People have a need to express themselves and to be unique from others. However, this need might conflict with other goals, such as the need to belong and fit in with others. Recent research and polling suggest that people may be more reluctant to express themselves and stand out than in previous years, but few studies have examined such societal trends in a systematic way. We examined changes in need for uniqueness among 1,339,160 Internet respondents (Mage = 21.09, SD = 9.69; 65.8% women) from 2000 and 2020. Across the 20-year period, participants who completed the survey more recently reported a lower need for uniqueness, particularly in terms of not wanting to defend their beliefs in public forums and caring more about what others think about them. Results are discussed in the context of possible causes of changes in uniqueness desires and the possible societal implications.
People have a need and desire to express themselves—to be seen as unique from other people (Snyder & Fromkin, 2012). However, this need to be unique and distinct from others inherently conflicts with other motivations people have, such as the need to belong (Baumeister & Leary, 1995; Brewer, 1993; Leonardelli et al., 2010). Negotiating competing needs can be difficult as it involves actively monitoring social situations and modulating behavior (C. Chan et al., 2012). Being too unique can alienate oneself from others and compromise an individual’s social standing and well-being (Brewer, 1993; Leonardelli et al., 2010; Snyder & Fromkin, 2012). Recent polling suggests that this balance is often on people’s minds—people are worried about expressing too controversial opinions, particularly in online spaces and on politically charged topics, while also wanting to feel included in experiences and groups (Ekins, 2017; Liu et al., 2017). Trends in other psychological characteristics suggest that social anxieties and ways of relating to others have also been changing (Twenge, 2000; Twenge et al., 2019). However, it is unclear whether these concerns have precedent, have intensified recently, or are entirely unfounded. The current study addressed these questions by examining year-to-year changes in people’s need for uniqueness between 2000 and 2020.
Modern environments, particularly with respect to social media and networking platforms, provide people with ample opportunities to express themselves (Moreira Aguirre et al., 2019). These opportunities are useful in people’s pursuit to satisfy their needs for uniqueness and authenticity—to be themselves and express who they are (Bailey et al., 2020; Snyder & Fromkin, 2012). Authenticity and expressing oneself are crucial for enjoying social interactions and being happy (Bailey et al., 2020). Moreover, concealing things from others can be a barrier to meaningful relationships and well-being, which is partly explained by the fact that others can spot inauthenticity in individuals (Le & Impett, 2016).
However, expressing oneself in an unfettered way puts people at risk of damaging their relationships, compromising their social standing, and putting themselves at risk to be ostracized (Le & Impett, 2016; Noelle-Neumann, 1993). Stating unpopular opinions or behaving in a potentially norm-defying way leaves people vulnerable to rejection and social exclusion (Norris, 2023). The fear of rejection and being ostracized is well-founded. Being ostracized is linked with depression, suicidal ideation and attempts, and eating disorders (Williams, 2007)—highlighting its potential to severely impact an individual’s mental health. To prevent being ostracized or isolated, people change their behavior in socially acceptable ways (Asch, 1956; Briggs et al., 1980). Because norm violations put people at risk for being ostracized, people may, for example, self-censor in an effort to avoid upsetting others and compromising their social standing (Noelle-Neumann, 1993; Petersen, 2012).
However, research also suggests that people may have become more sensitive to concerns about belongingness—affecting whether they seek out opportunities to be unique. For example, some research suggests that both general and social forms of anxiety have been increasing in recent years (Jefferies & Ungar, 2020; Kindred & Bates, 2023; Twenge, 2000). Social anxiety modulates people’s behavior in both virtual and non-virtual spaces—namely, socially anxious people are less likely to make use of opportunities to be unique and stand out from others (Lee-Won et al., 2015; Schlenker & Leary, 1982). Thus, the desire to be unique may have declined due to changes in other psychological characteristics, like social anxiety.
