View in EDS HTML Full Text PDF Full Text

Structuring First-Year Retention at a Regional Public Institution: Validating and Refining the Structure of Bowman's SEM Analysis

Saved in:
Bibliographic Details
Title: Structuring First-Year Retention at a Regional Public Institution: Validating and Refining the Structure of Bowman's SEM Analysis
Language: English
Authors: Collier, Daniel A. (ORCID 0000-0002-3164-0815), Fitzpatrick, Dan (ORCID 0000-0002-9095-0370), Brehm, Chelsea (ORCID 0000-0002-1808-8405), Hearit, Keith, Beach, Andrea
Source: Research in Higher Education. Dec 2020 61(8):917-942.
Availability: Springer. Available from: Springer Nature. One New York Plaza, Suite 4600, New York, NY 10004. Tel: 800-777-4643; Tel: 212-460-1500; Fax: 212-460-1700; e-mail: customerservice@springernature.com; Web site: https://link.springer.com/
Peer Reviewed: Y
Page Count: 26
Publication Date: 2020
Sponsoring Agency: Department of Education (ED)
Contract Number: P116F140353
Document Type: Journal Articles
Reports - Research
Education Level: Higher Education
Postsecondary Education
Descriptors: College Freshmen, School Holding Power, Public Colleges, Structural Equation Models, Academic Persistence, Student Characteristics, Financial Problems, Security (Psychology), Food, Social Adjustment
DOI: 10.1007/s11162-020-09612-w
ISSN: 0361-0365
Abstract: Structural equation modeling (SEM) considering how students' non-cognitive attributes influence first-year college student persistence remain extraordinarily rare--as are studies that test and expand upon published structural models or studies that include college student food security. This study addresses each. We surveyed "Beginner" Freshmen, capturing eight non-cognitive measurements and using institutional data on performance and fall-to-fall persistence measures, we then tested the structure of Bowman et al.'s (Res High Educ 60:135-152, 2019) SEM model. In Model 1, we mimic the Bowman model's financial variable by only including financial stress. We confirm that Bowman is a good structural model of student persistence, although our data were collected for another purpose, using different scales for non-cognitive elements and even one different non-cognitive measurement. We found students' non-cognitive attributes remain importantly influential to social adjustment (r=0.65), commitment to persist (r=0.40), college GPA (r=0.25), and fall-to-fall persistence (r=0.30). In Model 2, we generated a latent financial security variable incorporating financial stress and food security. Including food security generated a direct influence from the financial security variable to high-school GPA (r=0.25), not found in the Bowman model or Model 1, and a direct significant relationship from financial security to social adjustment (r=0.11)--not found in Model 1. Further changes are observed in the indirect relationship from financial security to college GPA from Model 1 (r=0.29) to Model 2 (r=0.51). We highlight the robustness of the Bowman model and that the inclusion of food security brings increased strength to several relationships without sacrificing optimal fit.
Abstractor: As Provided
Entry Date: 2020
Accession Number: EJ1273896
Database: ERIC
Full text is not displayed to guests.
FullText Links:
  – Type: pdflink
    Url: https://content.ebscohost.com/cds/retrieve?content=AQICAHj0k_4E0hTGH8RJwT4gCJyBsGNe_WN95AvKlDbXJGqwxwFm4frnAcTTHolmwz7L8JJZAAAA4zCB4AYJKoZIhvcNAQcGoIHSMIHPAgEAMIHJBgkqhkiG9w0BBwEwHgYJYIZIAWUDBAEuMBEEDMH7o4f7UyvBnRhX2wIBEICBm-VF47lz8MgeElG2Ak-WcmwkzDjbZwYSFlru3pZHM1eBn2O4xw6FmyeoFG4f-87Ieiv5snykhRv5S5xVv5ttJ1CGC-l9YHSnKQMoZeUucwbNgwjrHKOPK0r3fFoGMlTi7YVS8r_9q3pU-qwTCvrBGGN4sSOdR9MAmSdzRJTHo32ot5uO0MjffZI1YfMqSShjm5nT2UopwXXFZDvj
Text:
  Availability: 1
  Value: <anid>AN0146556866;rhe01dec.20;2020Oct23.04:50;v2.2.500</anid> <title id="AN0146556866-1">Structuring First-Year Retention at a Regional Public Institution: Validating and Refining the Structure of Bowman's SEM Analysis </title> <p>Structural equation modeling (SEM) considering how students' non-cognitive attributes influence first-year college student persistence remain extraordinarily rare—as are studies that test and expand upon published structural models or studies that include college student food security. This study addresses each. We surveyed "Beginner" Freshmen, capturing eight non-cognitive measurements and ussing institutional data on performance and fall-to-fall persistence measures, we then tested the structure of Bowman et al.'s (Res High Educ 60:135–152, 2019) SEM model. In Model 1, we mimic the Bowman model's financial variable by only including financial stress. We confirm that Bowman is a good structural model of student persistence, although our data were collected for another purpose, using different scales for non-cognitive elements and even one different non-cognitive measurement. We found students' non-cognitive attributes remain importantly influential to social adjustment (r =.65), commitment to persist (r =.40), college GPA (r =.25), and fall-to-fall persistence (r =.30). In Model 2, we generated a latent financial security variable incorporating financial stress and food security. Including food security generated a direct influence from the financial security variable to high-school GPA (r =.25), not found in the Bowman model or Model 1, and a direct significant relationship from financial security to social adjustment (r =.11)—not found in Model 1. Further changes are observed in the indirect relationship from financial security to college GPA from Model 1 (r =.29) to Model 2 (r =.51). We highlight the robustness of the Bowman model and that the inclusion of food security brings increased strength to several relationships without sacrificing optimal fit.</p> <p>Keywords: Structural equation modeling; Non-cognitive attributes; College persistence; Food security; First-year students</p> <p>Students' performance and persistence in postsecondary education is driven by a combination of prior academic success overlaid with students' non-cognitive attributes. Researchers and stakeholders have paid increased attention to the latter in recent decades. The increased attention on students' non-cognitive attributes has been integral in developing theories and models of student retention that guide much contemporary research (see Astin [<reflink idref="bib5" id="ref1">5</reflink>]; Bean [<reflink idref="bib10" id="ref2">10</reflink>]; Tinto [<reflink idref="bib88" id="ref3">88</reflink>]). These theories and models have been subjected to extensive modifications as faults are identified (Braxton et al. [<reflink idref="bib16" id="ref4">16</reflink>]) and new non-cognitive elements gain in popularity and power (Bowman et al. [<reflink idref="bib15" id="ref5">15</reflink>]). Grit (Duckworth et al. [<reflink idref="bib31" id="ref6">31</reflink>]) and food security (Goldrick-Rab et al. [<reflink idref="bib40" id="ref7">40</reflink>]) are two such elements. The expansion and modification of the theoretical underpinnings of earlier models sometimes manifest v[<reflink idref="bib15" id="ref8">15</reflink>]ia studies that utilize structural equation modeling (SEM; Cabrera et al. [<reflink idref="bib20" id="ref9">20</reflink>]; Sass et al. [<reflink idref="bib83" id="ref10">83</reflink>]; Bowman et al. [<reflink idref="bib15" id="ref11">15</reflink>]), a causal statistical method used to examine direct and indirect influences between variables and outcomes. Best practices include replicating and re-testing SEM structures. However, re-testing SEM structures remains difficult as SEM analyses require large sample sizes. Further, the research community is not uniform in the fielding of aligned survey constructs, or in agreement regarding which student attributes remain important to collect and examine; thus, making true replication and testing in new contexts difficult. Our study used a combination of institutional and survey data collected from incoming freshmen at a 4-year institution to test the robustness of the structure of Bowman, et al.'s ([<reflink idref="bib15" id="ref12">15</reflink>]) recent model, answering the following primary questions:</p> <p></p> <ulist> <item> Does the structure of the Bowman model accurately reflect independently-collected data at a 4-year regional public institution?</item> <p></p> <item> Does modifying the structure's financial means measurement into a latent variable combining financial stress and food security improve the model?</item> </ulist> <hd id="AN0146556866-2">The Bowman Model</hd> <p>Bowman et al. ([<reflink idref="bib15" id="ref13">15</reflink>]) developed their model in response to a persistent gap in understanding how students' non-cognitive attributes directly and indirectly affect key outcomes of performance and persistence. From here onward, we will refer to this model as the Bowman model. Guided by Astin's ([<reflink idref="bib4" id="ref14">4</reflink>]) input-environment-outcome model and Farrington et al.'s ([<reflink idref="bib34" id="ref15">34</reflink>]) framework for understanding the types of students' non-cognitive attributes, the Bowman model sought to test how students' incoming attributes affect student behavior, performance and persistence. The Bowman model's primary contribution is to test how students' non-cognitive characteristics influence college-related performance and persistence—see Fig. 1.</p> <p>Graph: Fig. 1 The Bowman Model ([<reflink idref="bib15" id="ref16">15</reflink>])</p> <p>Unlike prior descriptive studies examining how non-cognitive attributes relate to performance and persistence (like Collier et al. [<reflink idref="bib27" id="ref17">27</reflink>]; Laskey and Hetzel [<reflink idref="bib55" id="ref18">55</reflink>]), the Bowman model and SEM technique captures the direct and indirect influences of student attributes on the outcome of persistence. A notable strength of the Bowman model is that the findings were generated using multi-institutional data and a sample size of over 10,000 participants—which the authors suggested fills an important gap, as most SEM models use data from individual institutions. Because the study found the model accurate for a large sample of multi-institutional data, we wanted to test if it remained valid for an unrelated set of students and data. Furthermore, as a recent addition to the field, to our knowledge, the model is not yet tested beyond the original study, and likely would not be, considering that less than 1% of published education-related studies in top journals intend to replicate prior research (Makel and Plucker [<reflink idref="bib60" id="ref19">60</reflink>]).</p> <p>The Bowman model suggests that the relationships between financial means, students' non-cognitive attributes, social adjustment, and academic performance is complicated and entangled. The central thesis of the study suggests that students' non-cognitive attributes are more impactful than they were previously given credit for—mostly because the indirect influences of these attributes on performance was previously unknown and influences on persistence were untested (Bowman et al. [<reflink idref="bib15" id="ref20">15</reflink>]). The Bowman model illustrates that the outcome of retention is directly influenced by the combination of College GPA (<emph>r</emph> = 0.50), Commitment to Institution (<emph>r</emph> = 0.21), Social Adjustment (<emph>r</emph> = 0.16), Financial Means (<emph>r</emph> = 0.09), and Non-cognitive attributes (<emph>r</emph> = -0.15)—with indirect influences positively bolstering each of the variables. Of particular interest, indirect influences bring the non-cognitive attributes (<emph>r</emph> = 0.14) to a net-positive total influence on retention—aligning with the study's central assumption. See Table 2 for all reported direct and indirect findings of the Bowman model. Overall, Bowman et al. ([<reflink idref="bib15" id="ref21">15</reflink>]) concluded that because the higher education-focused literature has a dearth of SEM models and causal studies, the field generally remains in the dark regarding the full (direct and indirect) effects of students' non-cognitive attributes on social engagement performance, and importantly persistence—however, their study is one step in the right direction.</p> <p>One critique we have about the model is that Bowman et al. ([<reflink idref="bib15" id="ref22">15</reflink>]) did not necessarily measure "financial means" despite using that label for one construct. To capture "financial means," Bowman's model used a previously-untested 3-item scale measuring students' distress over paying for college and other expenses. This action is defensible for use in SEM (see Cabrera et al. [<reflink idref="bib20" id="ref23">20</reflink>]); yet, we believe using a validated financial stress scale could produce stronger models. Additionally, because the authors did not include actual elements of student or family financial <emph>means</emph>—being family income, Pell eligibility or receipt, or the amount of expected loan debt, we believe the variable label is <emph>unintentionally</emph> misleading and that it should be relabeled financial distress (or comfort if reverse-coded) to better represent what the variable measured.</p> <p>With the Bowman model's strengths and our critique noted, our study tested the structure of the Bowman model with the financial stress scale developed by Lim et al. ([<reflink idref="bib58" id="ref24">58</reflink>]). Additionally, in a subsequent model we reformed financial means to a latent variable including food security. As discussed later, both variables measure financially-related security. Recently, college students' food security has become a major topic of concern (Goldrick-Rab et al. [<reflink idref="bib40" id="ref25">40</reflink>]) with studies linking it to students' non-cognitive attributes, engagement, performance, and persistence (Collier et al. [<reflink idref="bib26" id="ref26">26</reflink>]; Martinez et al. [<reflink idref="bib65" id="ref27">65</reflink>]). Potentially, generating a latent financial security variable will modify the direct and indirect relationships between this factor and students' non-cognitive, social, performance, and persistence constructs. By incorporating food security—a better reflection of the real-world experiences of current college students and of current policy priorities—this revised structural model may be able to help researchers and practitioners generate improved future research and interventions.</p> <hd id="AN0146556866-3">Key Non-cognitive Attributes and Links to Performance and Persistence</hd> <p></p> <hd id="AN0146556866-4">Financial Security</hd> <p>Undeniably, students' financial security and stress link with their college experiences and their college performance (Goldrick-Rab [<reflink idref="bib39" id="ref28">39</reflink>]). Financially-related elements are included in many well-cited theoretical models (see Bean [<reflink idref="bib10" id="ref29">10</reflink>]; Tinto [<reflink idref="bib88" id="ref30">88</reflink>]) and credible SEM models examining student retention (Bowman et al. [<reflink idref="bib15" id="ref31">15</reflink>]; Cabrera et al. [<reflink idref="bib20" id="ref32">20</reflink>]; Sass et al. [<reflink idref="bib83" id="ref33">83</reflink>]). Although we have a litany of descriptive research establishing connections between financial security and students' ultimate performance and persistence, scholars continue to debate and examine the true importance of financial means on performance and persistence (see Goldrick-Rab [<reflink idref="bib39" id="ref34">39</reflink>]; Nazmi et al. [<reflink idref="bib68" id="ref35">68</reflink>]; Sass et al. [<reflink idref="bib83" id="ref36">83</reflink>]). Due to limited SEM studies, we do not fully comprehend the direct, indirect, and total effects of financial security on student persistence. As highlighted by Bowman et al. ([<reflink idref="bib15" id="ref37">15</reflink>]), of the SEM models currently published, most study only the effects of student attributes to GPA—not persistence. Furthermore, even when financial stress is included within SEM models exploring student performance and persistence—some SEM models like Cabrera et al. ([<reflink idref="bib20" id="ref38">20</reflink>]) and Bowman et al. ([<reflink idref="bib15" id="ref39">15</reflink>]) have picked individual items from scales or generated untested scales to measure what could be labeled as financial stress.