Article: Identity and Brand Congruence
Reviewed: November 2026 · Method version 2 · How these reviews work
Each factual claim in the article was checked against the source it cites, then challenged by reviewers whose job was to find the strongest evidence against it: replication attempts, meta-analyses, later papers that cite the original work, and research from other fields that study the same mechanism. A separate reviewer weighed both sides, and a script re-checked every quoted source. Where a claim was overstated or misattributed, the article was corrected. The research below supports the article as it now reads.
On the day before and the morning of the 2008 US presidential election, 88 registered California voters took a short online survey worded either about “voting” or “being a voter”; state records showed 95.5% of the noun group voted against 81.8% of the verb group.
The sample of 88, the survey timing and the turnout figures were confirmed through the Gerber et al. 2016 replication paper’s description of the original design and a Stanford SPARQ summary, because the PNAS full text is blocked to automated access. The replication record is handled in the edit to C4.
A second experiment during New Jersey’s 2009 governor’s race went the same way, 89.9% against 79.0%.
Figures confirmed through the Gerber et al. 2016 replication paper’s restatement of the original study and a Stanford SPARQ summary; the replication record is handled in the edit to C4.
Hold that result loosely. In 2016, a much larger field experiment repeated the design and found no reliable difference in turnout between the two wordings. Two more replications by the same team, including one in the 2016 presidential race, also found none, and a 2024 review of the noun research concluded the advantage has not been reliably shown. The original authors argue the replications differed in important ways, but for now the effect should be treated as unproven.
Added the two later failed replications, the 2024 meta-analysis and the original authors’ rebuttal, so readers see the full replication record.
People do things, and buy things, partly to be a certain kind of person.
Suggested Belk (1988) as a citation for the general identity-consumption idea.
Consumer researchers call the resulting bond a self-brand connection. Jennifer Escalas and James Bettman describe consumers buying brands partly to construct their self-concepts, and forming connections with the brands that do that work.
Corrected the term to ‘self-brand connection’, the construct the cited paper uses.
In their studies, a brand whose image matched a group people belonged to strengthened the connection. A brand tied to a group they did not belong to weakened it, especially for people who see themselves as independent rather than defined by their groups, and both effects were larger for symbolic brands.
Added the self-construal boundary condition the cited paper reports for the outgroup effect.
When yellow Livestrong wristbands were spreading fast as a sign of support for cancer research, research assistants went door to door in Stanford dorms selling them for a small donation: 38 to residents of one dorm, and 38 to a control dorm on the other side of campus.
Removed the unverified year.
Different assistants then asked, inside an unrelated-looking survey, who had worn a band in the past two days. In the week after the neighbors adopted it, the number of first-dorm residents reporting that they wore the wristband fell by 32%. In the control dorm, which had the same chance to get bored of it, the drop was 6%. It was a small sample and the difference only just reached statistical significance, but it matched what the same paper found in the lab.
Added the sample size and marginal significance of the field comparison. Also consider changing ‘The clearest field demonstration’ to ‘A well-known field demonstration’.
Brands carry a similar risk, especially products people use to show who they are. The people in your screenshots, testimonials and launch campaigns tell each visitor who the product is for, and courting a new audience can quietly change that message for the old one.
Limited the brand extrapolation to identity-signaling products and suggested a consumer-research citation.
On April 23, 1985, the Coca-Cola Company changed the formula of its flagship drink, in what it calls the first formula change in 99 years. The new recipe had been chosen on the strength of taste tests with nearly 200,000 consumers.
Attributed the ‘99 years’ framing to the company.
The research was right about taste. It measured the product and gave too little weight to what the product meant to the people who bought it; by several accounts, early surveys and focus groups did pick up loyal drinkers’ opposition and it was dismissed as a minority view.
Changed ‘missed’ to ‘gave too little weight to’, since accounts of the research say the opposition was detected and discounted.
Onboarding that lets people say who they are (“I run a team,” “I work alone”) supplies the kind of situational cue Oyserman describes, and gives later screens a self to speak to, though one question on its own may not shift much.
Marked the onboarding advice as an extension of the model and noted weak identity-priming replications.
These edits were made on November 12, 2026 after the review.
Claim reviewed: In 2016, a much larger field experiment repeated the design and found no difference in turnout between the two wordings.
