Abstract
Attraction-selection-attrition (ASA) theory proposes that “the people make the place” and has served as a foundation for many areas of organizational research. In this review, we take stock of the ASA literature to identify what we know and what we need to know about ASA processes and their effects on organizations. Based on a review of over 6,000 articles that cited ASA, we identified 321 studies that used the theory as a basis for their hypotheses. However, only 77 (24%) of those studies actually tested an aspect of the theory. For example, although ASA is an organizational-level theory, most studies used the theory to test phenomena at other levels, such as individuals, teams, or occupations. Among studies that did test the theory, very few directly assessed its core hypotheses. For instance, only one published study directly tested the central hypothesis that ASA processes lead to homogeneity. Moreover, some parts of the theory were not supported. As an example, although the theory suggests that ASA processes will reduce organizational effectiveness, several studies found that homogeneity was associated with better performance. Although the amount of empirical support for ASA theory was uneven, we believe it still has the potential to help understand organizations and the people that make them. To that end, we provide an agenda for future research that prioritizes how to best test ASA’s core hypotheses. We also highlight connections between ASA and other theories and literatures that examine similar phenomena to inspire future research opportunities.
Keywords
Attraction-selection-attrition (ASA) theory was built on the assertion that “the people make the place” (Schneider, 1987: 450). Specifically, the people an organization attracts, selects, and retains will collectively reflect the organization’s culture. Culture refers to the way an organization looks, feels, and functions and originates with the founder’s goals and characteristics. Attraction refers to the notion that alike individuals tend to gravitate toward organizations that show congruence with their own characteristics (e.g., personality, values). Selection refers to the notion that organizations tend to select people who possess characteristics consistent with the organization’s culture. Finally, attrition refers to the notion that individuals will leave organizations when they do not feel like they fit the culture. ASA theory predicts that these three processes constitute a cycle that makes organizations become increasingly homogenous in terms of employee personality, values, and goals. Furthermore, over time, this within-firm homogeneity may cause organizations to become stagnant and less effective.
ASA theory has been cited thousands of times, and Figure 1 shows that citations of ASA and Schneider (1987) have increased steadily with time. 1 Further, the theory has influenced many areas of organizational research. In the organizational behavior literature, ASA has been the central theory for research on person–organization (PO) fit (e.g., Kristof, 1996). In human resources management, ASA has been used to study employee selection (e.g., Ones & Viswesvaran, 2003) and turnover (e.g., McCulloch & Turban, 2007). In the teams literature, ASA has served as a basis for hypotheses involving group composition, dynamics, and outcomes (e.g., George, 1990). In strategic management, ASA has been used to understand entrepreneurship (e.g., Zhao & Seibert, 2006), top management teams (e.g., Jackson, Brett, Sessa, Cooper, Julin, & Peyronnin, 1991), and human capital (e.g., Ployhart, Weekley, & Baughman, 2006). Finally, ASA also has been used frequently by researchers in other fields, such as economics, education, and medicine.
Citations to ASA Theory (1987 to 2024)
Despite its popularity, important theoretical and methodological discrepancies exist between ASA theory and research that has used the theory. Theoretically, ASA is an organizational-level theory that seeks to understand differences between organizations. As Schneider, Goldstein, and Smith (1995: 766) noted, “research conducted on the ASA model must be at the organizational level of analysis.” However, most studies have used the theory to explain phenomena within organizations, using individuals or teams as the unit of analysis. As another example, the theory suggests that ASA processes eventually will reduce organizational innovation and effectiveness. However, studies often hypothesize that ASA processes lead to better performance (e.g., Oh, Kim, & Van Iddekinge, 2015). Methodologically, it appears that many studies have not tested ASA in ways consistent with its theoretical propositions. For example, studies often assume that homogeneity is due to attraction, selection, or attrition but do not measure these processes. Further, although the theory predicts that ASA processes will result in greater homogeneity over time, studies often use cross-sectional designs.
Thus, despite the venerable status of ASA theory in management and applied psychology, it is difficult to gauge the level and consistency of empirical support for the theory. Schneider et al. (1995) was arguably the last systematic review of the theory. However, ASA was not even a decade old at that time, and since 1995, thousands of articles have cited the theory. Thus, considerable research has since been published that cites ASA but has not been reviewed. A book chapter by Dickson, Resick, and Goldstein (2008) examined the theory, but the authors noted that a comprehensive review was “well beyond the scope of this chapter” (6). Some reviews and meta-analyses have examined elements of ASA, but not the entire theory. Specifically, Kristof-Brown, Schneider, and Su (2023) briefly revisited several key propositions of ASA theory, but their review focused on PO fit. Likewise, several meta-analyses examined attraction, selection, and/or attrition within the context of PO fit (e.g., Kristof-Brown, Zimmerman, & Johnson, 2005) but did not test homogeneity or other ASA hypotheses. Thus, a systematic review of empirical evidence for ASA theory is long overdue. As we will show, the theory is largely unchanged since its introduction nearly 40 years ago. This raises an important question: Is ASA theory unchanged because it has received strong support or because it has not been adequately tested?
The first goal of the present review is to take stock of the ASA literature to identify what we know, what we do not know, and what we need to know about ASA processes and their effects on organizations. Such a review is important because ASA is a foundational theory that has influenced many literatures (e.g., PO fit, recruitment, and selection). Although this suggests that the theory is impactful, the extensive citations to ASA may mask the amount of support for the theory. A systematic review of empirical evidence is needed to understand which ASA hypotheses are supported, which have received inadequate support, and which might need to be reconsidered.
After reviewing evidence for the theory, the second goal of the present review is to provide a framework to help guide future studies. We do so by identifying which ASA predictions require more research. We also highlight connections between ASA theory and other literatures (e.g., diversity, strategic management) and theories (e.g., resource-based view) that examine similar questions. In doing so, we seek to identify overlapping theories or concepts that could potentially be pruned, as well as stimulate new research opportunities and directions. Altogether, we leverage the review to generate new insights that will inform both ASA and other literatures.
Overview of ASA Theory
We begin by summarizing ASA theory’s core assumptions, hypotheses, and constructs. This is important given some of the ambiguities surrounding what ASA proposes. We relied primarily on Schneider (1987) and Schneider et al. (1995). We also consulted subsequent reviews of the theory, particularly Dickson et al. (2008), as well as reviews of adjacent literatures, such as Kristof-Brown et al.’s (2023) review of PO fit. In addition, we drew upon key empirical studies that review and test aspects of the theory (e.g., Oh, Han, Holtz, Kim, & Kim, 2018).
ASA Assumptions and Hypotheses
ASA theory is based on three assumptions that help define the boundaries and inclusion criteria of our review. First, the primary independent variable (i.e., organizational homogeneity) and outcomes (e.g., coordination, effectiveness) are at the organizational level. Indeed, “it is not the personal attributes of individuals within an organization that is the predictor of interest but the relative homogeneity of personality at the organizational level of analysis that is the predictor; the criterion is not individual performance within an organization but organizational effectiveness” (Schneider et al., 1995: 766-767). Second, although unstated in the original writing, ASA is a multilevel theory. The processes of attraction, selection, and attrition occur at the individual level and contribute to the emergence of homogeneity at the organizational level. Using the language of Klein and Kozlowski (2000), the level of measurement is at the individual level, but the theory is at the organizational level. Third, ASA is a temporal theory. That is, “ASA is a time-based conceptualization of homogeneity, one in which homogeneity is predicted to occur over time as the ASA cycle plays out for each individual” (Schneider et al., 1995: 763).
Table 1 presents the original ASA theory propositions from Schneider (1987). It also includes refinements to the original propositions based on Schneider et al.’s (1995) update of the theory and Kristof-Brown et al.’s (2023) PO fit review. These two articles are the only ones that proposed modest refinements to core ASA hypotheses and, thus, are included in the table. We note several points about the original propositions and refinements.
ASA Theory Propositions, Subsequent Refinements, and Hypotheses on Which the Present Review Focuses
Note: Schneider (1987) Propositions 5 and 7 overlap, and our homogeneity hypothesis captures both propositions.