</p> <p>College students' food security is an important topic. Recent data from a nationally-employed survey indicates that in 4-year institutions 48% of students experience food insecurity and at the site institution our team has uncovered similar trends, as 42% of incoming students reported some level of food insecurity (Collier et al. [<reflink idref="bib26" id="ref40">26</reflink>]). Taking cues from decades of research on K-12 students (see Alaimo et al. [<reflink idref="bib1" id="ref41">1</reflink>]; Coleman-Jensen et al. [<reflink idref="bib25" id="ref42">25</reflink>]), higher education researchers have placed greater emphasis on gauging college students' food security (Goldrick-Rab et al. [<reflink idref="bib40" id="ref43">40</reflink>]) to make explicit connections between food insecurity and both students' non-cognitive attributes (Collier et al. [<reflink idref="bib27" id="ref44">27</reflink>]; Mukigi and Brown [<reflink idref="bib67" id="ref45">67</reflink>]) and their academic outcomes (Nazmi, et al. [<reflink idref="bib68" id="ref46">68</reflink>]; Patton-Lopez et al. [<reflink idref="bib73" id="ref47">73</reflink>]).</p> <p>College students' food security is an element we are <emph>just</emph> beginning to see in structural equation models. At the time of writing this paper, our search yielded three SEM studies that included college student food security. Bowman and associates raised concerns that SEM models on college students are not generally housed in higher education journals and lack concrete connections to established higher education theory. Two of the SEM studies including food security were indeed published in a non-higher education journal (Bruening et al. [<reflink idref="bib18" id="ref48">18</reflink>]; Raskind, et al. [<reflink idref="bib76" id="ref49">76</reflink>]); whereas one was found in the <emph>Journal of College Student Development</emph> (Camelo and Elliott [<reflink idref="bib21" id="ref50">21</reflink>]). Bruening et al. ([<reflink idref="bib18" id="ref51">18</reflink>]) examined the food security of college students over time, finding that food insecurity possesses a causal effect on students' mental health and moods. Bruening et al. ([<reflink idref="bib18" id="ref52">18</reflink>]) did not examine any performance or persistence outcomes. Raskind et al. ([<reflink idref="bib76" id="ref53">76</reflink>]) examined the direct and indirect role of food insecurity on GPA, finding direct influences and mediated impacts through mental distress. Finally, Camelo and Elliott ([<reflink idref="bib21" id="ref54">21</reflink>]) explored the impacts of food insecurity to GPA, including mediation through student characteristics (e.g. race, Pell). Missing from all three are the robust set of non-cognitive attributes that the Bowman model and well-established theoretical higher education models (Astin [<reflink idref="bib5" id="ref55">5</reflink>]; Bean [<reflink idref="bib10" id="ref56">10</reflink>]; Tinto [<reflink idref="bib88" id="ref57">88</reflink>]) indicate are also important factors predicting GPA. Finally, again, none of these SEM analyses examine the outcome of persistence.</p> <hd id="AN0146556866-5">Character Traits</hd> <p>A notable inclusion of the Bowman model is Duckworth et al.'s ([<reflink idref="bib31" id="ref58">31</reflink>]) measurement of grit—a construct that measures students' perseverance and passions for long-term goals. Since the concept's inception, grit has gained traction with some researchers and has been used to influence K-12 and higher education policy (Cohen [<reflink idref="bib24" id="ref59">24</reflink>]; Sparks [<reflink idref="bib86" id="ref60">86</reflink>]). Although grit has gained in popularity, the concept itself is rather problematic as the foundations it was built upon possess the following flaws: (<reflink idref="bib1" id="ref61">1</reflink>) describing effect sizes larger than models indicate, (<reflink idref="bib2" id="ref62">2</reflink>) sample issues as initial studies primarily focused on high-achieving people to start, and (<reflink idref="bib3" id="ref63">3</reflink>) how closely grit is related to the Big Five's conceptualization of conscientiousness (Crede et al. [<reflink idref="bib29" id="ref64">29</reflink>]). Several recent studies have found weak links or no ties between grit and performance (Christensen et al. [<reflink idref="bib22" id="ref65">22</reflink>]; Ivcevic and Brackett [<reflink idref="bib48" id="ref66">48</reflink>]). Due to these critiques, when developing our instrument, we opted to capture conscientiousness instead of grit. Conscientiousness measures the degree to which individuals exhibit focus and organization, and are goal-oriented (Komarraju et al. [<reflink idref="bib52" id="ref67">52</reflink>]). Previous studies generally examined the Big Five together. Yet, the influences of the remaining four traits on college student performance are mixed; conscientiousness usually rises to significance when predicting positive learning outcomes and specifically college persistence (Beattie et al. [<reflink idref="bib13" id="ref68">13</reflink>]; Lievens et al. [<reflink idref="bib59" id="ref69">59</reflink>]; Komarraju et al. [<reflink idref="bib52" id="ref70">52</reflink>]).</p> <p>Cognitive engagement gauges students' attitudes and approaches to learning, as well as self-regulation (discipline and efficacy) and planning (time management; Appleton et al. [<reflink idref="bib3" id="ref71">3</reflink>]; Gunuc and Kuzu [<reflink idref="bib43" id="ref72">43</reflink>]; Krause and Coates [<reflink idref="bib53" id="ref73">53</reflink>]). Gauging this concept allows stakeholders to understand the degree to which students indicate a readiness to meet the academic challenges of college (Gunuc and Kuzu [<reflink idref="bib43" id="ref74">43</reflink>]). Cognitive engagement is strongly correlated with attending and engaging in class. Furthermore, cognitive engagement is linked with campus engagement and very strongly linked with behavioral engagement (Collier et al. [<reflink idref="bib27" id="ref75">27</reflink>])—the degree to which one attends and participates in class (Gunuc and Kuzu [<reflink idref="bib43" id="ref76">43</reflink>]).</p> <hd id="AN0146556866-6">Social Adjustment</hd> <p>Exploring students' interactions with peers and in building peer social networks has been an intense focus for researchers examining student performance and persistence (Kuh et al. [<reflink idref="bib54" id="ref77">54</reflink>]). Building strong bonds with peers relates to a stronger sense of belonging and overall comfort, which should translate into higher academic performance and increased chances to persist (Mihaly [<reflink idref="bib66" id="ref78">66</reflink>]; Strayhorn [<reflink idref="bib87" id="ref79">87</reflink>]). Astin's ([<reflink idref="bib6" id="ref80">6</reflink>]) seminal work highlights that peer-group interactions correlate with college students' sense of self, intellectual growth, confidence, and behaviors that tend to lead to stronger academic success. Connecting with concepts explained here, peer-group interactions were found as the only social engagement measure significantly correlated with extrinsic-external regulation motivation—suggesting that deeper interactions with peers help students identify how behaviors connect to rewards and consequences in college. Peer-group interactions are positively correlated with motivation, academic and faculty/staff engagement, and college GPA (Collier et al. [<reflink idref="bib27" id="ref81">27</reflink>]).</p> <p>Students usually place great importance on developing relationships with faculty (Tinto [<reflink idref="bib88" id="ref82">88</reflink>]) and staff (Harper [<reflink idref="bib44" id="ref83">44</reflink>]). Yet, the relationships students have with faculty or staff are often combined. Traditionally, many scholars studied only interactions with faculty and placed greater importance on the level of involvement of faculty towards student performance and persistence (Tinto [<reflink idref="bib88" id="ref84">88</reflink>]). Yet, institutional decisions to move away from hiring full-time, tenue-track faculty and become more reliant on part-time and adjunct faculty may have diminished faculty power in and understanding of their role in student success (Kezar and Maxey [<reflink idref="bib51" id="ref85">51</reflink>]). As institutions decided to change the faculty, they became more dependent on student services and academic affairs staff to address students' needs (Sandeen and Barr [<reflink idref="bib80" id="ref86">80</reflink>]). Therefore, these staff members could be crucial student allies, and relationships that students forge with staff may be influential in students' college experiences (Quaye and Harper [<reflink idref="bib75" id="ref87">75</reflink>]). Given that students identify relationships with staff as being critical to personal development, performance, and persistence (Harper [<reflink idref="bib44" id="ref88">44</reflink>]), the role of staff should be examined separately from faculty.</p> <hd id="AN0146556866-7">Amotivation</hd> <p>An important element in student performance and persistence is motivation (Tinto [<reflink idref="bib88" id="ref89">88</reflink>]; Warden and Myers [<reflink idref="bib93" id="ref90">93</reflink>]). Motivation itself is a widely examined construct within higher education research circles. However, motivation is a multi-faceted construct tied to extrinsic and intrinsic factors, and further to specific aptitudes within those factors. For example, the capability to identify rewards or consequences impact elements of extrinsic motivation (Ryan and Deci [<reflink idref="bib79" id="ref91">79</reflink>]). The complexities of motivation lead to mixed findings regarding what type of motivation is "best" to predict performance and persistence. However, one construct within motivation that is particularly useful in studying student persistence is amotivation—which is understood as the lack of focus and understanding of why one is engaged in or working towards a specific goal, and the lack of feeling autonomy while working towards goals (Ryan and Deci [<reflink idref="bib79" id="ref92">79</reflink>]). Here, the goal is to persist toward a degree—for college students, amotivation could be defined as lacking purpose in college (Vallerand et al. [<reflink idref="bib92" id="ref93">92</reflink>]). Related to the Bowman model, prior research correlated amotivation with college GPA (Warden and Myers [<reflink idref="bib93" id="ref94">93</reflink>]), students' non-cognitive attributes, and stopping out (Collier et al. [<reflink idref="bib27" id="ref95">27</reflink>]). As amotivation has previously been linked with persistence and beliefs centered on whether students intend to persist, we have substituted amotivation to hold the place of intention to persist in the structure of Bowman's model.</p> <hd id="AN0146556866-8">Analysis Plan</hd> <p></p> <hd id="AN0146556866-9">Site and Sample</hd> <p>This study was conducted at Western Michigan University[<reflink idref="bib1" id="ref96">1</reflink>]—a large, public, research intensive, doctoral granting university in an urban area—and was funded by a First in the World Grant (P116F140353) provided by the U.S. Department of Education. Similar to data used by Bowman et al. ([<reflink idref="bib15" id="ref97">15</reflink>]), survey data were collected in the first month of the Fall Semester—late August to mid-September in 2018. Our sample included only incoming, domestic "Beginner" or first time in any college (FTIAC) Freshman students. In total, <emph>N</emph> = 2351 students were invited to participate in the survey and those who completed the survey were awarded gift cards worth up to $20 and entered into a lottery-style drawing with additional gift cards ranging from $25 to $250. With the initial call for participation and two additional reminders we gained <emph>N</emph> = 487 (21% response rate for FTIAC students) participants with full profiles. Our sample is statistically significantly less advantaged when compared to non-participants regarding neighborhood AGI ($71,000 v. $76,000) and high-school free and reduced lunch percentage (31% v. 29%), while possessing higher high school GPA (3.56 v. 3.40). Additionally, the sample consists of a higher percentage of non-White (33% v. 27%) and female students (63% v. 48%). The sample also has a higher overall college GPA (3.20 v. 3.00) and a higher percentage enrolled for Fall'19 (80% v. 74%). In summary, our sample is generally less economically advantaged but higher-performing, with more underrepresented minority and female students, than the WMU population overall.</p> <hd id="AN0146556866-10">Instrument Constructs and Connections to the Bowman Model</hd> <p>Initially guided by Tinto's ([<reflink idref="bib88" id="ref98">88</reflink>]) model of student departure, we generated a 68-item survey consisting of previously-tested scales to capture a combination of students' amotivation, social engagement, readiness to perform academically, financial and psychological distress, and food security. See Table 1 for more details on the scales, including the number of items and study alphas.