Text samples contain a massive amount of information that is currently not available to researchers nor practitioners.
Traditionally, both psychological research and decision making (e.g., college admissions and job applications), have depended on quantifiable behavioral tasks, and self-reported questionnaires.
These have a number of limitations
Reference bias
Awareness limitations
Faking
Social desirability bias
Part of these limitations can be addressed by a complementary approach. Machine learning assisted analysis of student writing.
Educational Data Science
In general terms, educational data science is defined as the application of statistical, artificial intelligence, and machine learning techniques to educational data (cite).
One important subfield of data science is text mining. Text mining is defined as the extraction of non-trivial information from unstructured text data, rather than structured datasets, as input (Tan, 1999).
There are simple approaches which use features of language as inputs for prediction and classification.
For example, bag-of-words approaches take the frequency of words in a document as input in a prediction or classification task. As such, it disregards grammar and structure, focusing solely on the content of the document (cite).
Often, instead of working with single words, models will employ n-grams, or combinations two, three, or more words. These n-gram approaches include some contextual information on the semantic information surrounding words. In that way they might distinguish between the bank of a river and the bank as a financial institution.
In recent years, more sophisticated models of text mining have been developed. Since its introduction (Vaswani et al., 2017) transformer architecture has quickly become the dominant approach for text mining tasks (Wolf et al., 2020).
Even though several alternatives exist (e.g., RoBERTa, GPT-2) an algorithm called Bidirectional encoder representations from transformers (BERT) is the optimal model for this use case. This is because….
BERT can be computationally less expensive than GPT, since it has smaller versions (?)
BERT is fully open access (?)
It’s available in more languages and is more widely used worldwide, providing a test more comparable to other potential users (?)
The fact that it follows the convention of naming language models after muppets gives it more street cred.
Prior research has used BERT in educational contexts.
To evaluate objectivity in students self-evaluations (Nikolovski, 2020).
To assess teacher effectiveness (Wang et al., 2020)
To summarize text and deliver it via a chatbot application (Bathija et al., 2020)
Text mining, particularly the more recently developed methods like BERT, would seem promising for testing the possibility of using text samples to extract valuable information.
The Problem of Predicting College Graduation
Knowing who will or will not graduate from college is an important, yet complex, question.
Understanding who graduates and why is important, because graduating from college is associated with better health, societal, and economic outcomes.
The college admissions process tries, and sometimes fails, to answer this question correctly.
In the false negative case, students who would graduate are denied the possibility of attending college, because their applications suggest that they will not benefit from the resources allocated to them.
In the false positive case, students are offered admission, and then drop out. In the process, resources allocated to them are wasted. Perhaps more damagingly, they preclude other students from benefiting from a college education.
Several predictors of college graduation have been identified
Cognitive ability and academic ability.
High-school Grades (Galla et al., 2019) and first semester GPA (Gershenfeld, et al., 2016)
Despite the evidence in their favor, leaning on high-school GPA and standardized test scores is not without limitation.
They explain a modest amount of variance
They might be culturally biased
The are correlated with socioeconomic status
However, important psychological non-cognitive predictors have often been overlooked.
An important reason why these are overlooked is that they are qualitative in nature, so they don’t lend themselves to simple sorting of applicants. However, advances in text mining, a subdiscipline of data science makes the analysis of this data possible.
Different indicators of student motivation are theoretically interesting potential predictors of college graduation.
Prosocial motivation might impact college graduation by providing self-transcendent motivation which fosters academic self regulation (Yeager et al., 2014).
Leadership
An orientation towards learning is definitely useful in predicting college graduation. Schmitt et al., (2009), show that students with intellectual curiosity and being interested in learning is a significant predictor of college GPA, and graduation.
Setting goals and planning for them
Self-concordance might lead to college graduation because self-concordant autonomous motivation is linked with deeper learning (cite)
Teamwork
Perseverance
Beyond Grades and Test Scores
The Current Study
In this study we merged data from the common application during the 2008-2009 admissions cycle with graduation data form the National Student Clearinghouse data set.
We coded a subset of students essays about out-of-school activities to identify seven motivational themes
We then trained transformer models to perform document classification on students essays about out-of-school activities.
We then input the computed generated estimates into logistic regression models predicting 4- and 6- year graduation.
[Insert somewhere] Prediction vs. Explanation.