First, Schneider’s (1987) first three propositions summarized interactionalist psychology theory and research. They were intended to “set the stage” for the next four propositions, which are specific to ASA. Second, Schneider et al. (1995) did not reiterate or revise the original propositions but rather highlighted “two central propositions of the ASA model” (752): (a) the role of the founders and (b) the homogeneity of attributes that results from the ASA cycle. Third, Kristof-Brown et al. (2023) affirmed the original ASA propositions, as well as distinguished between within-organization homogeneity and between-organization differences. Kristof-Brown et al. also called out a separate hypothesis about the negative effects of homogeneity.
Integrating these sets of propositions and hypotheses suggests five unique hypotheses that will be the focus of our review. First, the founders hypothesis predicts that the personality, values, and goals of an organization’s founder(s) will influence the kinds of people who are attracted to, are selected by, and stay with the organization. This is Proposition 6 in Schneider (1987; see our Table 1).
Second, the ASA processes hypothesis predicts that people are attracted to, selected by, and choose to remain in organizations that match their personality, values, and goals. This is Proposition 4 in Schneider (1987). Specifically, job applicants are attracted to organizations they think match their personality, values, and goals. Based on information such as how companies represent themselves in job ads and other recruitment materials and interactions with recruiters, applicants infer what an organization is like and how they might fit in (Neville & Schneider, 2021). People who perceive fit with an organization then apply for job openings. This idea has served as a primary basis for research on PO fit (Kristof, 1996). Next, applicants are selected into organizations based on their fit with the organization’s personality, values, or goals. This ASA process involves two aspects. Organizations use selection procedures such as personality tests and interviews and make job offers to applicants who are considered a good fit. Then, applicants accept or reject job offers based on their own assessments of fit. Finally, ASA predicts that employees whose personality, values, and goals match those of others in the organizations will tend to remain. In contrast, employees who do not fit will tend to attrit, either voluntarily due to dissatisfaction or involuntarily due to low job performance.
Third, the homogeneity hypothesis predicts that through the processes of attraction, selection, and attrition, people within organizations become relatively homogenous in terms of characteristics such as personality, values, and goals. This reflects Propositions 5 and 7 in Schneider (1987) and Hypothesis (a) in Kristof-Brown et al. (2023). As noted, homogeneity on these characteristics is assumed to be at the firm level but their measurement is at the individual level.
Fourth, the organizational differences hypothesis is a corollary of the homogeneity hypothesis. It predicts that, because of organizational homogeneity, the characteristics of employees within each firm will differ from the characteristics of employees in other firms. This is Hypothesis (b) in Kristof-Brown et al. (2023). As Schneider (1987: 440) noted, “Different kinds of organizations attract, select, and retain different kinds of people, and it is the outcome of the ASA cycle that determines why organizations look and feel different from each other.”
Finally, the effects of homogeneity hypothesis predicts that homogeneity eventually will cause organizations to become less successful. This is Hypothesis (c) in Kristof-Brown et al. (2023). Although Schneider (1987) cautioned against the potential dangers of homogeneity, he did not make this specific proposition. Schneider et al. (1995) discussed that homogeneity can have both positive and negative consequences. Specifically, “the primary positive consequences include . . . higher levels of satisfaction, communication, and cooperativeness, and fewer interpersonal conflicts. The primary negative consequence of such homogeneity is seen as an inability to change when the environment demands and, concomitantly, the demise of competitiveness through unchanging and easily predictable decision making” (765). Schneider et al. also suggested that homogeneity may be positive in the early stages of organizations because it can facilitate communication and coordination. However, as organizations become more established, homogeneity can produce “inflexibility, an inability to adapt and change, and eventual organizational demise (765-766).” Thus, the effects of the homogeneity hypothesis are perhaps the most tentative and least understood ASA hypothesis.
ASA Constructs
Schneider (1987) did not specify the precise constructs on which ASA processes exert influence but suggested that ASA processes restrict the range of the “types of people” (444) in organizations. He did highlight goals, which are determined by the founders. In fact, organizational goals are central to the original ASA framework (see Figure 1 in Schneider, 1987). Schneider et al. (1995) focused on personality, attitudes, and values. They also reiterated the important role of the founder’s goals. However, Schneider et al. mentioned attitudes only in passing and primarily emphasized the other attributes. Therefore, ASA generally seems to focus on homogeneity with respect to three broad domains of individual differences: goals, personality, and values. 2
ASA theory proposes that attraction, selection, and attrition will cause organizations to become increasingly homogenous (i.e., the homogeneity hypothesis). Thus, homogeneity is the key and most proximal outcome of ASA. As discussed, the theory is indefinite about the effects of homogeneity on subsequent outcomes. For example, homogeneity may exert a positive influence on factors such as coordination, cohesion, and efficiency, particularly in the early stages of the firm. However, Schneider et al. (1995: 760) suggested that “the tendency towards homogeneity can be dangerous so far as long-term organizational effectiveness is concerned.” Thus, based on the theory, we focus on the effects of homogeneity on firm-level psychological (e.g., coordination and cohesion) and performance (e.g., productivity and financial) outcomes.
Method
Literature Search
Our search for relevant articles involved four steps. First, we searched articles that included the phrases “attraction selection attrition” and “framework or model or process or theory” within the Business Source Complete, ProQuest, PsycINFO, and Web of Science databases. Second, we searched these same databases for articles that cited Schneider (1983), Schneider (1987), or Schneider et al. (1995). Third, we scanned all available online programs from the Academy of Management and SIOP annual meetings to identify potentially relevant conference papers. Finally, we reviewed the reference sections of the studies we obtained to identify additional articles. In total, we examined the titles and abstracts of 6,027 articles.
Inclusion Criteria
Our review of the 6,027 articles involved two main stages. In the first stage, we used three criteria to identify articles to include in the review. First, given that our review focuses on evidence for ASA theory, we included empirical studies, including both primary and meta-analytic studies. We excluded non-empirical works such as reviews and conceptual papers (though we read many of these articles to inform our work). Second, we included articles that used ASA theory as a theoretical basis to ensure that our review focused on studies that have used or tested ASA theory’s core hypotheses. We erred on the side of inclusion and, for example, incorporated studies for which ASA was not the primary basis but was mentioned in the hypothesis development. Third, we included independent samples only and excluded studies that reported results based on the same or a highly similar previous sample. A total of 299 articles met these criteria. However, 22 articles could not be accessed online or through interlibrary loan, leaving 277 articles, which included 321 independent studies.
In the second stage, we applied three criteria to the 321 studies we identified. First, we included studies that directly or indirectly tested at least one ASA hypothesis. We differentiated this set of studies from those that used ASA as a theoretical basis but did not test any of its core hypotheses. For example, a study that measured the Big Five personality factors (which is an ASA construct domain) but did not assess homogeneity regarding personality (e.g., Danaeefard, Boustani, Khaefelahi, & Delkhah, 2018).
Second, we included studies that measured construct domains ASA theory specifies, namely personality, values, and goals. We excluded studies that examined other constructs, such as affective states (e.g., job satisfaction, stress), perceptions of the organization (e.g., climate, culture), demographic characteristics (e.g., age, gender, tenure), organizational practices (e.g., hiring practices, leadership), and political ideology.
Third, because ASA is an organizational-level theory, we included studies that used the organization as the referent. In addition, ASA processes occur at the individual level, which are then predicted to produce homogeneity at the firm level. Therefore, we also included individual-level studies that measured attraction, selection, or attrition by comparing applicant or employee characteristics (e.g., values) to characteristics of the organization. We excluded studies that focused on other levels, such as groups, occupations, or industries. A total of 77 articles and 91 independent studies met all the criteria and were the focus of our review. We used this set of studies to assess empirical support for the theory.
Coding Procedure
We developed an initial coding form that we used to code several sets of 5-10 articles. After each set, we met to discuss the form and make improvements. After we finalized the coding procedure and form, we proceeded to code the remaining articles. Two authors independently coded each study. Intercoder agreement across the 321 studies we evaluated for final inclusion in the review was 91.8%. Agreement on the other main codes was as follows: 93.8% to 99.2% for identifying which of the ASA hypotheses were tested; 87.8% to 93.3% for which of the ASA processes were tested; 95.2% for identifying the ASA construct domain(s) used; and 89.1% for the level of support for each hypothesis. We also coded several other characteristics to help understand how ASA theory has been tested, including the discipline in which the article was published, as well as the study sample, setting, design, and analyses.