</p> <p>Data sources and survey constructs</p> <p> <ephtml> <table frame="hsides" rules="groups"><thead><tr><th align="left" colspan="4"><p>Institutional research variables</p></th></tr></thead><tbody><tr><td align="left" colspan="4"><p>High school GPA</p></td></tr><tr><td align="left" colspan="4"><p>Overall college GPA</p></td></tr><tr><td align="left" colspan="4"><p>Enrollment in fall 2019 courses</p></td></tr></tbody></table> </ephtml> </p> <p>Data sources and survey constructs</p> <p> <ephtml> <table frame="hsides" rules="groups"><thead><tr><th align="left" colspan="3"><p>Survey Scales</p></th><th align="left"><p>Study's Cronbach's Alpha</p></th></tr></thead><tbody><tr><td align="left"><p>AMS-C—Amotivation Sub-Scale</p></td><td align="left"><p>4-item, 7pt Likert</p></td><td align="left"><p>Vallerand et al. (<xref ref-type="bibr" rid="bibr92">1992</xref>)</p></td><td align="left"><p>0.85</p></td></tr><tr><td align="left"><p>BFI-2-S—Conscientiousness</p></td><td align="left"><p>6-item, 5pt Likert</p></td><td align="left"><p>Soto and John (<xref ref-type="bibr" rid="bibr85">2017</xref>)</p></td><td align="left"><p>0.75</p></td></tr><tr><td align="left"><p>Cognitive Engagement</p></td><td align="left"><p>10-item, 5pt Likert</p></td><td align="left"><p>Gunuc and Kuzu (<xref ref-type="bibr" rid="bibr43">2015</xref>)</p></td><td align="left"><p>0.81</p></td></tr><tr><td align="left"><p>Peer-Group Interaction</p></td><td align="left"><p>7-item, 5pt Likert</p></td><td align="left"><p>Pascarella and Terenzini (<xref ref-type="bibr" rid="bibr72">1980</xref>)</p></td><td align="left"><p>0.84</p></td></tr><tr><td align="left"><p>Faculty Interaction</p></td><td align="left"><p>4-item, 5pt Likert</p></td><td align="left"><p>Pascarella and Terenzini (<xref ref-type="bibr" rid="bibr72">1980</xref>)</p></td><td align="left"><p>0.83</p></td></tr><tr><td align="left"><p>Staff Interaction<sup>1</sup></p></td><td align="left"><p>4-item, 5pt Likert</p></td><td align="left" /><td align="left"><p>0.86</p></td></tr><tr><td align="left"><p>Financial Stress</p></td><td align="left"><p>6-item, 5pt Likert</p></td><td align="left"><p>Lim, et al. (<xref ref-type="bibr" rid="bibr58">2014</xref>)</p></td><td align="left"><p>0.87</p></td></tr><tr><td align="left"><p>Food Security Scale</p></td><td align="left"><p>6-item, Affirmative<sup>2</sup></p></td><td align="left"><p>U.S. Department of Agriculture (<xref ref-type="bibr" rid="bibr91">2012</xref>)</p></td><td align="left"><p>0.79</p></td></tr></tbody></table> </ephtml> </p> <p> <sups>1</sups>We modified the 4-item construct replacing "faculty" with "staff" to separately gauge intent to interact with non-faculty <sups>2</sups>Scoring is based on the number of affirmative responses, the categorizations are as follows: 0 = high food security, 1 = marginal security, 2–4 = low security, 5–6 = very low food security. The U.S. Department of Agriculture ([<reflink idref="bib90" id="ref99">90</reflink>]) considers those with Marginal Security to be food secure—the categorization illustrates anxiety over the amount of food available but limited changes to diet or food intake. Due to the distress and anxiety students may experience, we included "marginal security" in reports of <emph>any degree</emph> of food insecurity but also report categories of severity</p> <hd id="AN0146556866-11">Financial Security</hd> <p>We used a previously tested college student financial stress scale. The financial scale measures students' stress associated with the ability for pay for college and with the debt-loads students have assumed (e.g. student loans, credit card debt). Prior research indicates that the scale is highly reliable (<emph>a</emph> = 0.83) and that increased levels of financial stress are correlated with increased likelihoods of seeking financial help (Lim et al. [<reflink idref="bib58" id="ref100">58</reflink>]). A sample item is: <emph>I worry about having enough money to pay for school.</emph> These items match the sentiment found in Bowman and associates' items—"to what degree are you confident that you can pay for the next term's tuition and fees" (p. 141). The relatively new scale has been used in prior research (e.g. Farrell et al. [<reflink idref="bib33" id="ref101">33</reflink>]) with previous examinations illustrating correlations between increased financial anxiety and lowered extrinsic motivation, academic and social engagement, and performance and persistence (Collier et al. [<reflink idref="bib27" id="ref102">27</reflink>]). Financial stress is included in both of our models.</p> <p>To measure food security, the second component of the financial security latent variable, we aligned with a litany of prior studies in higher education (e.g. Gaines et al. [<reflink idref="bib38" id="ref103">38</reflink>]; Goldrick-Rab et al. [<reflink idref="bib40" id="ref104">40</reflink>]) and used the 6-item food security scale (U.S. Department of Agriculture [<reflink idref="bib91" id="ref105">91</reflink>]). The scale measures food security within the past 12-months. A sample item is: <emph>How often are you hungry but do not eat because you cannot afford enough food.</emph> The scale is scored by number of affirmative responses (0–6), producing four unique categories ranging from high food security (zero responses) to very low food security (five to six responses)—see Table 1 for more details. Prior research indicates the scale is highly reliable with alphas between 0.86 and 0.88 (Collier et al. [<reflink idref="bib26" id="ref106">26</reflink>]; Martinez et al. [<reflink idref="bib65" id="ref107">65</reflink>]). Food security is the only construct that is included exclusively in Model 2.</p> <hd id="AN0146556866-12">Non-cognitive Constructs</hd> <p>To measure students' conscientiousness, we used the Big Five Inventory-2 Short Form (BFI-2-S) that houses 6 items, scored along a 5-point Likert scale, for conscientiousness (<emph>a</emph> = 0.78, Soto and John [<reflink idref="bib85" id="ref108">85</reflink>]). A sample item is: <emph>I am someone who is reliable, can always be counted on</emph>. Related to the Bowman model, based upon Crede et al.'s ([<reflink idref="bib29" id="ref109">29</reflink>]) critiques, we argue that conscientiousness is a suitable replacement for grit, which also gauges elements of students' self-discipline. A similar point is made by Bowman et al. ([<reflink idref="bib15" id="ref110">15</reflink>]) given the strong correlations between the concepts of grit and self-discipline.</p> <p>The facets within cognitive engagement align well with the other three constructs that were included in Bowman et al.'s ([<reflink idref="bib15" id="ref111">15</reflink>]) non-cognitive attributes latent factor: academic self-efficacy, self-discipline, and time management. To capture cognitive engagement, we used Gunuc and Kuzu's ([<reflink idref="bib43" id="ref112">43</reflink>]) reliable (<emph>a</emph> = 0.82-0.88) scale. A sample item of the cognitive engagement scale is: <emph>I believe I spend enough time to make enough effort to academically thrive.</emph> Again, making the direct link to the Bowman model, their 3-item academic self-efficacy scale measured, "student's beliefs that they can succeed academically, even in their most difficult coursework" (p. 142) and "to what degree are you certain that you can do well on all problems and tasks assigned to your courses" (p. 140). As highlighted by Gunuc and Kuzu ([<reflink idref="bib43" id="ref113">43</reflink>]), cognitive engagement captures students' abilities to self-regulate—which can generally be defined as having the skillset such as self-efficacy and discipline to act in one's best interests and in pursuing goals (Bassett et al. [<reflink idref="bib9" id="ref114">9</reflink>])—and measures students' capabilities in planning ahead for academic success; therefore, we are capturing similar elements and student mentalities as did Bowman et al. ([<reflink idref="bib15" id="ref115">15</reflink>]) for the non-cognitive attributes latent variable.</p> <hd id="AN0146556866-13">Social Adjustment</hd> <p>To obtain elements of students' social adjustment and engagement we used three scales. First, to gauge intended interactions and developed relationships with peers we used Pascarella and Terenzini's ([<reflink idref="bib72" id="ref116">72</reflink>]) peer-group interaction scale. This scale measures the degree to which students believe that developing relationships with peers will result in closeness, friendships, and personal growth. A sample item from this scale is: <emph>I expect to develop close personal relationships with other students</emph> (Pascarella and Terenzini [<reflink idref="bib72" id="ref117">72</reflink>]). We also used Pascarella and Terenzini's ([<reflink idref="bib72" id="ref118">72</reflink>]) faculty interaction subscale. This scale captures the degree to which students believe their relationships will correlate to benefits and personal growth—which the original authors linked to performance and persistence. In more recent research, this scale has illustrated high reliability ranging from 0.82 to 0.86 (Baker et al. [<reflink idref="bib7" id="ref119">7</reflink>]; French and Oakes [<reflink idref="bib37" id="ref120">37</reflink>])—a sample item is: <emph>My interactions with faculty will have a positive influence on my personal growth, values, and attitudes</emph>. To capture the influence of staff interactions the faculty interaction scale was modified by changing "faculty" to staff—thus, creating a mirrored staff interaction scale which is highly reliable (<emph>a</emph> = 0.88, Collier et al. [<reflink idref="bib27" id="ref121">27</reflink>]).</p> <p>Altogether the Bowman model's 'social adjustment' latent construct captured peer connections and social integration. The connections between our peer-group interaction scale and peer connections are obvious; the important connections between our scales and integration are less intuitive. However, each of the scales house items that implicitly and explicitly gauge satisfaction with the relationships, for example the peer-group interaction scale includes: <emph>I expect that student friendships I will develop at this university will be personally satisfying.</emph> Each of these scales are included in both models. As in Bowman's model, because we captured these data early in the first semester, interaction and cognitive engagement scales may be capturing intent only, and not necessarily how students had engaged throughout the academic year.</p> <hd id="AN0146556866-14">Amotivation (Intent to Persist Substitute)</hd> <p>To determine students' commitment to the institution and commitment to persist to degree, we gauged amotivation by using the amotivation sub-scale from the Academic Motivation Scale for College Students (AMS-C, Vallerand et al. [<reflink idref="bib92" id="ref122">92</reflink>]). Prior reliability for the scale is high (<emph>a</emph> = 0.85, Collier et al. [<reflink idref="bib27" id="ref123">27</reflink>]). A sample item is: <emph>I don't know why I attend class, I really feel that I am wasting my time.</emph> Our measurement departs from Bowman and associates' measurement in that they used a 3-item scale to capture the degree to which students specifically intended to persist to degree at the institution at which they were enrolled. The amotivation scale does not capture that exact sentiment. In addition to being a validated and widely used scale, amotivation shows empirical links to collegiate performance and persistence net of other academic and noncognitive characteristics (Collier et al. [<reflink idref="bib27" id="ref124">27</reflink>]; Warden and Myers [<reflink idref="bib93" id="ref125">93</reflink>]). As a result, we argue that amotivation <emph>may</emph> fill the same structural role in the model as intention to persist.</p> <hd id="AN0146556866-15">Institutional Data</hd> <p>Finally, to populate the rest of the Bowman model, we accessed institutional reports for students' high school GPA, overall college GPA, and class enrollments for Fall 2019—which indicates a traditional measurement of first-year to second-year persistence. The high school GPA and overall college GPA are scored on a 4.0 scale and we denote enrollment in Fall 2019 as a binary outcome of enrolling in at least 1 credit hour: 0 = no, 1 = yes.</p> <hd id="AN0146556866-16">Data Manipulation and SEM Analysis</hd> <p></p> <hd id="AN0146556866-17">Missing Data</hd> <p>We calculated survey constructs (means) allowing a single item to have been skipped (or a pair of items for particularly large batteries), but if a respondent skipped several items on a given construct, that student would have a missing value for that construct instead of a mean based on only the questions that they answered. After applying this method, we had <emph>N</emph> = 498 records of domestic first-time students with values for all of the psychosocial constructs. Of our survey respondents otherwise eligible for inclusion, <emph>n</emph> = 11 participants (2.3%) had missing high school-related institutional data. As the sample of those with missing data were small, we opted to not include them in this analysis—therefore, we listwise deleted those cases. Although listwise deletion is an accepted method of dealing with missing data, there are some drawbacks. One drawback is the reduction of sample size and another is the potential to generate biased parameter estimates (Wothke [<reflink idref="bib95" id="ref126">95</reflink>]) and larger standard errors (Allison [<reflink idref="bib2" id="ref127">2</reflink>]). Yet, given how small our missing data sample size was, we were comfortable in engaging this technique.</p> <hd id="AN0146556866-18">Reverse Scoring</hd> <p>Our instrument contained several scales in which higher scores represent an undesirable outcome—for example, an increased score of amotivation represents a lack of purpose in college, which typically correlates with negative performance or persistence. For ease of interpretation within the SEM model and subsequent tables, we reversed-scored these scales. Now <emph>the higher score represents the more desirable outcome</emph> for: (<reflink idref="bib1" id="ref128">1</reflink>) amotivation, (<reflink idref="bib2" id="ref129">2</reflink>) financial stress, and (<reflink idref="bib3" id="ref130">3</reflink>) food insecurity (security). As we have reverse coded these variables, from here onward we will refer to amotivation as 'motivation' and the latent variable predicting financial stress and food security as 'financial security'—given that in combination the financial stress and food security scales measure the degree to which students feel and exhibit confidence or anxiety regarding their financial situations[<reflink idref="bib2" id="ref131">2</reflink>].</p> <hd id="AN0146556866-19">SEM Analysis</hd> <p>To generate the SEM models, we used the JASP software, which contains a Lavaan module—a SEM library found in R and previously used by Bowman et al. ([<reflink idref="bib15" id="ref132">15</reflink>]). Again, in accordance with Bowman and to deal with the binary outcome of persistence, our SEM models were generated using a weighted least square means and variance-adjusted (WLSMV) approach. A WLSMV approach is also a diagonally-weighted least squares (DWLS) estimator with robust standard errors (Guenole and Brown [<reflink idref="bib41" id="ref133">41</reflink>]). This technique was selected in part to follow the Bowman study, but mostly because a robust DWLS estimator outperforms a maximum likelihood estimator, especially for binary outcome variables, by producing more accurate factor loading estimates, interfactor correlations, and structural coefficient estimates (DiStefano and Morgan [<reflink idref="bib30" id="ref134">30</reflink>]; see also Beauducel and Herzberg [<reflink idref="bib12" id="ref135">12</reflink>]; Flora and Curran [<reflink idref="bib36" id="ref136">36</reflink>]; Li [<reflink idref="bib56" id="ref137">56</reflink>]). Finally, for our sample of under <emph>N</emph> = 1,000, WLSMV should produce more precise Chi-square statistics than the more default maximum likelihood estimation would (Li [<reflink idref="bib57" id="ref138">57</reflink>]). For direct, indirect, and total effects, we report the robust standardized estimates.</p> <p>We reported five model fit statistics for each model. Unlike most areas of quantitative modeling and analysis, there is no commonly agreed-upon single test for what SEM model is better or best. Instead, more than one model may have identical fit statistics, multiple measures may disagree with each other about which of two dissimilar models better fits the data, and selecting among models may include more subjectivity and judgment than in other areas of statistics (Marsh et al. [<reflink idref="bib62" id="ref139">62</reflink>]; Tomarken and Waller [<reflink idref="bib89" id="ref140">89</reflink>]). We reported the Chi-square statistic and associated <emph>p</emph>-value for each model—in which the Lavaan module automatically adjusts for the WLSMV model producing a scaled statistic (Satorra [<reflink idref="bib81" id="ref141">81</reflink>]).