View in EDS HTML Full Text PDF Full Text

A Meta-Analysis of the Relationship between Wonderlic Test Scores and School Success

Saved in:
Bibliographic Details
Title: A Meta-Analysis of the Relationship between Wonderlic Test Scores and School Success
Language: English
Authors: Chet Robie (ORCID 0000-0003-4521-6924), Sabah Rashe, Stephen D. Risavy, Piers Steel
Source: International Journal of Testing. 2024 24(2):169-189.
Availability: Routledge. Available from: Taylor & Francis, Ltd. 530 Walnut Street Suite 850, Philadelphia, PA 19106. Tel: 800-354-1420; Tel: 215-625-8900; Fax: 215-207-0050; Web site: http://www.tandf.co.uk/journals
Peer Reviewed: Y
Page Count: 21
Publication Date: 2024
Document Type: Journal Articles
Information Analyses
Education Level: Higher Education
Postsecondary Education
Descriptors: Meta Analysis, Test Validity, Alternative Assessment, Scores, Cognitive Measurement, Cognitive Ability, Academic Achievement, Undergraduate Students, Grade Point Average, College Entrance Examinations, Comparative Analysis, Predictive Validity, Test Preparation, Timed Tests
Assessment and Survey Identifiers: SAT (College Admission Test), ACT Assessment
DOI: 10.1080/15305058.2024.2318424
ISSN: 1530-5058
1532-7574
Abstract: This meta-analysis examined the validity of an alternative to traditional assessments called the Wonderlic which is a brief measure of general mental ability. Our results showed significant, positive correlations between Wonderlic scores and academic performance in general ([r-bar] = 0.26), between Wonderlic scores and undergraduate GPA in particular ([r-bar] = 0.27, [p-bar] = 0.33), and between Wonderlic scores and retention ([r-bar] = 0.09, [p-bar] = 0.12). We also identified several significant moderators of the relationship between Wonderlic scores and relevant outcomes (e.g., test publisher reported coefficients were larger than those reported by other sources). Subgroup differences in test scores were in the same range as other post-secondary admissions assessments (e.g., ACT and SAT scores). Overall, the Wonderlic has similar levels of subgroup differences and is less strongly related to GPA than traditional assessments but still retains useful levels of predictiveness and is a shorter, less expensive assessment that requires less preparation than the ACT or SAT.
Abstractor: As Provided
Entry Date: 2024
Accession Number: EJ1422761
Database: ERIC
Full text is not displayed to guests.
FullText Links:
  – Type: pdflink
    Url: https://content.ebscohost.com/cds/retrieve?content=AQICAHj0k_4E0hTGH8RJwT4gCJyBsGNe_WN95AvKlDbXJGqwxwFA5yyw40slckwgTTIdHL84AAAA4zCB4AYJKoZIhvcNAQcGoIHSMIHPAgEAMIHJBgkqhkiG9w0BBwEwHgYJYIZIAWUDBAEuMBEEDA6X4CjZ3QjDfvUsnAIBEICBm4FbbQ6JJdFDazgEwO87tKX8tslK0l2xAAX0Dm-qCZiBSePc3dkPhZr3-3CIVSCzxC-b3SnOVLusYNVgCPexJ4BAkmhLGJcL3bo3i2jJgZ78E3l1bpzwe3rV4vx4-UGKQ1GnYf-rNsW6_Vn2hTeCKtztlP-YDfyY16KSJkFWOL2QACzBU3LB6aPEELLly4qGu2hkrSCieVLfqF5W
Text:
  Availability: 1
  Value: <anid>AN0176845951;k0301apr.24;2024Apr29.06:04;v2.2.500</anid> <title id="AN0176845951-1">A meta-analysis of the relationship between Wonderlic test scores and school success </title> <p>This meta-analysis examined the validity of an alternative to traditional assessments called the Wonderlic which is a brief measure of general mental ability. Our results showed significant, positive correlations between Wonderlic scores and academic performance in general (r̅ =.26), between Wonderlic scores and undergraduate GPA in particular (r̅ =.27, ρ ¯ =.33), and between Wonderlic scores and retention (r̅ =.09, ρ ¯ =.12). We also identified several significant moderators of the relationship between Wonderlic scores and relevant outcomes (e.g., test publisher reported coefficients were larger than those reported by other sources). Subgroup differences in test scores were in the same range as other post-secondary admissions assessments (e.g., ACT and SAT scores). Overall, the Wonderlic has similar levels of subgroup differences and is less strongly related to GPA than traditional assessments but still retains useful levels of predictiveness and is a shorter, less expensive assessment that requires less preparation than the ACT or SAT.</p> <p>Keywords: Wonderlic; meta-analysis; admissions</p> <p>Beyond high school or undergraduate grades, post-secondary institutions often require their applicants to complete assessments to be admitted into their undergraduate or graduate programs. Many of the popular assessment options (e.g., Scholastic Assessment Test [SAT], The ACT Test [ACT]) have issues regarding the time and cost required for students to complete them as well as the need for domain-specific (i.e., non-general) knowledge. These issues may hinder the assessment of students from lower socioeconomic status (SES) as well as negatively impact the quality of students that higher education institutions are admitting into their programs. Although standardized admissions assessments such as the SAT or ACT are less used in countries outside of the United States, the high knowledge requirements and cost of these assessments (particularly for developing countries) could perpetuate income inequality by limiting educational opportunities for individuals in the lower income brackets. Furthermore, grade inflation (which leads to grades being less able to differentiate among applicants) which was accelerated as a result of the recent COVID-19 pandemic and the recent increased emphasis on equity, diversity, and inclusion (EDI) have introduced new variables into post-secondary admissions. Thus, there is an imperative need to assess alternative admission testing options that may help to differentiate amongst applicants while minimizing subgroup (e.g., visible minorities, Indigenous Peoples, lower SES) differences compared with other popular post-secondary admissions tests (e.g., SAT, ACT).</p> <p>The Wonderlic is an especially quick (50 items; 12 min), general, reliable, and inexpensive test of general mental ability (GMA) that has been used in numerous post-secondary settings. However, the Wonderlic does not appear to be as widely used as the SAT or ACT, either in the United States or in other countries throughout the world. This appears to be a missed opportunity to level the playing field for students around the world who possess a high level of intellectual potential but who are not able to afford to participate in highly enriched primary and secondary educational milieus. We are particularly interested in the Wonderlic because it appears to be marketed for use in post-secondary contexts but unlike the SAT and ACT, no cumulative study has attempted to organize this research to examine the overall ability of this test to predict important post-secondary student outcomes as well as its impact on subgroups in this context. This knowledge gap limits the ability of post-secondary institutions to understand whether the Wonderlic should be part of the solution to the ongoing grade inflation and EDI concerns that are of significant import for all post-secondary institutions. The explicit objectives of this research are to:</p> <p></p> <ulist> <item> Quantitatively examine the validity of the Wonderlic's association with relevant post-secondary student outcomes (e.g., grades, retention); and</item> <p></p> <item> Assess the Wonderlic's subgroup differences with the goal of comparing the Wonderlic's subgroup difference metrics with those of traditional post-secondary admission assessments (e.g., grades, essays, interviews).</item> </ulist> <hd id="AN0176845951-2">Context</hd> <p>Colleges and universities throughout the world need to make decisions on which applicants to admit to their institutions. In order to do so, they employ a myriad of different decision aids, such as standardized scholastic aptitude tests, high school grades, letters of recommendation, personal statements, interviews, personality and interest assessments, situational judgment tests, and biographical data (Kuncel et al., [<reflink idref="bib15" id="ref1">15</reflink>]). Unfortunately, standardized scholastic aptitude tests such as the SAT or ACT often evidence mean differences between majority and some minority groups and those who differ in SES even though the assessments show strong criterion-related validity (Thomas, [<reflink idref="bib27" id="ref2">27</reflink>]). Indeed, some have blamed the testing industry itself for not doing enough to balance predictive validity and social responsibility (Koljatic et al., [<reflink idref="bib14" id="ref3">14</reflink>]). This has led to some universities making those tests optional. At least one study found positive effects on diversity for test-optional policies (Bennett, [<reflink idref="bib2" id="ref4">2</reflink>]); whereas another study found no positive effects of test optional policies on diversity or student quality (Saboe & Terrizzi, [<reflink idref="bib19" id="ref5">19</reflink>]). Some schools that went test-optional have reinstated their standardized test requirement. For example, the Massachusetts Institute of Technology (MIT) reinstated their requirement recently and noted:</p> <p>...we have decided to reinstate our SAT/ACT requirement for future admissions cycles. Our research shows standardized tests help us better assess the academic preparedness of all applicants, and also help us identify socioeconomically disadvantaged students who lack access to advanced coursework or other enrichment opportunities that would otherwise demonstrate their readiness for MIT. (Schmill, [<reflink idref="bib22" id="ref6">22</reflink>])</p> <p>Clearly, there is a debate surrounding the use of which standardized tests to use (if any) in college and university admissions. Our research is designed to examine whether a specific assessment (the Wonderlic) that shares many of the properties of scholastic aptitude tests such as SAT or ACT (but differs in important ways) has similar levels of predictive validity and perhaps smaller subgroup mean differences than these scholastic aptitude tests. We will review the research on college admissions tests in terms of their predictive validity and subgroup differences. Next, we will briefly describe what the Wonderlic measures and our rationale for investigating its possible role in post-secondary admissions. Finally, we will outline how we will investigate whether the Wonderlic may be a useful assessment to use in college admissions.</p> <hd id="AN0176845951-3">Previous research on admissions assessments</hd> <p>Several standardized assessments have been used in the post-secondary admissions process to screen for scholastic aptitude. The most regularly used assessments at the undergraduate level are the SAT and ACT tests. The SAT assesses basic cognitive abilities such as verbal, quantitative, and logical reasoning, whereas the ACT directly assesses knowledge of content areas from high school (which is highly 'g' loaded as well).</p> <p>These assessments share several characteristics. First, research suggests that these tests tap into general mental ability as they all contain subdimensions that are moderately to highly correlated (i.e., if an applicant does well or poorly on one subdimension, they tend to do well or poorly on the other subdimensions; Frey & Detterman, [<reflink idref="bib9" id="ref7">9</reflink>]; Koenig et al., [<reflink idref="bib13" id="ref8">13</reflink>]). Second, many of these tests require a firm understanding of advanced algebra, geometry, and pre-calculus or have a writing component. This could disadvantage applicants from lower quality or poorly resourced secondary or post-secondary institutions and non-native English speakers. Third, all these assessments require applicants to sit for over three hours to complete the test and the test fees range from $60 to over $200 (USD). These factors could lead to a chilling effect for applicants who suffer from disabilities, such as attention deficit hyperactivity disorder or anxiety, and applicants who cannot easily afford the sitting fee.