Results
Characteristics of Studies That Cited ASA Theory
Figure 2 summarizes the articles we evaluated for potential inclusion in the review. This figure provides a high-level view of how the field has used ASA. Of the 6,027 studies that cited the theory, only 4.5% (277) used ASA as a foundation for the study’s hypotheses. Further, of the 277 articles that were eligible for inclusion, 43.7% did not test an ASA hypothesis, 42.1% did not include a relevant ASA construct domain, and 33.9% did not use an organizational referent. These findings suggest that very little of the research that cites ASA tested aspects of the theory.
Flow Diagram of the Search and Review of Articles for Potential Inclusion in the Review
Table 2 reports characteristics of the 321 independent studies from 277 articles that met our initial inclusion criterion of using ASA theory as part of the theoretical basis. 3 Several noteworthy observations emerged. First, ASA theory has been used across many disciplines, including various subfields of psychology and management, as well as other subfields within business (e.g., economics, marketing) and the social and natural sciences (e.g., education, medicine; see Table 2 note). Psychology and management are the most dominant literatures, accounting for 83.4% of the studies.
Frequencies for Coded Variables Among Studies That Used ASA Theory
Note: k = 277 articles for academic discipline, which reflects the number of articles that used ASA theory. k = 321 for all other variables, which reflects the number of independent studies across the 277 articles. For some variables, the total k exceeds 321 because studies could fit into more than one category (e.g., Attraction and Multiple processes).
Psychology includes I/O psychology, vocational psychology, and general psychology. Management includes organizational behavior and human resources, general management, public management, strategy, industrial relations, and entrepreneurship. Business includes economics, international business, marketing, accounting, and hospitality. Other social sciences includes education, library science, and safety science. Natural sciences includes medicine and general science.
Other constructs researchers used to test ASA theory included variables such as interests, demographics, climate perceptions, job satisfaction, reward orientations, and perceptions of selection procedures.
Second, although ASA was intended to be an organizational-level theory, only 45 studies (14%) were conducted at the firm level. Most studies (70.4%) collected and analyzed data at the individual level only. Studies at the occupation, group/team, or industry levels were prevalent. Third, although the theory assumes that ASA processes occur over time, most studies (70%) used cross-sectional designs. Only 38 studies (11.8%) used longitudinal designs that could test whether and how homogeneity—or the ASA processes thought to produce it—changed over time. Thus, relatively little research has tested ASA in a manner consistent with its core assumptions.
Fourth, 46.4% of studies tested one or more ASA hypotheses. The ASA processes hypothesis was by far the most frequently tested (33.0%), and attraction was the most studied process (17.8%). Only 3.4% of studies examined multiple ASA processes. Only 19 studies (5.9%) tested the homogeneity hypothesis, which arguably is the core proposition of the theory. Further, very few studies tested the founders (1.2%) or the effects of homogeneity (2.5%) hypotheses (4 and 8 studies, respectively).
Finally, of the construct domains on which ASA process are thought to act, values (31.8%) and personality (30.8%) were the most prevalently studied. Far fewer studies examined goals (2.5%). Interestingly, the largest category (41.1%) comprised studies that used constructs ASA theory did not specify. Examples include vocational interests, demographics, climate perceptions, job satisfaction, and perceptions of selection procedures. Thus, many researchers have been using ASA to develop research questions quite different from the focus of the theory.
Findings From Studies That Tested an ASA Hypothesis
Of the 321 independent studies that used ASA as part of their theoretical basis, only 77 (24%) tested one of the theory’s hypotheses. In the rest of the results, we summarize empirical evidence for each ASA hypothesis based on these 77 studies. We describe representative studies and, when available, meta-analyses that cumulated that research.
Table 3 overviews the ASA hypotheses these studies examined, and Appendix B reports the details for each study. For each variable, we report the number and percentage of studies that directly or indirectly tested each hypothesis. Direct studies are those whose results provided direct evidence for a hypothesis, such as studies that assessed within-firm homogeneity or tested whether an ASA process was associated with greater homogeneity. Indirect studies are those whose results provided more ancillary evidence for an ASA hypothesis. An example of ancillary evidence would be studies that related PO fit regarding ASA constructs (e.g., values) to employees’ intentions to leave the company. Table 3 also notes the level of support (i.e., none, partial, or full support) for the hypotheses (also see Figure 3). We based level of support on the significance tests and effect sizes (as available) that most directly tested each hypothesis. 4 We also perused the studies for any moderators of ASA processes or their effects and incorporated these results into the review.
Summary of Empirical Support From Studies That Tested an ASA Hypothesis
Percentages reflect the number of studies that directly or indirectly tested each hypothesis compared to the total number of studies that tested ASA theory (k = 77).
Percentages in this section reflect the number of studies that found no support, partial support, or full support compared to the number of studies that tested the hypothesis.
The numbers and percentages in this section are based on 107 unique tests (within the 77 primary studies) because some studies examined the same ASA construct domain(s) across multiple hypotheses.
Summary of Empirical Support for ASA Hypotheses
Founders Hypothesis
The founders hypothesis predicts that characteristics of an organization’s founder(s) will influence the kinds of people who are attracted to, selected by, and stay with the organization. Our review uncovered only two studies that related founders’ characteristics to those of organizational members. Giberson, Resick, and Dickson (2005) correlated the Big Five factors and 10 personal values of CEOs from 32 firms with the mean personality and values of their employees. A total of 18 of the 32 CEOs were founders, and relations did not depend on whether CEOs were founders or not. Observed rs between CEO and employee personality ranged from −.25 for openness to .35 for agreeableness and conscientiousness (Mr = .14). For values, rs ranged from −.13 for status to .42 for benevolence (Mr = .20). Using hierarchical linear modeling, the authors regressed the mean employee personality and values scores onto CEOs’ personality and values across organizations. The estimated slope coefficient (for the Level 1 analysis) of .23 was statistically significant, which suggested that leaders’ personality and values profiles were related to employee profiles. To calculate an effect size, the authors computed a pseudo R2, which indicated that leaders’ scores accounted for 21% of the variance in employee scores.
In a similar study, Kyser (2017) reported correlations between 23 founders and samples of employees from each firm on three of the Big Five factors. The founder-employees r was significant for openness (.23) but not for conscientiousness (.02) or agreeableness (.08).
Summary
Very little evidence exists for what Schneider et al. (1995) considered one of the key propositions of ASA theory. We found only two studies that related the personality and values of a small group of founders to the attributes of employees in their firms. Relations generally were modest and sometimes in opposing directions.
ASA Processes Hypothesis
This hypothesis predicts that people are attracted to, selected by, and choose to remain with organizations that match their personality, values, and goals. We summarize research that provided evidence for each ASA process.
Attraction
The attraction process has been the most frequently studied aspect of ASA theory. Of the 77 studies that tested ASA, 38 (49.4%) directly or indirectly assessed attraction, and 73.7% of these studies found full or partial support for the process. The typical study design assessed the fit between applicants (or students in an experiment) and an organization and correlated fit with self-reported attraction to the company (e.g., intentions to apply). Nearly all studies examined attraction with respect to personality or values; no studies examined goals. Kristof-Brown et al. (2005) meta-analyzed some of this research and found a mean estimated true-score correlation (ρ) of .46 (SDρ = .10) between PO fit measures and applicant attraction (k = 11, n = 9,001). Mean estimates of ρ were notably larger among studies that measured applicants’ perceptions of fit (i.e., “subjective fit”, .62) than fit assessed by measuring the similarity between applicant and organization member ratings of the same values, personality, etc. (i.e., “objective fit”, .22).