Other factors might also explain why individuals may not have as strong a desire to be unique as in the past. According to the spiral of silence theory, people who perceive their opinion to be the minority opinion are less likely to speak out (Noelle-Neumann, 1993; Petersen, 2012). Indeed, expressing minority opinions, particularly among those wanting to be included in a group, can be a distressing experience and make people reluctant to share (Rios, 2012). The silencing has broader implications as well, for both individuals and the groups they are not sharing their opinions with (Rios, 2012; Rios & Chen, 2014). This is especially important in the realm of social networking sites and public discourse—settings that may curtail people behaving in unique ways or communicating unpopular or controversial opinions that go against the norms of a social environment (Liu et al., 2017). Of the many ways that people can stand out and be unique from others is their willingness to do so by standing firm and defending their beliefs publicly—a subscale of the most popular need for uniqueness scale (Snyder & Fromkin, 1977, 2012). Somewhat adjacent work shows that one of the primary reasons why people are likely to self-censor their political beliefs (even those beliefs held by majority groups) is out of fear of isolation—again highlighting the balance people seek between striving to be unique and fitting in (Burnett et al., 2022; M. Chan, 2018; Hampton et al., 2017; Liu et al., 2017; Wells et al., 2017). Recent polling suggests that this might be an increasingly common concern—anywhere between 60-71% of Americans feel the need to self-censor in online and in public settings, and they feel that restrictions on their ability to express their beliefs is harming public discourse (Ekins, 2017, 2020).
The Current Study
The need to be unique while fulfilling belonging needs is something that nearly all people grapple with (Brewer, 1993, 2012; Burnett et al., 2022; Leonardelli et al., 2010; Noelle-Neumann, 1993; Rios & Chen, 2014). Considering preliminary evidence suggesting increases in social anxiety and evidence for an environment that is particularly punitive to individuals expressing uniqueness in terms of risks for social isolation (Ekins, 2017, 2020; Jefferies & Ungar, 2020; Kindred & Bates, 2023; Twenge, 2000), we hypothesized that people’s self-reported need for uniqueness would have declined in recent years. This is particularly relevant for facets of uniqueness that violate perceived group norms and attitudes (e.g., defending controversial beliefs and not caring what others think about you). To test this hypothesis, we examined changes in need for uniqueness in a large sample of over 1 million people over the past 20 years.
Open Practices and Data Accessibility Statement
Data, syntax, and materials for the current project can be found at https://osf.io/jm8c3/. This study was not pre-registered.
Method
Participants
Participants were Internet respondents from the Gosling-Potter Personality Project between 2000 and 2020 (Gosling et al., 2004). A total of 1,339,160 participants completed a need for uniqueness measure. Participation rates ranged from 4,578 (2019) to 234,817 (2002), but were relatively high across the years (M = 63,770.52, SD = 80,947.80).
Demographic information was available for only a subset (24.17%) of participants as it was added to the survey late in 2004 and removed in early 2009 (Mage = 21.09, SD = 9.69; 65.8% women). At the time of the survey, the median level of education was a high school education (20.0% of the sample; with the remaining 39.1% having less than a high school education, 17.9% having some college, 13.2% being college graduates, and 9.8% having gone to at least some graduate or professional school). The median social class level was “middle class” (41.5% of the sample, with the remaining 16.7% identifying as working class, 13.9% lower-middle class, 23.3% upper-middle class, and 4.7% upper class). Although these demographic distributions largely match other samples collected through this mechanism (Chopik et al., 2023; Srivastava et al., 2003), our sample was a little younger, despite having large numbers of people over the age of 30.1
Measure
Participants completed the need for uniqueness questionnaire, a 32-item scale assessing different facets that comprise the need for uniqueness construct (Snyder & Fromkin, 1977). Need for uniqueness is comprised of three components—lack of concern regarding others’ reactions, desire to not always follow the rules, and a willingness to publicly defend one’s beliefs. Each of these dimensions taps into attitudes and behaviors that have been linked to individual differences in social/group behavior and consumer choices (Lynn & Snyder, 2002; Snyder & Fromkin, 2012; Tepper, 1996). For example, scoring high in these dimensions is associated with behaving in non-conforming ways, bucking statistical norms, providing novel responses on creativity tests, membership in niche clubs and organizations (e.g., Mensa), and negative emotional responses when similarities with others are noted (Snyder & Fromkin, 1977, 2012; Tepper & Hoyle, 1996).
