CoChat
Sign Up
Cited by 34Decision-Making and Behavioral EconomicsInnovative Human-Technology InteractionBehavioral Health and Interventions

On or Off Track: How (Broken) Streaks Affect Consumer Decisions

Open access
Silverman, Jackie, Barasch, Alixandra·2023·Journal of Consumer Research
DOI 10.1093/jcr/ucac029
Open this paper in CoChat — free account

Abstract

New technologies increasingly enable consumers to track their behaviors over time, making them more aware of their “streaks”—behaviors performed consecutively three or more times—than ever before. Our research explores how these logged streaks affect consumers’ decisions to engage in the same behavior subsequently. In seven studies, we find that intact streaks highlighted via behavioral logs increase consumers’ subsequent engagement in that behavior, relative to when broken streaks are highlighted. Importantly, this effect is independent of actual past behavior and depends solely on how that behavior is represented within the log. This is because consumers consider maintaining a logged streak to be a meaningful goal in and of itself. In line with this theory, the effect of intact (vs. broken) logged streaks is amplified when consumers attribute a break in the streak to themselves rather than to external factors, and attenuated when consumers can “repair” a broken streak. Our research provides actionable insights for companies seeking to benefit from highlighting consumers’ streaks in various consequential domains (e.g., fitness, learning) without incurring a cost (e.g., reduced engagement or abandonment) when those streaks are broken.

AI snapshot

Written from the paper’s full text.

TL;DR

Across seven studies, consumers are more likely to continue a behavior when a behavioral log shows an intact streak than when it shows a broken streak, even when actual past behavior is held constant. The effect is driven by treating streak maintenance as a goal in itself, amplified when consumers attribute the break to themselves and attenuated when they can repair the broken streak.

Key findings

  • In a field study of 980 employees in a 7,000-step fitness challenge, intact vs. broken logged streaks predicted meeting the challenge (b = 0.38, SE = .03, Z = 11.27, p < .001); separate models showed intact b = 0.25 and broken b = -1.01.
  • In experiments, intact logged streaks increased target behavior vs. broken logged streaks: Study 2, 66.23% vs. 57.86% (OR = 1.43); Study 3, 78.33% vs. 53.31%, with highlighting intact streaks 92.47% vs. no log 64.94% and broken streaks 45.21% vs. no log 60.90%; Study 4, 65.84% vs. 47.98% (no log 54.73%); Study 5, 82.55% vs. 67.92% (no log 77.78%); Study 6, intact 53.17%, externally attributed broken 42.00%, self-attributed broken 28.87%; Study 7, intact 93.14%, broken 68.66%, repairable broken 85.20%.
  • Mechanism: streak-maintenance goal mediated the effect in Study 4 (indirect effect = 0.20, 95% CI [0.05, 0.37]); sense of accomplishment mediated effects in Studies 4, 5, and 7, while negative emotion did not mediate in Studies 5 and 7.
  • Moderation: self-attributed breaks amplified the effect (Study 6: intact vs. self-attributed broken OR = 2.80; self- vs. externally attributed broken OR = 0.56), and repair opportunities attenuated it (Study 7). Pilot data: 59% reported going out of their way to maintain a streak, and 27% engaged in offline behavior to maintain one.

Methods

  • Pilot descriptive survey (N = 100 MTurk) plus seven main studies. Study 1 was correlational field data from a university 30-day fitness challenge: 980 employees, daily step counts, logistic models clustered by individual and controlling for overall rate of meeting the challenge.
  • Studies 2–7 were MTurk between-subjects experiments using app-like behavioral logs; they manipulated intact vs. broken logged streaks, sometimes with no-log controls, categorization of streak-contributing behaviors, attribution of the break, or repair opportunity. Dependent variables were real choices (e.g., continue exercise/language/game vs. switch or stop). Target samples were 400–600; Study 5 had 805 recruited and 452 final after 353 stopped before manipulation; Study 6 had 802 recruited and 418 final. Studies 4–7 were preregistered.
  • Analyses used chi-square tests, binary logit models, ANOVA, and mediation with bootstrap procedures (10,000 samples). Data, materials, and preregistrations are in the OSF repository.

Limitations

  • Stated: Study 1 is correlational and allows self-selection, limiting causal isolation. The no-log contrasts in Studies 3–5 were not consistently significant and the studies were not powered specifically for those individual contrasts. Study 6’s final sample was lower than targeted due to lower completion, though it had over 130 per condition and full-sample analyses replicated effects. Study 7’s repair opportunity did not fully eliminate the effect, possibly because repaired streaks seem less valuable or authentic. Future work notes streak lengths beyond 20 were not tested, independent intact/broken effects were inconsistent, and log features such as effort and public/private visibility need more study.
  • Evident: Most causal evidence comes from MTurk lab/app paradigms rather than field settings, which may limit generalizability. Study 5 had substantial pre-manipulation attrition (353 of 805), though attrition did not differ by condition.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1093/jcr/ucac029

Cites

76 of 76 references verified0 citation errors · 0 not checkable

Statistics

31 tests recomputed · 0 with the reported and recomputed p on opposite sides of .05

Claims the paper makes

8 of 49 cited sources checked · 5 supported · 0 flagged · accuracy not yet measured

  • Premisenone given
    § INTRODUCTION · p. 6

    Prior research does not consider how highlighting specific patterns of goal-consistent behavior can become motivating in itself and affect consumers' subsequent decisions.