Our investigation capitalized on a national dataset of 311,308 students who completed the Common Application during the 2008-2009 admissions cycle and for whom objective longitudinal data on college graduation were available via the National Student Clearinghouse. Our main analytic sample comprised N = 47,303 students who submitted online rather than hardcopy applications and therefore had report card grades that were possible to de-identify. As a robustness check, however, we replicated analyses in the full sample. Approximately 1% of this sample constituted a stratified random training sample (n = 3,999), and the remaining 99% of data became the holdout sample (n = 43,667). Our multi-step process is summarized in Figure 1, and details on methodology and results, including robustness analyses, are available in the Online Supplement.
For each essay in the training sample, qualified research assistants manually assigned binary codes for each of seven motivational orientations that in prior research have been shown to have benefits in the education context: prosocial purpose, leadership, learning, goal pursuit, self-concordant motivation, teamwork, and perseverance. See Table 1 for coding rules and fictionalized examples.
Table 1
Motivational themes, coding rules and fictionalized example responses
| Theme | Criteria for coding by human raters | Fictionalized example response with theme-relevant phrases underlined |
|---|---|---|
Prosocial purpose
|
Applicant reports helping others or reports a desire to help others, or applicant reports working in the service of others who occupy a disadvantaged social position (e.g., children, the elderly, low-income populations, disenfranchised populations, sick or disabled individuals, animals); applicant may discuss how the help will benefit others, why the applicant wants to help, or provide evidence that the applicant finds the experience of helping others to be enjoyable or rewarding | Every summer for the last three years, I have worked as camp counselor at a camp for young children from underprivileged families. Helping children realize their hidden talents is one of the most rewarding experiences I have ever had. I’ve been so fulfilled by watching these children develop confidence in their abilities. This experience has been so important to me, and shown me that a career in education is where I belong.
|
Leadership
|
Applicant reports serving in a leadership role; applicant may comment on what he or she did in his or her capacity as a leader, or discuss the value, meaning, or importance of leadership | I was chosen to be cheerleading captain during my senior year. It was one of the most rewarding experiences I have ever had. My freshman year captain had a huge impact on my life, and I felt like it was my time to pay it forward. I am so proud of everything I did for the girls: creating a mentorship system, organizing events and fundraisers, and encouraging everyone to work as hard as they could. At the end of the year, a few girls thanked me and told me how much fun they had had and how much they learned. I was completely overcome with emotion. I’ve never felt so gratified in my life. |
| Learning | Applicant reports that knowledge, skills, abilities have been learned, developed, or improved
|
I love softball and played in high school, but when I started I was not a very strong player. My freshman, sophomore, and junior years on the J.V. team were pretty average. When I finally made the varsity team my senior year, I was determined to have a better season. I worked constantly to improve my game – during practice and on my own time. My skills grew so much. Because of my hard work, I finished the year with the best record on my team!
|
Goal pursuit
|
Applicant reports having a goal and/or a plan
|
I have been playing soccer since I was six years old. Unfortunately, last year I injured my knee, and it has been a struggle to get back to the level I was playing at before my injury. It has been really challenging, but I’ve been doing physical therapy and practicing everyday so that I can be a varsity starter this year. |
Self-concordant motivation
|
Applicant refers to the activity as enjoyable or interesting (or other synonyms) and makes statements of affinity for the activity (e.g., “I love tennis), or statements of identification with the activity (e.g.,”Drama club is my life") | Running track is so much more than a sport to me. It’s a challenge and an adventure, and I put everything I have into it. I love every aspect of it, even the afternoons I spend drenched in sweat in the scorching heat. |
Teamwork
|
Applicant mentions working with others or learning from others and/or the value that fellow participants bring to the activity
|
I’ve been on my school’s debate team since my freshman year, and was elected co-captain because of my commitment to the team’s success. My fellow co-captains and I worked together get our team ready for competitions. We knew that a strong team performance was more important than the successes of a few individuals. We stressed teamwork and cooperation between our teammates. Because we focused on team effort, we earned first place at the state meet. |
| Perseverance | Applicant discusses persisting in the face of challenge
|
I’ve learned to become a gracious victor and to grow from defeat. Track has helped me overcome my fear of losing, and even helped me put my life in perspective. I’ve learned to keep working and fighting even when the odds seem impossible to beat. There were many times that I found myself lagging behind, but I pulled ahead at the end because I never gave up. The most important thing I’ve learned is to never let anything stand in my way. |
Using Bidirectional Encoder Representations from Transformers (BERT), the computer algorithm “learned” from the manually-coded essays to assign continuous probabilities for each of these seven themes for each of the 3,999 essays in the training sample. BERT is an advanced approach to analyzing natural language recently developed by researchers at Google, the advantages of which are detailed in supporting online material.
Finally, in the holdout sample, we used these computer-coded probabilities to predict college graduation in separate binary logistic regression models that controlled for a suite of covariates, including demographics, high school GPA, and college entrance exam scores.