Participants were instructed to indicate how much they agree with 32 statements on a scale ranging from 1(strongly disagree) to 5(strongly agree). We adopted the refined factor solution proposed by Tepper and Hoyle (1996), which removed three poorly fitting items from the final solution.2 Responses were averaged to yield composites for lack of concern (14 items; α = .82; sample item: “It is better to always agree with the opinion of others than to be considered a disagreeable person.” M = 3.44, SD = .71), desire to not follow the rules (10 items; α = .72; sample item: “I always try to follow rules.” [reverse scored]; M = 3.35, SD = .69), willingness to defend beliefs publicly (5 items; α = .67; sample item: “I tend to express my opinions publicly, regardless of what others say.” M = 3.43, SD = .82), and total need for uniqueness (29 items; α = .86; M = 3.40, SD = .57). Correlations among the three subscales were moderate (ranging from r = .36 to .52).3
Results
Centered year was entered as a predictor of need for uniqueness and its three subscales (lack of concern regarding others’ reactions, desire to not always follow the rules, and willingness to publicly defend one’s beliefs). All indicators of need for uniqueness were lower in more recent years: need for uniqueness overall (b = -.008, 95% CI [-.008, -.008], β = -.051, p < .001), lack of concern (b = -.007, 95% CI [-.007, -.007], β = -.036, p < .001), desire to not follow the rules (b = -.006, 95% CI [-.007, -.006], β = -.033, p < .001), and willingness to publicly defend beliefs (b = -.014, 95% CI [-.008, -.008], β = -.063, p < .001). This pattern of results can be seen in Figure 1.
Note. 95% confidence intervals around the predicted linear effect are provided in red. Observed changes shown in grey.
Note. 95% confidence intervals around the predicted linear effect are provided in red. Observed changes shown in grey.
Demographic Characteristics and Their Effects on Need for Uniqueness
We also ran an additional analysis controlling for demographic characteristics (i.e., age, gender, education) for the years we had this information available (2004-2009; see Supplementary Table 2). All indicators of need for uniqueness were lower in more recent years: need for uniqueness overall (b = -.010, 95% CI [-.011, -.008], β = -.021, p < .001), lack of concern (b = -.016, 95% CI [-.018, -.014], β = -.027, p < .001), desire to not follow the rules (b = -.003, 95% CI [-.004, -.001], β = -.004, p = .012), and willingness to publicly defend beliefs (b = -.006, 95% CI [-.009, -.004], β = -.009, p < .001).
These three demographic characteristics rarely moderated the association of year on the uniqueness outcomes (i.e., only three significant moderation tests out of twelve). Decomposing these three interactions revealed that education moderated the associations between year on (#1) overall need for uniqueness (β = .006, p = .036) and (#2) desire to not follow the rules (β = .007, p = .022). Specifically, the slope of year was slightly larger among people with lower levels of education (overall need for uniqueness: β = -.024, p < .001; desire to not follow the rules: β = -.008, p = .004) compared to people with higher levels of education (overall need for uniqueness: β = -.015, p < .001; desire to not follow the rules: β = .001, p = .644). Age also moderated the association between year on the desire to not follow the rules (#3; β = -.014, p < .001). Specifically, the slope of year was negative and significant among older adults (β = -.019, p < .001) but positive and significant for younger adults (β = .006, p = .045).
It may be tempting to conclude that declines in need for uniqueness are more present among people with lower levels of education (possibly because deviations from norms might be more consequential for them and educational disparities in autonomy; Shifrer, 2019) and older participants (who are more socially conservative; Chopik, 2016; Jackson et al., 2009). However, most moderation tests were not significant, and the few that were significant were relatively small in magnitude or the p-values were close to p = .05 (Benjamin et al., 2018). Finally, these moderation tests were based only on participants who had demographic information (i.e., about a quarter of the sample and only from parts of 2004 until 2009). Future research can more deliberately test if yearly differences in need for uniqueness vary across different demographic characteristics. Full regression tables and moderation tests can be found in Supplementary Table 2.
Testing Non-linear Trends
Finally, upon inspecting the figures, we noticed that there might be some non-linearity to the declines in need for uniqueness over time. Non-linear fluctuations would suggest that there were periods over the last 20 years where differences in need for uniqueness might have flattened out and stabilized or even moved in the opposite direction. Perhaps there was a particular point in recent history where these declines were disrupted that a linear estimate would miss. Thus, we tested this possibility—that the association between year and the need for uniqueness scales were curvilinear—in two exploratory analyses. First, we formally modeled the quadratic transformation of the year variable on each need for uniqueness outcome. Second, we used generalized additive models to model data-dependent patterns of non-linearity.
For the polynomial regressions, we entered a quadratic term into the first set of regressions (see Chopik et al., 2013; Srivastava et al., 2003). We found that the inclusion of the quadratic term did not contribute much to the models for need for uniqueness total (ΔR2 < .001), lack of concern regarding others’ reactions (ΔR2 = .001), desire to not always follow the rules (ΔR2 < .001), and willingness to publicly defend one’s beliefs (ΔR2 < .001).