    “Although prior research has explored how progress towards an end goal (or lack thereof) affects subsequent goal-consistent behavior (Cochran and Tesser 1996; Kivetz, Urminsky and Zheng 2006; Soman and Cheema 2004), it does not consider how highlighting specific patterns of goal-consistent behavior can become motivating in itself and affect consumers' subsequent decisions.”
    Quote in the paperSubstantial claim1 of 3 sources checked
    • cites Cochran and Tesser 1996

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites Kivetz, Urminsky and Zheng 2006 source
    • cites Soman and Cheema 2004 source
  • Premisecitation
    § INTRODUCTION · p. 4

    No prior research has explored how highlighting specific patterns of past behavior through tracking technology affects subsequent behavior.

    “While recent research has examined the general impact of behavioral tracking on different dimensions of consumer wellbeing (e.g., Etkin 2016; Karapanos et al. 2016), none has explored how highlighting specific patterns of past behavior through such technology actually affects subsequent behavior.”
    Quote in the paperSubstantial claim1 of 2 sources checked
    • cites Etkin 2016 source
    • cites Karapanos et al. 2016 source
  • Interpretationown data
    § Future Directions · p. 50

    The activation of a streak goal (from an intact logged streak) may be somewhat categorical and insensitive to streak length.

    “These results are consistent with our theory, suggesting that the activation of a streak goal (from an intact logged streak) may be somewhat categorical and insensitive to streak length (cf. Sheeran, Webb and Gollwitzer 2005).”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Sheeran, Webb and Gollwitzer 2005 source

      In the General Discussion (page unavailable), Study 2 reports that implementation intentions affected performance when the relevant goal was activated, but not when it was not activated. The paper does not study logged streaks or test whether goal activation is insensitive to streak length. It also reports that the strength of a goal intention mattered for performance, so it does not establish the proposed insensitivity. (LLM, conf=0.83)

      Experimental — accuracy not yet measured
  • Interpretationcitation
    § Theoretical Contributions · p. 48

    The symbols used in behavioral tracking (e.g., checkmarks in behavioral logs) can have a special influence on consumers' behavior.

    “Furthermore, our findings inform our understanding of how the symbols used in behavioral tracking (e.g., the checkmarks in behavioral logs) can have a special influence on consumers' behavior.”
    Quote in the paperSubstantial claim1 of 2 sources checked
    • cites Hamari et al. 2014

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites Hsee et al. 2003 source
  • Findingreasoning
    § Abstract · p. 3

    The effect of intact (vs. broken) logged streaks is amplified when consumers attribute a break in the streak to themselves rather than to external factors.

    “the effect of intact (vs. broken) logged streaks is amplified when consumers attribute a break in the streak to themselves rather than to external factors”
    Quote in the paperSubstantial claim
  • May combine two claims. 35
  • Classified as probably not a substantial claim. 38
  • The quote only partly states this claim. 1

  • May combine two claims.

    Premisecitation
    § The Current Research · p. 8

    People value streaks and seek to preserve them.

    “Findings from several literatures support the notion that people value streaks and seek to preserve them.”
    Quote in the paperSubstantial claim1 of 3 sources checked
    • cites Singer 1966 source
    • cites Spence 1956 source

      The indexed copy’s Abstract (page unavailable) discusses stimulus control and reinforced choices, not whether people value streaks or try to preserve them. Searches for streaks and for maintaining a sequence of reinforced responses found no passage addressing that claim. The indexed passages also appear under a title different from the cited paper, so they cannot verify what Spence’s work says. (LLM, conf=0.75)

      Experimental — accuracy not yet measured
    • cites Markus and Wurf 1987 source

  • May combine two claims.

    Findingown data
    § Discussion · p. 38

    The no log condition always fell between the two logged streak conditions for all measures, and there were no differences in the estimated effect sizes of these contrasts.

    “That said, although the significance levels of these contrasts differed for these two studies, the no log condition always fell between the two logged streak conditions for all measures (as predicted), and there were no differences in the estimated effect sizes of these contrasts (cf. Gelman and Stern 2006).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Gelman and Stern 2006 source

  • Classified as probably not a substantial claim.

    Premisecitation
    § Theoretical Contributions · p. 48

    Consumers continue certain behaviors after several past instances.

    “Additionally, we contribute to past work showing that consumers continue certain behaviors after several past instances (e.g., collecting two or three items: Gao et al. 2014; acquiring a medium like loyalty points: Hsee et al. 2003)”
    Quote in the paperSubstantial claim1 of 2 sources checked
    • cites Gao et al. 2014 source
    • cites Hsee et al. 2003 source

      Study 5 (Results and Discussion; page unavailable) found that participants continued an activity longer when they earned points as a medium. The activity was listening to noise, however, not repeating a behavior after several past instances. The Implications section discusses loyalty-point programs as an application, not as a tested setting. (LLM, conf=0.86)

      Experimental — accuracy not yet measured

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § Discussion · p. 43

    Consumers often incur effort and costs to avoid watching advertisements.

    “This result is notable given the effort and costs consumers often incur to avoid watching advertisements (e.g., Anderson and Gans 2011).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Anderson and Gans 2011 source

  • May combine two claims.

    Findingown data
    § CONSUMER LOGGING PILOT STUDY · p. 14

    The largest portion of participants (48%) named gaming apps, followed by social media/messaging apps (40%), health and fitness apps (39%), and language-learning apps (9%); 13% named other types of apps.