As shown in Table 2, Leadership and learning demonstrated incremental predictive validity for both six- and four- year college graduation over and above traditional predictors and demographics—but goal pursuit, self-concordant motivation, teamwork, prosocial purpose, and perseverance did not. Similar evidence for these predictive validities were obtained in a series of robustness checks included in the Online Supplement.
Frequencies for all seven manually-coded motivational themes were similar to corresponding mean computer-coded estimates in the training sample, with computers being slightly more likely to assign any given code. However, the motivational themes of leadership and perseverance were less commonly observed by both research assistants and the machine learning algorithm (i.e., human coders and the machine learning algorithm identified them in fewer than 20% and 25% of essays, respectively) than others.
The correlation between manually- and computer-coded estimates, though somewhat variable, were acceptable, ranging from r = .53*** for goal pursuit and teamwork to r = .79*** for prosocial purpose. Compared to other motivational themes, goal pursuit and teamwork were less reliably detected by human coders (αs < .60).
Table 2
Reliability, Prevalence, and Predictive Validity of Motivational Themes
| Motivational theme | Manually-coded frequency | Computer-estimated likelihood | Reliability of human raters | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
|---|---|---|---|---|---|---|---|---|---|---|
| f | M | α | ||||||||
| Self-concordant motivation | .41 | .51 | .66 | .68*** | .00 | -.14*** | .03*** | -.01 | .03*** | .02*** |
| Goal pursuit | .30 | .38 | .54 | .01 | .53*** | -.04*** | -.10*** | .00 | .00 | .12*** |
| Leadership | .17 | .20 | .75 | -.07*** | .04** | .75*** | -.05*** | .01† | .01* | .00 |
| Learning | .41 | .47 | .71 | -.02 | .01 | -.03† | .72*** | .06*** | .07*** | .02*** |
| Perseverance | .18 | .21 | .61 | .02 | .03* | .00 | .09*** | .63*** | -.08*** | .05*** |
| Prosocial purpose | .33 | .35 | .74 | -.04* | .05*** | .00 | -.07*** | -.11*** | .79*** | .09*** |
| Teamwork | .23 | .29 | .44 | -.01 | .03† | .15*** | .07*** | .06*** | -.03† | .53*** |
Notes. N = 3,999 for stratified training sample used by human coders and 43,677 for the holdout sample used for computer coding. f frequency. M mean. α Krippendorf’s alpha. rpb Point biserial correlation. OR odds ratios. Odds ratios are based on standardized predictors for comparability and are interpreted as a positive effect if greater than one, and a negative effect if the value is less than 1.
*p < .05, **p < .01, ***p < .001.
Correlations among the codes were small, ranging in absolute value from .00 to .15 in the human coded samples, and from .00 to .20 in the computer codes. As such, multicollinearity was not an issue in our models, with variance inflation factors (i.e., VIF) below 1.76 for all predictors.
Finally, our results are not explainable by the machine learning codes simply capturing pre-existing socioeconomic differences among students in the sample. The computer generated codes were unrelated to indicators of SES and demographic variables. We regressed each code on indicators of SES (i.e, school type and parental education), and found that these indicators explained on average 0.17%, ranging from 0.02% to 0.33%. Further additional sociodemographic factors (i.e., student gender, race/ethnicity, parent marital status, and English language learner status) still weren’t predictive of machine learning codes: variance explained across the seven codes averaged 0.51%, ranging from 0.14% to 1.36%.
Collectively, these findings demonstrate the potential of supervised machine learning for detecting motivational themes from college application essays at scale. Specifically, students whose descriptions of out-of-school activities were coded by our computer algorithm as exemplifying prosocial purpose, leadership, or learning were more likely to go on to earn their college diplomas, even when controlling for a rich set of covariates including grades and admissions test scores. The relevance of prosocial orientation to college graduation is not surprising given the benefits of beyond-the-self motivation for academic engagement and achievement (Yeager et al., 2014). Likewise, holding leadership positions during high school has been shown to predict total years of postsecondary education (Rouse, 2012). Finally, it is not surprising that students who wrote about developing new skills and knowledge in their out-of-school activities were better able to complete their college education.