Generalized additive models assess non-linear change in a flexible way that does not assume the shape of year-to-year differences a priori (Hastie, 2017; Jones et al., 2022; McKeown & Sneddon, 2014; Wood, 2017). Rather, this approach allows for non-linear smooth functions (via cubic splines) that best represent an association between a predictor (year) and an outcome (need for uniqueness). As seen in Supplementary Figures 1-4, the year-to-year differences were largely linear in their differences such that need for uniqueness was higher in 2000 compared to 2020, particularly for the overall need for uniqueness and the willingness to publicly defend one’s beliefs (Supplementary Figures 1 and 4, respectively). Lack of concern departed from this trend in that it increased from 2000 to 2005 before declining relatively linearly until 2020 (Supplementary Figure 2). Finally, the desire to not always follow the rules showed the most variable trend (Supplementary Figure 3). Specifically, this factor declined from 2000 until 2012 before increasing from 2012 until 2015, before decreasing again from 2015 until 2018 before increasing again. This variable “up and down” can also be seen in Figure 1, although simple linear and quadratic trends masked such variations. Exactly why this one facet (and not the others) showed such variations is unclear. Our lack of explanatory measures (and measures at the fidelity required to explain such variations) prevent us from making strong attributions for why rule-breaking needs would be higher in 2015 and 2020. However, we hope future research generates explanations and formal tests of these variations over the past 20+ years in the desire to not always follow the rules.4
Discussion
People have a desire to express themselves and be unique (Snyder & Fromkin, 2012), but actually doing so might conflict with other innate needs and desires, like the need to belong to a broader social group (Baumeister & Leary, 1995; Brewer, 2012). In our study of over one million participants surveyed from 2000 to 2020, we found that need for uniqueness was lowest among participants who took the survey most recently in 2020 compared to those in 2000.
Although each facet of need for uniqueness was lower in more recent years, the difference from previous years was most dramatic for the facet of whether people are willing to defend their beliefs publicly. The fact that people do not feel comfortable defending their beliefs publicly is consistent with the observation that people may prioritize fitting in or avoiding social isolation from their communities (Burnett et al., 2022; M. Chan, 2018; Ekins, 2020). Further, the willingness to break rules is lower at more recent times in the survey, and people are more concerned with what others think about them. Collectively, the patterns of differences across the study window for each of these facets suggest that people may be more conforming and do not desire to cause great perturbations in their social and public environments.
But why would need for uniqueness be declining in recent years? There are many possible reasons why people who took the survey more recently might not express as strong a desire to be unique. For example, some of this reticence might stem from concurrent increases in social anxiety and social monitoring more generally (Jefferies & Ungar, 2020; Kindred & Bates, 2023; Twenge, 2000). Because socially anxious people are more likely to have belongingness needs (Leary et al., 2013), this heightened social monitoring might make people particularly sensitive to their social standing and, as a result, people may be reluctant to strive for uniqueness. But, of course, there are several alternative explanations. In addition to the explanation provided above, it is also possible that the need for uniqueness has declined in recent years because this need has been satisfied. Such a possibility is consistent with motivational accounts of need for uniqueness—that people are motivated to seek out opportunities to satisfy this need (Brewer, 1993; Leonardelli et al., 2010; Snyder & Fromkin, 2012). In the many public reports decrying the influence of social media and technology on mental health (Haidt, 2024), there is a relative lack of acknowledgement that online platforms can provide opportunities for people, particularly from marginalized groups, to seek out community and ways of expressing themselves (although media scholars highlight benefits and costs to these opportunities; Littman, 2022; Mehra et al., 2004; Miller, 2017; Sweet et al., 2020). This observation provides an alternative explanation for our findings. Namely, given the ostensibly wide variety and availability of ways to express oneself as a unique agent, it makes sense that people report being less motivated to stand out, express unpopular beliefs, and break rules (because they are already doing all of those things in a less encumbered way than in the early 2000s). The wording of many of the items could undermine this possibility (e.g., “Others’ disagreement makes me uncomfortable” and “I find that criticism affects my self-esteem” likely tap more into a concern about perceptions, which of course could stem from an underlying need). However, the satiation of a need (to stand out and be unique) would likely reduce participants’ endorsements of statements reflecting a comfort with and desire to be seen as unique.