</p> <p>We also reported two goodness-of-fit indices, the Comparative Fit Index (CFI) and the Tucker-Lewis Index (TLI). Additionally, we reported two badness-of-fit measures: the Root Mean Square Error of Approximation (RMSEA) and the Standardized Root Mean Square Residual (SRMR). Note that badness-of-fit measures may increase with no decrease (or even with increases) in goodness-of-fit. We reported each of these measurements because debate remains regarding which fit statistic is "best" or most accurate. For example, although the Chi-square statistic is a traditional means of reporting model fit, the statistic is highly sensitive to sample sizes of over <emph>N</emph> = 200 (Bentler and Bonett [<reflink idref="bib14" id="ref142">14</reflink>]), and normality—which may result in model rejection even if the goodness and badness of fit indices are strong (Brown [<reflink idref="bib17" id="ref143">17</reflink>]; Kenny and McCoach [<reflink idref="bib50" id="ref144">50</reflink>]). For these reasons, researchers employing SEM with larger sample should not solely or heavily rely on the Chi-square statistic. As such, we also used the combination of commonly agreed upon goodness and badness of fit statistics to determine model acceptance, examine changes in the models, and to discuss which model may be most optimal. For an in-depth understanding of these fit indices see Hooper et al. ([<reflink idref="bib45" id="ref145">45</reflink>]). Best practice in SEM analysis in general is to consider a model to have good fit if it meets thresholds across all four indices: TLI ≥ 0.95, CFI ≥ 0.95, RMSEA ≤ 0.06, and SRMR ≤ 0.07 (Bagozzi and Yi [<reflink idref="bib8" id="ref146">8</reflink>]; Fan and Sivo [<reflink idref="bib32" id="ref147">32</reflink>]; Hu and Bentler [<reflink idref="bib46" id="ref148">46</reflink>], [<reflink idref="bib47" id="ref149">47</reflink>]; Shi et al. [<reflink idref="bib84" id="ref150">84</reflink>]). Recent analysis indicates that perhaps the cutoffs need to be more stringent when using WLSMV instead of MLE (see Xia and Yang [<reflink idref="bib96" id="ref151">96</reflink>]); as a result, we refer to the 'optimal' range as TLI ≥ 0.98, CFI ≥ 0.98, RMSEA ≤ 0.03, and SRMR ≤ 0.05.</p> <p>As stated, we generated two SEM models for comparison and to understand how the inclusion of food security may impact the model. The first model we generated emulates the Bowman model in that the financial security construct consisted solely of the reverse coded financial stress scale score. The second model included food security in a latent financial security variable. Both of our models fit within the "optimal" range and the Chi-square statistic is non-significant, meaning the models should be accepted. Descriptively comparing the fit statistics suggests that Model 1, the model without food security, is of marginally stronger fit as determined by the values of the TLI, RMSEA, and SRMR (see Table 2).</p> <p>The Bowman model findings and robustly estimated DWLS (WLSMV) SEM models</p> <p> <ephtml> <table frame="hsides" rules="groups"><thead><tr><th align="left" /><th align="left" colspan="3"><p>Bowman model</p></th><th align="left" colspan="3"><p>Model 1</p></th><th align="left" colspan="3"><p>Model 2</p></th></tr><tr><th align="left" /><th align="left"><p>Direct</p></th><th align="left"><p>Indirect</p></th><th align="left"><p>Total</p></th><th align="left"><p>Direct</p></th><th align="left"><p>Indirect</p></th><th align="left"><p>Total</p></th><th align="left"><p>Direct</p></th><th align="left"><p>Indirect</p></th><th align="left"><p>Total</p></th></tr></thead><tbody><tr><td align="left" colspan="4"><p>Non-cognitive attributes</p></td><td char="." align="char" /><td char="." align="char" /><td char="." align="char" /><td align="left" /><td char="." align="char" /><td char="." align="char" /></tr><tr><td align="left"><p> Financial security (means)</p></td><td align="left"><p>0.25***</p></td><td align="left" /><td align="left"><p>0.25***</p></td><td char="." align="char"><p>0.03</p></td><td char="." align="char" /><td char="." align="char"><p>0.03</p></td><td align="left"><p>0.08</p></td><td char="." align="char" /><td char="." align="char"><p>0.08</p></td></tr><tr><td align="left"><p>High school GPA</p></td><td align="left" /><td align="left" /><td align="left" /><td char="." align="char" /><td char="." align="char" /><td char="." align="char" /><td align="left" /><td char="." align="char" /><td char="." align="char" /></tr><tr><td align="left"><p> Financial security</p></td><td align="left"><p>0.12***</p></td><td align="left"><p>0.03***</p></td><td align="left"><p>0.16***</p></td><td char="." align="char"><p>0.09<sup>+</sup></p></td><td char="." align="char"><p>0.01</p></td><td char="." align="char"><p>0.09<sup>+</sup></p></td><td align="left"><p>0.25***</p></td><td char="." align="char"><p>0.01</p></td><td char="." align="char"><p>0.25***</p></td></tr><tr><td align="left"><p> Non-cognitive attributes</p></td><td align="left"><p>0.13***</p></td><td align="left" /><td align="left"><p>0.13***</p></td><td char="." align="char"><p>0.10<sup>+</sup></p></td><td char="." align="char" /><td char="." align="char"><p>0.10<sup>+</sup></p></td><td align="left"><p>0.08</p></td><td char="." align="char" /><td char="." align="char"><p>0.08</p></td></tr><tr><td align="left"><p>Social adjustment</p></td><td align="left" /><td align="left" /><td align="left" /><td char="." align="char" /><td char="." align="char" /><td char="." align="char" /><td align="left" /><td char="." align="char" /><td char="." align="char" /></tr><tr><td align="left"><p> Financial security</p></td><td align="left"><p>0.13***</p></td><td align="left"><p>0.10***</p></td><td align="left"><p>0.23***</p></td><td char="." align="char"><p>− 0.05</p></td><td char="." align="char"><p>0.07+</p></td><td char="." align="char"><p>0.03</p></td><td align="left"><p>0.11*</p></td><td char="." align="char"><p>0.07<sup>+</sup></p></td><td char="." align="char"><p>0.17**</p></td></tr><tr><td align="left"><p> Non-cognitive attributes</p></td><td align="left"><p>0.41***</p></td><td align="left"><p>− 0.00</p></td><td align="left"><p>0.40***</p></td><td char="." align="char"><p>0.64***</p></td><td char="." align="char"><p>0.01</p></td><td char="." align="char"><p>0.65***</p></td><td align="left"><p>0.62***</p></td><td char="." align="char"><p>0.01</p></td><td char="." align="char"><p>0.63***</p></td></tr><tr><td align="left"><p> High school GPA</p></td><td align="left"><p>− 0.03</p></td><td align="left" /><td align="left"><p>− 0.03</p></td><td char="." align="char"><p>0.10*</p></td><td char="." align="char" /><td char="." align="char"><p>0.10*</p></td><td align="left"><p>0.07</p></td><td char="." align="char" /><td char="." align="char"><p>0.07</p></td></tr><tr><td align="left"><p>Motivation (commitment to institution)</p></td><td align="left" /><td align="left" /><td align="left" /><td char="." align="char" /><td char="." align="char" /><td char="." align="char" /><td align="left" /><td char="." align="char" /><td char="." align="char" /></tr><tr><td align="left"><p> Financial security</p></td><td align="left" /><td align="left"><p>0.13***</p></td><td align="left"><p>0.13***</p></td><td char="." align="char" /><td char="." align="char"><p>− 0.01</p></td><td char="." align="char"><p>− 0.01</p></td><td align="left" /><td char="." align="char"><p>0.05*</p></td><td char="." align="char"><p>0.05*</p></td></tr><tr><td align="left"><p> Non-cognitive attributes</p></td><td align="left"><p>0.13***</p></td><td align="left"><p>0.17***</p></td><td align="left"><p>0.28***</p></td><td char="." align="char"><p>0.30***</p></td><td char="." align="char"><p>0.10*</p></td><td char="." align="char"><p>0.40***</p></td><td align="left"><p>0.22**</p></td><td char="." align="char"><p>0.14**</p></td><td char="." align="char"><p>0.36***</p></td></tr><tr><td align="left"><p> High school GPA</p></td><td align="left"><p>0.09***</p></td><td align="left"><p>− 0.01</p></td><td align="left"><p>0.08***</p></td><td char="." align="char"><p>0.01</p></td><td char="." align="char"><p>0.01</p></td><td char="." align="char"><p>0.03</p></td><td align="left"><p>0.04</p></td><td char="." align="char"><p>0.01</p></td><td char="." align="char"><p>0.05</p></td></tr><tr><td align="left"><p> Social adjustment</p></td><td align="left"><p>0.38***</p></td><td align="left" /><td align="left"><p>0.38***</p></td><td char="." align="char"><p>0.19*</p></td><td char="." align="char" /><td char="." align="char"><p>0.19*</p></td><td align="left"><p>0.21**</p></td><td char="." align="char" /><td char="." align="char"><p>0.21*</p></td></tr><tr><td align="left"><p>College GPA</p></td><td align="left" /><td align="left" /><td align="left" /><td char="." align="char" /><td char="." align="char" /><td char="." align="char" /><td align="left" /><td char="." align="char" /><td char="." align="char" /></tr><tr><td align="left"><p> Financial security</p></td><td align="left" /><td align="left"><p>0.11***</p></td><td align="left"><p>0.11***</p></td><td char="." align="char" /><td char="." align="char"><p>0.29***</p></td><td char="." align="char"><p>0.29***</p></td><td align="left" /><td char="." align="char"><p>0.51***</p></td><td char="." align="char"><p>0.51***</p></td></tr><tr><td align="left"><p> Non-cognitive attributes</p></td><td align="left"><p>0.28***</p></td><td align="left"><p>0.00</p></td><td align="left"><p>0.28***</p></td><td char="." align="char"><p>0.21**</p></td><td char="." align="char"><p>0.04</p></td><td char="." align="char"><p>0.25***</p></td><td align="left"><p>22**</p></td><td char="." align="char"><p>0.03</p></td><td char="." align="char"><p>0.25***</p></td></tr><tr><td align="left"><p> High school GPA</p></td><td align="left"><p>0.46***</p></td><td align="left"><p>0.00</p></td><td align="left"><p>0.47***</p></td><td char="." align="char"><p>0.49***</p></td><td char="." align="char"><p>− 0.00</p></td><td char="." align="char"><p>0.48***</p></td><td align="left"><p>0.49***</p></td><td char="." align="char"><p>− 0.00</p></td><td char="." align="char"><p>0.48***</p></td></tr><tr><td align="left"><p> Social adjustment</p></td><td align="left"><p>− 0.14***</p></td><td align="left" /><td align="left"><p>− 0.14***</p></td><td char="." align="char"><p>− 0.02</p></td><td char="." align="char" /><td char="." align="char"><p>− 0.02</p></td><td align="left"><p>− 0.02</p></td><td char="." align="char" /><td char="." align="char"><p>− 0.02</p></td></tr><tr><td align="left"><p>Fall to fall retention</p></td><td align="left" /><td align="left" /><td align="left" /><td char="." align="char" /><td char="." align="char" /><td char="." align="char" /><td align="left" /><td char="." align="char" /><td char="." align="char" /></tr><tr><td align="left"><p> Financial security</p></td><td align="left"><p>0.09***</p></td><td align="left"><p>0.08***</p></td><td align="left"><p>0.17***</p></td><td char="." align="char"><p>0.16*</p></td><td char="." align="char"><p>− 0.03</p></td><td char="." align="char"><p>0.14<sup>+</sup></p></td><td align="left"><p>0.12</p></td><td char="." align="char"><p>0.02</p></td><td char="." align="char"><p>0.13<sup>+</sup></p></td></tr><tr><td align="left"><p> Non-cognitive attributes</p></td><td align="left"><p>− 0.15***</p></td><td align="left"><p>0.29***</p></td><td align="left"><p>0.14***</p></td><td char="." align="char"><p>− 0.14</p></td><td char="." align="char"><p>0.44***</p></td><td char="." align="char"><p>0.30*</p></td><td align="left"><p>− 0.15</p></td><td char="." align="char"><p>0.36***</p></td><td char="." align="char"><p>0.21<sup>+</sup></p></td></tr><tr><td align="left"><p> High school GPA</p></td><td align="left" /><td align="left"><p>0.25***</p></td><td align="left"><p>0.25***</p></td><td char="." align="char" /><td char="." align="char"><p>0.50*</p></td><td char="." align="char"><p>0.50*</p></td><td align="left" /><td char="." align="char"><p>0.46*</p></td><td char="." align="char"><p>0.46*</p></td></tr><tr><td align="left"><p> Social adjustment</p></td><td align="left"><p>0.16***</p></td><td align="left"><p>0.01</p></td><td align="left"><p>0.17***</p></td><td char="." align="char"><p>0.26*</p></td><td char="." align="char"><p>− 0.03</p></td><td char="." align="char"><p>0.23</p></td><td align="left"><p>0.25*</p></td><td char="." align="char"><p>− 0.03</p></td><td char="." align="char"><p>0.22<sup>+</sup></p></td></tr><tr><td align="left"><p> Motivation</p></td><td align="left"><p>0.21***</p></td><td align="left" /><td align="left"><p>0.21***</p></td><td char="." align="char"><p>− 0.10</p></td><td char="." align="char" /><td char="." align="char"><p>− 0.10</p></td><td align="left"><p>− 0.11</p></td><td char="." align="char" /><td char="." align="char"><p>− 0.11</p></td></tr><tr><td align="left"><p> College GPA</p></td><td align="left"><p>0.50***</p></td><td align="left" /><td align="left"><p>0.50***</p></td><td char="." align="char"><p>0.47***</p></td><td char="." align="char" /><td char="." align="char"><p>0.47***</p></td><td align="left"><p>0.46***</p></td><td char="." align="char" /><td char="." align="char"><p>0.46***</p></td></tr><tr><td align="left" /><td align="left" colspan="3"><p>CFI = 0.98</p><p>TLI = 0.97</p><p>RMSEA = 0.05</p><p>SRMR = 0.03</p></td><td align="left" colspan="3"><p>X<sup>2</sup> = 28.65</p><p>df = 23</p><p>CFI = 0.99</p><p>TLI = 0.99</p><p>RMSEA = 0.02</p><p>SRMR = 0.04</p><p><italic>X</italic><sup><italic>2</italic></sup> p = 0.192</p></td><td align="left" colspan="3"><p>X<sup>2</sup> = 43.39</p><p>df = 31</p><p>CFI = 0.99</p><p>TLI = 0.98</p><p>RMSEA = 0.03</p><p>SRMR = 0.05</p><p><italic>X</italic><sup><italic>2</italic></sup><italic> p</italic> = 0.069</p></td></tr></tbody></table> </ephtml> </p> <p> <sups>+</sups>≤ 0.10, *≤ 0.05, **≤ 0.01. ***≤ 0.001</p> <p>However, the goodness-of-fit statistics do not measure whether the two models are <emph>statistically</emph> different. To directly compare our WLSMV models, we had to use the DIFFtest function found in Mplus (neither JASP nor the Lavaan module allows for this test). We used the SBDIFF.EXE (Crawford and Henry [<reflink idref="bib28" id="ref152">28</reflink>]) program which allowed us to manually input the unadjusted and the scaled Satorra-Bentler ([<reflink idref="bib82" id="ref153">82</reflink>]) Chi-square statistics and <emph>df</emph> of both models to calculate whether these models are statistically different—which they were not (<emph>p</emph> = 0.053).[<reflink idref="bib3" id="ref154">3</reflink>] Based upon this test, and the other goodness of fit statistics, we conclude that both models are in the optimal range for all goodness-of-fit indices and that Model 1 is marginally but not significantly stronger than the model with food security included.