</p> <p>Previous meta-analyses have found many of these tests to be related to academic success as measured by post-secondary grades and retention. Sackett et al. ([<reflink idref="bib20" id="ref9">20</reflink>]) found the correlation between SAT scores and undergraduate GPA to be <emph>r̅</emph> =.35 (</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mrow><mover accent="true"><mi>ρ</mi><mo stretchy="true">¯</mo></mover></mrow></mrow></math> </ephtml> =.47). Westrick et al. ([<reflink idref="bib30" id="ref10">30</reflink>]) found the correlation between ACT scores and 2nd-year cumulative GPA to be <emph>r̅</emph> =.45 (</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mrow><mover accent="true"><mi>ρ</mi><mo stretchy="true">¯</mo></mover></mrow></mrow></math> </ephtml> =.55) and the relation between ACT scores and 3rd-year retention to be <emph>r̅</emph> =.10 (</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mrow><mover accent="true"><mi>ρ</mi><mo stretchy="true">¯</mo></mover></mrow></mrow></math> </ephtml> =.18).</p> <p>Unfortunately, many of these admissions test evidence sex and racial/ethnic subgroup differences or are correlated with SES. For instance, SES has been found to relate to SAT scores (<emph>r̅</emph> =.22; Sackett et al., [<reflink idref="bib20" id="ref11">20</reflink>]). Additionally, the Black-White subgroup difference for the SAT and ACT is approximately one standard deviation and approximately half to three-quarters of a standard deviation for White-Hispanic comparisons (Roth et al., [<reflink idref="bib18" id="ref12">18</reflink>]). Perhaps due to the selection effect, however, the Black-White subgroup difference is lower for college students (<emph>d̅</emph> = 0.69) than for applicants (<emph>d̅</emph> = 0.98); this is likely true for White-Hispanic subgroup differences, although Roth et al. ([<reflink idref="bib18" id="ref13">18</reflink>]) did not report that for college students. Further, Mau and Lynn ([<reflink idref="bib16" id="ref14">16</reflink>]) found that males tended to have higher SAT math (<emph>d</emph> = 0.41), SAT verbal (<emph>d</emph> = 0.14), and ACT (<emph>d</emph> = 0.23) scores than females.</p> <p>It should be noted that admissions assessments are likely not differentially valid—using Cleary's ([<reflink idref="bib5" id="ref15">5</reflink>]) definition—by different levels of SES (Higdem et al., [<reflink idref="bib11" id="ref16">11</reflink>]; Sackett et al., [<reflink idref="bib20" id="ref17">20</reflink>], 2012). For example, when controlling for SES, the relationship between SAT scores and grades reduces from <emph>r̅</emph> =.47 to <emph>r̅</emph> =.44 (Sackett et al., [<reflink idref="bib20" id="ref18">20</reflink>]). There is evidence, however, that some admissions tests may be differentially valid for certain ethnic/racial subgroups. For example, the SAT tends to be more valid for White samples than Black and Hispanic samples (cf. Berry et al., [<reflink idref="bib3" id="ref19">3</reflink>]), although this tends to be reduced when controlling for course-taking patterns and differential range restriction (Dahlke et al., [<reflink idref="bib6" id="ref20">6</reflink>]). Dahlke et al. ([<reflink idref="bib6" id="ref21">6</reflink>]) also found that the SAT was less valid for males than females in predicting college GPA.</p> <hd id="AN0176845951-4">Rationale for studying the Wonderlic in post-secondary admissions</hd> <p>The full-version of the Wonderlic is a timed 12-minute assessment of general mental ability that contains 50 items which incorporate a wide variety of problem types such as patterns, ordering, deductive reasoning, spatial reasoning, synonyms, analogies, word problems, averages, and basic algebra (e.g., rearrange an equation to find the one unknown number from two known numbers; Wonderlic & Hovland, [<reflink idref="bib32" id="ref22">32</reflink>]).[<reflink idref="bib1" id="ref23">1</reflink>] The knowledge requirement to complete the Wonderlic is approximately Grade 6–8 level in comparison to the Grade 10–12 level knowledge requirements of the ACT and SAT (Wonderlic, [<reflink idref="bib31" id="ref24">31</reflink>]) which would lead to a potentially broader applicant pool. The sitting fee is a fraction of the other assessments at $25 (USD). Finally, large-scale admissions decisions often utilize only the overall score from the typical standardized tests instead of their subcomponents. If only overall scores are being used in decision-making, it makes more sense to simply utilize a less expensive, shorter assessment with potentially less saturation with variance related to SES. Despite the Wonderlic's potential for use beyond pre-employment industrial settings, no meta-analysis to date has been conducted to examine its criterion-related validity and potential subgroup differences in a post-secondary admissions context.</p> <hd id="AN0176845951-5">Method</hd> <p></p> <hd id="AN0176845951-6">Literature search</hd> <p>The literature search was conducted in January 2023. We searched Google Scholar, Proquest, Web of Science, and EBSCOhost. Our search terms were "Wonderlic" and "Scholastic Level Exam" (anywhere in the text). Using only these two search terms (and not constraining to the title or abstract of the paper) led to a large number of papers because many were identified that made mention of the Wonderlic or Scholastic Level Exam but either did not measure it or did not have useful variables to code. The initial search after automatic removal of duplications using a web-based meta-analytic data management system called HubMeta (www.hubmeta.com; Steel et al., [<reflink idref="bib25" id="ref25">25</reflink>]) resulted in 2,992 papers in a variety of formats such as peer-reviewed journal articles, unpublished dissertations and theses, technical reports, working papers, and conference papers. A PRISMA chart of the literature search is shown in Figure 1.</p> <p>PHOTO (COLOR): Figure 1. PRISMA Diagram.</p> <hd id="AN0176845951-7">Inclusion and exclusion criteria</hd> <p>Paper titles and abstracts were initially screened by one coauthor to ensure that they: (<reflink idref="bib1" id="ref26">1</reflink>) included a student sample, (<reflink idref="bib2" id="ref27">2</reflink>) had correlations of the Wonderlic with academic performance, gender, or race, (<reflink idref="bib3" id="ref28">3</reflink>) used one of the following versions of the test: WPT, WPT-Q, SLE, or SLE-Q, (<reflink idref="bib4" id="ref29">4</reflink>) were written in English, and (<reflink idref="bib5" id="ref30">5</reflink>) did not focus on specialized samples with reported psychological impairment (e.g., attention deficit disorder). After title screening, we were left with 224 potentially codable articles.</p> <p>Three of the coauthors then independently screened the full texts of the papers to ensure that they met the inclusion criteria and that they included codable data. Papers were excluded for the following reasons: (<reflink idref="bib1" id="ref31">1</reflink>) data were gathered but not reported (e.g., the Wonderlic was administered but correlations with relevant variables were not reported) (<emph>n</emph><bold></bold>=<bold></bold>52), (<reflink idref="bib2" id="ref32">2</reflink>) wrong measures or outcomes (e.g., the Wonderlic was cited in the study but not actually used) (<emph>n</emph><bold></bold>=<bold></bold>23), (<reflink idref="bib3" id="ref33">3</reflink>) population not of interest (e.g., the sample was comprised of employees not students) (<emph>n</emph><bold></bold>=<bold></bold>14), (<reflink idref="bib4" id="ref34">4</reflink>) duplicate studies that were not caught <emph>via</emph> automation (<emph>n</emph><bold></bold>=<bold></bold>7), and (<reflink idref="bib5" id="ref35">5</reflink>) records could not be retrieved (<emph>n</emph><bold></bold>=<bold></bold>5). The kappa inter-rater reliability coefficient for the full text screening was high (.96). The few discrepancies in ratings were discussed among the raters and successfully resolved. A final sample of 123 papers were included in the meta-analysis (see Table S1 in the supplemental online material) of which 81 were peer-reviewed journal or chapter articles, 39 were dissertations or theses, two were working papers, and one was the Wonderlic test manual (which included a large number of different effect sizes). Several studies that we identified did not report the proper statistics for coding purposes; however, the authors provided us with the raw data upon request.</p> <hd id="AN0176845951-8">Data extraction</hd> <p>Data were independently coded by two members of the research team. The inter-rater agreement was 94%. Again, the few discrepancies in coding were discussed among the raters and successfully resolved.</p> <p>Raw correlations and sample sizes were extracted from each article. The following intercorrelations were extracted from the primary studies where available: (<reflink idref="bib1" id="ref36">1</reflink>) White versus nonwhite; (<reflink idref="bib2" id="ref37">2</reflink>) White versus Asian; (<reflink idref="bib3" id="ref38">3</reflink>) White versus Hispanic; (<reflink idref="bib4" id="ref39">4</reflink>) White versus Black; (<reflink idref="bib5" id="ref40">5</reflink>) age; (<reflink idref="bib6" id="ref41">6</reflink>) gender; (<reflink idref="bib7" id="ref42">7</reflink>) socioeconomic status; (<reflink idref="bib8" id="ref43">8</reflink>) Wonderlic scores; (<reflink idref="bib9" id="ref44">9</reflink>) academic performance; and (<reflink idref="bib10" id="ref45">10</reflink>) retention.</p> <p>A variety of moderator variables were also extracted from the primary studies where available: (<reflink idref="bib1" id="ref46">1</reflink>) Wonderlic sample mean and standard deviation [<emph>k</emph><bold></bold>=<bold></bold>66 with both <emph>M</emph>and<emph>SD</emph> and <emph>k</emph><bold></bold>=<bold></bold>11 with <emph>M</emph>only]; (<reflink idref="bib2" id="ref47">2</reflink>) date published (pre-2000 [<emph>k</emph><bold></bold>=<bold></bold>29], 2000-present [<emph>k</emph><bold></bold>=<bold></bold>62]); (<reflink idref="bib3" id="ref48">3</reflink>) source (peer-reviewed [<emph>k</emph><bold></bold>=<bold></bold>52], dissertation/thesis [<emph>k</emph><bold></bold>=<bold></bold>24], manual [<emph>k</emph><bold></bold>=<bold></bold>14], working paper [<emph>k</emph><bold></bold>=<bold></bold>1]); (<reflink idref="bib4" id="ref49">4</reflink>) program of study (Art [<emph>k</emph><bold></bold>=<bold></bold>1], Business [<emph>k</emph><bold></bold>=<bold></bold>21], Computer Science [<emph>k</emph><bold></bold>=<bold></bold>1], Engineering [<emph>k</emph><bold></bold>=<bold></bold>2], Interdisciplinary [<emph>k</emph><bold></bold>=<bold></bold>29], Physical Therapy [<emph>k</emph><bold></bold>=<bold></bold>1], Psychology [<emph>k</emph><bold></bold>=<bold></bold>20], Vocational [<emph>k</emph><bold></bold>=<bold></bold>16]); (<reflink idref="bib5" id="ref50">5</reflink>) Wonderlic language version (English [<emph>k</emph><bold></bold>=<bold></bold>85], Spanish [<emph>k</emph><bold></bold>=<bold></bold>5], Chinese [<emph>k</emph><bold></bold>=<bold></bold>1]); (<reflink idref="bib6" id="ref51">6</reflink>) country in which study was conducted (United States [<emph>k</emph><bold></bold>=<bold></bold>69], Canada [<emph>k</emph><bold></bold>=<bold></bold>12], UK [<emph>k</emph><bold></bold>=<bold></bold>5], China [<emph>k</emph><bold></bold>=<bold></bold>2], Germany [<emph>k</emph><bold></bold>=<bold></bold>1], Nigeria [<emph>k</emph><bold></bold>=<bold></bold>1], multiple countries [<emph>k</emph><bold></bold>=<bold></bold>1]); (<reflink idref="bib7" id="ref52">7</reflink>) type of academic performance (exam grade [<emph>k</emph><bold></bold>=<bold></bold>8], single course grade [<emph>k</emph><bold></bold>=<bold></bold>7], average grade across multiple courses [<emph>k</emph><bold></bold>=<bold></bold>76]); (<reflink idref="bib8" id="ref53">8</reflink>) percentage male (<emph>k</emph><bold></bold>=<bold></bold>65); (<reflink idref="bib9" id="ref54">9</reflink>) percentage White (<emph>k</emph><bold></bold>=<bold></bold>34); (<reflink idref="bib10" id="ref55">10</reflink>) percentage Black (<emph>k</emph><bold></bold>=<bold></bold>25); (<reflink idref="bib11" id="ref56">11</reflink>) percentage Hispanic (<emph>k</emph><bold></bold>=<bold></bold>21); (<reflink idref="bib12" id="ref57">12</reflink>) percentage Asian (<emph>k</emph><bold></bold>=<bold></bold>22); (<reflink idref="bib13" id="ref58">13</reflink>) Wonderlic length version (short-form [<emph>k</emph><bold></bold>=<bold></bold>6], long-form [<emph>k</emph><bold></bold>=<bold></bold>85]); and (<reflink idref="bib14" id="ref59">14</reflink>) educational level (post-secondary [<emph>k</emph><bold></bold>=<bold></bold>83], graduate [<emph>k</emph><bold></bold>=<bold></bold>5], both [<emph>k</emph><bold></bold>=<bold></bold>3]). Means and standard deviations for continuous moderators can be found in Table S2 in the supplemental online material. Participant age was not used as a moderator as its variance was highly constrained—the median participant age across 66 codable studies was 20.9 with an interquartile range of 4.1.