    “The largest portion of participants (48%) named gaming apps, followed by social media/messaging apps (40%), health and fitness apps (39%), and language-learning apps (9%); 13% named other types of apps”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § INTRODUCTION · p. 6

    Logged streaks affect consumer decisions not only in prototypical goal-oriented domains (e.g., exercising, learning a language) but also for activities that do not necessarily contribute directly to a quantitative, specific end goal (e.g., playing games).

    “Indeed, we find that logged streaks not only affect consumer decisions in prototypical goal-oriented domains (e.g., exercising, learning a language) but also for activities that do not necessarily contribute directly to a quantitative, specific end goal (e.g., playing games).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data
    § STUDY 1: THE EFFECT OF STREAKS ON STEP COUNTS IN THE FIELD · p. 15

    Participants did not report feeling overwhelmed, bothered, or distracted by such information, or that it took away from their experience in using the app (Ms < 2.60, ts > 7.75, ps < .001, ds > 0.75).

    “Participants did not report feeling overwhelmed, bothered, or distracted by such information, or that it took away from their experience in using the app ( M s < 2.60, t s > 7.75, p s < .001, d s > 0.75).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § CONSUMER LOGGING PILOT STUDY · p. 15

    Participants reported that seeing information about their streaks increased their awareness of their streaks, enhanced their overall experience, and motivated them to keep using the app (Ms > 4.90, ts > 5.00, ps < .001, ds > 0.50).

    “and that seeing information about their streaks increased their awareness of their streaks, enhanced their overall experience, and motivated them to keep using the app ( M s > 4.90, t s > 5.00, p s < .001, d s > 0.50)”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingreasoning
    § Abstract · p. 3

    The effect of intact vs broken logged streaks on subsequent engagement is independent of actual past behavior and depends solely on how that behavior is represented within the log.

    “Importantly, this effect is independent of actual past behavior, and depends solely on how that behavior is represented within the log.”
    • Also stated in: § INTRODUCTION · p. 5 · plain
      “These effects hold even when consumers' past behavior is held constant, and the only difference is whether that behavior and its contribution to a (broken) streak is displayed within a behavioral log.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data
    § CONSUMER LOGGING PILOT STUDY · p. 14

    Participants named 101 unique apps that highlight streaks, with total listed N=202, per person M=2.02, SD=1.52.

    “Together, they named 101 unique apps (total listed: N = 202; per person: M = 2.02, SD = 1.52)”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § Results · p. 31

    In Study 4, participants in the intact logged streak condition felt a directionally greater sense of accomplishment than the no log condition (M = 7.24, SD = 2.39; t(401) = 0.99, p = .32; d = 0.10).

    “participants in the intact logged streak condition felt a directionally greater sense of accomplishment relative to this control ( M = 7.24, SD = 2.39; t (401) = 0.99, p = .32; d = 0.10)”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § Future Directions · p. 50

    Neither Study S3a nor Study S3b revealed a significant interaction between the streak and length conditions (ps > .40).

    “and neither study revealed a significant interaction between the streak and length conditions ( p s &gt; .40).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data
    § Results · p. 36

    Participants in the broken logged streak condition felt a significantly lower sense of accomplishment and significantly greater negative emotion than participants in the no log condition (t > 2.80, p < .006, d > 0.32).

    “Additionally, the no log condition fell between these two conditions for both measures ( Maccomplishment = 8.18, SD = 2.10; Memotion = 3.25, SD = 3.22), with participants in the intact logged streak condition feeling a directionally greater sense of accomplishment and directionally lower negative emotion ( t s &gt; 1.25, p s &lt; .22, d s &gt; 0.14), and participants in the broken logged streak”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingown data
    § Future Directions · p. 51

    Studies 3, 4, and 5 found some evidence for the positive effect of an intact logged streak and the negative effect of a broken logged streak, relative to the absence of a behavioral log, but the statistical significance of these contrasts was not consistent across studies.

    “In particular, Studies 3, 4, and 5 found some evidence for the positive effect of an intact logged streak and the negative effect of a broken logged streak, relative to the absence of a behavioral log, but the statistical significance of these contrasts was not consistent across studies.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § Discussion · p. 38

    Relative to the absence of a behavioral log, highlighting an intact logged streak had a stronger positive effect in Study 4, while highlighting a broken logged streak had a stronger negative effect in Study 5.

    “Specifically, relative to the absence of a behavioral log, highlighting an intact logged streak had a stronger positive effect in Study 4, while highlighting a broken logged streak had a stronger negative effect in Study 5.”
    Quote in the paperNot a substantial claim

  • The quote only partly states this claim.

    Findingnone given
    § Theoretical Contributions · p. 48

    Highlighting a streak (making it salient through behavioral tracking) affects consumers' subsequent behavior.

    “We build on this research by examining the effect of highlighting a specific, commonly occurring pattern of behavior - a streak - on consumers' subsequent decisions.”
    • Also stated in: § Theoretical Contributions · p. 48 · boosted
      “by demonstrating the effects of highlighting multiple consecutive past behaviors on subsequent behavior.”
    • Also stated in: § Theoretical Contributions · p. 48 · plain
      “we investigate how streaks made salient through such tracking influence subsequent behavior.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § Future Directions · p. 53

    The effect of intact versus broken streaks was directionally stronger when the log would be made public (OR=1.72) versus kept private (OR=1.52).