Why weren’t students whose essays were coded as reflecting goal pursuit, team orientation, or perseverance more likely to graduate from college when accounting for traditional predictors? After all, these motivational constructs have been associated with academic achievement in prior research (e.g., Zimmerman et al., 1992). We cannot know for certain, but these motivational themes were less frequent and less reliably detected by human raters, suggesting that predictive validities may have been compromised by noisy measurement. In contrast, computer-coded self-concordant motivation failed to predict college graduation despite relatively high frequency and reliability. Why? Experience sampling studies suggest that in general, students are more intrinsically motivated during out-of-school activities than during academic classes (Larson, 2000). It seems plausible that the intrinsic rewards many students derive from sports, paid work, and other out-of-school pursuits are distinct from the gratifications found in the college experience.
Though reliable, the effect sizes for the three most reliable predictors of college graduation were small—both in absolute terms and when compared to the predictive validities of high school grades and admissions test scores. In our view, it is therefore premature for colleges to add supervised machine learning to their admissions processes. At the same time, we are sanguine about the potential of artificial intelligence to revolutionize college admissions in the not too distant future. Compared to the intuitive judgments of human raters, algorithms are dramatically less noisy (Kahneman et al., 2016), more easily tested for fairness across gender, ethnicity, or other subgroups (Hutt et al., 2019), and dramatically more cost-effective, particularly at scale. Indeed, it is our hope that as a complement to existing metrics, automated assessments of writing samples will one day make the admissions process both more efficient and more equitable.
Note for us: While the original manuscript states that 315 essays were coded twice, in the raw data, 381 essays were coded more than once. Of those, 377 were coded two times, and 4 were coded three times. This discrepancy in the numbers is likely due to the fact one coder coded 20 essays twice. To calculate interrater reliability, we have used all available information, and have considered coders as independent (even when one of the coders coded the same essays twice).
Interrater reliability was adequate for 5 out of 7 codes, but was low for goal pursuit and teamwork. They are shown in the diagonal of the table below. Correlations among human codes were quite modest, as seen in the table below. The largest correlations are those between leadership and teamwork (r = .15), and between perseverance and self-transcendence (r* = -.11*).
| Variable | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
|---|---|---|---|---|---|---|---|
| 1. Self-concordant motivation | .66 | .00 | -.14*** | .03*** | -.01 | .03*** | .02*** |
| 2. Goal pursuit | .01 | .54 | -.04*** | -.10*** | .00 | .00 | .12*** |
| 3. Leadership | -.07*** | .04** | .75 | -.05*** | .01† | .01* | .00 |
| 4. Learning | -.02 | .01 | -.03† | .71 | .06*** | .07*** | .02*** |
| 5. Perseverance | .02 | .03* | .00 | .09*** | .61 | -.08*** | .05*** |
| 6. Prosocial purpose | -.04* | .05*** | .00 | -.07*** | -.11*** | .74 | .09*** |
| 7. Teamwork | -.01 | .03† | .15*** | .07*** | .06*** | -.03† | .44 |
| M | 40.54% | 30.32% | 16.7% | 40.79% | 18.42% | 32.89% | 22.85% |
| SD | |||||||
| n | 3,937 | 3,945 | 3,945 | 3,947 | 3,946 | 3,941 | 3,922 |
Computer generated codes match adequately codes given by human coders (rs > .60), albeit with lower values for Goal Orientation, and Teamwork. The following table shows correlations between human codes and computer-generated binary predictions (0s or 1s depending on a threshold of .5) and continuous probabilities.
| Code | Binary | Probability |
|---|---|---|
| Type | .77*** | .79*** |
| Accolades | .70*** | .71*** |
| Self-concordant motivation | .65*** | .68*** |
| Goal pursuit | .49*** | .53*** |
| Leadership | .72*** | .75*** |
| Learning | .70*** | .72*** |
| Perseverance | .60*** | .63*** |
| Prosocial purpose | .77*** | .79*** |
| Teamwork | .50*** | .53*** |
The following table shows means, and sds for the training sample, the holdout sample and the robustness check sample.