</p> <hd id="AN0146556866-20">Limitations</hd> <p>First, unlike Bowman et al. ([<reflink idref="bib15" id="ref155">15</reflink>]) research, this is a single-institution study. We concede that this study does not possess the same degree of generalizability as does the Bowman study, and that our findings may only be generalizable to large, public institutions that are like the study site. That limitation is of greater importance to our second research question than to our first. Next, our study possesses the same limitations as the Bowman study: (<reflink idref="bib1" id="ref156">1</reflink>) our grade and retention data were collected well after the instrument was deployed; therefore, students engaged with the all constructs of this study at the same time, and (<reflink idref="bib2" id="ref157">2</reflink>) we only measure persistence within WMU and do not capture whether students persist in institutions outside of the site institution. Finally, recent comparative analysis revealed that fielding the USDA food security scale with two screening items designates about half as many students who are food insecure as when such a screener—which we did not employ—is not used (Nikolaus et al. [<reflink idref="bib69" id="ref158">69</reflink>]). This may be due to differences in how student populations interpret question phrasing (Nikolaus et al. [<reflink idref="bib70" id="ref159">70</reflink>]). However, given that the true prevalence of food insecurity among college students remains unknown, that empirical work has established important links between student responses on the 6-item USDA food security scale and a host of negative outcomes such as psychosocial health (Raskind et al. [<reflink idref="bib76" id="ref160">76</reflink>]), lower GPA and considering disruptions to studies (Collier et al. [<reflink idref="bib26" id="ref161">26</reflink>]; Phillips et al. [<reflink idref="bib74" id="ref162">74</reflink>]) and that the USDA scale without screener has been the dominant tool in higher education research (Nikolaus et al. [<reflink idref="bib69" id="ref163">69</reflink>], [<reflink idref="bib70" id="ref164">70</reflink>], [<reflink idref="bib71" id="ref165">71</reflink>]; Regan [<reflink idref="bib77" id="ref166">77</reflink>]) and remains so (e.g., Coffino et al. [<reflink idref="bib23" id="ref167">23</reflink>]; Innis et al. [<reflink idref="bib49" id="ref168">49</reflink>]; White [<reflink idref="bib94" id="ref169">94</reflink>]), we remain comfortable with its use in the SEM context, even if an argument could be made that the construct is anxiety about food security rather than food security.</p> <hd id="AN0146556866-21">Findings and Discussion</hd> <p>Table 2 presents the direct, indirect, and total relationships among the factors we examined. Some readers may find it easier to interpret the direct pathways presented visually in Figs. 2 and 3. Model 1 (and Fig. 2) are our test of Bowman's model in a different context. Table 3 shows the R<sups>2</sups> of variables within the model for both Models 1 and 2, revealing minimal differences between the two. Across multiple measures, the model is an optimally good fit for the observed relationships in the data. Each of the model's fit statistics are in the optimal range, illustrating better goodness and badness of fit statistics than the original Bowman model—except for the SRMR statistic. It is common for multiple structures to have good fit when conducting SEM (Marsh et al. [<reflink idref="bib63" id="ref170">63</reflink>], [<reflink idref="bib64" id="ref171">64</reflink>]). This robust finding for data that were collected primarily for other research purposes at a university unrelated to Bowman and associates' work, using different collection instruments than Bowman's team, and with some constructs adjusted (such as the exchange of conscientiousness for grit) emphasizes that the structure is an adaptable and likely suitable structural model for college student persistence.</p> <p>Graph: Fig. 2 Tested Model 1: Solid lines denote significant direct relationships</p> <p>Graph: Fig. 3 Tested Model 2: Solid lines denote significant direct relationships</p> <p>Observables' R<sups>2</sups></p> <p> <ephtml> <table frame="hsides" rules="groups"><thead><tr><th align="left" /><th align="left"><p>Model 1</p></th><th align="left"><p>Model 2</p></th></tr></thead><tbody><tr><td align="left"><p>Cognitive engagement</p></td><td char="." align="char"><p>0.89</p></td><td char="." align="char"><p>0.89</p></td></tr><tr><td align="left"><p>Conscientiousness</p></td><td char="." align="char"><p>0.24</p></td><td char="." align="char"><p>0.24</p></td></tr><tr><td align="left"><p>Faculty interaction</p></td><td char="." align="char"><p>0.57</p></td><td char="." align="char"><p>0.55</p></td></tr><tr><td align="left"><p>Staff interaction</p></td><td char="." align="char"><p>0.64</p></td><td char="." align="char"><p>0.62</p></td></tr><tr><td align="left"><p>Peer group interaction</p></td><td char="." align="char"><p>0.33</p></td><td char="." align="char"><p>0.33</p></td></tr><tr><td align="left"><p>High school GPA</p></td><td char="." align="char"><p>0.02</p></td><td char="." align="char"><p>0.07</p></td></tr><tr><td align="left"><p>(A)motivation</p></td><td char="." align="char"><p>0.15</p></td><td char="." align="char"><p>0.16</p></td></tr><tr><td align="left"><p>Overall college GPA</p></td><td char="." align="char"><p>0.30</p></td><td char="." align="char"><p>0.30</p></td></tr><tr><td align="left"><p>Persist to spring</p></td><td char="." align="char"><p>0.30</p></td><td char="." align="char"><p>0.29</p></td></tr></tbody></table> </ephtml> </p> <hd id="AN0146556866-22">Bowman v. Model 1</hd> <p>Model 1 illustrates that our testing confirms the core finding of the Bowman et al. ([<reflink idref="bib15" id="ref172">15</reflink>]) paper, in that students' non-cognitive attributes are significantly influential to social adjustment (<emph>r</emph> = 0.64), commitment to persist (<emph>r</emph> = 0.30), and college GPA (<emph>r</emph> = 0.21). Although we did not find significant, direct relationships from non-cognitive attributes to persistence (<emph>r</emph> = − 0.14) the degree of correlation to the outcome mimics the Bowman model (<emph>r</emph> = − 0.15). Like the Bowman model (<emph>r</emph> = 0.29), in Model 1 the indirect influence of non-cognitive attributes on persistence is much stronger (<emph>r</emph> = 0.44) than the direct influence and flips the total influence positive: to <emph>r</emph> = 0.14 for Bowman and <emph>r</emph> = 0.30 for Model 1 (the indirect and total influence of the non-cognitive attributes on persistence in Model 1 are both statistically significant). Within Model 1 the total influence of non-cognitive attributes on other variables examined are generally stronger than what Bowman found, suggesting the non-cognitive measurements we included could be a better fit than those which Bowman et al. ([<reflink idref="bib15" id="ref173">15</reflink>]) included. Our larger estimates could stem from ensuring each of the non-cognitive attributes were tabulated via proven scales—notably the inclusion of cognitive engagement produced <emph>R</emph><sups><emph>2</emph></sups> = 0.89. Alternatively, the stronger finding could result from the inclusion of conscientiousness (<emph>R</emph><sups><emph>2</emph></sups> = 0.24) over grit; however, since we did not collect grit we cannot empirically test which construct would be better.</p> <p>Overall, Model 1 confirms longstanding theorizations and descriptive findings that students' non-cognitive attributes directly affect social adjustment (Astin [<reflink idref="bib6" id="ref174">6</reflink>]; Tinto [<reflink idref="bib88" id="ref175">88</reflink>]), and performance and persistence (Beattie et al. [<reflink idref="bib13" id="ref176">13</reflink>]; Gunuc and Kuzu [<reflink idref="bib43" id="ref177">43</reflink>]) while highlighting lesser understood direct and total influences on students' first-year experience that we must better understand. Additionally, our findings of good fit on data collected for another primary purpose, using some different instruments and even exchanging conscientiousness for grit emphasizes that the Bowman structure is a good representation of the experience of first-year college students. Although Model 1 confirms Bowman's central test, this model does not exactly align with the Bowman model—we will elaborate on the unobserved relationships later in this paper.</p> <hd id="AN0146556866-23">Model 1 v. Model 2—the Inclusion of Food Security</hd> <p>Model 2 (and Fig. 3) expands our analysis with food security incorporated into the financial means latent factor. Many of the individual coefficients remain very similar in Model 2 as in Model 1, but two direct relationship changes emphasize the validity and importance of considering food security in the college student experience. In Model 1, the total relationship between financial means and HSGPA is an insignificant <emph>r</emph> = 0.09; in Model 2, the relationship is a significant <emph>r</emph> = 0.25, which is more consistent with the extensive body of research on the relationship between SES and high school performance (Faught et al. [<reflink idref="bib35" id="ref178">35</reflink>]; Maroto et al. [<reflink idref="bib61" id="ref179">61</reflink>]). Model 1 also shows a statistically insignificant negative (<emph>r</emph> = − 0.05), direct relationship between higher financial security and the latent social adjustment construct, a counterintuitive finding. Model 2 instead shows a direct significant, positive relationship (<emph>r</emph> = 0.11) between the latent financial security and social adjustment constructs. Again, this finding better aligns with our general understanding of the relationship between engagement and socio-economic status (Rubin [<reflink idref="bib78" id="ref180">78</reflink>]).</p> <p>The most noticeable increase in indirect influence is between financial security and college GPA. As a reference to the original design, both of our models illustrate higher indirect correlation to college GPA than did Bowman (<emph>r</emph> = 0.11). Consisting of only financial stress, Model 1 illustrated and indirect influence of <emph>r</emph> = 0.29. In Model 2, the correlation strength is increased to <emph>r</emph> = 0.51—becoming a stronger influence on college GPA than high school GPA.</p> <p>Recall that college students' food security is rarely included in published SEM models and has not previously been included in SEM models that predict college persistence. Our findings suggest that at minimum food security is an important indirect influence on first-year overall GPA (bounded by our intent to replicate the structure of the Bowman model, we do not test for a direct relationship). Given the emergence of research linking food security to student non-cognitive attributes and experiences, this finding was expected (Bruening et al. [<reflink idref="bib18" id="ref181">18</reflink>]; Collier et al. [<reflink idref="bib26" id="ref182">26</reflink>]; Raskind et al. [<reflink idref="bib76" id="ref183">76</reflink>]).</p> <p>Our findings continue the conversation that illustrates value in bringing food security into SEM analysis. Given that prior research directly connects food insecurity to performance (Collier et al. [<reflink idref="bib26" id="ref184">26</reflink>]; Martinez et al. [<reflink idref="bib65" id="ref185">65</reflink>]), we should probably expect to see significant, direct influences on college GPA in future models. The optimal fit statistics of Model 2 were marginally weaker than the fit statistics of Model 1, but in direct comparison, Model 1 was not a better-fit model. In addition to aligning with college student experiences and matching policy priorities including food security in the SEM models helped the overall observed direct and indirect relationships align with large bodies of research on how financial security relates to students' behaviors and performance. These substantive theoretical and empirical reasons lead us to propose that Model 2, which includes food security, is overall a beneficial adjustment in structurally modeling students' transition into university.</p> <hd id="AN0146556866-24">Relationships Not Observed</hd> <p>Although we find many similarities in direct influences between variables and to retention—our models are not completely aligned with Bowman's. In both models generated, we did not find direct, significant linkages between financial security and non-cognitive attributes, between the latent non-cognitive variable or commitment to the institution (as substituted by motivation) to first-year persistence. The Bowman model established a relatively strong direct correlation between financial security and students' non-cognitive attributes (<emph>r</emph> = 0.25), but we were unable to do so, with our strongest result yielding <emph>r</emph> = 0.08. Our finding aligns with descriptive research that neither this financial stress scale nor food security were significant correlates of conscientiousness or cognitive engagement (Collier et al. [<reflink idref="bib26" id="ref186">26</reflink>]). It is also possible that both the absence of a significant, direct connection between financial security and students' non-cognitive attributes are a result of the specific non-cognitive attributes we collected.</p> <p>Next, prior work using the conscientiousness and cognitive engagement scales have correlated both measurements with college GPA (Beattie et al. [<reflink idref="bib13" id="ref187">13</reflink>]; Gunuc, [<reflink idref="bib42" id="ref188">42</reflink>]; Komarraju et al. [<reflink idref="bib52" id="ref189">52</reflink>]), but we were unable to identify research that correlates these scales directly to a first-year fall-to-fall retention. Supporting understood trends, we find the non-cognitive latent variable directly influenced college GPA (<emph>r</emph> = 0.21, <emph>r</emph> = 0.22), to a similar degree as found in the Bowman model (<emph>r</emph> = 0.28). Both of our models also align in magnitude of direct relationship between the non-cognitive attributes and persistence (<emph>r</emph> = − 0.14, <emph>r</emph> = − 0.15) though without the statistical significance in Bowman's work (<emph>r</emph> = − 0.15). Confirming a negative direct influence of attributes that research overwhelmingly suggests are extremely beneficial for college students on academic outcomes requires additional examination. Overall, in both models we find the non-cognitive attributes possess a strong, positive indirect influence on retention (<emph>r</emph> = 0.44, <emph>r</emph> = 0.36) but in Model 2 the total influence slides just outside of significance despite possessing a larger magnitude than in Bowman's model (<emph>r</emph> = 0.21). Following Crede et al.'s ([<reflink idref="bib29" id="ref190">29</reflink>]) advice, possibly collecting a combination of grit and conscientiousness may help us understand the unique attributes each scale brings to the model and could help generate better latent variables, which may help build a model that more accurately reflects the experience of first-year students.</p> <hd id="AN0146556866-25">Issues with Motivation as a Measurement of Intent to Persist</hd> <p>We did not find a significant, direct influence from motivation to persistence in either model. Prior theoretical foundations suggest that motivation is a critical component of student performance (Tinto [<reflink idref="bib88" id="ref191">88</reflink>]; Bean [<reflink idref="bib11" id="ref192">11</reflink>]) and related to motivation only one study we have identified suggests the amotivation scale is directly correlated to a general persistence measurement (Collier et al. [<reflink idref="bib27" id="ref193">27</reflink>]), though not a first-year fall-to-fall retention rate. Although we made a theoretical case for motivation to be a proxy for intention to persist, likely collecting intention to persist via motivation only in the first few weeks of the semester is not an ideal way to gauge an influence on persistence—nor does the measurement exactly match an intention to remain at an <emph>institution</emph>. Following Tinto's ([<reflink idref="bib88" id="ref194">88</reflink>]) model structure capturing changes from pre-to-post experience motivation might be a better signal of intention to persist instead of relying on only the pre-experience measurement. A further confounding factor is student swirl—students sometimes start at one institution with the intent to transfer to another. Therefore, many first-year students may intend to persist but not necessarily at this institution, which could be problematic to our models considering that the outcome measurement is institutional persistence. Addressing this issue may include tracking students who transfer to another institution, in a broader measure of persistence. However, we also understand that not all states make this data readily available to individual institutions and that surveying students who have left their first institution may not produce enough data to appropriately capture this more generalized measure of persistence.