Only two studies attempted to test the effects of attraction (as well as selection and attrition) on homogeneity. Meyer (2008) compared variance in seven scales from the Hogan Personality Inventory between population norms and job applicants in each of 10 organizations (i.e., 70 total comparisons). Attraction was supported in 48/70 cases (69%), such that variance was significantly lower for applicants compared to the population. Oh et al. (2018) used a similar approach to compare variance in the Big Five factors in the workforce population to variance among applicants from three organizations. To assess the effects of attraction, the researchers calculated range restriction values (u) by dividing the standard deviation (SD) of personality scores among applicants by the SD of scores for the workforce population (based on normative data). The resulting mean u values across the Big Five were .97, .98, and .98 across the organizations. This suggests that applicant pools for these firms were only slightly more homogeneous than the general workforce.
We found only one study that examined a potential moderator of relations between an ASA construct and attraction. Yu and Verma (2019) tested whether goal orientation moderated the effects of values-based PO fit on attraction. Results revealed an effect for learning goal orientation but not for performance orientation. Specifically, the fit-attraction relationship was stronger when learning goal orientation was low and was not significant when learning orientation was high. This finding supported the hypothesis that people with higher learning goal orientations will be more tolerant of differences between themselves and potential employers.
Selection
A total of 11 studies directly or indirectly tested the selection element of the ASA processes hypothesis, and 72.7% found partial or full support for this element. The first type of study that sheds light on the selection process involves relating fit between applicants and the organization (e.g., values) to whether applicants received or accepted offers from the organization. Kristof-Brown et al. (2005) reported a mean estimated ρ of .32 (SDρ = .24) across eight studies (n = 1,556) that correlated PO fit with whether applicants received a job offer. As before, relations were much larger using subjective versus objective fit measures (.50 vs. .03). In addition, across four studies (n = 1,829), the researchers found a mean ρ of .24 (SDρ = .09) between PO fit and job acceptance.
Meyer (2008) and Oh et al. (2018) were the only studies to test the effects of selection on homogeneity. Meyers found that personality score variance was lower among new hires than among job applicants in 56 of 70 cases (80%). Similarly, Oh et al. compared variance in the Big Five in the applicant pool to variance in these traits among newly hired employees. Across three organizations, u values were .88, .84, and .94. Thus, personality variance generally was reduced (i.e., increased homogeneity) after the selection process.
Attrition
A total of 27 studies examined the attrition element of the ASA processes hypothesis, and 85.2% found full or partial support for this process. The most prevalent design involved correlating a measure of PO fit with turnover intentions or actual turnover. Kristof-Brown et al. (2005) found mean ρs of −.46 (SDρ = .13) between values-based fit and turnover intentions (k = 32, n = 18,222) and −.14 (SDρ = .12) between fit (constructs not specified) and turnover (k = 10, n = 2,157). A meta-analysis by Arthur, Bell, Villado, and Doverspike (2006) included five studies (n = 942) that related values-based PO fit to turnover, and the mean ρ was .38 (SDρ = .19; whereby positive rs reflected less turnover). Thus, in both meta-analyses, better fit (e.g., in terms of values) was related to lower attrition.
Once again, Meyer (2008) and Oh et al. (2018) were the only studies that addressed whether attrition contributes to homogeneity. Meyers reported that personality variance was higher among job incumbents (after excluding new hires who attrited) than among new hires, which is the opposite of what ASA predicts. Oh et al. compared variance in the Big Five among new hires to variance among employees who were still in the organization one or more years later. Unlike selection (which consistently reduced variance relative to the applicant pool) attrition appeared to exert minimal effects on homogeneity. Indeed, SDs between new hires and seasoned employees largely were unchanged, even four or five years later.
Finally, these two studies were also the only ones that addressed the influence of all three ASA processes on homogeneity. Meyer (2008) examined the extent to which variance in personality decreased across all four groups: a population sample, applicants, new hires, and job incumbents. In only 14/70 cases (20%) did variance consistently decrease moving from the population to incumbents. Oh et al. (2018) suggested that selection may exert the strongest effect when organizations hire applicants based on an ASA construct. Indeed, many organizations use personality tests and fit assessments based on ASA constructs for selection (e.g., Kantrowitz, Tuzinski, & Raines, 2018). In contrast, myriad factors can influence attrition, including factors that may be less related to ASA constructs, such as low pay or lack of supervisor support. In line with this rationale, Oh et al. found evidence that selection contributed most to within-organization homogeneity. Selection effects were particularly notable because the organizations did not select applicants based on their personality test scores, but rather used interviews and other selection procedures to make decisions. As such, these findings may underestimate the influence of selection on homogeneity compared to firms that base hiring decisions on personality or other ASA constructs (e.g., values).
Oh et al. (2018) assessed changes in variance in personality as individuals progressed from applicants to new hires to seasoned employees. To assess homogeneity, they calculated u values that compared the SD in the restricted sample to the SD in the population from which the sample was drawn. For instance, in a banking organization, they found u values for conscientiousness of .99 for applicants (compared to the general workforce), .86 for new hires (compared to applicants), and .80 for incumbents two years later (compared to new hires). Thus, variance in conscientiousness became more restricted (i.e., more homogeneity) over time. Overall, the researchers estimated that ASA processes reduced variance by 10% and 19% in two samples over six years. In a third sample, there was an 8% reduction in variance over two years.
Summary
In support of ASA theory, research typically finds that people are attracted to organizations they perceive possess similar values or personality. This conclusion is based primarily on studies that related values-based PO fit to intentions to apply to an organization or actual application decisions. In support of the selection process, organizations often offer jobs to applicants whose values appear to fit those of the firm, and applicants tend to accept jobs when they also perceive a good fit with the firm. Finally, there is some evidence that people whose values-based fit is poor often intend to or actually leave the organization.
However, tests of the individual processes often were based on designs that prevent strong inferences. For example, many tests of attraction used cross-sectional designs (for an exception, see Judge & Cable, 1997), and studies of attrition often relied on self-reported intentions to quit rather than actual turnover (for an exception, see De Cooman et al., 2009). Further, most studies used brief measures that assessed global, perceived fit with the organization. For example, the widely used values congruence measure of Cable and Judge (1996) includes three items (e.g., “To what degree do you feel your values ‘match’ or fit this organization and the current employees in this organization?”). Fewer studies have examined the fit between individuals and firms regarding specific values, personality traits, or goals.
Moreover, evidence that these individual-level ASA processes lead to homogeneity at the organization level is very sparse. We uncovered just two studies that addressed whether the processes increased homogeneity (Meyer, 2008; Oh et al., 2018). In both studies, selection appeared to exert the strongest influence on homogeneity, even in organizations that did not directly select or reject applicants based on their personality test scores. Effects for attraction and attrition appeared relatively minimal, and even opposing to what ASA predicts in Meyer (2008). Overall, there is very little direct evidence for this key proposition of ASA theory.
Homogeneity Hypothesis
The homogeneity hypothesis predicts that—through the processes of attraction, selection, and attrition—people within organizations become relatively homogenous in terms of personality, values, and goals. In the ASA literature, there is little consistency regarding how to test homogeneity (Dickson et al., 2008). The most prevalent approach we encountered was MANOVA in which the organizational membership was the independent variable and the ASA constructs (e.g., the Big Five factors) were the dependent variables. A significant F-value indicates mean differences exist among organizations with respect to the construct. However, MANOVA tests whether mean differences exist between organizations relative to differences within the organizations. As such, MANOVA does not directly test the homogeneity hypothesis (Bradley-Geist & Landis, 2012; Oh et al., 2018). For instance, large between-firm differences in personality or values can mask notable within-firm heterogeneity in the constructs.
Similarly, some studies reported intraclass correlation coefficients (ICCs). ICC(1) estimates the portion of total variance in an ASA construct (e.g., personality) due to the unit (e.g., organization), whereas ICC(2) provides an estimate of the reliability of the unit means (Bliese, 2000). Like MANOVA, neither statistic directly assesses within-firm homogeneity. ICC(1) is an estimate of within- versus between-unit variance and thus suffers from some of the same limitations as MANOVA. Some studies used ICC(2) to assess homogeneity (e.g., Schneider & Bartram, 2017). However, this statistic focuses on interrater reliability rather than on interrater agreement. ICC(2) also is influenced by the number of raters and unit size.