</p> <hd id="AN0176845951-9">Data analytic strategy</hd> <p>In the case of SES, where studies included multiple measures of the construct (e.g., family income, mother's education, father's education), we computed composite scores (Schmidt & Hunter, [<reflink idref="bib21" id="ref60">21</reflink>]; as implemented in HubMeta). This approach appropriately accounts for dependencies inherent in using multiple measures of the same construct within a given study.</p> <p>Raw correlations were used as input. We quantified heterogeneity and uncertainty in effect sizes using <emph>τ</emph> estimated restricted maximum likelihood (REML), confidence intervals, and prediction intervals. Sampling variance was computed with the usual large-sample approximation but plugging in the sample-size weighted average of the (bias-corrected) correlation coefficients into the equation. We used the Knapp and Hartung ([<reflink idref="bib12" id="ref61">12</reflink>]) method to adjust standard errors when testing individual coefficients. Publication bias was examined using a funnel plot and Egger's regression test (Sterne & Egger, [<reflink idref="bib26" id="ref62">26</reflink>]). To identify outliers and influential studies, we used a variety of outlier and influence case diagnostics. The primary meta-analyses were performed using the Hedges and Olkin ([<reflink idref="bib10" id="ref63">10</reflink>]) approach with the <emph>metafor</emph> package in R (version 4.0-0; Viechtbauer, [<reflink idref="bib29" id="ref64">29</reflink>]) and HubMeta (https://hubmeta.com/).</p> <p>As a supplemental set of analyses, we used the <emph>psychmeta</emph> package in R (Dahlke & Wiernik, [<reflink idref="bib7" id="ref65">7</reflink>]) to correct the raw correlations for artifacts using the Hunter and Schmidt artifact distribution method (Schmidt & Hunter, [<reflink idref="bib21" id="ref66">21</reflink>]).[<reflink idref="bib2" id="ref67">2</reflink>] We conducted these supplemental analyses so that direct comparisons could be made to corrected average sample size-weighted correlations from meta-analyses that have examined other predictors of academic achievement (e.g., SAT, ACT). Corrections for unreliability in the predictor were not utilized because we were interested in the operational validity of the Wonderlic.[<reflink idref="bib3" id="ref68">3</reflink>] Corrections for range restriction were implemented by using the average standard deviation of population norms for the Wonderlic of high school graduates (ages 16–30) (<emph>SD</emph><bold></bold>=<bold></bold>6.8) and college graduates (ages 20–30) (<emph>SD</emph><bold></bold>=<bold></bold>6.3) (Wonderlic, [<reflink idref="bib31" id="ref69">31</reflink>], p. 36) as the unrestricted (i.e., unselected) standard deviation (i.e., <emph>SD</emph><bold></bold>=<bold></bold>6.55) and the sample standard deviations from the meta-analytic samples as the restricted (i.e., selected) standard deviations.[<reflink idref="bib4" id="ref70">4</reflink>] The average ratio of restricted standard deviations to unrestricted standard deviations (i.e., U values) in our artifact distribution (<emph>k</emph><bold></bold>=<bold></bold>66) was.81. Corrections for unreliability in GPA were made by using the distribution of mean estimates of stepped-up GPA composites from Beatty et al. ([<reflink idref="bib1" id="ref71">1</reflink>], p. 36). We matched all nine of the estimates in Beatty et al. ([<reflink idref="bib1" id="ref72">1</reflink>], p. 36) (three of which were cumulative GPA and six of which were first-year GPA) to our distribution of GPA because the majority of our estimates were cumulative GPA (<emph>k</emph><bold></bold>=<bold></bold>58) but some were first-year GPA (<emph>k</emph><bold></bold>=<bold></bold>10) which tend to be lower (see Beatty et al., [<reflink idref="bib1" id="ref73">1</reflink>], p. 36). The average GPA correction in our study was.92.[<reflink idref="bib5" id="ref74">5</reflink>] Finally, we also corrected the correlations of Wonderlic scores with retention, gender, and Black versus White for range restriction in Wonderlic scores. Note that we used indirect range restriction corrections (Case IV; see Carretta & Ree, [<reflink idref="bib4" id="ref75">4</reflink>]) because it is possible that for a small minority of the institutions that are reflected in our meta-analytic database, range restriction occurred directly through the use of the Wonderlic to make admissions decisions. However, range restriction likely occurred indirectly through use of the ACT or SAT to make admission decisions in the overwhelming majority of the studies reflected in our meta-analytic database.</p> <hd id="AN0176845951-10">Results</hd> <p></p> <hd id="AN0176845951-11">Meta-analytic correlation Matrix</hd> <p>We first computed a meta-analytic correlation matrix using the functionality provided by HubMeta, results of which can be found in Table 1. Other than correlations of Wonderlic scores with academic performance and retention (which we will discuss below), several correlations of Wonderlic scores with other variables were of interest. First, the correlations of Wonderlic scores with several ethnicity variables found that Whites scored higher than nonwhites (<emph>r̅</emph> = −0.22, <emph><emph>d̅</emph></emph><bold></bold>=<bold></bold>−0.45), Whites and Asians scored similarly to one another (<emph>r̅</emph> =.05 [95% CI = −0.08,.18], <emph><emph>d̅</emph></emph> =.10), Whites scored higher than Hispanics (<emph>r̅</emph> = −0.23, <emph><emph>d̅</emph></emph> =<bold></bold>−0.47), and Whites scored higher than Blacks (<emph>r̅</emph> = −0.32, <emph><emph>d̅</emph></emph><bold></bold>=<bold></bold>−0.68). Second, men tended to score slightly higher than women on the Wonderlic (<emph>r̅</emph> =.11, <emph><emph>d̅</emph></emph> =.22). Third, age was largely uncorrelated with Wonderlic scores (<emph>r̅</emph> =.04 [95% CI = −0.01,.08], <emph>d̅</emph> =.08). Finally, SES was positively correlated with Wonderlic scores (<emph>r̅</emph> =.20) such that individuals who were from higher SES backgrounds tended to score higher on the Wonderlic.</p> <p>Table 1. Meta-analytic correlation matrix.</p> <p> <ephtml> <table><thead><tr><td /><td>N_W</td><td>A_W</td><td>H_W</td><td>B_W</td><td>Retention</td><td>SES</td><td>Age</td><td>Gender</td><td>AP</td><td>Wonderlic</td></tr></thead><tbody valign="top"><tr><td>N_W</td><td /><td>—</td><td>—</td><td>—</td><td char=".">1(2,450)</td><td>—</td><td char=".">3(705)</td><td char=".">6(3,889)</td><td char=".">1(2,450)</td><td char=".">6(3,889)</td></tr><tr><td>A_W</td><td>—</td><td /><td>—</td><td>—</td><td>—</td><td>—</td><td>—</td><td>—</td><td>—</td><td char=".">1(227)</td></tr><tr><td>H_W</td><td>—</td><td>—</td><td /><td>—</td><td>—</td><td char=".">1(343)</td><td char=".">1(343)</td><td char=".">1(343)</td><td char=".">1(343)</td><td char=".">4(1,981)</td></tr><tr><td>B_W</td><td>—</td><td>—</td><td>—</td><td /><td>—</td><td char=".">1(207)</td><td char=".">4(777)</td><td char=".">6(1,140)</td><td char=".">2(301)</td><td char=".">12(3,673)</td></tr><tr><td>Retention</td><td char=".">−.08[–.12,–.04]</td><td>—</td><td>—</td><td>—</td><td /><td>—</td><td>—</td><td char=".">1 (2,450)</td><td char=".">1 (2,450)</td><td char=".">7 (2,974)</td></tr><tr><td>SES</td><td>—</td><td>—</td><td char=".">−.32 [–.41,–.22]</td><td char=".">−.32 [–.44,–.20]</td><td>—</td><td /><td char=".">3(744)</td><td char=".">3 (744)</td><td char=".">2 (537)</td><td char=".">3 (744)</td></tr><tr><td>Age</td><td char=".">−.07[–.15,.01]</td><td>—</td><td char=".">−.23[–.33,–.13]</td><td char="."><bold>.02</bold>[–.12,.16]</td><td>—</td><td char="."><bold>.08</bold>[–.28,.43]</td><td /><td char=".">32(6,714)</td><td char=".">13(3,025)</td><td char=".">37(8,125)</td></tr><tr><td>Gender</td><td char=".">−.03[–.09,.03]</td><td>—</td><td char=".">.07[–.04,.18]</td><td char="."><bold>−.06</bold>[–.20,.07]</td><td char=".">.10[.06,.14]</td><td char="."><bold>.00</bold>[–.16,.16]</td><td char="."><bold>.06</bold>[.02,.10]</td><td /><td char=".">16(6,185)</td><td char=".">63(26,405)</td></tr><tr><td>AP</td><td char=".">−.15[–.19,–.11]</td><td>—</td><td char=".">−.16[–.26,–.06]</td><td char=".">−.20[–.42,.02]</td><td char=".">.47[.44,.50]</td><td char="."><bold>.20</bold>[–.01,.41]</td><td char="."><bold>.01</bold>[–.09,.10]</td><td char="."><bold>.00</bold>[–.10,.11]</td><td /><td char=".">91(19,317)</td></tr><tr><td>Wonderlic</td><td char="."><bold>−.22</bold>[–.30,–.14]</td><td char=".">.05[–.08,.18]</td><td char="."><bold>−.23</bold>[–.34,–.13]</td><td char="."><bold>−.32</bold>[–.40,–.24]</td><td char="."><bold>.09</bold>[–.17,.35]</td><td char="."><bold>.20</bold>[.05,.35]</td><td char="."><bold>.04</bold>[–.01,.08]</td><td char="."><bold>.11</bold>[.03,.20]</td><td char="."><bold>.26</bold>[.22,.29]</td><td /></tr></tbody></table> </ephtml> </p> <p>1 <emph>Note</emph>. Above the diagonal are the total number of correlations that comprise the average followed by the total sample sizes in parentheses. Missing correlates are denoted by dashes. N_W (White 0 versus nonwhite 1). A_W (White 0 versus Asian 1). H_W (White 0 versus Hispanic 1). B_W (White 0 versus Black 1). Retention (did not graduate 0 versus graduated 1). SES (socioeconomic status). Age (age in years). Gender (female 0 versus male 1). AP (academic performance). Wonderlic (Wonderlic test score). 95% confidence intervals in brackets. Bolded correlations denote significant (<emph>p</emph> <.05) heterogeneity using the <emph>Q</emph> homogeneity statistic.</p> <hd id="AN0176845951-12">Publication bias</hd> <p>We only examined publication bias and influence and outlier diagnostics for the relationship between Wonderlic scores and academic performance (<emph>k</emph><bold></bold>=<bold></bold>91) because of the relatively few correlations that we were able to locate for the relationship between Wonderlic scores and retention (<emph>k</emph><bold></bold>=<bold></bold>7). Egger's regression test for funnel plot asymmetry was non-significant (<emph>t</emph>[<reflink idref="bib89" id="ref76">89</reflink>] = 1.43, <emph>p</emph> =.16) and the funnel plot appeared to be fairly symmetrical (see Figure S3 in the supplemental online material) suggesting a lack of publication bias. This apparent lack of publication bias is not surprising given that over one-quarter of the correlation coefficients that we coded for in the relationship between Wonderlic scores and academic performance were sourced from unpublished dissertations and theses.[<reflink idref="bib6" id="ref77">6</reflink>]</p> <hd id="AN0176845951-13">Outlier and influence case diagnostics</hd> <p>Outlier and influence case diagnostic plots can be found in Figure S4 in the supplemental online material. One estimate was consistently identified across diagnostics—this was case number 37 (as illustrated in the plots) (<emph>r</emph> =.18, <emph>n</emph><bold></bold>=<bold></bold>2,450) which accounted for 13% of the total meta-analytic sample size. We computed the overall meta-analytic relationship between Wonderlic scores and academic performance without this estimate which only led to a.01 increase in average correlation (<emph>r̅</emph> =.26 to <emph>r̅</emph> =.27). We chose not to delete this outlier given that we derived this estimate from the primary data (and therefore were confident it was not a transcription error) and that its inclusion did not overly influence the results from an effect size or heterogeneity perspective.