    “the effect of intact versus broken streaks was directionally stronger when the log would be made public ( OR = 1.72) versus kept private ( OR = 1.52).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § Results · p. 36

    Participants in the intact logged streak condition felt a directionally greater sense of accomplishment and directionally lower negative emotion than participants in the no log condition (t > 1.25, p < .22, d > 0.14).

    “Additionally, the no log condition fell between these two conditions for both measures ( Maccomplishment = 8.18, SD = 2.10; Memotion = 3.25, SD = 3.22), with participants in the intact logged streak condition feeling a directionally greater sense of accomplishment and directionally lower negative emotion ( t s &gt; 1.25, p s &lt; .22, d s &gt; 0.14), and participants in the broken logged streak”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § GENERAL DISCUSSION · p. 47

    All of these studies showed effects on real consumer choices, both in the field (Study 1) and controlled lab paradigms (Studies 2-7), as well as for a variety of consumer-relevant, consequential domains, including exercise (Studies 1 and 2), language learning (Studies 3 and 4), and games (Studies 5, 6, and 7).

    “Notably, all of these studies showed effects on real consumer choices, both in the field (Study 1) and controlled lab paradigms (Studies 2-7), as well as for a variety of consumer-relevant, consequential domains, including exercise (Studies 1 and 2), language learning (Studies 3 and 4), and games (Studies 5, 6, and 7).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data
    § Discussion · p. 46

    Study 7 replicated the effect of intact versus broken logged streaks on subsequent behavior, and also demonstrated that the opportunity for streak repair attenuated this effect.

    “Study 7 replicated the effect of intact versus broken logged streaks on subsequent behavior, and also demonstrated that the opportunity for streak repair attenuated this effect.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data
    § Future Directions · p. 50

    Both Study S3a and Study S3b replicated the effects of intact versus broken logged streaks on subsequent behavior, regardless of streak length (ps < .001; Study S3a ORs > 2.40; Study S3b ds > 0.45).

    “Interestingly, both studies replicated the effects of intact versus broken logged streaks on subsequent behavior, regardless of streak length ( p s &lt; .001; Study S3a ORs &gt; 2.40; Study S3b d s &gt; 0.45)”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Premisecitation
    § Streaks · p. 7

    People overinterpret noise within small samples, seeing random streaks of past events as meaningful patterns that shape their subsequent predictions.

    “This past work suggests that people overinterpret noise within small samples, seeing random streaks of past events (e.g., getting heads for three consecutive coin flips) as meaningful patterns that shape their subsequent predictions.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Gilovich, Vallone and Tversky 1985 source

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Study design and conductnone given
    § Study Overview · p. 14

    We report all measures assessed, and no conditions or participants were dropped from any of the analyses.

    “We report all measures assessed, and no conditions or participants were dropped from any of the analyses.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning
    § Logged Streaks as Goals · p. 9

    An intact streak should increase consumers' sense of accomplishment and motivate them to continue engaging in that behavior.

    “Thus, if consumers view maintaining a streak of logged behavior as a goal in and of itself, an intact streak should increase their sense of accomplishment and motivate them to continue engaging in that behavior.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Study design and conductreasoning
    § Study Overview · p. 14

    All experimental studies involved sample sizes between 400 and 600 participants, determined in advance to provide at least 80% power to detect the focal effect in each study, based on preliminary effect size estimates from pilot studies.

    “All experimental studies involved sample sizes between 400 and 600 participants. These were determined in advance to provide at least 80% power to detect the focal effect in each study, based on preliminary effect size estimates from pilot studies.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Interpretationreasoning
    § Logged Streaks as Goals · p. 10

    Subsequent decisions will be influenced by logged streaks themselves, even when holding actual behavior constant.

    “Notably, our theory predicts that subsequent decisions will be influenced by logged streaks themselves , even when holding actual behavior constant.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation
    § Logged Streaks as Goals · p. 9

    Encountering failure during goal pursuit triggers feelings of discouragement and undermines belief in one's ability to ultimately reach that goal.

    “Encountering failure during goal pursuit triggers feelings of discouragement (Heath, Larrick and Wu 1999) and undermines belief in one's ability to ultimately reach that goal (Bandura and Locke 2003; Latham and Locke 2006).”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites Heath, Larrick and Wu 1999 source
    • cites Bandura and Locke 2003 source
    • cites Latham and Locke 2006 source

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given
    § Study Overview · p. 12

    All studies test the effects of logged streaks on real (i.e., not hypothetical) consumer decisions with consequences for participants (e.g., how they spend their time).

    “All studies test the effects of logged streaks on real (i.e., not hypothetical) consumer decisions with consequences for participants (e.g., how they spend their time).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given
    § Study Overview · p. 14

    Studies 4, 5, 6, and 7 were preregistered.

    “Studies 4, 5, 6, and 7 were preregistered.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductreasoning
    § Methods · p. 34

    All participants completed the exact same series of games, but the fourth game either maintained or broke their logged streak depending on condition.

    “Thus, all participants completed the exact same series of games (e.g., three Word Jumbles and then one Number Sum), but the fourth game either maintained or broke their logged streak, depending on condition.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Premisecitation
    § STUDY 2: THE EFFECT OF LOGGED STREAKS ON EXERCISE · p. 19

    The exercises were chosen because they are meant to be performed at a desk or on a couch, making them accessible to many adults who often face difficulties in finding time or energy to exercise.

