and Convenient - 3 Hours of Warmth and Burn Time - Great for Camping,...", four and a half stars with 542 ratings, price $18.99 (list $19.99), Prime Two-Day free delivery.] *Note: Stimulus image described by AI (Claude Fable 5.1) from the Web Appendix image. The image itself is not included in this machine-readable version; see the published Web Appendix for the visual.* **Web Appendix References** Campbell, Margaret C. (1995), "When Attention-Getting Advertising Tactics Elicit Consumer Inferences of Manipulative Intent: The Importance of Balancing Benefits and Investments," *Journal of Consumer Psychology*, 4 (3), 225--54. Campbell, Margaret C. and Amna Kirmani (2000), "Consumers' Use of Persuasion Knowledge: The Effects of Accessibility and Cognitive Capacity on Perceptions of an Influence Agent," *Journal of Consumer Research*, 27 (1), 69--83. Castelo, Noah, Maarten W. Bos, and Donald R. Lehmann (2019), "Task-Dependent Algorithm Aversion," *Journal of Marketing Research*, 56 (5), 809--25. Edwards, Jeffrey R. and Lisa Schurer Lambert (2007), "Methods for Integrating Moderation and Mediation: A General Analytical Framework Using Moderated Path Analysis," *Psychological Methods*, 12 (1), 1--22. Hayes, Andrew F. (2022), *Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach*, 3rd ed. Guilford Press. Mookerjee, Siddhanth (Sid), Yann Cornil, and JoAndrea Hoegg (2021), "From Waste to Taste: How 'Ugly' Labels Can Increase Purchase of Unattractive Produce," *Journal of Marketing*, 85 (3), 62--77. Muller, Dominique, Charles M. Judd, and Vincent Y. Yzerbyt (2005), "When Moderation Is Mediated and Mediation Is Moderated," *Journal of Personality and Social Psychology*, 89 (6), 852--63. Pieters, Rik (2017), "Meaningful Mediation Analysis: Plausible Causal Inference and Informative Communication," *Journal of Consumer Research*, 44 (3), 692--716. Spencer, Steven J., Mark P. Zanna, and Geoffrey T. Fong (2005), "Establishing a Causal Chain: Why Experiments Are Often More Effective Than Mediational Analyses in Examining Psychological Processes," *Journal of Personality and Social Psychology*, 89 (6), 845--51.