</p> <p>Several other outliers were identified (case numbers 80, 81, 82, 83, and 84) with correlations ranging from.47 to.70. These estimates were all sourced from the publisher's manual and were not likely influential as their combined sample size was <emph>N</emph><bold></bold>=<bold></bold>435. Indeed, removing these five studies only reduced the average correlation by.01 (<emph>r̅</emph> =.26 to <emph>r̅</emph> =.25). We did not remove these correlations from the analysis because we treated source as a moderator and these estimates did not survive later meta-regression analyses due to some moderator variables not being provided by the publisher. Moreover, the two sets of outliers (case number 37 versus case numbers 80, 81, 82, 83, and 84) canceled each other out in estimating the average sample-size weighted effect size (i.e., when removing all outliers, <emph>r̅</emph> =.26).</p> <hd id="AN0176845951-14">Wonderlic scores and academic performance</hd> <p>The overall relationship between Wonderlic scores and academic performance can be found in Table 2 (<emph>k</emph><bold></bold>=<bold></bold>91, <emph>N</emph><bold></bold>=<bold></bold>19,317, <emph>r̅</emph> =.26). The test for heterogeneity was statistically significant (<emph>Q</emph>[<reflink idref="bib90" id="ref78">90</reflink>] = 258.32, <emph>p</emph> <.001) and the 95% prediction interval (PI) was relatively wide (95% PI [.07,.45]), all of which suggests that a search for potential moderating variables would be worthwhile.</p> <p>Table 2. Meta-analysis results for the relationship between wonderlic scores and academic performance and retention.</p> <p> <ephtml> <table><thead><tr><td /><td /><td /><td /><td>95% CI</td><td>95% PI</td><td /><td /><td /></tr><tr><td>Model</td><td><italic>k</italic></td><td><italic>N</italic></td><td><italic>r̅</italic></td><td>LB</td><td>UB</td><td>LB</td><td>UB</td><td><italic>Q</italic></td><td><italic>τ</italic></td><td><italic>R<sup>2</sup></italic></td></tr></thead><tbody valign="top"><tr><td>Wonderlic-AP</td><td char=".">91</td><td char=".">19,317</td><td char=".">.26</td><td char=".">.22</td><td char=".">.29</td><td char=".">.07</td><td char=".">.45</td><td char=".">258.32**</td><td char=".">.10</td><td>–</td></tr><tr><td>Wonderlic-R</td><td char=".">7</td><td char=".">2,974</td><td char=".">.09</td><td char=".">−.17</td><td char=".">.35</td><td char=".">−.32</td><td char=".">.50</td><td char=".">29.47**</td><td char=".">.16</td><td>–</td></tr><tr><td>Wonderlic-R (Manual)</td><td char=".">4</td><td char=".">294</td><td char=".">.38</td><td char=".">.29</td><td char=".">.47</td><td char=".">.29</td><td char=".">.47</td><td>–</td><td char=".">.00</td><td char=".">100.00**</td></tr><tr><td>Wonderlic-R (Other)</td><td char=".">3</td><td char=".">2,680</td><td char=".">.06</td><td char=".">.03</td><td char=".">.09</td><td char=".">.03</td><td char=".">.09</td><td>–</td></tr></tbody></table> </ephtml> </p> <p>2 <emph>Note</emph>. **<emph>p</emph> <.001. AP = academic performance. R = retention. Manual = effect sizes sourced from the publisher test manual. Other = effect sizes not sourced from the publisher test manual. <emph>k</emph> = number of effect sizes. <emph>N</emph> = total sample size. <emph>r̅</emph> = mean sample-size weighted correlation. LB = lower bound. UB = upper bound. 95% CI = 95% confidence interval. 95% PI = 95% prediction interval. <emph>Q</emph> = homogeneity statistic. <emph>τ</emph> = square root of the variance of true effect sizes. <emph>R<sups>2</sups></emph> = amount of heterogeneity accounted for by the moderator.</p> <p>To increase the generalizability of the findings, in the primary set of analyses we only included non-continuous moderators with at least <emph>k</emph><bold></bold>≥<bold></bold>10 in each level. Meta-regression results for each potential moderator variable between Wonderlic scores and academic performance can be found in Table 3. Date of publication, source (publisher's manual, dissertation/thesis, peer-reviewed), program of study (business, interdisciplinary, psychology, vocational), and percentage male were all statistically significant (<emph>p</emph> <.05) moderators. On the other hand, AP_type (GPA versus exams or grades), %White, %Black, %Asian, and country (US versus Canada) were not significant moderators. Percentage male was positively related to effect size such that the higher the percentage male, the higher the effect size. Effect sizes from the publisher's manual were higher than those from dissertations/theses and peer-reviewed studies. Effect sizes from vocational programs of study were higher than those for business, interdisciplinary, and psychology programs of study. However, 14 of the 16 vocational effect sizes were sourced from the publisher's manual so results were somewhat confounded for publication source and program of study.</p> <p>Table 3. Meta-regression results for each potential moderator variable between wonderlic scores and academic performance.</p> <p> <ephtml> <table><thead><tr><td /><td /><td /><td /><td /><td /><td /><td /><td>95% CI</td></tr><tr><td>Potential Moderator</td><td><italic>k</italic></td><td><italic>τ</italic></td><td><italic>R<sup>2</sup></italic></td><td><italic>Q<sub>residual</sub></italic></td><td><italic>F<sub>moderator</sub></italic></td><td><italic>b</italic></td><td><italic>t</italic></td><td>LB</td><td>UB</td></tr></thead><tbody valign="top"><tr><td>AP_type (<italic>b<sub>0</sub></italic>)</td><td char=".">91</td><td char=".">.10</td><td char=".">1.37</td><td char=".">254.49***</td><td char=".">1.18</td><td char=".">0.234</td><td char=".">6.81***</td><td char=".">0.165</td><td char=".">0.302</td></tr><tr><td>AP_type_GPA (<italic>b<sub>1</sub></italic>)</td><td /><td /><td /><td /><td /><td char=".">0.040</td><td char=".">1.09</td><td char=".">−0.033</td><td char=".">0.114</td></tr><tr><td>Date (<italic>b<sub>0</sub></italic>)</td><td char=".">91</td><td char=".">.09</td><td char=".">12.14</td><td char=".">234.54***</td><td char=".">9.70**</td><td char=".">0.243</td><td char=".">16.28***</td><td char=".">0.213</td><td char=".">0.273</td></tr><tr><td>Date_pre-2000 (<italic>b<sub>1</sub></italic>)</td><td /><td /><td /><td /><td /><td char=".">0.088</td><td char=".">3.11**</td><td char=".">0.032</td><td char=".">0.144</td></tr><tr><td>Source (<italic>b<sub>0</sub></italic>)</td><td char=".">90</td><td char=".">.08</td><td char=".">32.03</td><td char=".">198.28***</td><td char=".">13.98***</td><td char=".">0.409</td><td char=".">12.75***</td><td char=".">0.346</td><td char=".">0.473</td></tr><tr><td>Source_DT (<italic>b<sub>1</sub></italic>)</td><td /><td /><td /><td /><td /><td char=".">−0.209</td><td char=".">−5.28***</td><td char=".">−0.287</td><td char=".">−0.130</td></tr><tr><td>Source_PR (<italic>b<sub>2</sub></italic>)</td><td /><td /><td /><td /><td /><td char=".">−0.146</td><td char=".">−4.12***</td><td char=".">−0.216</td><td char=".">−0.075</td></tr><tr><td>PS (<italic>b<sub>0</sub></italic>)</td><td char=".">86</td><td char=".">.09</td><td char=".">13.89</td><td char=".">236.42***</td><td char=".">6.20***</td><td char=".">0.388</td><td char=".">12.78***</td><td char=".">0.328</td><td char=".">0.448</td></tr><tr><td>PS_business (<italic>b<sub>1</sub></italic>)</td><td /><td /><td /><td /><td /><td char=".">−0.136</td><td char=".">−3.54***</td><td char=".">−0.213</td><td char=".">−0.060</td></tr><tr><td>PS_interdisciplinary (<italic>b<sub>2</sub></italic>)</td><td /><td /><td /><td /><td /><td char=".">−0.144</td><td char=".">−3.88***</td><td char=".">−0.219</td><td char=".">−0.070</td></tr><tr><td>PS_psychology (<italic>b<sub>3</sub></italic>)</td><td /><td /><td /><td /><td /><td char=".">−0.149</td><td char=".">−3.72***</td><td char=".">−0.228</td><td char=".">−0.069</td></tr><tr><td>%Male (<italic>b<sub>0</sub></italic>)</td><td char=".">65</td><td char=".">.08</td><td char=".">11.77</td><td char=".">147.48***</td><td char=".">4.54*</td><td char=".">0.180</td><td char=".">5.06***</td><td char=".">0.109</td><td char=".">0.251</td></tr><tr><td>%Male (<italic>b<sub>1</sub></italic>)</td><td /><td /><td /><td /><td /><td char=".">0.002</td><td char=".">2.13*</td><td char=".">0.000</td><td char=".">0.003</td></tr><tr><td>%White (<italic>b<sub>0</sub></italic>)</td><td char=".">34</td><td char=".">.09</td><td char=".">0.00</td><td char=".">99.23***</td><td char=".">0.16</td><td char=".">0.267</td><td char=".">4.97***</td><td char=".">0.158</td><td char=".">0.376</td></tr><tr><td>%White (<italic>b<sub>1</sub></italic>)</td><td /><td /><td /><td /><td /><td char=".">−0.000</td><td char=".">−0.40</td><td char=".">−0.002</td><td char=".">0.001</td></tr><tr><td>%Black (<italic>b<sub>0</sub></italic>)</td><td char=".">25</td><td char=".">.09</td><td char=".">0.00</td><td char=".">70.86***</td><td char=".">1.08</td><td char=".">0.236</td><td char=".">8.61***</td><td char=".">0.179</td><td char=".">0.293</td></tr><tr><td>%Black (<italic>b<sub>1</sub></italic>)</td><td /><td /><td /><td /><td /><td char=".">0.001</td><td char=".">1.04</td><td char=".">−0.001</td><td char=".">0.003</td></tr><tr><td>%Asian (<italic>b<sub>0</sub></italic>)</td><td char=".">22</td><td char=".">.09</td><td char=".">0.00</td><td char=".">67.10***</td><td char=".">0.07</td><td char=".">0.247</td><td char=".">8.64***</td><td char=".">0.187</td><td char=".">0.307</td></tr><tr><td>%Asian (<italic>b<sub>1</sub></italic>)</td><td /><td /><td /><td /><td /><td char=".">−0.000</td><td char=".">−0.26</td><td char=".">−0.002</td><td char=".">0.002</td></tr><tr><td>Country (<italic>b<sub>0</sub></italic>)</td><td char=".">81</td><td char=".">.09</td><td char=".">5.28</td><td char=".">219.16***</td><td char=".">1.98</td><td char=".">0.322</td><td char=".">8.61***</td><td char=".">0.247</td><td char=".">0.396</td></tr><tr><td>Country_US (<italic>b<sub>1</sub></italic>)</td><td /><td /><td /><td /><td /><td char=".">−0.057</td><td char=".">−1.41</td><td char=".">−0.137</td><td char=".">0.023</td></tr></tbody></table> </ephtml> </p> <p>3 <emph>Note</emph>. * <emph>p</emph> <.05. ** <emph>p</emph> <.01. *** <emph>p</emph> <.001. AP_type (GPA versus exam or course grades). Date = publication date (pre-2000 versus 2000–2023). Source = publication source (dissertation/thesis, peer-reviewed, publisher's manual). PS = program of study (business, interdisciplinary, psychology, vocational). %Male/White/Black/Asian = percentage of participants in the study who identified as male, White, Black, Asian. Country = country where study was conducted (Canada, US). <emph>k</emph> = number of effect sizes. <emph>τ</emph> = square root of the variance of true effect sizes. <emph>R<sups>2</sups></emph> = amount of heterogeneity accounted for by the moderator. <emph>Q<subs>residual</subs></emph> = test for residual heterogeneity. <emph>F<subs>moderator</subs></emph> = test of moderator. <emph>b</emph> = unstandardized regression coefficient. <emph>t</emph> = test of regression coefficient. LB = lower bound. UB = upper bound. 95% CI = 95% confidence interval.</p> <p>We then included a second meta-regression analysis (see Table 4) where we included all statistically significant moderators in a single analysis to determine which were unique moderators. Unfortunately, publication source and program of study dropped out of this analysis because the publisher's manual did not furnish %Male. Given that the publisher's effect sizes appeared to be outliers, we did not consider this to be a major limitation. Of the two variables in this meta-regression (date of publication and percentage male), percentage of males was statistically significant (<emph>p</emph> <.05) whereas date of publication was not. The non-significance of date of publication when included in a meta-regression with percentage of males is probably accounted for by the negative correlation between the two variables (<emph>r</emph><bold></bold>=<bold></bold>−0.35, <emph>p</emph> <.01, <emph>k</emph><bold></bold>=<bold></bold>65) such that the samples post-2000 contained a larger proportion of women than those from pre-2000.