    “We chose these exercises because they are meant to be performed at a desk or on a couch, making them accessible to many adults who often face difficulties in finding time or energy to exercise (Schmall 2019).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Schmall 2019

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductown data
    § Methods · p. 28

    A post-test confirmed that participants preferred watching fun videos over completing language-learning questions (M = 6.40, SD = 3.32, vs. scale midpoint: t(600) = 2.95, p = .003, d = 0.12).

    “In a post-test at the end of the survey, we confirmed that participants indeed preferred watching fun videos over completing language-learning questions (when asked which they liked more, from 1 'Definitely language-learning' to 11 'Definitely fun videos'; M = 6.40, SD = 3.32, vs. scale midpoint: t(600) = 2.95, p = .003, d = 0.12).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Study design and conductreasoning
    § Discussion · p. 25

    Because participants in both broken streak conditions in Study 3 received the same quota message, any negative inferences about the app are constant across conditions and cannot explain the negative effect of highlighting a broken streak within a behavioral log.

    “Because participants in both broken streak conditions received the same quota message in this study, any such inferences are constant across conditions and cannot explain the negative effect of simply highlighting that broken streak within a behavioral log.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductreasoning
    § Discussion · p. 18

    Allowing for self-selection in Study 1 enhances ecological validity and realism but limits the ability to isolate the causal effects of users' logged streaks on subsequent decisions.

    “However, while allowing for self-selection enhances this study's ecological validity and realism, it limits our ability to isolate the causal effects of users' logged streaks themselves on subsequent decisions.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Study design and conductown data
    § Methods · p. 23

    In Study 3, participants were randomly assigned to condition in a 2 (streak: intact or broken) by 2 (behavioral log: absent or present) between-subjects design.

    “Participants were then randomly assigned to condition in a 2 ( streak: intact or broken ) by 2 ( behavioral log: absent or present ) between-subjects design.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Study design and conductown data
    § Results · p. 17

    The models included a variable for each individual's overall rate of meeting the challenge to control for the possibility that users with a higher rate of meeting the challenge may also have been more likely to exhibit streaks.

    “The models also included a variable for each individual's overall rate of meeting the challenge (i.e., a value from 0 to 1 representing the percentage of days the individual met the challenge), allowing us to control for the possibility that users who had a higher rate of meeting the challenge may also have been more likely to exhibit streaks.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data
    § Discussion · p. 25

    The effect was documented in a language-learning app, providing initial support for the generalizability and persistence of these effects.

    “We documented this effect in the context of a language-learning app, where consumers are accustomed to seeing their behavioral streaks visualized in behavioral logs, providing initial support for the generalizability and persistence of these effects.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § Streaks · p. 8

    The market is flooded with apps to help users quantify their behaviors and highlight their streaks.

    “The market is flooded with apps to help users quantify their behaviors and highlight their streaks (Nield 2019)”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Nield 2019

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductcitation
    § Results · p. 35

    This left a final sample of 452 participants, as targeted.

    “This left a final sample of 452 participants, as targeted (see preregistration).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites preregistration

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § INTRODUCTION · p. 4

    Consumers seek technologies that allow them to track and log their behavior across various domains, from health to finance.

    “Consumers, too, seek technologies that allow them to track and log their behavior across various domains, from health (Fox 2013) to finance (Malcolm 2015).”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites Fox 2013

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites Malcolm 2015

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductown data
    § Methods · p. 20

    In Study 2, the majority of participants (76.71%) reported completing all four exercises, while only 1.83% said they did none.

    “Participants also reported how many exercises they actually did during the study; the majority of participants (76.71%) reported completing all four exercises, while only 1.83% said that they did none of them.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Premisecitation
    § Methods · p. 28

    Successful goal pursuit typically leads to a greater sense of achievement and increased motivation to engage in goal-consistent behaviors, whereas goal failure precipitates discouragement and demotivation.

    “Second, they completed several measures capturing additional indicators of goal progress drawn from prior literature showing that successful goal pursuit typically leads to a greater sense of achievement and increased motivation to engage in goal-consistent behaviors (Bandura 1977; Hollenbeck et al. 1989; Scott and Nowlis 2013), whereas goal failure precipitates discouragement and demotivation (B”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites Bandura 1977 source
    • cites Hollenbeck et al. 1989 source
    • cites Scott and Nowlis 2013 source

  • Classified as probably not a substantial claim.

    Study design and conductown data
    § Results · p. 17

    The models clustered standard errors by individual to account for non-independence of observations due to the repeated measures nature of the data.

    “Due to the repeated measures nature of our data, the models clustered standard errors by individual, thus accounting for non-independence of observations.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § STREAK AFFECTS BEHAVIOR · p. 33

    Over 60% of Americans play games on their smartphones.

    “with over 60% of Americans playing games on their smartphones (Lynkova 2020).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Lynkova 2020

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § Streaks · p. 8

    The proliferation of tracking technology has been widely discussed in the popular press.

    “and the proliferation of tracking technology has been widely discussed in the popular press (Fox 2013; Austin 2019).”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites Fox 2013

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites Austin 2019

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductown data
    § Results · p. 35

    Three hundred fifty-three participants chose to stop playing games before the manipulation and thus could not be included in the final sample or analyses.

    “Three hundred fifty-three participants chose to stop playing games before our manipulation and thus could not be included in our final sample or analyses.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § STUDY 3: EXAMINING THE EFFECTS OF LOGGED STREAKS IN BOTH DIRECTIONS · p. 22

    Language-learning apps commonly log consumers' patterns of behavior.