Regardless, these psychological changes we saw are also occurring within a broader context in which people communicate a fear of expressing and defending unpopular beliefs in both online and offline spaces (Ekins, 2017, 2020; Liu et al., 2017). This fear of reprisal and social isolation from those they disagree with is consistent with the spiral of silence theory and its extensions regarding when people with minority opinions do or do not speak out (Noelle-Neumann, 1993; Petersen, 2012; Rios, 2012). Thus, collectively, lower levels of need for uniqueness in recent years might have stemmed from increases in social anxiety and a perception of a punitive environment for those who are too unique or violate group norms. For individuals, these two factors then might have resulted in lower levels of rule breaking tendencies, a heightened concern about what others think about them, and a lower willingness to defend their beliefs publicly.
Limitations and Future Direction
The study had many strengths. We examined a large sample of people who reported on their need for uniqueness over a long window of time (i.e., 20 years). Nevertheless, some limitations should be acknowledged.
First, few or no explanatory variables were collected in the study window. Having access to this information would have allowed us to more precisely discern the reasons why need for uniqueness might have declined in recent years. Some likely candidate explanatory variables include social anxiety, authenticity, the need to belong, and perceptions of and amount of usage of online and offline social environments. Future research should take a broader perspective and examine how these variables and others might explain why need for uniqueness has changed in recent years. Likewise, future research could experimentally manipulate some of these characteristics to more formally describe the origins of need for uniqueness (Snyder & Fromkin, 1977, 2012).
Second, our survey was limited to only a 20-year period. One of the many implications this limitation has is that we cannot quantify longer-term social and societal changes that might have affected need for uniqueness. For example, a longer design might have been able to quantify the impact that the introduction of the Internet had on people’s need for uniqueness, which we assume is non-negligible. It could be the case that the introduction of social media sites might have brought about a culture that penalizes individuals for not adopting socially sanctioned perspectives (Norris, 2023). Unfortunately, we cannot test this possibility with the data here. Future research could examine variability in need for uniqueness across different settings to provide some indication of how context shapes uniqueness. For example, some forums (both online and offline) provide people with the opportunity to maintain anonymity, which might reduce the reluctance to defend beliefs publicly, particularly if they do not conform to a majority opinion or violate a norm (Bargh et al., 2002; Gollwitzer et al., 2023). Examining variation in both naturally occurring and laboratory settings could shed light on when and where people feel comfortable being unique.
Conclusion
This study assessed how need for uniqueness differed between 2000 and 2020. Compared to respondents earlier in the survey, those who took the survey most recently reported lower levels of need for uniqueness, particularly the degree to which they are willing to publicly defend their beliefs. Theoretical insights (Noelle-Neumann, 1993), studies of other adjacent psychological characteristics (Jefferies & Ungar, 2020), and public polling (Ekins, 2017, 2020) all point to the possibility that individuals feel that being unique and expressing uniqueness might compromise their ability to fit in with others and lead to being ostracized. Self-censoring and restricting opportunities to express oneself authentically likely have important consequences for promoting the welfare of both individuals and societies (Rios, 2012). Future research can more deliberately examine the contexts under which uniqueness needs are expressed more freely and investigate the consequences of these changes in need for uniqueness.
Competing Interests
William Chopik is an editor for the Developmental Psychology section of Collabra: Psychology. He was not involved in the review process of this manuscript. The authors have no commercial or professional interests to disclose.
Author Contributions
WJC conceptualized the study idea. JP collected the data. WJC analyzed the data and interpreted the results. WJC and KG drafted the manuscript. All authors provided critical feedback and approved the content of this paper.
Data Accessibility Statement
Data, syntax, and materials for the current project can be found at https://osf.io/jm8c3/. This study was not pre-registered.
Footnotes
Additional measures were added, such as open-ended trait descriptions (e.g., how would your friends describe you?) and some behavioral indicators (e.g., do you own a bowling ball?) were introduced intermittently throughout the study window. However, these measures had small sample sizes (e.g., < 200,000), and are beyond the scope of the current report.