We identified nine studies that tested the homogeneity hypothesis using average deviation (AD; Burke, Finkelstein, & Dusig, 1999) and/or rwg (James, Demaree, & Wolf, 1984). In contrast to MANOVA and ICCs, these statistics focus on the degree of within-group agreement and are independent of between-group differences. AD and rwg focus on the level of homogeneity and not whether it is statistically significant. Moreover, these statistics can be calculated for each organization and enable researchers to assess the variability in homogeneity across firms. Of the studies that used one of these statistics, 100% found partial or full support for within-firm homogeneity. For instance, Bradley-Geist and Landis (2012) used AD to assess homogeneity within 25 organizations on the four Meyers-Briggs Type Indicator (MBTI) dimensions. They concluded that evidence of homogeneity existed in 83% of the cases. Chu, Fu, and Liu (2019) used rwg to assess homogeneity regarding the Big Five across nine Chinese railway organizations. Values ranged from .84 to .90 across the five factors, which would be considered “strong agreement” based on LeBreton and Senter’s (2008) benchmarks.
Finally, the homogeneity hypothesis predicts that ASA processes will cause organizations to become more homogeneous over time. Only one study examined changes in homogeneity. Haudek (2001) compared 12 values at two points (1 to 4 years apart) using data from over 5,000 employees across 10 firms and 41 units. Results revealed that within-firm variance in values decreased in 42.5% of cases and increased in 57.5% of cases. Similar results were found when comparing changes at the unit level. Interestingly, the time between value measurements did not appear to affect within-firm changes in variance.
Summary
There is some evidence of homogeneity within organizations on personality and values, whereas there are little or no data concerning homogeneity in goals. Further, ASA theory does not specify “how much” homogeneity is expected to exist, and it is often difficult to assess the degree of homogeneity within firms. Only one study tested this hypothesis by examining changes in homogeneity over time. Several studies used methods that do not appropriately test homogeneity, leaving some uncertainty about the actual level of support.
Organizational Differences Hypothesis
This hypothesis predicts that mean levels of ASA constructs (e.g., personality) will differ between organizations. A total of 16 studies tested this hypothesis, and 93.8% of studies found full or partial support for it. For example, Schaubroeck, Ganster, and Jones (1998) examined scores on the Eysenck Personality Test from 681 employees across five organizations. Regression analyses revealed a significant main effect for organization for 7 of 10 personality dimensions. Schneider, Smith, Taylor, and Fleenor (1998) analyzed MBTI data from 12,739 managers across 142 firms. MANOVA showed a significant organization effect across the four personality dimensions.
As with evidence for within-firm homogeneity, it was not always clear how large or pervasive differences were across firms given the focus on statistical significance. This is particularly the case for studies that analyzed data from many employees and/or organizations and, thus, had the statistical power to detect even small between-firm differences. Some studies reported squared canonical correlations, which reflect the variance in the ASA constructs due to organizational membership (Schneider et al., 1998). For instance, in Satterwhite, Fleenor, Braddy, Feldman, and Hoopes (2009), the summed canonical correlation was .12, which indicates that 12% of the variance in personality resided at the organization level. Eight studies reported ICC(1), and we calculated the mean value within and across these studies. The values ranged from .02 (D’Amato & Michaelides, 2021) to .16 (Oh et al., 2015) with a mean of .07. Thus, on average, organizational membership accounted for only 7% of the variance in the ASA constructs (which were the Big Five in 7 of the 8 studies).
ASA theory predicts that homogeneity occurs at the organization level, and several studies compared the amount of firm-level variance in ASA constructs to variance at other levels. For example, Ployhart et al. (2006) analyzed data on four of the Big Five factors from 9,603 employees in 85 jobs across 12 firms. Job-level differences accounted for an average of 21.3% of the variance across the Big Five, whereas organizational-level differences accounted for an average of 13.8%. Similarly, Satterwhite et al. (2009) reported slightly larger canonical correlations for occupation-level compared to firm-level differences (.15 vs. .12, respectively).
Two studies reported information to compare the variance in ASA constructs at the organization versus individual levels. King, Ott-Holland, Ryan, Huang, Wadlington, and Elizondo (2017) analyzed data for three of the Big Five factors from a sample of nearly 24,000 employees across 40 organizations. Analyses based on ICC(1) revealed that organization membership accounted for only 2% to 4% of the variance across the personality scales (and occupation accounted for only 3% to 6% of the variance). We calculated ICC(1) for the residual variance (which reflects variance due to individuals and other unmodeled sources of variance), which was 91% to 93%. D’Amato and Michaelides (2021) provided one of the most comprehensive tests of variance across levels. They collected MBTI data from 2,745 managers from 165 firms in 51 industries across 30 countries. Based on ICC(1), approximately 93% to 97% of the variance was at the individual (i.e., manager) level. The remaining 3% to 7% of the variance was due to organizations, industries, and countries.
Summary
Research generally has found evidence that organizations differ in the mean level of ASA constructs, particularly personality. There are little or no data for values or goals. Moreover, this conclusion often is based on significance tests (e.g., F-values from MANOVA) that possess substantial statistical power due to large number of firms and/or employees within each firm. When studies report effect size information (e.g., ICCs), the amount of variance due to organization often is small (M = 7% in our data set). A couple of studies that partitioned variance across various levels found that nearly all the variance was at the individual level. Thus, although organizations do vary on ASA constructs, the effect of organizational membership appears to pale in comparison to between-employee variability in the constructs. This raises questions about the extent to which variance in ASA constructs resides at the organization versus other levels.
Effects of Homogeneity Hypothesis
This final hypothesis focuses on the effects of within-firm homogeneity on other organizational outcomes. As discussed, Schneider (1987) did not clearly predict specific outcomes. Schneider et al. (1995) later suggested that homogeneity may have positive effects, such as higher job satisfaction, better communication and cooperation, and fewer interpersonal conflicts. We found only two studies that related homogeneity to such outcomes, both of which focused on job satisfaction. Ployhart et al. (2006) found that homogeneity in all four Big Five factors they examined was unrelated to employee-level satisfaction. In a sample of 71 South Korean firms, Oh et al. (2015) found that greater homogeneity (i.e., smaller SDs) in emotional stability and extraversion was related to higher firm-level managerial satisfaction. Controlling for firm-level mean personality, regression coefficients were −.34 for emotional stability and −.25 for extraversion. Although the other three Big Five factors also followed this trend, the relations were not significant.
Schneider et al. (1995) also suggested the homogeneity may have negative effects, including “an inability to change when the environment demands and, concomitantly, the demise of competitiveness through unchanging and easily predictable decision making” (765). Four studies related homogeneity to organizational effectiveness. Shepherd (1996) examined firm-employee congruence on six values (e.g., customer orientation, innovation, teamwork) across 32 locations of a manufacturing firm in relation to unit turnover and financial performance. Values congruence at the unit level was negatively related to turnover (r = −.64) and positively related to financial performance outcomes, such as sales (r = .28) and return on assets (r = .55). The results are somewhat limited by the small sample size (i.e., 32 units). Additionally, the author measured congruence between employees’ perceptions of the organization’s values and employees’ enactment of those values (i.e., espoused values) rather than the extent to which employees shared the firm’s values.
The other three studies focused on the Big Five factors. Ployhart et al. (2006) related homogeneity to individual-level job performance and found that that greater homogeneity in emotional stability was positively related to performance. Homogeneity in agreeableness, conscientiousness, and extraversion also followed this trend but the relations to performance were not significant. Similarly, Oh et al. (2015) found greater homogeneity in emotional stability was related to higher firm labor productivity and financial performance as measured by return on equity (ROE). Homogeneity in extraversion also was positively related to these outcomes. For example, the effects of within-firm SDs for emotional stability on labor productivity and ROE one year later (controlling for mean emotional stability) were −.22 and −.30, respectively. In contrast, greater homogeneity in agreeableness was related to lower firm ROE (b = .25). Using data from 167 firms, Schneider and Bartram (2017) related homogeneity in the Big Five to financial performance as measured by return on investment (ROI) and return on assets (ROA). Except for extraversion, all within-firm SDs correlated significantly with firm performance. Observed rs ranged from −.17 for conscientiousness to −.23 for emotional stability and suggested that greater homogeneity was related to better performance.