Now reads: Hold that result loosely. In 2016, a much larger field experiment repeated the design and found no reliable difference in turnout between the two wordings. Two more replications by the same team, including one in the 2016 presidential race, also found none, and a 2024 review of the noun research concluded the advantage has not been reliably shown. The original authors argue the replications differed in important ways, but for now the effect should be treated as unproven.
Why: The 2016 study did find no reliable difference, as the article says. But the article understates the case: the same team failed to replicate the effect twice more, including in the 2016 presidential election, and a 2024 meta-analysis concluded the noun advantage has not been reliably shown. ‘Hold the size of that effect loosely’ suggests the effect is real but smaller, which the evidence does not support. The original authors dispute the replications, so the honest framing is that the effect is unproven. Now cites Gerber et al., ‘Voting behavior is unaffected by subtle linguistic cues: evidence from a psychologically authentic replication’, Behavioural Public Policy (2020).
Claim reviewed: Consumer researchers call this fit self-brand congruence; Escalas and Bettman describe consumers buying brands partly to construct their self-concepts and forming connections with the brands that do that work.
Now reads: Consumer researchers call the resulting bond a self-brand connection. Jennifer Escalas and James Bettman describe consumers buying brands partly to construct their self-concepts, and forming connections with the brands that do that work.
Why: Escalas and Bettman do say consumers buy brands partly to build their self-concepts and form connections with them. But their term is ‘self-brand connection’. ‘Self-congruity’ is a separate research tradition, and ‘self-brand congruence’ blends the two.
Claim reviewed: In Escalas and Bettman’s studies, a brand whose image matched a group people belonged to strengthened the connection, a brand tied to a group they did not belong to weakened it, and both effects were larger for symbolic brands.
Now reads: In their studies, a brand whose image matched a group people belonged to strengthened the connection. A brand tied to a group they did not belong to weakened it, especially for people who see themselves as independent rather than defined by their groups, and both effects were larger for symbolic brands.
Why: The ingroup and symbolic-brand findings are accurate. The outgroup finding leaves out the paper’s main qualifier: the negative effect was stronger for people with an independent sense of self than for people who define themselves through their groups. Other work finds similar limits, such as public versus private use.
Claim reviewed: In 2004, research assistants sold yellow Livestrong wristbands door to door in Stanford dorms for a small donation, 38 to one dorm and 38 to a control dorm on the other side of campus; a week later they sold 36 to a ‘geeky’ academic-focus dorm directly across the courtyard.
Now reads: When yellow Livestrong wristbands were spreading fast as a sign of support for cancer research, research assistants went door to door in Stanford dorms selling them for a small donation: 38 to residents of one dorm, and 38 to a control dorm on the other side of campus.
Why: The design details (38 and 38 bands, 36 to the academic dorm a week later, the courtyard and the control dorm across campus) all match the authors’ manuscript. The year 2004 is not stated in the version checked; it is plausible but unverified.
Claim reviewed: In the week after the neighbors adopted it, first-dorm residents reporting wearing the wristband in the past two days fell by 32%, against a 6% drop in the control dorm.
Now reads: Different assistants then asked, inside an unrelated-looking survey, who had worn a band in the past two days. In the week after the neighbors adopted it, the number of first-dorm residents reporting that they wore the wristband fell by 32%. In the control dorm, which had the same chance to get bored of it, the drop was 6%. It was a small sample and the difference only just reached statistical significance, but it matched what the same paper found in the lab.
Why: The 32% and 6% figures are right. The article leaves out that the comparison rests on 36 people, used self-reported wearing, and only just reached significance (p = .05). Calling it ‘the clearest field demonstration’ reads stronger than one marginal test supports, although the lab studies in the same paper point the same way.
Claim reviewed: Brands carry the same risk: the people in screenshots, testimonials and campaigns tell visitors who the product is for, and courting a new audience can quietly change that message for the old one.
Now reads: Brands carry a similar risk, especially products people use to show who they are. The people in your screenshots, testimonials and launch campaigns tell each visitor who the product is for, and courting a new audience can quietly change that message for the old one.