    “Like exercise apps, language-learning apps have grown increasingly popular (the most widely used app, Duolingo, has over 500 million registered users; de León 2020) and commonly log consumers' patterns of behavior (see Consumer Logging Pilot Study).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Consumer Logging Pilot Study

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Classified as probably not a substantial claim.

    Study design and conductown data
    § Methods · p. 27

    In Study 4, participants were randomly assigned to one of three between-subjects conditions.

    “Participants were randomly assigned to one of three between-subjects conditions.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Study design and conductown data
    § Methods · p. 28

    As preregistered, all ten items were combined into one measure of sense of accomplishment, and results were similar when the sub-scales were analyzed separately.

    “As preregistered, we combined all ten items into one measure of sense of accomplishment , but the results were similar when these sub-scales were analyzed separately (see Web Appendix for analyses with these sub-scales for all relevant studies, including serial mediation).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Study design and conductreasoning
    § Methods · p. 20

    The Study 2 design controlled for participants' actual behavior and manipulated only what participants viewed in their behavioral log.

    “This design thus controlled for participants' actual behavior (i.e., everyone completed the same four exercises), and manipulated only what participants viewed in their behavioral log (i.e., whether the logged streak representing that series of behaviors was intact or broken; see Figure 1).”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Study design and conductown data
    § Methods · p. 19

    In Study 2, participants were randomly assigned to one of two between-subjects conditions.

    “Participants were randomly assigned to one of two between-subjects conditions.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductown data
    § Results · p. 17

    Robustness check models with additional controls (e.g., day effects, state dependence) and using various streak lengths replicate the effects.

    “Our Web Appendix includes several robustness check models with additional controls (e.g., day effects, state dependence) and using various streak lengths, all of which replicate these effects.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Study design and conductreasoning
    § STREAK AFFECTS BEHAVIOR · p. 33

    The design held participants' actual behavior constant across conditions and varied only how it was logged, thereby enabling a clean test of how logged streaks affect subsequent behavior.

    “As in Studies 2 and 4, this design held participants' actual behavior constant across conditions and varied only how it was logged, thereby enabling a clean test of how logged streaks affect subsequent behavior.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation
    § Logged Streaks as Goals · p. 9

    Failure is demotivating and reduces the likelihood of future goal-consistent behaviors.

    “Consequently, failure is demotivating and reduces the likelihood of future goal-consistent behaviors (e.g., Soman and Cheema 2004).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Soman and Cheema 2004 source

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation
    § STUDY 3: EXAMINING THE EFFECTS OF LOGGED STREAKS IN BOTH DIRECTIONS · p. 22

    Language-learning apps have grown increasingly popular, with the most widely used app, Duolingo, having over 500 million registered users.

    “Like exercise apps, language-learning apps have grown increasingly popular (the most widely used app, Duolingo, has over 500 million registered users; de León 2020) and commonly log consumers' patterns of behavior (see Consumer Logging Pilot Study).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites de León 2020

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S4 (based).

    Interpretationreasoning
    § Practical Implications · p. 50

    Helping consumers stay (or get back) on track by highlighting streaks should increase their persistence towards personal goals.

    “Helping consumers stay (or get back) on track by highlighting streaks should increase their persistence towards personal goals”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: Companies have almost total discretion over whether, when, and how to employ tracking technologies that log and display information about those behaviors to consumers.

  • Classified as probably not a substantial claim.

    Study design and conductreasoning
    § Results · p. 40

    This left a final sample of 418 participants.

    “This left a final sample of 418 participants.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationreasoning
    § Theoretical Contributions · p. 47

    The goal identified by a logged streak is defined by an ongoing process without a clear end state, since streaks can only be maintained, not completed.

    “our research identifies a goal that is defined by an ongoing process without a clear end state (since streaks can only be maintained, not 'completed').”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Study design and conductown data
    § Results · p. 40

    Given that the final sample contained over 130 per condition, the authors felt it was adequately powered.

    “Given that the final sample contained over 130 per condition, we felt it was adequately powered and therefore stopped data collection (see Web Appendix for analyses with the full sample, which replicated the effects).”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Interpretationreasoning
    § Practical Implications · p. 49

    Highlighting intact streaks via behavioral logs can encourage consumers to maintain them.

    “While highlighting intact streaks via behavioral logs can encourage consumers to maintain them, companies might want to avoid highlighting consumers' broken streaks”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Interpretationreasoning
    § Practical Implications · p. 50

    Offering behavioral logs that highlight streaks may improve consumer welfare.

    “Beyond the company benefits of increased consumer engagement, offering behavioral logs that highlight streaks may improve consumer welfare as well.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: Companies have almost total discretion over whether, when, and how to employ tracking technologies that log and display information about those behaviors to consumers.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductreasoning
    § Discussion · p. 38

    Because these studies were not powered specifically to estimate these individual contrasts, these comparisons yielded mixed evidence.

    “However, because these studies were not powered specifically to estimate these individual contrasts (as per our preregistration), these comparisons yielded mixed evidence.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites preregistration

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • May combine two claims.

    Premisecitation
    § STUDY 6: THE EFFECT OF LOGGED STREAKS IS AMPLIFIED WHEN CONSUMERS ATTRIBUTE RESPONSIBILITY FOR THE BREAK TO THEMSELVES · p. 38

    Attributing goal failure to one's own actions reduces self-efficacy and future goal pursuit.