These studies also tested moderators of the effects of homogeneity on performance. First, Oh et al. (2015) explored the breadth of the ASA construct as a potential moderator. Specifically, they examined relations between homogeneity and firm performance for each Big Five factor and six facets of each factor. Although there was some variance across facets (e.g., rs = −.25 vs. −.09 for the anxiety and self-conscientiousness facets of emotional stability), facet SDs did not predict labor productivity or ROE to a significantly greater extent than the SDs of their respective broad factor (e.g., r = −.21 for emotional stability).
Second, all three studies tested whether mean levels of the Big Five moderated relations between homogeneity and organizational effectiveness. Oh et al. (2015) found evidence of moderation for three of the Big Five factors. For example, the effects of homogeneity in emotional stability on productivity and ROE depended on firms’ mean levels of emotional stability. Specifically, the positive effects of homogeneity on outcomes were significant only when firm personality means also were high. That is, firm performance was highest when managers were uniformly high in emotional stability and not when they were uniformly low in this trait. The case was reversed for openness, such that a high firm mean accentuated the positive effect of heterogeneity on labor productivity. Thus, firm performance was best when employees were generally high but varied in levels of openness. Ployhart et al. (2006) found the same pattern for emotional stability. Specifically, a significant SD by mean interaction suggested that individual performance was the highest when employees were consistently high in this trait. Finally, Schneider and Bartram (2017) conducted similar analyses but did not find evidence that firm-level means of the Big Five moderated relations between firm-level SDs and performance.
Summary
No studies examined outcomes that Schneider et al. (1995) suggested homogeneity might positively impact, such as coordination or cohesion. Further, contrary to Schneider (1987), who suggested that homogeneity “can be dangerous to long-term organizational health” (446), all four studies that related homogeneity to employee or firm performance found positive effects for homogeneity in personality. Thus, rather than harming organizational effectiveness, homogeneity tends to be associated with better outcomes. Interestingly, in three studies, homogeneity in emotional stability was the most strongly related to performance. Moreover, in both Oh et al. (2015) and Ployhart et al. (2006), the highest levels of performance occurred when there was a high mean and low variance in emotional stability.
These studies focused on personality, so we do not know whether similar or different results might be found for values or goals. Moreover, although Oh et al. (2015) and Schneider and Bartram (2017) included a lag between personality and the outcomes (Ployhart et al. [2006] used a cross-sectional design), neither study tested whether changes in homogeneity related to changes in outcomes. In this sense, no studies have tested this hypothesis, nor the possibility that homogeneity may be beneficial in the early stages of a firm but eventually become problematic.
Discussion and Implications
Nearly 40 years ago, Schneider (1987) suggested that “the people make the place.” Since then, thousands of studies have cited ASA, and researchers in fields such as psychology, management, economics, and education have used the theory as a basis for their work. Several factors may contribute to the popularity of the theory. First, ASA is a broad theory that can be applied to many literatures and research questions. Second, ASA ties together the HR practices of recruitment, selection, and retention into a single framework. Third, the main ideas of the theory are very intuitive, and its pithy title (“the people make the place”) attracts attention. Overall, it is easy to see why so many studies refer to ASA when discussing why people are attracted to, are selected by, and leave organizations or other entities.
Although widely cited, direct tests of ASA hypotheses are surprisingly sparse. Our purpose was to critically and systematically review ASA research to take stock of the empirical evidence for its core hypotheses. Table 4 summarizes our main conclusions. Figure 4 provides a model of the relationships ASA theory predicts, the level of support based on our review, and potential moderators to be explored in future research.
Main Conclusions From Review and Directions for Future Research
Model of ASA Theory Relationships, Existing Evidence, and Moderators to Examine in Future Research
At the start of this article, we noted that ASA theory is largely unchanged since its introduction. We then asked whether the theory is unchanged because it has received strong support or because it has not been adequately tested. The answer appears to be somewhere in between. On the one hand, some aspects of the theory have been tested and supported. First, in support of the ASA processes hypothesis, people tend to be attracted to, selected by, and remain with organizations that fit their values. Most of this evidence is based on research relating measures of perceived PO fit with applicant attraction, job acceptance, and actual or intended turnover. Second, in support of the homogeneity hypothesis, there is some evidence for homogeneity within firms, particularly on personality traits. Third, there is evidence of between-firm mean differences in ASA constructs, which supports the organizational differences hypothesis. However, the magnitude of differences between organizations typically is modest.
On the other hand, some of the central hypotheses of ASA have not been adequately tested. Specifically, only two studies directly tested the founders hypothesis, and only two other studies directly tested whether ASA processes lead to homogeneity over time. Thus, there is almost no evidence for what Schneider et al. (1995) referred to as the two core propositions of the theory. In addition, only a few studies tested whether homogeneity in ASA constructs relates to firm-level outcomes. In all cases, homogeneity was associated with better firm performance, which is contrary to the general implication of ASA theory that homogeneity is detrimental to the long-term health of organizations. Moreover, none of the studies examined the effects of homogeneity longitudinally, which is assumed in ASA theory. In fact, many ASA studies employed designs (e.g., cross-sectional designs, same source data) that limited their ability to draw inferences about ASA’s temporal or causal hypotheses.
Several factors may contribute to the dearth of direct tests of the theory. First, many aspects of ASA theory are challenging to test. For example, testing the founders hypothesis requires identifying company founders, obtaining data from them, as well as from applicants or employees in the firms they founded. Directly testing the ASA processes and homogeneity hypotheses requires longitudinal data, and testing the between-firm differences hypothesis requires data from multiple organizations and a representative sample of employees within each firm. Further, testing the effects of the homogeneity hypothesis requires data on multiple organizations, representative samples within each firm, and relevant criterion measures. Methodological challenges such as these likely explain why most ASA research has used cross-sectional designs and individual-level data to test processes such as attraction. Second, ambiguity about the specific constructs on which ASA processes operate likely has contributed to the fact that most research citing ASA has focused on constructs other than personality, values, and goals (see Table 2). Third, many studies cite ASA as a foundational theory but then develop their hypotheses using theory more directly relevant to the specific research question. That is, ASA sets a broad foundation (e.g., ASA processes create homogeneity) upon which theories tailored to a specific phenomenon (e.g., PO fit) or context (e.g., recruitment) are built and tested. As a result, ASA is often cited as a theoretical backdrop but is less frequently the focus of investigation.
To highlight this latter possibility, Table 5 presents the related theories we identified and their overlap with ASA. For example, organizational demography (Pfeffer, 1983) and upper echelons theories (Hambrick & Mason, 1984) are like ASA in their focus on the organization level. Organizational demography is perhaps the most like ASA given its focus on similarity (i.e., homogeneity). However, this theory focuses on demographics rather than on personality, values, and goals. Upper echelons theory proposes that the composition and actions of TMTs influence the entire organization, including its culture and performance. This proposition is similar to the founders hypothesis. The other theories focus on different levels, including occupations (Holland, 1959), jobs (Dawis & Lofquist, 1984; McCormick, Jeanneret, & Mecham 1972), groups (Tsui & O’Reilly, 1989), dyads (Byrne, 1971), and individuals (Tajfel, 1978).
Theories That Make Predictions Similar to ASA Theory
Like ASA, several theories hypothesize that people will be attracted to entities that match their characteristics. For instance, Holland’s theory focuses on fit with occupations based on interests, the gravitational model focuses on fit with jobs based on abilities, and relational demography theory focuses on fit with dyads or groups based on demographics. Similarly, several of the theories refer to attrition based on a lack of fit (e.g., the gravitational model). Most or all the other theories predict positive outcomes from homogeneity, such as better communication and perceptions of fairness. This dovetails with Schneider et al.’s (1995) suggestion that homogeneity may be helpful early in an organization’s development. However, these theories diverge from ASA, which suggests that over time, homogeneity can lead to stagnation and lower effectiveness.