</p> <p>Table 4. Meta-regression results for a combined set of moderator variables between wonderlic scores and academic performance.</p> <p> <ephtml> <table><thead><tr><td /><td /><td /><td /><td /><td /><td /><td /><td>95% CI</td></tr><tr><td>Potential Moderator</td><td><italic>k</italic></td><td><italic>τ</italic></td><td><italic>R<sup>2</sup></italic></td><td><italic>Q<sub>residual</sub></italic></td><td><italic>F<sub>moderators</sub></italic></td><td><italic>b</italic></td><td><italic>t</italic></td><td>LB</td><td>UB</td></tr></thead><tbody valign="top"><tr><td>Intercept (<italic>b<sub>0</sub></italic>)</td><td char=".">65</td><td char=".">.08</td><td char=".">10.46</td><td char=".">146.11***</td><td char=".">2.70<sup>†</sup></td><td char=".">0.174</td><td char=".">4.79</td><td char=".">0.101</td><td char=".">0.246</td></tr><tr><td>Date_pre-2000 (<italic>b<sub>2</sub></italic>)</td><td /><td /><td /><td /><td /><td char=".">−0.042</td><td char=".">−0.94</td><td char=".">−0.132</td><td char=".">0.048</td></tr><tr><td>%Male (<italic>b<sub>3</sub></italic>)</td><td /><td /><td /><td /><td /><td char=".">0.002</td><td char=".">2.32*</td><td char=".">0.000</td><td char=".">0.003</td></tr></tbody></table> </ephtml> </p> <p>4 <emph>Note</emph>. <sups>†</sups><emph>p</emph> <.10. * <emph>p</emph> <.05. **<emph>p</emph> <.01. ***<emph>p</emph> <.001. Date = publication date (pre-2000 versus 2000–2023). %Male = percentage of participants in the study who identified as male. <emph>k</emph> = number of effect sizes. <emph>τ</emph> = square root of the variance of true effect sizes. <emph>R<sups>2</sups></emph> = amount of heterogeneity accounted for by the moderator. <emph>Q<subs>residual</subs></emph> = test for residual heterogeneity. <emph>F<subs>moderators</subs></emph> = omnibus test of moderators. <emph>b</emph> = unstandardized regression coefficient. <emph>t</emph> = test of regression coefficient. LB = lower bound. UB = upper bound. 95% CI = 95% confidence interval.</p> <hd id="AN0176845951-15">Wonderlic scores and retention</hd> <p>Retention as a variable was able to be coded for in seven studies. Retention was measured as 0 = did not graduate versus 1 = graduated. The most appropriate analysis to use in this case was logistic regression. However, we only had access to the primary data for one study and all other studies reported correlations between Wonderlic scores and retention as a binary outcome. We therefore used raw correlations as input to the meta-analysis for all seven studies (total <emph>N</emph><bold></bold>=<bold></bold>2,974). Most of the studies (<emph>k</emph><bold></bold>=<bold></bold>4) were sourced from the publisher's manual, whereas the remaining studies (<emph>k</emph><bold></bold>=<bold></bold>3) were sourced from three dissertations. A forest plot of the relationship between Wonderlic scores and retention can be found in Figure S5 in the supplemental online material.</p> <p>As can be seen in Table 2, the test of heterogeneity was statistically significant (<emph>Q</emph>[<reflink idref="bib6" id="ref79">6</reflink>] = 29.47, <emph>p</emph> <.001) which suggested that there were potential moderators that could better explain variability in this relationship. Source was a significant moderator in the meta-regression of the relationship between Wonderlic scores and retention. The test of the moderator was statistically significant <emph>F<subs>moderator</subs></emph> (<reflink idref="bib1" id="ref80">1</reflink>, 5) = 70.66, <emph>p</emph> <.001 and the residual test of heterogeneity was non-significant <emph>Q<subs>residual</subs></emph> (<reflink idref="bib5" id="ref81">5</reflink>) = 1.99, <emph>p</emph><bold></bold>=<bold></bold>0.85. As also can be seen in Table 2, 100% of the variance was accounted for by the source moderator and thus <emph>τ</emph> was zero. The average correlation was much higher when the correlations were sourced from the publisher's manual (<emph>r̅</emph> =.38) than from when the correlations were sourced from dissertations (<emph>r̅</emph> =.06). The results for retention should be regarded as preliminary and provisional given the small number of studies that we could locate that measured this outcome variable.</p> <hd id="AN0176845951-16">Hunter and Schmidt method results</hd> <p>The corrected correlations based on the Hunter and Schmidt method can be found in Tables 5 and 6. The corrected correlation between Wonderlic scores and GPA was.33 with a higher corrected correlation for those studies that had a greater than 50% concentration of males (</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mover accent="true"><mi>ρ</mi><mo>¯</mo></mover></mrow></math> </ephtml> =.37) than for those studies that had less than 50% concentration of males (</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mover accent="true"><mi>ρ</mi><mo>¯</mo></mover></mrow></math> </ephtml> =.28). Although not statistically significant when included with other moderators in meta-regression, the relationship between Wonderlic scores and GPA appeared to be higher for those effect sizes sourced from the publisher's manual (</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mover accent="true"><mi>ρ</mi><mo>¯</mo></mover></mrow></math> </ephtml> =.47) in comparison to peer-reviewed (</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mover accent="true"><mi>ρ</mi><mo>¯</mo></mover></mrow></math> </ephtml> =.35) or dissertation/thesis (</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mover accent="true"><mi>ρ</mi><mo>¯</mo></mover></mrow></math> </ephtml> =.26) sources. The corrected correlation between Wonderlic scores and retention was.12 with a higher corrected correlation from estimates sourced from the publisher's manual (</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mover accent="true"><mi>ρ</mi><mo>¯</mo></mover></mrow></math> </ephtml> =.47) than from other sources (</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mover accent="true"><mi>ρ</mi><mo>¯</mo></mover></mrow></math> </ephtml> =.07) (with 100% of the variance accounted for by source as in the analyses using the Hedges and Olkin [[<reflink idref="bib10" id="ref82">10</reflink>]] meta-analytic approach as implemented in the primary analyses). The corrected correlation between White (0) versus Black (<reflink idref="bib1" id="ref83">1</reflink>) was −0.39 which translates to a Cohen's <emph>d̅</emph> of −.85. The corrected correlation between Female (0) and Male (<reflink idref="bib1" id="ref84">1</reflink>) was.14 which translates to a Cohen's <emph>d̅</emph> of.28.</p> <p>Table 5. Meta-analysis results for the relationship between wonderlic scores and undergraduate GPA and retention (Hunter & Schmidt).</p> <p> <ephtml> <table><thead><tr><td>Model</td><td><italic>k</italic></td><td><italic>N</italic></td><td><p id="ilm0018"><graphic href="hijt_a_2318424_ilm0018.gif" content-type="Graph" /><math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mover accent="true" xmlns=""><mi>r</mi><mo>¯</mo></mover></math></p></td><td><p id="ilm0019"><graphic href="hijt_a_2318424_ilm0019.gif" content-type="Graph" /><math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mi xmlns="">S</mi><msub xmlns=""><mrow><mi>D</mi></mrow><mrow><mi>r</mi></mrow></msub></math></p></td><td><p id="ilm0020"><graphic href="hijt_a_2318424_ilm0020.gif" content-type="Graph" /><math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mi xmlns="">S</mi><msub xmlns=""><mrow><mi>D</mi></mrow><mrow><mtext mathvariant="italic">res</mtext></mrow></msub></math></p></td><td><p id="ilm0021"><graphic href="hijt_a_2318424_ilm0021.gif" content-type="Graph" /><math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow xmlns=""><mover accent="true"><mi>ρ</mi><mo stretchy="true">¯</mo></mover></mrow></math></p></td><td><p id="ilm0022"><graphic href="hijt_a_2318424_ilm0022.gif" content-type="Graph" /><math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mi xmlns="">S</mi><msub xmlns=""><mrow><mi>D</mi></mrow><mrow><msub><mrow><mi>r</mi></mrow><mrow><mi>c</mi></mrow></msub></mrow></msub></math></p></td><td><p id="ilm0023"><graphic href="hijt_a_2318424_ilm0023.gif" content-type="Graph" /><math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow xmlns=""><mi>S</mi><msub><mi>D</mi><mi>ρ</mi></msub></mrow></math></p></td><td>95% CI</td><td>80% CR</td></tr></thead><tbody valign="top"><tr><td>Wonderlic-GPA</td><td char=".">70</td><td char=".">16,378</td><td char=".">.27</td><td char=".">.10</td><td char=".">.08</td><td char=".">.33</td><td char=".">.12</td><td char=".">.09</td><td>[.30,.36]</td><td>[.21,.45]</td></tr><tr><td char="."><50% Male</td><td char=".">29</td><td char=".">7,339</td><td char=".">.22</td><td char=".">.09</td><td char=".">.06</td><td char=".">.28</td><td char=".">.11</td><td char=".">.07</td><td>[.24,.32]</td><td>[.18,.37]</td></tr><tr><td char=".">≥50% Male</td><td char=".">19</td><td char=".">5,369</td><td char=".">.30</td><td char=".">.08</td><td char=".">.04</td><td char=".">.37</td><td char=".">.09</td><td char=".">.05</td><td>[.33,.42]</td><td>[.30,.44]</td></tr><tr><td>Peer-Reviewed</td><td char=".">35</td><td char=".">8,616</td><td char=".">.28</td><td char=".">.09</td><td char=".">.05</td><td char=".">.35</td><td char=".">.10</td><td char=".">.06</td><td>[.31,.38]</td><td>[.26,.43]</td></tr><tr><td>Dissertation/Thesis</td><td char=".">21</td><td char=".">5,800</td><td char=".">.21</td><td char=".">.09</td><td char=".">.06</td><td char=".">.26</td><td char=".">.10</td><td char=".">.07</td><td>[.21,.31]</td><td>[.17,.35]</td></tr><tr><td>Publisher's Manual</td><td char=".">14</td><td char=".">1,962</td><td char=".">.39</td><td char=".">.11</td><td char=".">.06</td><td char=".">.47</td><td char=".">.12</td><td char=".">.07</td><td>[.40,.54]</td><td>[.38,.56]</td></tr><tr><td>Wonderlic-Retention</td><td char=".">7</td><td char=".">2,974</td><td char=".">.09</td><td char=".">.11</td><td char=".">.10</td><td char=".">.12</td><td char=".">.14</td><td char=".">.12</td><td>[−.01,.24]</td><td>[−.06,.29]</td></tr><tr><td>Publisher's Manual</td><td char=".">4</td><td char=".">294</td><td char=".">.38</td><td char=".">.08</td><td char=".">.00</td><td char=".">.47</td><td char=".">.09</td><td char=".">.00</td><td>[.32,.60]</td><td>[.47,.47]</td></tr><tr><td>Other</td><td char=".">3</td><td char=".">2,680</td><td char=".">.06</td><td char=".">.02</td><td char=".">.00</td><td char=".">.07</td><td char=".">.02</td><td char=".">.00</td><td>[.02,.13]</td><td>[.07,.07]</td></tr></tbody></table> </ephtml> </p> <p>5 <emph>Note.</emph> GPA = grade point average. <emph>k</emph> = number of studies contributing to meta-analysis; <emph>N</emph> = total sample size;</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mrow><mover accent="false"><mrow><mi>r</mi></mrow><mo>¯</mo></mover></mrow></mrow></math> </ephtml> = mean observed correlation;</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mi>S</mi><msub><mrow><mi>D</mi></mrow><mrow><mi>r</mi></mrow></msub></math> </ephtml> = observed standard deviation of</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mi>r</mi></mrow></math> </ephtml> ;</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mi>S</mi><msub><mrow><mi>D</mi></mrow><mrow><mtext mathvariant="italic">res</mtext></mrow></msub></math> </ephtml> = residual standard deviation of</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mi>r</mi></mrow></math> </ephtml> ;</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mover accent="true"><mi>ρ</mi><mo stretchy="true">¯</mo></mover></mrow></math> </ephtml> = mean true-score correlation;</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mi>S</mi><msub><mrow><mi>D</mi></mrow><mrow><msub><mrow><mi>r</mi></mrow><mrow><mi>c</mi></mrow></msub></mrow></msub></math> </ephtml> = observed standard deviation of corrected correlations (</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><msub><mrow><mi>r</mi></mrow><mrow><mi>c</mi></mrow></msub></math> </ephtml> );</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mi>S</mi><msub><mi>D</mi><mi>ρ</mi></msub></mrow></math> </ephtml> = residual standard deviation of ρ; CI = confidence interval around</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mover accent="true"><mi>ρ</mi><mo stretchy="true">¯</mo></mover></mrow></math> </ephtml> ; CR = credibility interval around</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mover accent="true"><mi>ρ</mi><mo stretchy="true">¯</mo></mover></mrow></math> </ephtml> . Correlations corrected using artifact distributions.