Why: Research on dissociative reference groups backs the idea that a product tied to an unwanted group loses appeal, mainly for symbolic and publicly used products. But the article gives no source, and one study finds loyal customers with a strong connection to the brand keep their good view of it when unexpected people use it. Now cites White & Dahl, ‘To Be or Not Be? The Influence of Dissociative Reference Groups on Consumer Preferences’, Journal of Consumer Psychology (2006).
Claim reviewed: On April 23, 1985, the Coca-Cola Company changed its flagship formula for the first time in 99 years, choosing the new recipe on the strength of taste tests with nearly 200,000 consumers.
Now reads: On April 23, 1985, the Coca-Cola Company changed the formula of its flagship drink, in what it calls the first formula change in 99 years. The new recipe had been chosen on the strength of taste tests with nearly 200,000 consumers.
Why: The date and the nearly 200,000 taste tests are on the company’s history page. ‘First time in 99 years’ is the company’s own framing, and the only source is the company itself.
Claim reviewed: The research was right about taste; it measured the product and missed what the product meant to the people who bought it.
Now reads: The research was right about taste. It measured the product and gave too little weight to what the product meant to the people who bought it; by several accounts, early surveys and focus groups did pick up loyal drinkers’ opposition and it was dismissed as a minority view.
Why: The company’s own page says the tests missed the bond. Retellings of the research say Coca-Cola’s surveys and focus groups did pick up strong opposition among loyal drinkers, and management dismissed it as a minority view. ‘Underweighted’ is closer than ‘missed’, though the counter-evidence here is a secondary source.
Claim reviewed: Digital redesigns repeat the error: a new navigation wins on task completion in testing, then loyal users revolt because the name, icon or tone they identified with has gone.
Now reads: Digital redesigns can repeat the error at smaller scale: a new navigation wins on task completion in testing, then loyal users push back because a name, icon or tone they were attached to has gone. Product teams call this change aversion, and it can be reduced by explaining changes and phasing them in.
Why: No source is given. Industry research on ‘change aversion’ confirms users often react badly to sudden interface changes, but it does not show the specific pattern of a test win followed by a revolt over identity, and it adds that the reaction can be managed. Now cites Sedley & Müller, ‘Minimizing change aversion for the Google Drive launch’, CHI Extended Abstracts (2013).
Claim reviewed: Copy can offer an identity rather than a task (“Join other independent bookshops” instead of “Sign up”), provided the identity is one the reader wants.
Now reads: Copy can offer an identity rather than a task, as in “Join other independent bookshops” instead of “Sign up,” provided the identity is one the reader wants and the claim is true. Treat this as an idea to test rather than a proven lift: the voting result behind it has not held up in larger replications, and in consumer studies, messages that spelled out what a purchase says about the buyer lowered purchase intent compared with messages that only hinted at it.
Why: The advice rests on the noun-versus-verb voting result, which three larger replications and a 2024 meta-analysis did not support, while the related ‘don’t be a cheater’ effect did replicate in a registered report, so the evidence is split. Outside politics the picture is also mixed: identity nouns made children help more but backfired after setbacks, verb wording beat noun wording in a 2025 waste study, a ‘don’t be a spreader’ message did nothing in a COVID study, and consumer research found that spelling out what a purchase says about the buyer lowered purchase intent. No study tests identity nouns in signup or checkout copy, so the article should present this as a hypothesis to test rather than a known lift. Now cites Bhattacharjee, Berger & Menon, ‘When Identity Marketing Backfires: Consumer Agency in Identity Expression’, Journal of Consumer Research 41(2), 2014.
Claim reviewed: Onboarding that lets people say who they are supplies the situational cue Oyserman describes, so later screens can speak to that self.
Now reads: Onboarding that lets people say who they are (“I run a team,” “I work alone”) supplies the kind of situational cue Oyserman describes, and gives later screens a self to speak to, though one question on its own may not shift much.
Why: Oyserman’s model supports the idea that situational cues make an identity active, and an identity-based school program had lasting effects. But the cited work does not study onboarding prompts, and some replications of simple identity priming found no effect, so a single question may do less than implied.
Claim reviewed: Consumer psychology starts with identity because it sits upstream of most other motives.
Now reads: Consumer psychology often starts with identity because it shapes many other motives.
Why: No source ranks identity above other consumer motives, and none was found. Identity is one major framework in consumer psychology alongside motivation, attitudes and decision-making.