</p> <p>After developing the two SEM models designed to follow the Bowman model's structure, we ran an exploratory factor analysis (EFA) to examine whether it would make more sense for motivation to be a non-cognitive attribute. The EFA did not group motivation with conscientiousness or cognitive engagement—nor with the social or financial security variables. Based upon prior findings (Collier et al. [<reflink idref="bib27" id="ref195">27</reflink>]; Warden and Myers [<reflink idref="bib93" id="ref196">93</reflink>]), we also tested whether motivation was an indirect influence on persistence through college GPA and found no significant direct relationship between motivation and overall college GPA (<emph>r</emph> = − 0.04) and no significant indirect relationship between motivation and persistence through college GPA (<emph>r</emph> = − 0.02)—with optimal fit statistics. Recall that we reverse-coded amotivation where higher scores are "better" (or demonstrate lower amotivation), so these point estimates indicate a (statistically insignificant) inverse relationship between the desired lower levels of amotivation and the intended outcomes of higher performance and persistence—see Appendix Table 4. We remain unsure where motivation should be positioned in a SEM predicting first-year fall-to-fall persistence. Future testing is clearly warranted here. In the meantime, we suggest that future studies that include an intention to persist measurement default to the constructs used by Bowman et al. ([<reflink idref="bib15" id="ref197">15</reflink>]) or Cabrera et al. ([<reflink idref="bib20" id="ref198">20</reflink>]).</p> <hd id="AN0146556866-26">Conclusion and Implications</hd> <p>Structural equation models in higher education that examine persistence remain relatively rare. However, when SEM models are developed, they are done so with the intention to be evaluated, replicated, and expanded upon. Despite this intent, almost no structural models are re-tested on data other than that in their original article, more rarely do studies expand upon existing models. This study is one of a few in higher education that tested the structure of an existing SEM model and, based on a combination of theoretical and empirical research, expanded elements of the established model to assess the result of changes. We adjusted two constructs in the model, used different data collection instruments for some constructs, and gathered data at a different institution; but in testing the structure of the Bowman model, found that it fit our data at least as well as Bowman et al. ([<reflink idref="bib15" id="ref199">15</reflink>]), with many of the same direct and indirect pathways. These statistically re-affirmed pathways suggest that the core elements of the Bowman model may not be deeply influenced by institutional setting or type of similar (but not exact) non-cognitive data collected. Our study supports Bowman et al. ([<reflink idref="bib15" id="ref200">15</reflink>]) central understanding that non-cognitive attributes directly and indirectly influence many aspects of students' college experience and performance (namely overall GPA) and that interventions attending to non-cognitive attributes should help students feel more supported—which should result in higher college GPA and (indirectly) first year persistence. Our aligned findings illustrated an impressive amount of accuracy and flexibility in the Bowman model, which suggests that the model should be a strong comparison point for future tests of structures of first-year college student experiences.</p> <p>Although many of the relationships remain intact, our models do not completely confirm the Bowman model's structure. One possibility is that, despite Bowman et al.'s ([<reflink idref="bib15" id="ref201">15</reflink>]) cross-institutional testing across thousands of student profiles, that the Bowman structure does not exactly work for regional 4-year institutions that are predominantly White and tend to enroll students that hail from neighborhoods positioned well above the state and national median income. Another possibility is that the latent noncognitive factor has differing relationships because it does not contain the exact same set of noncognitive traits. The non-cognitive variables we opted to collect versus those found in the Bowman model likely modified relationships, pushing these relationships outside of significance—for example non-cognitive attributes' influences on high school GPA. Until more researchers test the Bowman model, we are left unable to answer definitively why our results differed; whether the model does not necessarily fit the institution, the relationships are sensitive to the type of data used, a combination of both, or neither.</p> <p>Finally, the introduction of food security to the financial security variable is one step towards understanding the direct and indirect influence of this variable on students' non-cognitive attributes, performance, and persistence. The inclusion of food security directly influenced social adjustment and provided a significant boost in correlation to the indirect impact of the financial security variable on college GPA. Overall, the inclusion of this variable suggests the total influence of food security on the college experience is largely left unknown but that by easing students' food insecurity we may produce positive, direct impacts to students' social adjustment and positive, indirect impacts to their overall college GPA. Within the Bowman model's structure, food security is positioned as an influence of social adjustment and performance. Therefore, if this influence is not addressed, interventions that address only the attributes or behaviors we hope students will change may not be very effective.</p> <hd id="AN0146556866-27">Funding</hd> <p>Funding was provided by U.S. Department of Education (Grant No. P116F140353).</p> <hd id="AN0146556866-28">Appendix</hd> <p>See Table 4.</p> <p>Robust estimated DWLS (WLSMV) SEM models</p> <p> <ephtml> <table frame="hsides" rules="groups"><thead><tr><th align="left" /><th align="left" colspan="3"><p>Model 2a</p></th></tr><tr><th align="left" /><th align="left"><p>Direct</p></th><th align="left"><p>Indirect</p></th><th align="left"><p>Total</p></th></tr></thead><tbody><tr><td align="left"><p>Non-cognitive attributes</p></td><td align="left" /><td char="." align="char" /><td char="." align="char" /></tr><tr><td align="left"><p> Financial security</p></td><td align="left"><p>0.07</p></td><td char="." align="char" /><td char="." align="char"><p>0.07</p></td></tr><tr><td align="left"><p>High school GPA</p></td><td align="left" /><td char="." align="char" /><td char="." align="char" /></tr><tr><td align="left"><p> Financial security</p></td><td align="left"><p>0.25***</p></td><td char="." align="char"><p>0.00</p></td><td char="." align="char"><p>0.25***</p></td></tr><tr><td align="left"><p> Non-cognitive attributes</p></td><td align="left"><p>0.09<sup>+</sup></p></td><td char="." align="char" /><td char="." align="char"><p>0.09<sup>+</sup></p></td></tr><tr><td align="left"><p>Social adjustment</p></td><td align="left" /><td char="." align="char" /><td char="." align="char" /></tr><tr><td align="left"><p> Financial security</p></td><td align="left"><p>0.10<sup>+</sup></p></td><td char="." align="char"><p>0.06<sup>+</sup></p></td><td char="." align="char"><p>0.16**</p></td></tr><tr><td align="left"><p> Non-cognitive attributes</p></td><td align="left"><p>62***</p></td><td char="." align="char"><p>0.00</p></td><td char="." align="char"><p>0.63***</p></td></tr><tr><td align="left"><p> High school GPA</p></td><td align="left"><p>0.07</p></td><td char="." align="char" /><td char="." align="char"><p>0.07</p></td></tr><tr><td align="left"><p>Motivation</p></td><td align="left" /><td char="." align="char" /><td char="." align="char" /></tr><tr><td align="left"><p> Financial security</p></td><td align="left" /><td char="." align="char"><p>0.03<sup>+</sup></p></td><td char="." align="char"><p>0.03<sup>+</sup></p></td></tr><tr><td align="left"><p> Non-cognitive attributes</p></td><td align="left"><p>0.26***</p></td><td char="." align="char"><p>0.10*</p></td><td char="." align="char"><p>0.36***</p></td></tr><tr><td align="left"><p> High school GPA</p></td><td align="left"><p>0.00</p></td><td char="." align="char"><p>0.01</p></td><td char="." align="char"><p>0.01</p></td></tr><tr><td align="left"><p> Social adjustment</p></td><td align="left"><p>0.16*</p></td><td char="." align="char" /><td char="." align="char"><p>0.16*</p></td></tr><tr><td align="left"><p>College GPA</p></td><td align="left" /><td char="." align="char" /><td char="." align="char" /></tr><tr><td align="left"><p> Financial security</p></td><td align="left" /><td char="." align="char"><p>0.51***</p></td><td char="." align="char"><p>0.51***</p></td></tr><tr><td align="left"><p> Non-cognitive attributes</p></td><td align="left"><p>0.24**</p></td><td char="." align="char"><p>0.03</p></td><td char="." align="char"><p>0.27***</p></td></tr><tr><td align="left"><p> High school GPA</p></td><td align="left"><p>0.49**</p></td><td char="." align="char"><p>0.00</p></td><td char="." align="char"><p>0.49***</p></td></tr><tr><td align="left"><p> Social adjustment</p></td><td align="left"><p>− 0.01</p></td><td char="." align="char"><p>− 0.01</p></td><td char="." align="char"><p>− 0.02</p></td></tr><tr><td align="left"><p> Motivation</p></td><td align="left"><p>− 04</p></td><td char="." align="char" /><td char="." align="char"><p>− 0.04</p></td></tr><tr><td align="left"><p>Spring to fall retention</p></td><td align="left" /><td char="." align="char" /><td char="." align="char" /></tr><tr><td align="left"><p> Financial security</p></td><td align="left"><p>0.14<sup>+</sup></p></td><td char="." align="char"><p>0.00</p></td><td char="." align="char"><p>0.14*</p></td></tr><tr><td align="left"><p> Non-cognitive attributes</p></td><td align="left"><p>− 0.15</p></td><td char="." align="char"><p>0.41***</p></td><td char="." align="char"><p>0.26*</p></td></tr><tr><td align="left"><p> High school GPA</p></td><td align="left" /><td char="." align="char"><p>0.43*</p></td><td char="." align="char"><p>0.43*</p></td></tr><tr><td align="left"><p> Social adjustment</p></td><td align="left"><p>0.24*</p></td><td char="." align="char"><p>− 0.02</p></td><td char="." align="char"><p>0.22<sup>+</sup></p></td></tr><tr><td align="left"><p> Motivation</p></td><td align="left"><p>− 0.10</p></td><td char="." align="char"><p>− 0.02</p></td><td char="." align="char"><p>− 0.12</p></td></tr><tr><td align="left"><p> College GPA</p></td><td align="left"><p>0.45***</p></td><td char="." align="char" /><td char="." align="char"><p>0.45***</p></td></tr><tr><td align="left" /><td align="left" colspan="3"><p>X<sup>2</sup> = 37.20</p><p>df = 29</p><p>CFI = 0.99</p><p>TLI = 0.99</p><p>RMSEA = 0.03</p><p>SRMR = 0.04</p><p><italic>X</italic><sup><italic>2</italic></sup><italic> p</italic> = 0.141</p></td></tr></tbody></table> </ephtml> </p> <p> <sups>+</sups>≤ 0.10, *≤ 0.05, **≤ 0.01. ***≤ 0.001</p> <hd id="AN0146556866-29">Publisher's Note</hd> <p>Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.</p> <ref id="AN0146556866-30"> <title> References </title> <blist> <bibl id="bib1" idref="ref41" type="bt">1</bibl> <bibtext> Alaimo K, Olson CM, Frongillo EA. Food insufficiency and the American school-aged children's cognitive, academic, and psychosocial development. Pediatrics. 2001; 108; 1: 44-53</bibtext> </blist> <blist> <bibl id="bib2" idref="ref62" type="bt">2</bibl> <bibtext> Allison PD. Missing data. 2001: Thousand Oaks, CA; Sage Publications</bibtext> </blist> <blist> <bibl id="bib3" idref="ref63" type="bt">3</bibl> <bibtext> Appleton JJ, Christenson SL, Kim D, Reschly AL. Measuring cognitive and psychological engagement: Validation of the student engagement instrument. Journal of School Psychology. 2006; 44; 5: 427-445</bibtext> </blist> <blist> <bibl id="bib4" idref="ref14" type="bt">4</bibl> <bibtext> Astin AW. The methodology of research on college impact, part one. Sociology of Education. 1970; 43; 3: 223-254</bibtext> </blist> <blist> <bibl id="bib5" idref="ref1" type="bt">5</bibl> <bibtext> Astin AW. Student involvement: A development theory for higher education. Journal of College Student Development. 1984; 40: 518-529</bibtext> </blist> <blist> <bibl id="bib6" idref="ref80" type="bt">6</bibl> <bibtext> Astin AW. What matters in college: Four critical years revisited. 1993: San Francisco; Jossey-Bass</bibtext> </blist> <blist> <bibl id="bib7" idref="ref119" type="bt">7</bibl> <bibtext> Baker BA, Caison AL, Meade AW. Assessing gender-related differential item functioning and predictive validity with the institutional integration scale. Educational and psychological measurement. 2007; 67; 3: 545-559</bibtext> </blist> <blist> <bibl id="bib8" idref="ref146" type="bt">8</bibl> <bibtext> Bagozzi RP, Yi Y. Specification, evaluation, and interpretation of structural equation models. Journal of the Academy of Marketing Science. 2012; 40; 1: 8-34</bibtext> </blist> <blist> <bibl id="bib9" idref="ref114" type="bt">9</bibl> <bibtext> Bassett HH, Denham S, Wyatt TM, Warren-Khot HK. Refining the preschool self-regulation assessment for use in preschool classrooms. Infant and Child Development. 2012; 21; 6: 596-616</bibtext> </blist> <blist> <bibtext> Bean JP. Dropouts and turnover: The synthesis and test of a causal model of student attrition. Research in Higher Education. 1980; 12; 2: 155-187</bibtext> </blist> <blist> <bibtext> Bean JP. Nine themes of college student retention. College student retention: Formula for student success. 2005: Westport, CT; Praeger: 215-244</bibtext> </blist> <blist> <bibtext> Beauducel A, Herzberg PY. On the performance of maximum likelihood versus means and variance adjusted weighted least squares estimation in CFA. Structural Equation Modeling. 2006; 13: 186-203</bibtext> </blist> <blist> <bibtext> Beattie G, Laliberte JWP, Oreopoulos P. Thrivers and divers: Using non-academic measures to predict college success and failure. Economics of Education. 2018; 62: 170-180</bibtext> </blist> <blist> <bibtext> Bentler PM, Bonett DG. Significance tests and goodness of fit in the analysis of covariance structures. Psychological Bulletin. 1980; 88; 3: 588-606</bibtext> </blist> <blist> <bibtext> Bowman NA, Miller A, Woosley S, Maxwell NP, Kolze MJ. Understanding the link between noncognitive attributes and college retention. Research in Higher Education. 2019; 60: 135-152</bibtext> </blist> <blist> <bibtext> Braxton JM, Doyle WR, Hartley HV III, Hirschy AS, Jones WA, McLendon MK. Rethinking college student retention. 2014: San Francisco, CA; Jossey-Bass</bibtext> </blist> <blist> <bibtext> Brown TA. Confirmatory factor analysis for applied research. 20152: New York, NY; Guilford Press</bibtext> </blist> <blist> <bibtext> Bruening M, van Woerden I, Todd M, Laska MN. Hungry to learn: The prevalence and effects of food insecurity on health behaviors and outcomes over time among a diverse sample of university freshmen. Journal of Behavioral Nutrition and Physical Activity. 2018; 15; 9: 1-10</bibtext> </blist> <blist> <bibtext> Bryant, F. B, & Satorra, A. (2013). EXCEL macro file for conducting scaled difference Chi-square tests via LISREL 8, LISRAEL 9, EQS, and Mplus. Retrieved September 9, 2020, from https://<ulink href="http://www.econ.upf.edu/~satorra/dades/BryantSatorraScaledDifferenceTestsForLISREL8LISREL9EQSandMplus.xls">www.econ.upf.edu/~satorra/dades/BryantSatorraScaledDifferenceTestsForLISREL8LISREL9EQSandMplus.xls</ulink>.