</p> <p>Table 6. Meta-analysis results for the relationship between Wonderlic Scores and Black versus White and Sex (Hunter & Schmidt).</p> <p> <ephtml> <table><thead><tr><td>Model</td><td><italic>k</italic></td><td><italic>N</italic></td><td><p id="ilm0036"><graphic href="hijt_a_2318424_ilm0036.gif" content-type="Graph" /><math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow xmlns=""><mrow><mrow><mover accent="false"><mrow><mi>r</mi></mrow><mo>¯</mo></mover></mrow></mrow></mrow></math></p></td><td><p id="ilm0037"><graphic href="hijt_a_2318424_ilm0037.gif" content-type="Graph" /><math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mi xmlns="">S</mi><mrow xmlns=""><msub><mrow><mi>D</mi></mrow><mrow><mi>r</mi></mrow></msub></mrow></math></p></td><td><p id="ilm0038"><graphic href="hijt_a_2318424_ilm0038.gif" content-type="Graph" /><math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mi xmlns="">S</mi><mrow xmlns=""><msub><mrow><mi>D</mi></mrow><mrow><mtext mathvariant="italic">res</mtext></mrow></msub></mrow></math></p></td><td><p id="ilm0039"><graphic href="hijt_a_2318424_ilm0039.gif" content-type="Graph" /><math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow xmlns=""><mrow><mover accent="true"><mi>ρ</mi><mo stretchy="true">¯</mo></mover></mrow></mrow></math></p></td><td><p id="ilm0040"><graphic href="hijt_a_2318424_ilm0040.gif" content-type="Graph" /><math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mi xmlns="">S</mi><mrow xmlns=""><msub><mrow><mi>D</mi></mrow><mrow><mrow><msub><mrow><mi>r</mi></mrow><mrow><mi>c</mi></mrow></msub></mrow></mrow></msub></mrow></math></p></td><td><p id="ilm0041"><graphic href="hijt_a_2318424_ilm0041.gif" content-type="Graph" /><math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow xmlns=""><mi>S</mi><msub><mi>D</mi><mi>ρ</mi></msub></mrow></math></p></td><td>95% CI</td><td>80% CR</td></tr></thead><tbody valign="top"><tr><td>White (0) versus Black (1)</td><td char=".">12</td><td char=".">3,673</td><td char=".">−.32</td><td char=".">.08</td><td char=".">.05</td><td char=".">−.39</td><td char=".">.09</td><td char=".">.06</td><td>[−.45, −.33]</td><td>[−.47, −.32]</td></tr><tr><td>Female (0) versus Male (1)</td><td char=".">63</td><td char=".">26,405</td><td char=".">.11</td><td char=".">.09</td><td char=".">.07</td><td char=".">.14</td><td char=".">.11</td><td char=".">.09</td><td>[.12,.17]</td><td>[.03,.26]</td></tr></tbody></table> </ephtml> </p> <p>6 <emph>Note. k</emph> = number of studies contributing to meta-analysis; <emph>N</emph> = total sample size;</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mrow><mrow><mover accent="false"><mrow><mi>r</mi></mrow><mo>¯</mo></mover></mrow></mrow></mrow></math> </ephtml> = mean observed correlation;</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mi>S</mi><mrow><msub><mrow><mi>D</mi></mrow><mrow><mi>r</mi></mrow></msub></mrow></math> </ephtml> = observed standard deviation of</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mi>r</mi></mrow></math> </ephtml> ;</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mi>S</mi><mrow><msub><mrow><mi>D</mi></mrow><mrow><mtext mathvariant="italic">res</mtext></mrow></msub></mrow></math> </ephtml> = residual standard deviation of</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mi>r</mi></mrow></math> </ephtml> ;</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mrow><mover accent="true"><mi>ρ</mi><mo stretchy="true">¯</mo></mover></mrow></mrow></math> </ephtml> = mean true-score correlation;</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mi>S</mi><mrow><msub><mrow><mi>D</mi></mrow><mrow><mrow><msub><mrow><mi>r</mi></mrow><mrow><mi>c</mi></mrow></msub></mrow></mrow></msub></mrow></math> </ephtml> = observed standard deviation of corrected correlations (</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><msub><mrow><mi>r</mi></mrow><mrow><mi>c</mi></mrow></msub></mrow></math> </ephtml> );</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mi>S</mi><msub><mi>D</mi><mi>ρ</mi></msub></mrow></math> </ephtml> = residual standard deviation of ρ; CI = confidence interval around</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mrow><mover accent="true"><mi>ρ</mi><mo stretchy="true">¯</mo></mover></mrow></mrow></math> </ephtml> ; CR = credibility interval around</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mrow><mover accent="true"><mi>ρ</mi><mo stretchy="true">¯</mo></mover></mrow></mrow></math> </ephtml> . Correlations corrected using artifact distributions.</p> <hd id="AN0176845951-17">Discussion</hd> <p>The main purpose of the present meta-analysis was to assess the validity of the Wonderlic (a brief measure of GMA) for the prediction of academic performance in general, undergraduate GPA, and student retention. We found that the Wonderlic was positively correlated with academic performance in general (<emph>r̅</emph> =.26), undergraduate GPA (<emph>r̅</emph> =.27,</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mover accent="true"><mi>ρ</mi><mo>¯</mo></mover></mrow></math> </ephtml> =.33), and retention (<emph>r̅</emph> =.09,</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mover accent="true"><mi>ρ</mi><mo>¯</mo></mover></mrow></math> </ephtml> =.12); however, we found a significant degree of heterogeneity which suggested the presence of moderators. The presence of moderators was unsurprising as previous research that has assessed the validity of other cognitive tests, such as the SAT has also found important moderating relationships (e.g., Shen et al., [<reflink idref="bib24" id="ref85">24</reflink>]).</p> <p>Interestingly, the heterogeneity found for the relationship between the Wonderlic and retention was entirely resolved when we assessed the correlations reported in the publisher's test manual and included publication source as a moderator. This suggests that the correlations reported by the publisher (<emph>r̅</emph> =.38) were not replicated in the independent research studies (<emph>r̅</emph> =.06) included in this meta-analysis. Although there is a chance that this could be due to bias associated with the Wonderlic publisher's results, it also could have been due to differences in how retention was defined or measured across studies (e.g., measurement time period), differential sample characteristics, or sample size (Manual <emph>N</emph><bold></bold>=<bold></bold>294; Other <emph>N</emph><bold></bold>=<bold></bold>2,680). Differences in findings could also be due to whether corrections for the dichotomous nature of the variables were made to the effect sizes sourced from the Wonderlic manual. Although such corrections did not seem to be applied to the effect sizes not sourced from the Wonderlic manual, it was less clear that such corrections were or were not made from the effect sizes sourced from the Wonderlic manual. Unfortunately, information needed to make such corrections was missing from the majority of studies and we therefore could not empirically test this proposition. Finally, given that we only found three non-publisher coefficients for the relationship between Wonderlic and retention (compared to four from the publisher), we believe that further replication studies are required to understand why the test publisher's results appear to differ so drastically from other studies.</p> <p>The most robust source of heterogeneity for the relationship between the Wonderlic and academic performance and undergraduate GPA was the percentage of males in the samples such that the relationship between Wonderlic scores and academic performance and undergraduate GPA was stronger when samples contained a higher percentage of males (controlling for date of publication). This is an interesting counterpoint to the results of Dahlke et al. ([<reflink idref="bib6" id="ref86">6</reflink>]) who found found that the SAT was less valid for males than females in predicting college GPA. Our results in this regard are only suggestive as they may be due to unmeasured third variables and are not a true test of differential validity. Future research should more definitively explore this issue.</p> <p>A secondary goal of the meta-analysis was to investigate the degree of subgroup differences that Wonderlic scores evidenced. We found that participants who identified themselves as male had moderately higher Wonderlic scores than those who identified themselves as female (corrected <emph>d̅</emph> =.28). This was not surprising given that previous research has shown that males tend to score higher than females on comparable entrance tests like the SAT (math <emph>d</emph> = 0.41; verbal <emph>d</emph> = 0.14) and ACT (<emph>d</emph> = 0.23; Mau & Lynn, [<reflink idref="bib16" id="ref87">16</reflink>]). Additionally, our results showed similar subgroup differences between Black and White students (corrected <emph>d̅</emph> = |0.85|) as previous research on the ACT and SAT scores (college students <emph>d̅</emph> = 0.69; college applicants <emph>d̅</emph> = 0.98; Roth et al., [<reflink idref="bib18" id="ref88">18</reflink>]). Despite these subgroup differences, institutions have persistently employed such admissions assessments which could be a tribute to the relatively higher validity, efficiency, and objectivity of such assessments compared to alternative assessments such as essays and interviews.</p> <hd id="AN0176845951-18">Practical implications</hd> <p>Although the Wonderlic was found to evidence similar subgroup differences along racial and gender lines in comparison to the SAT and ACT, it was found to be less predictively valid in relation to undergraduate GPA compared to these other admissions assessment options. As previously noted, the corrected validities for other commonly used undergraduate university admissions assessments with GPA were</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mover accent="true"><mi>ρ</mi><mo>¯</mo></mover></mrow></math> </ephtml> =.55 for the ACT (Westrick et al., [<reflink idref="bib30" id="ref89">30</reflink>]) and</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mover accent="true"><mi>ρ</mi><mo>¯</mo></mover></mrow></math> </ephtml> =.47 for the SAT (Sackett et al., [<reflink idref="bib20" id="ref90">20</reflink>]) compared to (in our study) the Wonderlic with</p> <p>Graph</p> <p> <ephtml> <math display="inline" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mover accent="true"><mi>ρ</mi><mo>¯</mo></mover></mrow></math> </ephtml> =.33.</p> <p>Aside from the lower criterion-related validity, the potential benefits of using the Wonderlic in undergraduate admissions are that it: (<reflink idref="bib1" id="ref91">1</reflink>) is a much shorter assessment (Wonderlic 12 min, SAT 3 h 15 min, ACT 2 h 55 min), (<reflink idref="bib2" id="ref92">2</reflink>) does not require expensive or time-consuming preparation on the applicant's part, and (<reflink idref="bib3" id="ref93">3</reflink>) is less expensive than the SAT and ACT. Critical drawbacks for using the Wonderlic include: (<reflink idref="bib1" id="ref94">1</reflink>) limited reliable differentiation of candidates through the test score range compared to longer assessments, (<reflink idref="bib2" id="ref95">2</reflink>) a limited pool of items with no computer adaptive option (for which we are aware) which could lead to violations of test security and easier coaching, and (<reflink idref="bib3" id="ref96">3</reflink>) assessments such as the SAT and ACT may be perceived as more relevant to college admissions and thus more fair from the perspective of the candidate (Schmitt et al., [<reflink idref="bib23" id="ref97">23</reflink>]).