In evaluating whether need for uniqueness differs from 2000 to 2020, we first sought to establish measurement invariance across years. To do so, we tested for weak/metric, strong/scalar, and strict/residual invariance in the need for uniqueness scale from 2000 until 2020 (21 years or 21 groups in the context of a multi-group structural equation model). For the most parsimonious test, we fit a three-factor model in which the three subscales of the measure were allowed to covary with each other. Admittedly, the basic model had mediocre to acceptable fit, depending on which criterion was examined (CFI = .793, RMSEA = .062, SRMR = .061). Worth noting, Rutkowski and Svetina (2014) have shown that traditional cut-offs for fit indices might be too conservative when the number of groups starts to exceed 10. This suggests that, given the large number of groups, our model fit statistics might be more acceptable than traditional contexts.
Nevertheless, ΔRMSEAs for weak (.000), strong (.011), and strict (.006) all fell below the traditional cut-offs (L. -t. Hu & Bentler, 1998; L. T. Hu & Bentler, 1999; MacCallum et al., 1996), suggesting that the scale is relatively non-invariant across these varying conditions and across years. The index of ΔSRMR was also favorable for concluding the scale was invariant across years (weak: .008, strong: .011, strict: .006). Nevertheless, other indices, like ΔCFI, were not as favorable given the traditional cut-offs but were more understandable given more liberal criteria provided by Rutkowski and Svetina (2014) (weak: .018, strong: .104, strict: .074). Worth noting, when testing measurement invariance of the subscales in isolation of each other (i.e., not as covarying subfactors of one overall construct), model fit (CFIs = .862-.956; RMSEAs = .065-.085, and SRMRs = .029-.050) and relative model fits were often more satisfactory.
Work on the nomological network of the need for uniqueness scale is limited (Tepper & Hoyle, 1996). To give readers an intuition about correlates of need for uniqueness, we ran a study assessing a battery of psychological scales among a separate sample of 1,234 undergraduate students (Mage = 19.36, SD = 1.99; 65.8% women, 33.4% men, .8% other; 64.4% White, 15.0% Asian, 8.3% Black/African American, 4.6% Hispanic/Latinx, and 7.7% other races/ethnicities. The results of these analyses (that used a planned missingness design) can be found in Supplementary Table 1. The largest correlates of the need for uniqueness scales were extraversion (r = .42), neuroticism (r = -.35), need for cognition (r = .40), and NPI narcissism (r = .36) with lack of concern, and conscientiousness with not wanting to follow the rules (r = -.30). Most associations were below r = |.30|, but it was generally the case that a high need for uniqueness was associated with being more extraverted, open to experience, more open-minded, and less neurotic. Need for uniqueness was positively correlated with need for cognition and narcissism but was negligibly related to life satisfaction. Finally, people high in need for uniqueness were more likely to endorse the statements, “I am a unique person.” (r = .33) and “I am a weird person.” (r = .18).
Another potential explanation for the findings seen here is that the characteristics of the sample may have changed significantly over time. Although it can be difficult to formally test this idea with our data, we provide three notes that speak to this possibility. First, we compared year-to-year differences in age (ƞ² = .013 [.012, .013]), education (ƞ² = .007 [.007, .008]), and social class (ƞ² = .001 [.001, .002]). We found that these differences were relatively miniscule. The same can be said for gender, such that most years fluctuated between 63.7% and 67.9% women (SD = 3.16%; 2009’s 59.2% women was a bit of an outlier, but very little demographic data was available for that year [N = 1114; 0.33% of the available data] compared to the other years [N = 340,057]).
Second, the demographics seen in our study are comparable to those seen in other internet samples, both from this same source and others (Chopik et al., 2023; Srivastava et al., 2003), albeit our sample is a bit younger. Although not as strong as evidence examining demographic changes, the fact that comparable demographics are seen in different data sources and over varying intervals suggests that our data are not disproportionately affected by selection processes (and our results are also consistent with nationally representative polling data).
Finally, we borrowed an approach from Srivastava et al. (2003) who examined whether the standard deviations of an outcome differed by demographic characteristics or by study year (in our case). If there was a large selection effect based on the variables at hand, one would expect differences in the standard deviations across levels of a variable (because if people particularly high or low in need for uniqueness were differentially absent over time, the variability [standard deviations] would shrink as extremes would be “cut off”). Comparing standard deviations across the study years showed a high degree of stability in how variable our outcome was. Specifically, the SDs ranged from .556 (in 2001) to .595 (in 2019) with little evidence of strictly linear changes (MSD = .575, SDSD = .009).
Thus, although selection effects are difficult to operationalize and test, we feel that these three bits of evidence somewhat reduce concerns over selection effects driving our results.
University of California Press