In sum, the methodological challenges, ambiguities in predictions surrounding the focal constructs, and overlap with other theories likely have led researchers to cite ASA theory without directly testing it. Ironically, the research ASA has stimulated has siphoned attention away from testing ASA. That is, supporting evidence specific to a phenomenon or literature (e.g., PO fit and recruitment) is likely taken as support for ASA. This is reasonable when ASA and another theory address the same issue (e.g., attraction). Yet, it leads to uneven coverage of ASA (e.g., lack of research on founders’ role) and potential over-estimation of support (e.g., assuming individual-level findings will generalize to the firm level). This scenario is not unique to ASA theory. For example, Keeler, Kong, Dalal, and Cortina (2019) reviewed research on Mischel’s (1973) theory of situational strength. Like our study, they found relatively few examples where research tested the theory in the manner originally proposed. Further, of the research that actually tested the theory, the effects were often small and in the opposite direction as predicted. We hope the present review provides a useful reminder of the importance of a very clear understanding of what a theory predicts and the level and consistency of empirical support for the theory.
Directions for Future Research
Our review led us to conclude that ASA is a useful theoretical framework that still has the potential to stimulate future research. However, to be a useful explanatory theory will require greater precision and additional empirical support for its core propositions. As noted, ASA is a firm-level theory based on a longitudinal process of homogeneity emergence. Therefore, we suggest research prioritize studying and testing ASA in the way it was intended. This will require multilevel research that examines how individual-level personality, values, or goals transform into a firm-level construct, as well as longitudinal research that examines how ASA processes contribute to this transformation. In addition, research should prioritize linking firm-level homogeneity to firm-level outcomes (i.e., the homogeneity, between-firm differences, and effects of homogeneity hypotheses) before focusing on antecedents of homogeneity. That is, we believe it is important for research to first demonstrate whether (and under what conditions) “homogeneity matters” for firm outcomes before attempting to identify potential antecedents of emergence (i.e., the founders hypotheses and ASA processes hypothesis). We now consider specific research directions related to each ASA hypothesis (see also Table 4), as well as connections with adjacent theories and literatures that may provide opportunities for theoretical pruning and future research.
Homogeneity Hypothesis
Given the centrality of homogeneity to ASA, we believe this topic should be a main priority for future research. First, ASA theory does not specify how much within-firm homogeneity should be expected. The field lacks sufficient evidence to provide a reasonable baseline, so a starting point for future research will be to conduct descriptive studies that document the degree of homogeneity on different ASA constructs. For instance, such data may be found in consulting organizations that use the same (or highly similar) assessments for multiple client organizations (e.g., Ployhart et al., 2006). It will be important for this research to be conducted longitudinally whenever possible, as homogeneity may fluctuate over time (e.g., periods of high worker mobility that occurred following the COVID pandemic).
Second, future studies should address the level(s) at which homogeneity is the strongest. ASA focuses on organizational-level homogeneity. However, multilevel principles suggest that composition-based homogeneity should be strongest at the level where it originates and weaken as one moves to farther levels because levels that are closer to each other will be more tightly connected (Klein & Kozlowski, 2000). For example, homogeneity may be strongest in a TMT and weaken as one moves to lower levels in the organization. Alternatively, homogeneity may be strong in lower-level teams and weaken at higher levels.
Third, we need to know much more about the constructs on which homogeneity is likely to emerge. Most research has focused on personality; we know much less about firm-level homogeneity regarding values and goals. This is an interesting gap, because PO fit research—which draws from the ASA theory—has found strong evidence that value congruence between applicants or employees and organizations relates to ASA processes such as attraction and attrition (Edwards & Cable, 2009). This suggests the possibility of high levels of within-firm homogeneity in values. Future research also needs to investigate the level at which ASA processes and homogeneity on a construct domain operate. As an example, perhaps personality primarily operates at the group level, values at the firm level, and interests at the occupation level. There also could be differences within ASA construct domains. For instance, Oh et al. (2018) found that changes in homogeneity were greater for extraversion than for the other Big Five traits. Future research may examine how factors such as the relative observability of a trait (e.g., Funder, 1995) influence the processes of attraction and selection.
Fourth, although ASA theory does not directly address demographic characteristics, future research needs to bridge ASA with diversity research. A particularly important question is whether ASA processes on personality, values, and goals contribute to homogeneity (i.e., lack of diversity) on characteristics such as race/ethnicity, sex, and age. The answer to this question will depend on organizational practices in recruiting and selection and ways in which fit regarding the ASA constructs are assessed in those processes. For example, organizations may select applicants based on individual attributes congruent with organizational characteristics as indicated by scores on personality or value assessments. If so, one would not expect a substantial impact of the selection process on demographic diversity given the relatively small subgroup differences on those constructs (e.g., Roth, Van Iddekinge, DeOrtentiis, Hackney, Zhang, & Buster, 2017). However, ASA processes often are influenced more heavily by perceived than by objectively assessed and calculated similarities on the ASA constructs. For instance, similar-to-me bias can influence recruiter perceptions of applicants and subsequent decisions (Rynes, 1991). If so, demographic diversity would be reduced as a by-product of the increasing homogeneity from ASA processes. Conversely, research on targeted recruiting (e.g., Newman & Lyon, 2009) has shown that organizations can attract minority applicants and enhance, rather than restrict, diversity by highlighting personality traits and values that the firms desires and that favor minority candidates (e.g., the achievement and dependability facets of conscientiousness; Foldes, Duehr, & Ones, 2008).
When diversity itself is a goal or value of the organization, we expect firms to attract and select job candidates who similarly value diversity. Further, employees whose diversity values do not match the values of the organization will be more likely to attrit. Research has shown that diversity practices can increase organizational attractiveness to minority candidates and help retain minority employees, though effectiveness of these practices depends on factors such as the explanation (or the lack thereof) given for the practice (for reviews, see Avery & McKay, 2006; Roberson, 2019). In sum, the relationship between within-firm homogeneity on ASA constructs and demographic diversity is variable and can be influenced by organizational practices. Future research should continue to address how organizations can develop and maintain competitive advantage in demographic diversity (Richards, 2000) through the ASA processes.
Fifth, studies often found variation in homogeneity across organizations. As an example, in Bradley-Geist and Landis (2012), rwg values for the E-I dimension of the MBTI ranged from .53 to .77 across the 25 firms in their sample. In instances where such between-firm variation is observed, there may be value in exploring factors that moderate the degree of homogeneity. For instance, smaller firms may be more susceptible to homogeneity than larger firms (Schneider et al., 1995). Many of the studies we reviewed had data on firm size, so the data exist to test this possibility.
Effects of Homogeneity Hypothesis
We suggest a second priority for future research is to examine the effects of homogeneity on organizations. Like concerns about the “black box” in the strategic HR literature (Becker & Huselid, 2006), ASA research needs to better articulate the proximal and distal outcomes of ASA processes and homogeneity. For instance, does homogeneity affect proximal outcomes such as communication and coordination as Schneider et al. (1995) suggested? In addition, ASA theory implies that changes in homogeneity should relate to changes in organizational outcomes, yet the literature lacks evidence for these change relationships at the within-firm level. One way to stimulate this type of research is to draw from the literature on strategic human capital resources (HCR; Ray, Essman, Nyberg, Ployhart, & Hale, 2023), because it has provided some evidence that changes in HCR contribute to changes in subsequent performance (e.g., Ployhart, Van Iddekinge, & MacKenzie 2011). It also would be interesting to investigate whether the effects of homogeneity are linear or curvilinear. For instance, do the positive effects of homogeneity early in a firm’s development (e.g., better coordination) level off and begin to become negative as the firm becomes more established and has difficulty adapting at very high levels of homogeneity? The literature on organizational life cycles may be useful in answering these questions given its focus on changes organizations experience as they move from one stage of development to the next (Quinn & Cameron, 1983).