</p> <p>Previous research has shown that for predictors with similar validities but substantial subgroup differences, administering predictors with higher adverse impact before assessments with a lower impact is associated with less adverse impact overall (De Corte et al., [<reflink idref="bib8" id="ref98">8</reflink>]). Essentially, this suggests that admissions testing should be used at an earlier stage of the admissions process (e.g., prior to administering structured interviews). Overall, using admissions tests in multi-staged admissions selection is still likely to be a cost-effective method given that only those who pass the first hurdle of this process would need to be assessed using more costly methods such as interviews. One additional practical caveat relates to the findings in the current paper regarding the Wonderlic test publisher's data. Because of the generally more favorable findings from the correlations and effect sizes that were sourced from the publisher's manual (e.g., the relationship between the Wonderlic and retention), it will be prudent for institutions to critically evaluate the information that is derived solely from test publishers, and it is also advisable that they collect their own data regarding the validity of their chosen testing option (i.e., local validation). Prior studies have found similar results to our study showing that test vendors tend to report higher criterion-related validities than those reported elsewhere in the literature (McDaniel et al., [<reflink idref="bib17" id="ref99">17</reflink>]; Van Iddekinge et al., [<reflink idref="bib28" id="ref100">28</reflink>]).</p> <hd id="AN0176845951-19">Potential limitations and future research directions</hd> <p>One potential limitation of the current study is associated with the relationship between Wonderlic scores and retention. Unlike the relationship between the Wonderlic and academic performance, there were relatively few correlations found between the Wonderlic and retention. Further, the studies that were found were all from either the publisher's manual or unpublished dissertations. Thus, future research endeavors should focus on conducting additional primary investigations into the relationship between the Wonderlic and retention in the context of institutions of higher education.</p> <p>Lastly, the accuracy of any meta-analysis is dependent upon the decisions that are made when searching the literature, determining and implementing the inclusion and exclusion criteria, and coding the data. However, our search of the literature was thorough; our initial search after automatic removal of duplications resulted in 2,992 papers. We were also quite thorough in contacting the authors of papers who were identified as not reporting the proper statistics for our coding purposes. Further, publication bias did not seem to be an issue as we were able to recover a substantial amount of relevant data from unpublished dissertations and theses. Nevertheless, one other area for future research to address would be to update our findings once a critical mass of new papers has emerged, especially pertaining to further meta-analyzing the Wonderlic and retention validity.</p> <hd id="AN0176845951-20">Conclusion</hd> <p>The present study found that Wonderlic scores were positively correlated with post-secondary academic performance, undergraduate GPA, and retention across studies, but at lower levels than other longer assessments such as the ACT and SAT. We also identified several moderators. Of note, publisher reported coefficients seemed to be larger than those reported by other sources for all outcomes. Further, the percent of male participants for each study were found to moderate the relationship between Wonderlic scores and academic performance in general and undergraduate GPA in particular. An assessment of subgroup differences also revealed that racial/ethnic subgroup differences were comparable to those previously reported for the SAT and ACT.</p> <hd id="AN0176845951-21">Disclosure statement</hd> <p>No potential conflict of interest was reported by the author(s).</p> <p>Correction Statement</p> <p>This article has been corrected with minor changes. These changes do not impact the academic content of the article.</p> <ref id="AN0176845951-22"> <title> Notes </title> <blist> <bibl id="bib1" idref="ref23" type="bt">1</bibl> <bibtext> There are several versions of the Wonderlic including the Wonderlic Personnel Test (WPT) and the Scholastic Level Exam (SLE). According to the publisher's manual: "The WPT is primarily used by business and governmental organizations to evaluate job applicants for employment and occupational training. The SLE is primarily used by postsecondary colleges and schools to evaluate student applicants and to provide a counseling aid in curriculum choices. All forms of both tests are designed to be equivalent and each form is expected to provide essentially the same score for applicants of the same ability" (Wonderlic, [31], p. 4). Shortened versions of the Wonderlic (WPT-Q or SLE-Q) have timing reduced to eight minutes and items reduced to 30. A regression equation is used to adjust scores to the 50-point scale used in the full measure. In the present meta-analysis, only six of the 91 validity coefficients used one of the short forms which precluded reliable separate analyses.</bibtext> </blist> <blist> <bibl id="bib2" idref="ref4" type="bt">2</bibl> <bibtext> Note that meta-regression cannot be performed when using the artifact distribution method.</bibtext> </blist> <blist> <bibl id="bib3" idref="ref19" type="bt">3</bibl> <bibtext> We did however extract a distribution of Wonderlic reliabilities from the literature which is required to correct for indirect range restriction using the Case IV method (see Caretta & Ree, 2022). The mean and variance of this distribution was 0.85 and 0.004, respectively.</bibtext> </blist> <blist> <bibl id="bib4" idref="ref29" type="bt">4</bibl> <bibtext> We used the average of the two standard deviations as the unrestricted population standard deviation because use of the high school graduate standard deviation would lead to an overcorrection and use of the college graduate standard deviation would lead to an undercorrection.</bibtext> </blist> <blist> <bibl id="bib5" idref="ref15" type="bt">5</bibl> <bibtext> We only included cases with undergraduate GPA (<emph>k</emph> = 70) and not other forms of academic performance such as exam or course grades or at the graduate level for these analyses because the artifact distribution was based on undergraduate GPA.</bibtext> </blist> <blist> <bibl id="bib6" idref="ref20" type="bt">6</bibl> <bibtext> Note that we were careful to not double-count results from unpublished work with that which had subsequently been published. In cases where dissertation and thesis work had been published, we reported the published results but not the unpublished results.</bibtext> </blist> <blist> <bibl id="bib7" idref="ref42" type="bt">7</bibl> <bibtext> Supplemental data for this article can be accessed online at https://doi.org/10.1080/15305058.2024.2318424.</bibtext> </blist> </ref> <ref id="AN0176845951-23"> <title> References </title> <blist> <bibtext> Beatty, A. S., Walmsley, P. T., Sackett, P. R., Kuncel, N. R., & Koch, A. J. (2015). The reliability of college grades. Educational Measurement, 34 (4), 31 – 40. https://doi.org/10.1111/emip.12096</bibtext> </blist> <blist> <bibtext> Bennett, C. T. (2022). Untested admissions: Examining changes in application behaviors and student demographics under test-optional policies. American Educational Research Journal, 59 (1), 180 – 216. https://doi.org/10.3102/00028312211003526</bibtext> </blist> <blist> <bibtext> Berry, C. M., Clark, M. A., & McClure, T. K. (2011). Racial/ethnic differences in criterion-related validity of cognitive ability tests: A qualitative and quantitative review. The Journal of Applied Psychology, 96 (5), 881 – 906. https://doi.org/10.1037/a0023222</bibtext> </blist> <blist> <bibtext> Carretta, T. R., & Ree, M. J. (2022). Corrections for range restriction: Lessons from 20 research scenarios. Military Psychology, 34 (5), 551 – 569. https://doi.org/10.1080/08995605.2021.2022067</bibtext> </blist> <blist> <bibtext> Cleary, T. A. (1968). Test bias: Prediction of grades of negro and white students in integrated colleges. Journal of Educational Measurement, 5 (2), 115 – 124. https://doi.org/10.1111/j.1745-3984.1968.tb00613.x</bibtext> </blist> <blist> <bibtext> Dahlke, J. A., Sackett, P. R., & Kuncel, N. R. (2019). Effects of range restriction and criterion contamination on differential validity of the SAT by race/ethnicity and sex. The Journal of Applied Psychology, 104 (6), 814 – 831. https://doi.org/10.1037/apl0000382</bibtext> </blist> <blist> <bibtext> Dahlke, J. A., & Wiernik, B. M. (2019). psychmeta : An R package for psychometric meta-analysis. Applied Psychological Measurement, 43 (5), 415 – 416. https://doi.org/10.1177/0146621618795933</bibtext> </blist> <blist> <bibl id="bib8" idref="ref43" type="bt">8</bibl> <bibtext> De Corte, W., Lievens, F., & Sackett, P. R. (2006). Predicting adverse impact and mean criterion performance in multistage selection. The Journal of Applied Psychology, 91 (3), 523 – 537. https://doi.org/10.1037/0021-9010.91.3.523</bibtext> </blist> <blist> <bibl id="bib9" idref="ref7" type="bt">9</bibl> <bibtext> Frey, M. C., & Detterman, D. K. (2004). Scholastic assessment or g ? The relationship between the Scholastic Assessment Test and general cognitive ability. Psychological Science, 15 (6), 373 – 378. https://doi.org/10.1111/j.0956-7976.2004.00687.x</bibtext> </blist> <blist> <bibtext> Hedges, L. V., & Olkin, I. (1985). Statistical methods for meta-analysis. Academic Press.</bibtext> </blist> <blist> <bibtext> Higdem, J. L., Kostal, J. W., Kuncel, N. R., Sackett, P. R., Shen, W., Beatty, A. S., & Kiger, T. B. (2016). The role of socioeconomic status in SAT–freshman grade relationships across gender and racial subgroups. Educational Measurement, 35 (1), 21 – 28. https://doi.org/10.1111/emip.12103</bibtext> </blist> <blist> <bibtext> Knapp, G., & Hartung, J. (2003). Improved tests for a random effects meta-regression with a single covariate. Statistics in Medicine, 22 (17), 2693 – 2710. https://doi.org/10.1002/sim.1482</bibtext> </blist> <blist> <bibtext> Koenig, K. A., Frey, M. C., & Detterman, D. K. (2008). ACT and general cognitive ability. Intelligence, 36 (2), 153 – 160. https://doi.org/10.1016/j.intell.2007.03.005</bibtext> </blist> <blist> <bibtext> Koljatic, M., Silva, M., & Sireci, S. G. (2021). College admissions tests and social responsibility. Educational Measurement, 40 (4), 22 – 27. https://doi.org/10.1111/emip.12425</bibtext> </blist> <blist> <bibtext> Kuncel, N., Tran, K., & Zhang, S. (2020). Measuring student character: Modernizing predictors of academic success. In M. Oliveri & C. Wendler (Eds.), Higher education admissions practices: An international perspective (pp. 276 – 302). Cambridge University Press. https://doi.org/10.1017/9781108559607.016</bibtext> </blist> <blist> <bibtext> Mau, W., & Lynn, R. (2001). Gender differences on the Scholastic Aptitude Test, the American College Test, and college grades. Educational Psychology, 21 (2), 133 – 136. https://doi.org/10.1080/01443410020043832</bibtext> </blist> <blist> <bibtext> McDaniel, M. A., Rothstein, H. R., & Whetzel, D. L. (2006). Publication bias: A case study of four test vendors. Personnel Psychology, 59 (4), 927 – 953. https://doi.org/10.1111/j.1744-6570.2006.00059.x</bibtext> </blist> <blist> <bibtext> Roth, P. L., Bevier, C. A., Bobko, P., Switzer, F. S. I., & Tyler, P. (2001). Ethnic group differences in cognitive ability in employment and educational settings: A meta-analysis. Personnel Psychology, 54 (2), 297 – 330. https://doi.org/10.1111/j.1744-6570.2001.tb00094.x</bibtext> </blist> <blist> <bibtext> Saboe, M., & Terrizzi, S. (2019). SAT optional policies? Do they influence graduate quality, selectivity, or diversity? Economics Letters, 174, 13 – 17. https://doi.org/10.1016/j.econlet.2018.10.017</bibtext> </blist> <blist> <bibtext> Sackett, P. R., Kuncel, N. R., Arneson, J. J., Cooper, S. R., & Waters, S. D. (2009). Does socioeconomic status explain the relationship between admissions tests and post-secondary academic performance? Psychological Bulletin, 135 (1), 1 – 22. https://doi.org/10.1037/a0013978</bibtext> </blist> <blist> <bibtext> Schmidt, F. L., & Hunter, J. E. (2015). Methods of meta-analysis: Correcting error and bias in research findings (3rd ed.). SAGE.