    “Past research suggests that attributing goal failure to one's own actions reduces self-efficacy and future goal pursuit (e.g., Bandura and Locke 2003).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Bandura and Locke 2003 source

  • May combine two claims.

    Study design and conductown data
    § Results · p. 40

    The final sample included all participants who correctly completed the first three games, which did not differ by condition (Χ2(2) = 3.52, p = .172).

    “As preregistered, our final sample included all participants who correctly completed the first three games (which did not differ by condition: Χ 2 (2) = 3.52, p = .172; see Web Appendix for details).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites preregistration

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § Theoretical Contributions · p. 48

    Technology affects consumers' lives.

    “Finally, our work contributes to growing research on how technology affects consumers' lives (e.g., Deighton et al. 2017; Schmitt 2019).”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites Deighton et al. 2017 source
    • cites Schmitt 2019 source
65 of 65

  • The paper relies on this without support.

    Companies use behavioral logs to highlight consumers' recent behavior within their platform.

    “One increasingly common strategy is to use behavioral logs to highlight consumers' recent behavior within the company's platform, capturing when a consumer has repeated the same desired behavior for multiple time periods.”

  • The paper relies on this without support.

    Companies highlight streaks via behavioral logs to motivate consumers to continue interacting with the product and ensure the desired behavior persists.

    “Companies presumably highlight streaks via behavioral logs to motivate consumers to continue interacting with the product and ensure the desired behavior persists.”

  • The paper relies on this without support.

    Consumers will inevitably experience interruptions to their streaks, despite their best intentions.

    “While displaying streaks may keep consumers engaged when they are on track, consumers will inevitably experience interruptions to their streaks, despite their best intentions.”

  • The paper relies on this without support.

    Almost everyone exhibits streaks of behavior at some point, whether purposefully or inadvertently.

    “This is surprising given the ubiquity of streaks in consumers' lives - almost everyone exhibits streaks of behavior at some point, whether purposefully or inadvertently.”

  • The paper relies on this without support.

    The effectiveness of highlighting streaks via behavioral logs (and when it might backfire) has not been investigated.

    “Yet the effectiveness of this strategy (and when it might backfire) has not been investigated.”

  • The paper relies on this without support.

    The question of how consumers' own streaks of freely chosen behaviors affect their subsequent decisions remains unexplored, as are the psychological mechanisms involved.

    “However, the question of how consumers' own streaks of freely chosen behaviors affect their subsequent decisions about whether to keep engaging in those behaviors remains unexplored, as are the psychological mechanisms involved.”

  • The paper relies on this without support.

    Many apps commonly used by consumers highlight users' current streak via an in-app behavioral log only visible to the user.

    “Like many other apps commonly used by consumers, it also highlighted users' current streak of 'meeting the challenge' via an in-app behavioral log only visible to the user which informed them of how many days in a row they had walked at least 7,000 steps (see Web Appendix for a sample screenshot).”

  • The paper relies on this without support.

    Consumers frequently encounter situations where they feel personally responsible for their broken streaks.

    “But of course, consumers frequently encounter situations where they feel personally responsible for their broken streaks, such as when they forget to log or run out of time before completing the relevant behavior.”

  • The paper relies on this without support.

    Companies have almost total discretion over whether, when, and how to employ tracking technologies that log and display information about those behaviors to consumers.

    “While firms have no direct control over the sequences of behavior consumers choose to engage in, they have almost total discretion over whether, when, and how to employ tracking technologies that log and display information about those behaviors to consumers.”
9 of 9

Statistics that don’t add up

Each reported p-value recomputed from its test statistic and degrees of freedom.

  • Inconsistent p-value: t (398) = 0.59, p = .256

    Minor

    The paper reports p = .256 here; recomputing from the test statistic and its degrees of freedom gives p = .556 (two-tailed). The check cannot tell whether the statistic or the p-value is the misprint.

    Reported p = .256 · recomputed p = .556

1 of 1

Statistics recomputed

Every reported test result, recomputed from its test statistic and degrees of freedom. A result shows as not adding up only once an independent check has confirmed it.

  • Doesn’t add upt(398) = 0.59reported p = .256 · recomputed p = .556

    “t (398) = 0.59, p = .256”

  • Consistentt(99) = 5.91reported p < .001 · recomputed p < .001

    “t (99) = 5.91, p < .001”

  • Consistentz = 11.27reported p < .001 · recomputed p < .001

    “Z = 11.27, p < .001”

  • Consistentz = 5.06reported p < .001 · recomputed p < .001

    “Z = 5.06, p < .001”

  • Consistentz = -14.94reported p < .001 · recomputed p < .001

    “Z = -14.94, p < .001”

  • Consistentχ²(1) = 4.47reported p = .035 · recomputed p = .034

    “Χ 2 (1) = 4.47, p = .035”

  • ConsistentF(1, 598) = 2.66reported p = .104 · recomputed p = .103

    “F (1, 598) = 2.66, p = .104”

  • Consistentχ²(1) = 7.46reported p = .006 · recomputed p = .006

    “Χ 2 (1) = 7.46, p = .006”

  • Consistentχ²(1) = 5.93reported p = .015 · recomputed p = .015

    “Χ 2 (1) = 5.93, p = .015”

  • Consistentt(600) = 2.95reported p = .003 · recomputed p = .003

    “t(600) = 2.95, p = .003”