</bibtext> </blist> <blist> <bibtext> Cabrera AF, Nora A, Castañeda MB. College persistence: Structural equations modeling test of an integrated model of student retention. Journal of Higher Education. 1993; 64; 2: 123-138</bibtext> </blist> <blist> <bibtext> Camelo K, Elliott M. Food insecurity and academic achievement among college students at a public university in the United States. Journal of College Student Development. 2019; 60; 3: 307-318</bibtext> </blist> <blist> <bibtext> Christensen R, Knezek G, Tyler-Wood T. Student perceptions of science, technology, engineering and mathematics (STEM) content and careers. Computers in Human Behavior. 2014; 34: 173-186</bibtext> </blist> <blist> <bibtext> Coffino JA, Spoor SP, Drach RD, Hormes JM. Food insecurity among graduate students: Prevalence and association with depression, anxiety and stress. Public Health Nutrition. 2020</bibtext> </blist> <blist> <bibtext> Cohen, R. M. (2015). Will students soon be tested for "grit"? The American Prospect. Retrieved September 9, 2020, from https://prospect.org/blog/tapped/will-students-soon-be-tested-grit.</bibtext> </blist> <blist> <bibtext> Coleman-Jensen, A, Rabbitt, M. P, Gregory, C. A, & Singh, A. (2016). Household food security in the United States in 2015 (EER-215). U.S. Department of Agriculture, Economic Research Service. Retrieved September 9, 2020, from https://<ulink href="http://www.basicknowledge101.com/pdf/health/FoodSecurity.pdf">www.basicknowledge101.com/pdf/health/FoodSecurity.pdf</ulink>.</bibtext> </blist> <blist> <bibtext> Collier, D. A, Fitzpatrick, D, Brehm, C, & Parnther, C. (2019a). Coming to college hungry: Correlating food insecurity to amotivation, stress, engagement and first-semester performance in a 4-year university (Working Paper No. 3490080). SSRN. 10.2139/ssrn.3490080</bibtext> </blist> <blist> <bibtext> Collier, D. A, Parnther, C, Fitzpatrick, D, Brehm, C, & Beach, A. (2019b). Helping students keep the promise: Exploring how Kalamazoo Promise scholars' basic needs, motivation, and engagement correlate to performance and persistence in a 4-year institution. Innovative Higher Education. Retrieved September 9, 2020, from 10.1007/s10755-019-9468-0.</bibtext> </blist> <blist> <bibtext> Crawford JR, Henry JD. The depression anxiety stress scales: Normative data and latent structure in a large non-clinical sample. British Journal of Clinical Psychology. 2003; 42: 111-131</bibtext> </blist> <blist> <bibtext> Crede M, Tynan MC, Harms PD. Much ado about grit: A meta-analytic synthesis of the grit literature. Journal of Personality and Social Psychology. 2017; 113; 3: 492-511</bibtext> </blist> <blist> <bibtext> DiStefano C, Morgan GB. A comparison of diagonal weighted least squares robust estimation techniques for ordinal data. Structural Equation Modeling: A Multidisciplinary Journal. 2014; 21; 3: 425-438</bibtext> </blist> <blist> <bibtext> Duckworth AL, Peterson C, Matthews MD, Kelly DR. Grit: Perseverance and passion for long-term goals. Journal of Personality and Social Psychology. 2007; 92; 6: 108</bibtext> </blist> <blist> <bibtext> Fan X, Sivo SA. Sensitivity of fit indices to model misspecification and model types. Multivariate Behavioral Research. 2007; 42; 3: 509-529</bibtext> </blist> <blist> <bibtext> Farrell L, Fry TR, Risse L. The significance of financial self-efficacy in explaining women's personal finance behaviour. Journal of Economic Psychology. 2016; 54: 85-99</bibtext> </blist> <blist> <bibtext> Farrington CA, Roderick M, Allensworth E, Nagaoka J, Keyes TS, Johnson DW. Teaching adolescents to become learners. The role of noncognitive factors in shaping school performance: A critical literature review. 2012: Chicago; University of Chicago Consortium on Chicago School Research</bibtext> </blist> <blist> <bibtext> Faught EL, Williams PL, Willows ND, Asbridge M, Veugelers PJ. The association between food insecurity and academic achievement in Canadian school-aged children. Public Health Nutrition. 2017; 20; 15: 2778-2785</bibtext> </blist> <blist> <bibtext> Flora DB, Curran PJ. An empirical evaluation of alternative methods of estimation for confirmatory factor analysis with ordinal data. Psychological Methods. 2004; 9: 466-491</bibtext> </blist> <blist> <bibtext> French BF, Oakes W. Reliability and validity evidence for the institutional integration scale. Educational and Psychological Measurement. 2004; 64; 1: 88-98</bibtext> </blist> <blist> <bibtext> Gaines A, Robb CA, Knol LL, Sickler S. Examining the role of financial factors, resources, and skills in predicting food security status among college students. International Journal of Consumer Studies. 2014; 38; 4: 374-383</bibtext> </blist> <blist> <bibtext> Goldrick-Rab S. Paying the price: College costs, financial aid, and the betrayal of the American dream. 2016: Chicago, IL; The University of Chicago Press</bibtext> </blist> <blist> <bibtext> Goldrick-Rab, S, Richardson, J, Schneider, J, Hernandez, A, & Cady, C. (2018). Still hungry and homeless in college. Madison, WI: Wisconsin Hope Lab. Retrieved September 9, 2020, from https://<ulink href="http://www.theotx.org/wp-content/uploads/2018/05/Wisconsin-HOPE-Lab-Still-Hungry-and-Homeless.pdf">www.theotx.org/wp-content/uploads/2018/05/Wisconsin-HOPE-Lab-Still-Hungry-and-Homeless.pdf</ulink>.</bibtext> </blist> <blist> <bibtext> Guenole N, Brown A. The consequences of ignoring measurement invariance for path coefficients in structural equation models. Frontier in Psychology. 2014</bibtext> </blist> <blist> <bibtext> Gunuc S. The relationships between student engagement and their academic achievement. International Journal on New Trends in Education and Their Implications. 2014; 5; 4: 199-214</bibtext> </blist> <blist> <bibtext> Gunuc S, Kuzu A. Student engagement scale: Development, reliability and validity. Assessment & Evaluation in Higher Education. 2015; 40; 4: 587-610</bibtext> </blist> <blist> <bibtext> Harper, S. R. (2012). Black male student success in higher education: A report from the National Black Male College Achievement Study. University of Pennsylvania, Graduate School of Education, Center for the Study of Race and Equity in Education. Retrieved September 9, 2020, from https://web-app.usc.edu/web/rossier/publications/231/Harper%2520(2012)%2520Black%2520Male%2520Success.pdf.</bibtext> </blist> <blist> <bibtext> Hooper D, Coughlan J, Mullen MR. Structural equation modeling: Guidelines for determining model fit. The Electronic Journal of Business Research Methods. 2008; 6: 53-60</bibtext> </blist> <blist> <bibtext> Hu LT, Bentler PM. Fit indices in covariance structure modeling: Sensitivity to underparameterized model misspecification. Psychological Methods. 1998; 3; 4: 424</bibtext> </blist> <blist> <bibtext> Hu LT, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: a Multidisciplinary Journal. 1999; 6; 1: 1-55</bibtext> </blist> <blist> <bibtext> Ivcevic Z, Brackett M. Predicting school success: Comparing conscientiousness, grit, and emotion regulation ability. Journal of Research in Personality. 2014; 52: 29-36</bibtext> </blist> <blist> <bibtext> Innis JA, Bishop M, Boloudakis S. Food insecurity and community college students. Community College Journal of Research and Practice. 2020; 44; 9: 694-699</bibtext> </blist> <blist> <bibtext> Kenny DA, McCoach DB. Effect of the number of variables on measures of fit in structural equation modeling. Structural Equation Modeling. 2003; 10; 3: 333-351</bibtext> </blist> <blist> <bibtext> Kezar A, Maxey D. Faculty matter: So why doesn't everyone think so?. Thought & Action. 2014; 2014: 29-44</bibtext> </blist> <blist> <bibtext> Komarraju M, Karau SJ, Schmeck RR, Avdic A. The big five personality traits, learning styles, and academic achievement. Personality and Individual Differences. 2011; 51; 4: 472-477</bibtext> </blist> <blist> <bibtext> Krause K, Coates H. Students' engagement in first-year university. Assessment and Evaluation in Higher Education. 2008; 33; 5: 493-505</bibtext> </blist> <blist> <bibtext> Kuh GD, Cruce TM, Shoup R, Kinzie J. Unmaking the effects of student engagement on first-year college grades and persistence. The Journal of Higher Education. 2008; 79; 5: 540-563</bibtext> </blist> <blist> <bibtext> Laskey ML, Hetzel CJ. Investigating factors related to retention of at-risk college students. Learning Assistance Review. 2011; 16; 1: 31-43</bibtext> </blist> <blist> <bibtext> Li CH. Confirmatory factor analysis with ordinal data: Comparing robust maximum likelihood and diagonally weighted least squares. Behavior Research Methods. 2016; 48; 3: 936-949</bibtext> </blist> <blist> <bibtext> Li CH. The performance of ML, DWLS, and USL estimation with robust corrections in structural equation models with ordinal variables. Psychological Methods. 2016; 21; 3: 369-387</bibtext> </blist> <blist> <bibtext> Lim H, Heckman SJ, Montalto CP, Letkiewicz J. Financial stress, self-efficacy, and financial help-seeking behavior of college students. Journal of Financial Counseling and Planning. 2014; 25; 2: 148-160</bibtext> </blist> <blist> <bibtext> Lievens F, Ones DS, Dilchert S. Personality scale validates increase throughout medical school. Journal of Applied Psychology. 2009; 94: 1514-1535</bibtext> </blist> <blist> <bibtext> Makel MC, Plucker JA. Facts are more important than novelty: Replication in the education sciences. Educational Researcher. 2014; 43; 6: 304-316</bibtext> </blist> <blist> <bibtext> Maroto ME, Snelling A, Linick H. Food insecurity among community college students. Community College Journal of Research & Practice. 2015; 39; 6: 515-526</bibtext> </blist> <blist> <bibtext> Marsh HW, Balla JR, Hau KT. An evaluation of incremental fit indices: A clarification of mathematical and empirical properties. Advanced structural equation modelling: Issues and techniques. 1996: Mahwah; Lawrence Erlbaum Associates: 315-353</bibtext> </blist> <blist> <bibtext> Marsh HW, Hau KT, Wen Z. In search of golden rules: Comment on hypothesis-testing approaches to setting cutoff values for fit indexes and dangers in overgeneralizing Hu and Bentler's (1999) findings. Structural Equation Modeling. 2004; 11; 3: 320-341</bibtext> </blist> <blist> <bibtext> Marsh HW, Wen Z, Hau KT. Structural equation models of latent interactions: Evaluation of alternative estimation strategies and indicator construction. Psychological Methods. 2004; 9; 3: 275</bibtext> </blist> <blist> <bibtext> Martinez SM, Webb K, Frongillo EA, Ritchie LD. Food insecurity in California's public university system: What are the risk factors?. Journal of Hunger and Environmental Nutrition. 2017; 13; 1: 1-18</bibtext> </blist> <blist> <bibtext> Mihaly, K. (2009). Do more friends mean better grades?: Student Popularity and Academic Achievement. Rand Labor and Population. Retrieved September 9, 2020, from https://pdfs.semanticscholar.org/9edf/ea12e6507cc32ed00045f06ceb64d3d9f5aa.pdf.</bibtext> </blist> <blist> <bibtext> Mukigi D, Brown O. Food insecurity among college students in the United States: A mini review. Annals of Reviews and Research. 2018; 1; 5: 1-6</bibtext> </blist> <blist> <bibtext> Nazmi A, Martinez S, Byrd A, Robinson D, Bianco S, Maguire J. A systematic review of food insecurity among US students in higher education. Journal of Hunger & Environmental Nutrition. 2018</bibtext> </blist> <blist> <bibtext> Nikolaus CJ, Ellison B, Nickols-Richardson SM. Are estimates of food insecurity among college students accurate? Comparison of assessment protocols. PLoS ONE. 2019; 14; 4: e0215161</bibtext> </blist> <blist> <bibtext> Nikolaus CJ, Ellison B, Nickols-Richardson SM. College students' interpretations of food security questions: results from cognitive interviews. BMC public health. 2019; 19; 1: 1282</bibtext> </blist> <blist> <bibtext> Nikolaus CJ, An R, Ellison B, Nickols-Richardson SM. Food insecurity among college students in the united states: A scoping review. Advances in Nutrition. 2020; 11; 2: 327-348</bibtext> </blist> <blist> <bibtext> Pascarella ET, Terenzini PT. Predicting freshman persistence and voluntary dropout decisions from a theoretical model. The Journal of Higher Education. 1980; 51; 1: 60-75</bibtext> </blist> <blist> <bibtext> Patton-Lopez MM, Lopez-Cevallos DF, Cancel-Tirado DI, Vazquez L. Prevalence and correlates of food insecurity among students attending a midsize rural university in Oregon. Journal of Nutrition Education and Behavior. 2014; 46; 3: 209-214</bibtext> </blist> <blist> <bibtext> Phillips E, McDaniel A, Croft A. Food insecurity and academic disruption among college students. Journal of Student Affairs Research and Practice. 2018; 55; 4: 353-372</bibtext> </blist> <blist> <bibtext> Quaye SJ, Harper SR. Student engagement in higher education: Theoretical perspectives and practical approaches for diverse populations. 2014: New York, NY; Routledge</bibtext> </blist> <blist> <bibtext> Raskind IG, Haardorfer R, Berg CJ. Food insecurity, psychosocial health, and academic performance among college and university students in Georgia, USA. Public Health Nutrition. 2019; 22; 3: 476-485</bibtext> </blist> <blist> <bibtext> Regan EP. Food insecurity among college students. Sociology Compass. 2020; 14; 6: e12790</bibtext> </blist> <blist> <bibtext> Rubin M. Social class differences in social integration among students in higher education: A meta-analysis and recommendation for future research. Journal of Diversity in Higher Education. 2012; 5; 1: 22-38</bibtext> </blist> <blist> <bibtext> Ryan RM, Deci EL. Self-determination theory and the facilitation of intrinsic motivation, social development, and well-being. American Psychologist. 2000; 55; 1: 68-78</bibtext> </blist> <blist> <bibtext> Sandeen A, Barr MJ. Critical issues for student affairs: challenges and opportunities. 2014: San Francisco, CA; John Wiley & Sons</bibtext> </blist> <blist> <bibtext> Satorra AHeijmans RDH, Pollock DSG, Satorra A. Scaled and adjusted restricted tests in multi-sample analysis of moment structures. Innovations in multivariate statistical analysis: Advanced studies in theoretical and applied econometrics. 2000: Boston; Springer: 233-247</bibtext> </blist> <blist> <bibtext> Satorra A, Bentler PM. A scaled difference Chi-square test statistic for moment structure analysis. Psychometrika. 2001; 66: 507-514</bibtext> </blist> <blist> <bibtext> Sass DA, Castro-Villareal F, Wilkerson S, Guerra N, Sullivan J. A structural model for predicting student retention. The Review of Higher Education. 2018; 42; 1: 103-135</bibtext> </blist> <blist> <bibtext> Shi D, Lee T, Maydeu-Olivares A. Understanding the model size effect on SEM fit indices. Educational and Psychological Measurement. 2019; 79; 2: 310-334</bibtext> </blist> <blist> <bibtext> Soto CJ, John OP. Short and extra-short forms of the Big Five Inventory-2: The BFI-2-S and the BFI-2-XS. Journal of Research in Personality. 2017; 68: 69-81</bibtext> </blist> <blist> <bibtext> Sparks, S. (2015). 'Nation's report card' to gather data on Grit, Mindset. Education Week. Retrieved September 9, 2020, from https://<ulink href="http://www.edweek.org/ew/articles/2015/06/03/nations-reportcard-to-gather-data-on.html">www.edweek.org/ew/articles/2015/06/03/nations-reportcard-to-gather-data-on.html</ulink>.</bibtext> </blist> <blist> <bibtext> Strayhorn TL. How college students' engagement affects personal and social learning outcomes. Journal of College and Character. 2008; 10; 2: 1-16</bibtext> </blist> <blist> <bibtext> Tinto V. Leaving college: Rethinking the causes and cures of student attrition. 