Although the effects of homogeneity on outcomes were somewhat modest, there was evidence of between-firm variance in such effects (e.g., Oh et al., 2015). In ASA processes where such variance is observed, it may be useful to identify and test potential moderators of the effects of homogeneity. In this regard, research on team diversity may be informative (e.g., Bell, Villado, Lukasik, Belau, & Briggs, 2011; Joshi & Roh, 2009). This research suggests that the direction and magnitude of the effects of team diversity depend on factors such as the construct (e.g., task- vs. relations-oriented characteristics), the outcome (e.g., cohesion vs. innovation), and the context (e.g., team tenure). In a similar way, homogeneity may be positively related to efficiency but negatively related to innovation and adaptability. The stage of the organization also could play a role. As discussed, homogeneity may benefit newly formed organizations through shared goals and enhanced coordination among the new workforce. On the other hand, established firms that undergo a CEO transition often reset their strategy (Sirmon, Hitt, & Ireland, 2007), suggesting that the homogeneity of the organization should change as well. The level within the firm also could be a moderator. For instance, homogeneity may be detrimental to organizational units responsible for strategic decision making, particularly TMTs. Indeed, Schneider (1987) suggested that homogeneity could lead to suboptimal decision-making due to groupthink.
ASA Processes Hypothesis
There is still much we need to know about the ASA processes themselves. For one, research needs to understand the extent to which ASA processes, both individually and collectively, produce homogeneity. Support for this key proposition largely has been inferred from studies that related PO fit to attraction, selection, or attrition. Only two studies related these ASA processes to homogeneity over time (Meyer, 2008; Oh et al., 2018). Given that most organizations utilize applicant tracking systems, this type of longitudinal data should be available.
Further, ASA proposes that employees who do not fit the values, goals, or personality of others in the organization eventually will leave, which, in turn, will lead to greater homogeneity. This view assumes that employees’ personal characteristics are relatively stable. In contrast, theories such as the Theory of Work Adjustment (Dawis & Lofquist, 1984) and the ASTMA model (attraction, selection, transformation, manipulation, and attrition; Roberts, 2006) suggest that, over time, individuals who do not fit perfectly may adapt to their jobs and organizations rather than choosing to leave. Indeed, recent research has demonstrated the dynamics of PO fit and the capacity of employees to change themselves or their work environment to achieve a better fit (for a review, see Kristof-Brown et al., 2023, Conundrum 6, pp. 401-403). Organizations may also attempt to shape employee values and goals through socialization (e.g., Cable & Parsons, 2001; Chatman, 1991). Even personality traits, previously thought as relative stable dispositions, have been shown to change due to work experience (e.g., Li et al., 2021). Future research should continue to address this issue by measuring ASA constructs over time and comparing variance reduction due to attrition versus changes in employees that make their values, goals, or personality more aligned with the organization.
Between-Firm Differences Hypothesis
The studies we reviewed generally found support for the hypothesis that organizations differ in their collective personalities, values, and goals. However, these differences tended to be modest and of unclear practical importance. Therefore, research must examine the extent to which between-firm differences in ASA constructs create meaningful differences in firm outcomes. In conducting this research, it will be important to carefully theorize the types of firm outcomes the constructs should affect. For example, between-firm differences in personality may create differences in innovation, whereas differences in values may create differences in the composition of TMTs or workforce diversity. Interestingly, strategic management research on resource-based theory (Barney, 1991) has argued for the presence of firm performance differences due to differences in internal resources (which can be homogeneity in personality, values, and goals). Thus, both ASA theory and resource-based theory emphasize that a firm’s internal resources will create organizational differences. However, whereas ASA emphasizes ASA processes and firm performance, strategic management emphasizes factor markets and competitive advantage. Future research should compare organizational differences in ASA constructs (e.g., values) to differences in other types of HCR (e.g., collective knowledge and skills) to determine which most strongly relate to different firm outcomes.
Founders Hypothesis
Finally, future research needs to examine whether and how founders shape ASA processes. The rationale for how founders influence their firms could also apply to CEOs and TMTs, who direct organizational strategy and determine policies that shape ASA processes and create homogeneity. For example, Colbert, Kristof-Brown, Bradley, and Barrick (2008) found that credit unions in which the CEO and TMTs had higher goal congruence demonstrated higher ROA one year later. Thus, examining ASA in the context of TMT and upper echelons research would be an interesting way to advance both ASA and strategic management literatures. Future research opportunities could also arise from connecting ASA with entrepreneurship theory and research. For instance, Brymer and Rocha (2024) used ASA theory as a basis to understand firms founded by entrepreneurial teams. They found that teams who shared affiliations among themselves (e.g., graduated from the same college) were more likely to hire employees who also shared those affiliations, particularly in the early years of the firm. This suggests that founders may influence the diversity (or lack thereof) within the firms they establish.
Implications for Design and Analysis
We conclude by noting some methodological implications of our review. In terms of design considerations, Schneider (1987: 440) proposed that lab studies “mask the display of individual differences” and thus are “inappropriate for studying the relative contributions of traits and situations to understanding behavior” (also see Proposition 1 in our Table 1). Future research might compare lab and field studies to see if they yield similar or different conclusions about ASA hypotheses. Answering this question could provide useful guidance for the designs and settings of future ASA studies, as well as clarify whether lab studies may be able to test some aspects of ASA (which could speed up research testing core ASA predictions).
In terms of data, many studies we reviewed had the data to test ASA hypotheses more rigorously. For example, studies that measured ASA constructs (e.g., values) in job applicants and then post-hire variables (e.g., job performance) could have shed light on the extent to which selection on ASA constructs produces homogeneity. Thus, we suggest future research better utilize available data to test ASA hypotheses. To test whether ASA processes cause homogeneity, studies need to collect data longitudinally as individuals progress from applicants to employees (e.g., Oh et al., 2018).
In terms of analyses, tests of ASA could benefit from more appropriate statistical models. For example, methods such as latent growth modeling (LGM) could provide valuable insights into ASA processes. LGM can assess the degree and nature of homogeneity changes over time, as well as factors that affect or covary with those changes (Ployhart & Vandenberg, 2010). Further, LGM allows estimation of baseline homogeneity via the intercept parameter, and the relative degree of homogeneity over time via the slope parameter. These intercept and slope terms can be related to outcomes to examine the importance of absolute versus relative homogeneity. Moreover, methods such as multi-level LGM could be used to test whether changes in homogeneity vary across team, unit, and/or firm levels. Modeling homogeneity will require researchers to use AD to test homogeneity and ICC(1) to assess the magnitude of the organization effect. In this regard, we encourage researchers to adopt methods like the consensus emergence model (e.g., Lang, Bliese, & de Voogt, 2018) to test how emergence occurs over time. Finally, we hope future research can compare the organization effect to effects at other levels, including units, teams, jobs, and individuals. Models for estimating variance within- and between-levels (e.g., Ployhart, Bliese, & Strizver, 2025) can be used to provide better tests and more informative insights about ASA theory.
Conclusion
ASA continues to be a widely used theoretical lens in many areas of organizational research. However, its core propositions are largely assumed and have not been sufficiently tested. Some of the key hypotheses regarding founders and the effects of ASA processes on homogeneity over time have been tested in only a couple of published articles. This was surprising to us, and we think it may surprise other researchers as well. We believe ASA may still have much to contribute to understanding organizations and the role of people in making them. However, to deliver such contributions, research needs to use methods that more directly test the processes and effects ASA predicts. To more quickly stimulate such research, we highlighted connections between ASA theory and other literatures and theories that seek to understand many of the same phenomena (e.g., homogeneity as a human capital resource, the relation between homogeneity and demographic diversity) and encourage other researchers to do so as well. We hope this review serves as a starting point to create deeper and yet more parsimonious explanations of phenomena across different areas, to not only advance those areas but also to create opportunities to test, extend, and apply ASA in new ways.
Supplemental Material
sj-docx-1-jom-10.1177_01492063251323858 – Supplemental material for Do the People Make the Place?: A 40-Year Review of Research on ASA Theory
Supplemental material, sj-docx-1-jom-10.1177_01492063251323858 for Do the People Make the Place?: A 40-Year Review of Research on ASA Theory by Chad H. Van Iddekinge, Jake T. Harrison, Rong Su and Robert E. Ployhart in Journal of Management
Footnotes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