</bibtext> </blist> <blist> <bibtext> Schmill, S. (2022, March 28). We are reinstating our SAT/ACT requirement for future admissions cycles. MIT Admissions. Retrieved from https://mitadmissions.org/blogs/entry/we-are-reinstating-our-sat-act-requirement-for-future-admissions-cycles/.</bibtext> </blist> <blist> <bibtext> Schmitt, N., Oswald, F. L., Kim, B. H., Gillespie, M. A., & Ramsay, L. J. (2004). The impact of justice and self-serving bias explanations of the perceived fairness of different types of selection tests. International Journal of Selection and Assessment, 12 (1–2), 160 – 171. https://doi.org/10.1111/j.0965-075X.2004.00271.x</bibtext> </blist> <blist> <bibtext> Shen, W., Sackett, P. R., Kuncel, N. R., Beatty, A. S., Rigdon, J. L., & Kiger, T. B. (2012). All validities are not created equal: Determinants of variation in SAT validity across schools. Applied Measurement in Education, 25 (3), 197 – 219. https://doi.org/10.1080/08957347.2012.687615</bibtext> </blist> <blist> <bibtext> Steel, P., Fariborzi, H., & Hendijani, R. (2023). An application of modern literature review methodology: Finding needles in ever-growing haystacks. In Sage research methods: Business. Sage. https://doi.org/10.4135/9781529667417</bibtext> </blist> <blist> <bibtext> Sterne, J. A. C., & Egger, M. (2005). Regression methods to detect publication and other bias in meta-analysis. In H. R. Rothstein, A. J. Sutton, & M. Borenstein (Eds.), Publication bias in meta-analysis: Prevention, assessment, and adjustments (pp. 99 – 110). Wiley.</bibtext> </blist> <blist> <bibtext> Thomas, M. K. (2004). The SAT II: Minority/majority test-score gaps and what they could mean for college admissions. Social Science Quarterly, 85 (5), 1318 – 1334. https://doi.org/10.1111/j.0038-4941.2004.00278.x</bibtext> </blist> <blist> <bibtext> Van Iddekinge, C. H., Roth, P. L., Raymark, P. H., & Odle-Dusseau, H. N. (2012). The criterion-related validity of integrity tests: An updated meta-analysis. The Journal of Applied Psychology, 97 (3), 499 – 530. https://doi.org/10.1037/a0021196</bibtext> </blist> <blist> <bibtext> Viechtbauer, W. (2010). Conducting meta-analyses in R with the metafor package. Journal of Statistical Software, 36 (3), 1 – 48. https://doi.org/10.18637/jss.v036.i03</bibtext> </blist> <blist> <bibtext> Westrick, P. A., Le, H., Robbins, S. B., Radunzel, J. M. R., & Schmidt, F. L. (2015). College performance and retention: A meta-analysis of the predictive validities of ACT® scores, high school grades, and SES. Educational Assessment, 20 (1), 23 – 45. https://doi.org/10.1080/10627197.2015.99761</bibtext> </blist> <blist> <bibtext> Wonderlic, E. F. (1992). Wonderlic Personnel Test and Scholastic Level Exam user's manual. Wonderlic and Associates.</bibtext> </blist> <blist> <bibtext> Wonderlic, E. F., & Hovland, C. I. (1939). The Personnel Test: A restandardized abridgment of the Otis S-A test for business and industrial use. Journal of Applied Psychology, 23 (6), 685 – 702. https://doi.org/10.1037/h0056432</bibtext> </blist> </ref> <aug> <p>By Chet Robie; Sabah Rasheed; Stephen D. Risavy and Piers Steel</p> <p>Reported by Author; Author; Author; Author</p> </aug> <nolink nlid="nl1" bibid="bib15" firstref="ref1"></nolink> <nolink nlid="nl2" bibid="bib27" firstref="ref2"></nolink> <nolink nlid="nl3" bibid="bib14" firstref="ref3"></nolink> <nolink nlid="nl4" bibid="bib19" firstref="ref5"></nolink> <nolink nlid="nl5" bibid="bib22" firstref="ref6"></nolink> <nolink nlid="nl6" bibid="bib13" firstref="ref8"></nolink> <nolink nlid="nl7" bibid="bib20" firstref="ref9"></nolink> <nolink nlid="nl8" bibid="bib30" firstref="ref10"></nolink> <nolink nlid="nl9" bibid="bib18" firstref="ref12"></nolink> <nolink nlid="nl10" bibid="bib16" firstref="ref14"></nolink> <nolink nlid="nl11" bibid="bib11" firstref="ref16"></nolink> <nolink nlid="nl12" bibid="bib32" firstref="ref22"></nolink> <nolink nlid="nl13" bibid="bib31" firstref="ref24"></nolink> <nolink nlid="nl14" bibid="bib25" firstref="ref25"></nolink> <nolink nlid="nl15" bibid="bib10" firstref="ref45"></nolink> <nolink nlid="nl16" bibid="bib12" firstref="ref57"></nolink> <nolink nlid="nl17" bibid="bib21" firstref="ref60"></nolink> <nolink nlid="nl18" bibid="bib26" firstref="ref62"></nolink> <nolink nlid="nl19" bibid="bib29" firstref="ref64"></nolink> <nolink nlid="nl20" bibid="bib89" firstref="ref76"></nolink> <nolink nlid="nl21" bibid="bib90" firstref="ref78"></nolink> <nolink nlid="nl22" bibid="bib24" firstref="ref85"></nolink> <nolink nlid="nl23" bibid="bib23" firstref="ref97"></nolink> <nolink nlid="nl24" bibid="bib17" firstref="ref99"></nolink> <nolink nlid="nl25" bibid="bib28" firstref="ref100"></nolink>
Header DbId: eric
DbLabel: ERIC
An: EJ1422761
AccessLevel: 3
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 0
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: A Meta-Analysis of the Relationship between Wonderlic Test Scores and School Success
– Name: Language
  Label: Language
  Group: Lang
  Data: English
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Chet+Robie%22">Chet Robie</searchLink> (ORCID <externalLink term="https://orcid.org/0000-0003-4521-6924">0000-0003-4521-6924</externalLink>)<br /><searchLink fieldCode="AR" term="%22Sabah+Rashe%22">Sabah Rashe</searchLink><br /><searchLink fieldCode="AR" term="%22Stephen+D%2E+Risavy%22">Stephen D. Risavy</searchLink><br /><searchLink fieldCode="AR" term="%22Piers+Steel%22">Piers Steel</searchLink>
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="SO" term="%22International+Journal+of+Testing%22"><i>International Journal of Testing</i></searchLink>. 2024 24(2):169-189.
– Name: Avail
  Label: Availability
  Group: Avail
  Data: Routledge. Available from: Taylor & Francis, Ltd. 530 Walnut Street Suite 850, Philadelphia, PA 19106. Tel: 800-354-1420; Tel: 215-625-8900; Fax: 215-207-0050; Web site: http://www.tandf.co.uk/journals
– Name: PeerReviewed
  Label: Peer Reviewed
  Group: SrcInfo
  Data: Y
– Name: Pages
  Label: Page Count
  Group: Src
  Data: 21
– Name: DatePubCY
  Label: Publication Date
  Group: Date
  Data: 2024
– Name: TypeDocument
  Label: Document Type
  Group: TypDoc
  Data: Journal Articles<br />Information Analyses
– Name: Audience
  Label: Education Level
  Group: Audnce
  Data: <searchLink fieldCode="EL" term="%22Higher+Education%22">Higher Education</searchLink><br /><searchLink fieldCode="EL" term="%22Postsecondary+Education%22">Postsecondary Education</searchLink>
– Name: Subject
  Label: Descriptors
  Group: Su
  Data: <searchLink fieldCode="DE" term="%22Meta+Analysis%22">Meta Analysis</searchLink><br /><searchLink fieldCode="DE" term="%22Test+Validity%22">Test Validity</searchLink><br /><searchLink fieldCode="DE" term="%22Alternative+Assessment%22">Alternative Assessment</searchLink><br /><searchLink fieldCode="DE" term="%22Scores%22">Scores</searchLink><br /><searchLink fieldCode="DE" term="%22Cognitive+Measurement%22">Cognitive Measurement</searchLink><br /><searchLink fieldCode="DE" term="%22Cognitive+Ability%22">Cognitive Ability</searchLink><br /><searchLink fieldCode="DE" term="%22Academic+Achievement%22">Academic Achievement</searchLink><br /><searchLink fieldCode="DE" term="%22Undergraduate+Students%22">Undergraduate Students</searchLink><br /><searchLink fieldCode="DE" term="%22Grade+Point+Average%22">Grade Point Average</searchLink><br /><searchLink fieldCode="DE" term="%22College+Entrance+Examinations%22">College Entrance Examinations</searchLink><br /><searchLink fieldCode="DE" term="%22Comparative+Analysis%22">Comparative Analysis</searchLink><br /><searchLink fieldCode="DE" term="%22Predictive+Validity%22">Predictive Validity</searchLink><br /><searchLink fieldCode="DE" term="%22Test+Preparation%22">Test Preparation</searchLink><br /><searchLink fieldCode="DE" term="%22Timed+Tests%22">Timed Tests</searchLink>
– Name: SubjectThesaurus
  Label: Assessment and Survey Identifiers
  Group: Su
  Data: <searchLink fieldCode="SU" term="%22SAT+%28College+Admission+Test%29%22">SAT (College Admission Test)</searchLink><br /><searchLink fieldCode="SU" term="%22ACT+Assessment%22">ACT Assessment</searchLink>
– Name: DOI
  Label: DOI
  Group: ID
  Data: 10.1080/15305058.2024.2318424
– Name: ISSN
  Label: ISSN
  Group: ISSN
  Data: 1530-5058<br />1532-7574
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: This meta-analysis examined the validity of an alternative to traditional assessments called the Wonderlic which is a brief measure of general mental ability. Our results showed significant, positive correlations between Wonderlic scores and academic performance in general ([r-bar] = 0.26), between Wonderlic scores and undergraduate GPA in particular ([r-bar] = 0.27, [p-bar] = 0.33), and between Wonderlic scores and retention ([r-bar] = 0.09, [p-bar] = 0.12). We also identified several significant moderators of the relationship between Wonderlic scores and relevant outcomes (e.g., test publisher reported coefficients were larger than those reported by other sources). Subgroup differences in test scores were in the same range as other post-secondary admissions assessments (e.g., ACT and SAT scores). Overall, the Wonderlic has similar levels of subgroup differences and is less strongly related to GPA than traditional assessments but still retains useful levels of predictiveness and is a shorter, less expensive assessment that requires less preparation than the ACT or SAT.
– Name: AbstractInfo
  Label: Abstractor
  Group: Ab
  Data: As Provided
– Name: DateEntry
  Label: Entry Date
  Group: Date
  Data: 2024
– Name: AN
  Label: Accession Number
  Group: ID
  Data: EJ1422761
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=eric&AN=EJ1422761
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.1080/15305058.2024.2318424
    Languages:
      – Text: English
    PhysicalDescription:
      Pagination:
        PageCount: 21
        StartPage: 169
    Subjects:
      – SubjectFull: Meta Analysis
        Type: general
      – SubjectFull: Test Validity
        Type: general
      – SubjectFull: Alternative Assessment
        Type: general
      – SubjectFull: Scores
        Type: general
      – SubjectFull: Cognitive Measurement
        Type: general
      – SubjectFull: Cognitive Ability
        Type: general
      – SubjectFull: Academic Achievement
        Type: general
      – SubjectFull: Undergraduate Students
        Type: general
      – SubjectFull: Grade Point Average
        Type: general
      – SubjectFull: College Entrance Examinations
        Type: general
      – SubjectFull: Comparative Analysis
        Type: general
      – SubjectFull: Predictive Validity
        Type: general
      – SubjectFull: Test Preparation
        Type: general
      – SubjectFull: Timed Tests
        Type: general
      – SubjectFull: SAT (College Admission Test)
        Type: general
      – SubjectFull: ACT Assessment
        Type: general
    Titles:
      – TitleFull: A Meta-Analysis of the Relationship between Wonderlic Test Scores and School Success
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Chet Robie
      – PersonEntity:
          Name:
            NameFull: Sabah Rashe
      – PersonEntity:
          Name:
            NameFull: Stephen D. Risavy
      – PersonEntity:
          Name:
            NameFull: Piers Steel
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 01
              M: 01
              Type: published
              Y: 2024
          Identifiers:
            – Type: issn-print
              Value: 1530-5058
            – Type: issn-electronic
              Value: 1532-7574
          Numbering:
            – Type: volume
              Value: 24
            – Type: issue
              Value: 2
          Titles:
            – TitleFull: International Journal of Testing
              Type: main
ResultId 1