19932: Chicago, IL; University of Chicago Press</bibtext> </blist> <blist> <bibtext> Tomarken AJ, Waller NG. Structural equation modeling: Strengths, limitations, and misconceptions. The Annual Review of Clinical Psychology. 2005; 1: 31-65</bibtext> </blist> <blist> <bibtext> U.S. Department of Agriculture. (2018). Definitions of food security. Retrieved August 3, 2019, from https://<ulink href="http://www.ers.usda.gov/topics/food-nutrition-assistance/food-security-in-the-us/definitions-of-food-security.aspx">www.ers.usda.gov/topics/food-nutrition-assistance/food-security-in-the-us/definitions-of-food-security.aspx</ulink>.</bibtext> </blist> <blist> <bibtext> U.S. Department of Agriculture Economic Research Service. (2012). U.S. adult food security survey module: Three-stage design, with screeners. Retrieved September 9, 2020, from https://<ulink href="http://www.ers.usda.gov/media/8279/ad2012.pdf">www.ers.usda.gov/media/8279/ad2012.pdf</ulink>.</bibtext> </blist> <blist> <bibtext> Vallerand RJ, Pelletier LJ, Blais MR, Briere NM, Senecal C, Vallieres EF. The academic motivation scale: A measure of intrinsic, extrinsic, and amotivation in education. Educational and Psychological Measurement. 1992; 52; 4: 1003-1017</bibtext> </blist> <blist> <bibtext> Warden DN, Myers CA. Nonintellective variables and nontraditional college students: A domain-based investigation of academic achievement. College Student Journal. 2017; 51; 3: 380-390</bibtext> </blist> <blist> <bibtext> White SS. Student food insecurity and the social equity pillar of campus sustainability. International Journal of Sustainability in Higher Education. 2020</bibtext> </blist> <blist> <bibtext> Wothke WLittle TD, Schnabel KU, Baumert J. Longitudinal and multigroup modeling with missing data. Modeling longitudinal and multilevel data: Practical issues, applied approaches, and specific examples. 2000: Mahwah; Lawrence Erlbaum Associates Publishers: 219-240</bibtext> </blist> <blist> <bibtext> Xia Y, Yang Y. RMSEA, CFI, and TLI in structural equation modeling with ordered categorical data: The story they tell depends on the estimation methods. Behavior Research Methods. 2019; 51; 1: 409-428</bibtext> </blist> </ref> <ref id="AN0146556866-31"> <title> Footnotes </title> <blist> <bibtext> We have IRB approval to identify the institution.</bibtext> </blist> <blist> <bibtext> We chose this term over other possibilities (including "financial stability," which seemed to conflate the issue of month-to-month income stability; or "financial comfort", which entailed implications of luxuries). As research incorporating multiple facets of students' financial situations is still nascent; we want to explicitly say that we remain open to alternative terminology in future work.</bibtext> </blist> <blist> <bibtext> We also accessed an Excel sheet provided by Bryant and Satorra ([19]) that allowed us to manually include the Chi-square statistics and calculate the DIFFtest. The findings were <emph>p</emph> = 0.060 and are aligned with the SBDIFF.EXE program in suggesting that the models are not statistically significantly different.</bibtext> </blist> </ref> <aug> <p>By Daniel A. Collier; Dan Fitzpatrick; Chelsea Brehm; Keith Hearit and Andrea Beach</p> <p>Reported by Author; Author; Author; Author; Author</p> </aug> <nolink nlid="nl1" bibid="bib10" firstref="ref2"></nolink> <nolink nlid="nl2" bibid="bib88" firstref="ref3"></nolink> <nolink nlid="nl3" bibid="bib16" firstref="ref4"></nolink> <nolink nlid="nl4" bibid="bib15" firstref="ref5"></nolink> <nolink nlid="nl5" bibid="bib31" firstref="ref6"></nolink> <nolink nlid="nl6" bibid="bib40" firstref="ref7"></nolink> <nolink nlid="nl7" bibid="bib20" firstref="ref9"></nolink> <nolink nlid="nl8" bibid="bib83" firstref="ref10"></nolink> <nolink nlid="nl9" bibid="bib34" firstref="ref15"></nolink> <nolink nlid="nl10" bibid="bib27" firstref="ref17"></nolink> <nolink nlid="nl11" bibid="bib55" firstref="ref18"></nolink> <nolink nlid="nl12" bibid="bib60" firstref="ref19"></nolink> <nolink nlid="nl13" bibid="bib58" firstref="ref24"></nolink> <nolink nlid="nl14" bibid="bib26" firstref="ref26"></nolink> <nolink nlid="nl15" bibid="bib65" firstref="ref27"></nolink> <nolink nlid="nl16" bibid="bib39" firstref="ref28"></nolink> <nolink nlid="nl17" bibid="bib68" firstref="ref35"></nolink> <nolink nlid="nl18" bibid="bib25" firstref="ref42"></nolink> <nolink nlid="nl19" bibid="bib67" firstref="ref45"></nolink> <nolink nlid="nl20" bibid="bib73" firstref="ref47"></nolink> <nolink nlid="nl21" bibid="bib18" firstref="ref48"></nolink> <nolink nlid="nl22" bibid="bib76" firstref="ref49"></nolink> <nolink nlid="nl23" bibid="bib21" firstref="ref50"></nolink> <nolink nlid="nl24" bibid="bib24" firstref="ref59"></nolink> <nolink nlid="nl25" bibid="bib86" firstref="ref60"></nolink> <nolink nlid="nl26" bibid="bib29" firstref="ref64"></nolink> <nolink nlid="nl27" bibid="bib22" firstref="ref65"></nolink> <nolink nlid="nl28" bibid="bib48" firstref="ref66"></nolink> <nolink nlid="nl29" bibid="bib52" firstref="ref67"></nolink> <nolink nlid="nl30" bibid="bib13" firstref="ref68"></nolink> <nolink nlid="nl31" bibid="bib59" firstref="ref69"></nolink> <nolink nlid="nl32" bibid="bib43" firstref="ref72"></nolink> <nolink nlid="nl33" bibid="bib53" firstref="ref73"></nolink> <nolink nlid="nl34" bibid="bib54" firstref="ref77"></nolink> <nolink nlid="nl35" bibid="bib66" firstref="ref78"></nolink> <nolink nlid="nl36" bibid="bib87" firstref="ref79"></nolink> <nolink nlid="nl37" bibid="bib44" firstref="ref83"></nolink> <nolink nlid="nl38" bibid="bib51" firstref="ref85"></nolink> <nolink nlid="nl39" bibid="bib80" firstref="ref86"></nolink> <nolink nlid="nl40" bibid="bib75" firstref="ref87"></nolink> <nolink nlid="nl41" bibid="bib93" firstref="ref90"></nolink> <nolink nlid="nl42" bibid="bib79" firstref="ref91"></nolink> <nolink nlid="nl43" bibid="bib92" firstref="ref93"></nolink> <nolink nlid="nl44" bibid="bib90" firstref="ref99"></nolink> <nolink nlid="nl45" bibid="bib33" firstref="ref101"></nolink> <nolink nlid="nl46" bibid="bib38" firstref="ref103"></nolink> <nolink nlid="nl47" bibid="bib91" firstref="ref105"></nolink> <nolink nlid="nl48" bibid="bib85" firstref="ref108"></nolink> <nolink nlid="nl49" bibid="bib72" firstref="ref116"></nolink> <nolink nlid="nl50" bibid="bib37" firstref="ref120"></nolink> <nolink nlid="nl51" bibid="bib95" firstref="ref126"></nolink> <nolink nlid="nl52" bibid="bib41" firstref="ref133"></nolink> <nolink nlid="nl53" bibid="bib30" firstref="ref134"></nolink> <nolink nlid="nl54" bibid="bib12" firstref="ref135"></nolink> <nolink nlid="nl55" bibid="bib36" firstref="ref136"></nolink> <nolink nlid="nl56" bibid="bib56" firstref="ref137"></nolink> <nolink nlid="nl57" bibid="bib57" firstref="ref138"></nolink> <nolink nlid="nl58" bibid="bib62" firstref="ref139"></nolink> <nolink nlid="nl59" bibid="bib89" firstref="ref140"></nolink> <nolink nlid="nl60" bibid="bib81" firstref="ref141"></nolink> <nolink nlid="nl61" bibid="bib14" firstref="ref142"></nolink> <nolink nlid="nl62" bibid="bib17" firstref="ref143"></nolink> <nolink nlid="nl63" bibid="bib50" firstref="ref144"></nolink> <nolink nlid="nl64" bibid="bib45" firstref="ref145"></nolink> <nolink nlid="nl65" bibid="bib32" firstref="ref147"></nolink> <nolink nlid="nl66" bibid="bib46" firstref="ref148"></nolink> <nolink nlid="nl67" bibid="bib47" firstref="ref149"></nolink> <nolink nlid="nl68" bibid="bib84" firstref="ref150"></nolink> <nolink nlid="nl69" bibid="bib96" firstref="ref151"></nolink> <nolink nlid="nl70" bibid="bib28" firstref="ref152"></nolink> <nolink nlid="nl71" bibid="bib82" firstref="ref153"></nolink> <nolink nlid="nl72" bibid="bib69" firstref="ref158"></nolink> <nolink nlid="nl73" bibid="bib70" firstref="ref159"></nolink> <nolink nlid="nl74" bibid="bib74" firstref="ref162"></nolink> <nolink nlid="nl75" bibid="bib71" firstref="ref165"></nolink> <nolink nlid="nl76" bibid="bib77" firstref="ref166"></nolink> <nolink nlid="nl77" bibid="bib23" firstref="ref167"></nolink> <nolink nlid="nl78" bibid="bib49" firstref="ref168"></nolink> <nolink nlid="nl79" bibid="bib94" firstref="ref169"></nolink> <nolink nlid="nl80" bibid="bib63" firstref="ref170"></nolink> <nolink nlid="nl81" bibid="bib64" firstref="ref171"></nolink> <nolink nlid="nl82" bibid="bib35" firstref="ref178"></nolink> <nolink nlid="nl83" bibid="bib61" firstref="ref179"></nolink> <nolink nlid="nl84" bibid="bib78" firstref="ref180"></nolink> <nolink nlid="nl85" bibid="bib42" firstref="ref188"></nolink> <nolink nlid="nl86" bibid="bib11" firstref="ref192"></nolink>
Header DbId: eric
DbLabel: ERIC
An: EJ1273896
AccessLevel: 3
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 0
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Structuring First-Year Retention at a Regional Public Institution: Validating and Refining the Structure of Bowman's SEM Analysis
– Name: Language
  Label: Language
  Group: Lang
  Data: English
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Collier%2C+Daniel+A%2E%22">Collier, Daniel A.</searchLink> (ORCID <externalLink term="http://orcid.org/0000-0002-3164-0815">0000-0002-3164-0815</externalLink>)<br /><searchLink fieldCode="AR" term="%22Fitzpatrick%2C+Dan%22">Fitzpatrick, Dan</searchLink> (ORCID <externalLink term="http://orcid.org/0000-0002-9095-0370">0000-0002-9095-0370</externalLink>)<br /><searchLink fieldCode="AR" term="%22Brehm%2C+Chelsea%22">Brehm, Chelsea</searchLink> (ORCID <externalLink term="http://orcid.org/0000-0002-1808-8405">0000-0002-1808-8405</externalLink>)<br /><searchLink fieldCode="AR" term="%22Hearit%2C+Keith%22">Hearit, Keith</searchLink><br /><searchLink fieldCode="AR" term="%22Beach%2C+Andrea%22">Beach, Andrea</searchLink>
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="SO" term="%22Research+in+Higher+Education%22"><i>Research in Higher Education</i></searchLink>. Dec 2020 61(8):917-942.
– Name: Avail
  Label: Availability
  Group: Avail
  Data: Springer. Available from: Springer Nature. One New York Plaza, Suite 4600, New York, NY 10004. Tel: 800-777-4643; Tel: 212-460-1500; Fax: 212-460-1700; e-mail: customerservice@springernature.com; Web site: https://link.springer.com/
– Name: PeerReviewed
  Label: Peer Reviewed
  Group: SrcInfo
  Data: Y
– Name: Pages
  Label: Page Count
  Group: Src
  Data: 26
– Name: DatePubCY
  Label: Publication Date
  Group: Date
  Data: 2020
– Name: SourceSuprt
  Label: Sponsoring Agency
  Group: SrcSuprt
  Data: Department of Education (ED)
– Name: NumberContract
  Label: Contract Number
  Group: NumCntrct
  Data: P116F140353
– Name: TypeDocument
  Label: Document Type
  Group: TypDoc
  Data: Journal Articles<br />Reports - Research
– 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="%22College+Freshmen%22">College Freshmen</searchLink><br /><searchLink fieldCode="DE" term="%22School+Holding+Power%22">School Holding Power</searchLink><br /><searchLink fieldCode="DE" term="%22Public+Colleges%22">Public Colleges</searchLink><br /><searchLink fieldCode="DE" term="%22Structural+Equation+Models%22">Structural Equation Models</searchLink><br /><searchLink fieldCode="DE" term="%22Academic+Persistence%22">Academic Persistence</searchLink><br /><searchLink fieldCode="DE" term="%22Student+Characteristics%22">Student Characteristics</searchLink><br /><searchLink fieldCode="DE" term="%22Financial+Problems%22">Financial Problems</searchLink><br /><searchLink fieldCode="DE" term="%22Security+%28Psychology%29%22">Security (Psychology)</searchLink><br /><searchLink fieldCode="DE" term="%22Food%22">Food</searchLink><br /><searchLink fieldCode="DE" term="%22Social+Adjustment%22">Social Adjustment</searchLink>
– Name: DOI
  Label: DOI
  Group: ID
  Data: 10.1007/s11162-020-09612-w
– Name: ISSN
  Label: ISSN
  Group: ISSN
  Data: 0361-0365
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: Structural equation modeling (SEM) considering how students' non-cognitive attributes influence first-year college student persistence remain extraordinarily rare--as are studies that test and expand upon published structural models or studies that include college student food security. This study addresses each. We surveyed "Beginner" Freshmen, capturing eight non-cognitive measurements and using institutional data on performance and fall-to-fall persistence measures, we then tested the structure of Bowman et al.'s (Res High Educ 60:135-152, 2019) SEM model. In Model 1, we mimic the Bowman model's financial variable by only including financial stress. We confirm that Bowman is a good structural model of student persistence, although our data were collected for another purpose, using different scales for non-cognitive elements and even one different non-cognitive measurement. We found students' non-cognitive attributes remain importantly influential to social adjustment (r=0.65), commitment to persist (r=0.40), college GPA (r=0.25), and fall-to-fall persistence (r=0.30). In Model 2, we generated a latent financial security variable incorporating financial stress and food security. Including food security generated a direct influence from the financial security variable to high-school GPA (r=0.25), not found in the Bowman model or Model 1, and a direct significant relationship from financial security to social adjustment (r=0.11)--not found in Model 1. Further changes are observed in the indirect relationship from financial security to college GPA from Model 1 (r=0.29) to Model 2 (r=0.51). We highlight the robustness of the Bowman model and that the inclusion of food security brings increased strength to several relationships without sacrificing optimal fit.
– Name: AbstractInfo
  Label: Abstractor
  Group: Ab
  Data: As Provided
– Name: DateEntry
  Label: Entry Date
  Group: Date
  Data: 2020
– Name: AN
  Label: Accession Number
  Group: ID
  Data: EJ1273896
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=eric&AN=EJ1273896
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.1007/s11162-020-09612-w
    Languages:
      – Text: English
    PhysicalDescription:
      Pagination:
        PageCount: 26
        StartPage: 917
    Subjects:
      – SubjectFull: College Freshmen
        Type: general
      – SubjectFull: School Holding Power
        Type: general
      – SubjectFull: Public Colleges
        Type: general
      – SubjectFull: Structural Equation Models
        Type: general
      – SubjectFull: Academic Persistence
        Type: general
      – SubjectFull: Student Characteristics
        Type: general
      – SubjectFull: Financial Problems
        Type: general
      – SubjectFull: Security (Psychology)
        Type: general
      – SubjectFull: Food
        Type: general
      – SubjectFull: Social Adjustment
        Type: general
    Titles:
      – TitleFull: Structuring First-Year Retention at a Regional Public Institution: Validating and Refining the Structure of Bowman's SEM Analysis
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Collier, Daniel A.
      – PersonEntity:
          Name:
            NameFull: Fitzpatrick, Dan
      – PersonEntity:
          Name:
            NameFull: Brehm, Chelsea
      – PersonEntity:
          Name:
            NameFull: Hearit, Keith
      – PersonEntity:
          Name:
            NameFull: Beach, Andrea
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 01
              M: 12
              Type: published
              Y: 2020
          Identifiers:
            – Type: issn-print
              Value: 0361-0365
          Numbering:
            – Type: volume
              Value: 61
            – Type: issue
              Value: 8
          Titles:
            – TitleFull: Research in Higher Education
              Type: main
ResultId 1