  • ConsistentF(2, 598) = 13.08reported p = .001 · recomputed p < .001

    “F (2, 598) = 13.08, p = .001”

    The paper prints p at a floor (such as p < .001)

  • Consistentχ²(1) = 5.2reported p = .023 · recomputed p = .023

    “Χ 2 (1) = 5.20, p = .023”

  • Consistentχ²(1) = 1.82reported p = .178 · recomputed p = .177

    “Χ 2 (1) = 1.82, p = .178”

  • ConsistentF(2, 598) = 3.35reported p = .036 · recomputed p = .036

    “F (2, 598) = 3.35, p = .036”

  • Consistentt(398) = 2.52reported p = .012 · recomputed p = .012

    “t (398) = 2.52, p = .012”

  • Consistentt(401) = 1.6reported p = .111 · recomputed p = .110

    “t (401) = 1.60, p = .111”

  • Consistentt(397) = 0.98reported p = .33 · recomputed p = .328

    “t (397) = 0.98, p = .33”

  • Consistentt(401) = 0.99reported p = .32 · recomputed p = .323

    “t (401) = 0.99, p = .32”

  • Consistentt(397) = 2.93reported p = .004 · recomputed p = .004

    “t (397) = 2.93, p = .004”

  • Consistentχ²(2) = 1.97reported p = .37 · recomputed p = .373

    “Χ 2 (2) = 1.97, p = .37”

  • Consistentt(306) = 0.71reported p = .48 · recomputed p = .478

    “t (306) = 0.71, p = .48”

  • ConsistentF(2, 449) = 4.61reported p = .01 · recomputed p = .010

    “F (2, 449) = 4.61, p = .010”

  • Consistentχ²(1) = 8.78reported p = .003 · recomputed p = .003

    “Χ 2 (1) = 8.78, p = .003”

  • Consistentχ²(1) = 1.05reported p = .31 · recomputed p = .306

    “Χ 2 (1) = 1.05, p = .31”

  • Consistentχ²(1) = 3.69reported p = .055 · recomputed p = .055

    “Χ 2 (1) = 3.69, p = .055”

  • Consistentχ²(2) = 3.52reported p = .172 · recomputed p = .172

    “Χ 2 (2) = 3.52, p = .172”

  • Consistentχ²(1) = 3.43reported p = .064 · recomputed p = .064

    “Χ 2 (1) = 3.43, p = .064”

  • Consistentχ²(1) = 5.48reported p = .019 · recomputed p = .019

    “Χ 2 (1) = 5.48, p = .019”

  • Consistentχ²(1) = 6.56reported p = .01 · recomputed p = .010

    “Χ 2 (1) = 6.56, p = .010”

  • Consistentχ²(1) = 8.58reported p = .003 · recomputed p = .003

    “X 2 (1) = 8.58, p = .003”

  • ConsistentF(1, 600) = 0.14reported p = .71 · recomputed p = .708

    “F (1, 600) = 0.14, p = .71”

31 of 31

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Study 3 exceeds the stated maximum experimental sample size

    Minor

    The overview says every experimental study had a sample size of at most 600, whereas Study 3 reports recruiting 602 participants.

  • Study 2 claims everyone completed four exercises despite contrary self-reports

    Minor

    The discussion treats completion of all four exercises as identical across participants, but the methods report that only 76.71% said they completed all four. Thus the stated control over actual exercise behavior did not hold for every participant.

2 of 2

Cites

Flagged first, then by citations

4 of the 80 references couldn’t be matched to a known paper.

Research works printed but not linked: 3

  1. Bitterly, 2015, The Psychology of Desire, 244
  2. Cochran, 1996, Striving and Feeling: Interactions among Goals, Affect, and Self-Regulation, 99
  3. Hayes, 2017, Introduction to Mediation, Moderation, and Conditional Process Analysis. Second Edition, a Regression-Based Approach

Cited by

Sorted by citations

Related

Authors
Silverman, J., & Barasch, A.
Affiliations
  • University of Delaware (US)
  • INSEAD (FR)
Corresponding author
Silverman, Jackie
ORCID
Silverman, Jackie
Venue
Journal of Consumer Research
Publisher
Oxford University Press (OUP)
Published
2022-06-30
Volume
49
Issue
6
Pages
1095-1117
ISSN
  • 0093-5301 (ISSN-L)
  • 1537-5277
DOI
10.1093/jcr/ucac029
OpenAlex
W4283735498
Cited by
34
References
80
Citations by year
2026: 4 · 2025: 10 · 2024: 11 · 2023: 8 · 2022: 1
FWCI
5.00
Type
journal article
Language
en
Open access
Licence
academic.oup.com/pages/standard-publication-reuse-rights (vor)
Locations
  • Journal of Consumer Research (publishedVersion)
  • SSRN Electronic Journal (publishedVersion)
  • Library, Museums and Press - UDSpace (University of Delaware) (submittedVersion)
  • OSF Preprints (OSF Preprints) (submittedVersion)
Funders
University of Pennsylvania
Topics
  • Social Sciences › Decision Sciences › General Decision Sciences › Decision-Making and Behavioral Economics
  • Physical Sciences › Computer Science › Human-Computer Interaction › Innovative Human-Technology Interaction
  • Social Sciences › Psychology › Applied Psychology › Behavioral Health and Interventions
Keywords
streaks, consumer behavior, goal maintenance, customer engagement, fitness tracking, learning

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-12.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-12.

Cited sources are checked by gpt-6-sol.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review