| Training Sample | Holdout Sample | Robustness Check Sample | |||||||
|---|---|---|---|---|---|---|---|---|---|
| n | mean | sd | n | mean | sd | n | mean | sd | |
| Demographics | |||||||||
| Female | 3,952.00 | 0.55 | 0.50 | 43,667.00 | 0.56 | 0.50 | 307,251.00 | 0.55 | 0.50 |
| No parents with college degree | 3,952.00 | 0.50 | 0.50 | 43,667.00 | 0.27 | 0.45 | 307,251.00 | 0.26 | 0.44 |
| One parent with college degree | 3,952.00 | 0.21 | 0.41 | 43,667.00 | 0.24 | 0.43 | 307,251.00 | 0.24 | 0.43 |
| Two parents with college degree | 3,952.00 | 0.28 | 0.45 | 43,667.00 | 0.48 | 0.50 | 307,251.00 | 0.50 | 0.50 |
| Married parents | 3,952.00 | 0.67 | 0.47 | 43,667.00 | 0.77 | 0.42 | 307,251.00 | 0.78 | 0.41 |
| English is first language | 3,952.00 | 0.75 | 0.43 | 43,667.00 | 0.88 | 0.33 | 307,251.00 | 0.88 | 0.32 |
| White | 3,952.00 | 0.28 | 0.45 | 43,667.00 | 0.51 | 0.50 | 307,251.00 | 0.52 | 0.50 |
| Black | 3,952.00 | 0.16 | 0.37 | 43,667.00 | 0.05 | 0.22 | 307,251.00 | 0.05 | 0.23 |
| Latino | 3,952.00 | 0.16 | 0.36 | 43,667.00 | 0.08 | 0.28 | 307,251.00 | 0.06 | 0.24 |
| Asian | 3,952.00 | 0.19 | 0.39 | 43,667.00 | 0.10 | 0.30 | 307,251.00 | 0.10 | 0.30 |
| Other | 3,952.00 | 0.09 | 0.29 | 43,667.00 | 0.09 | 0.28 | 307,251.00 | 0.08 | 0.27 |
| No race reported | 3,952.00 | 0.12 | 0.33 | 43,667.00 | 0.17 | 0.37 | 307,251.00 | 0.19 | 0.39 |
| Non-Title I public school | 3,425.00 | 0.41 | 0.49 | 41,379.00 | 0.19 | 0.39 | 285,481.00 | 0.16 | 0.36 |
| Title I public school | 3,425.00 | 0.59 | 0.49 | 41,379.00 | 0.46 | 0.50 | 285,481.00 | 0.51 | 0.50 |
| Private school | 3,425.00 | 0.00 | 0.00 | 41,379.00 | 0.34 | 0.47 | 285,481.00 | 0.33 | 0.47 |
| Homeschool | 3,425.00 | 0.00 | 0.00 | 41,379.00 | 0.01 | 0.08 | 285,481.00 | 0.00 | 0.05 |
| Essay codes | |||||||||
| Self-concordant motivation | 3,937.00 | 0.41 | 0.49 | 43,667.00 | 0.51 | 0.44 | 307,254.00 | 0.50 | 0.44 |
| Goal pursuit | 3,945.00 | 0.30 | 0.46 | 43,667.00 | 0.38 | 0.35 | 307,254.00 | 0.38 | 0.35 |
| Leadership | 3,945.00 | 0.17 | 0.37 | 43,667.00 | 0.20 | 0.36 | 307,254.00 | 0.19 | 0.35 |
| Learning | 3,947.00 | 0.41 | 0.49 | 43,667.00 | 0.47 | 0.45 | 307,254.00 | 0.47 | 0.45 |
| Perseverance | 3,946.00 | 0.18 | 0.39 | 43,667.00 | 0.21 | 0.34 | 307,254.00 | 0.21 | 0.34 |
| Prosocial purpose | 3,941.00 | 0.33 | 0.47 | 43,667.00 | 0.35 | 0.44 | 307,254.00 | 0.36 | 0.45 |
| Teamwork | 3,922.00 | 0.23 | 0.42 | 43,667.00 | 0.29 | 0.35 | 307,254.00 | 0.29 | 0.35 |
| Outcome and covariates | |||||||||
| Six-year graduation | 3,952.00 | 0.33 | 0.47 | 43,667.00 | 0.40 | 0.49 | 307,251.00 | 0.42 | 0.49 |
| Four-year graduation | 3,952.00 | 0.67 | 0.47 | 43,667.00 | 0.78 | 0.42 | 307,251.00 | 0.78 | 0.42 |
| Admissions tTest scores | 3,600.00 | 1,704.80 | 306.02 | 41,838.00 | 1,837.99 | 268.91 | 289,788.00 | 1,826.41 | 267.90 |
| HSGPA | 3,601.00 | 0.89 | 0.14 | 43,667.00 | 0.92 | 0.13 | 43,667.00 | 0.92 | 0.13 |
| Six-year college-level graduation rates | 3,233.00 | 0.48 | 0.24 | 38,388.00 | 0.57 | 0.22 | 269,878.00 | 0.58 | 0.22 |
| Four-year college-level graduation rates | 3,233.00 | 0.66 | 0.20 | 38,388.00 | 0.73 | 0.17 | 269,878.00 | 0.73 | 0.17 |
| Number of OSAs | 3,952.00 | 3.52 | 2.29 | 43,667.00 | 5.37 | 1.86 | 307,251.00 | 5.16 | 1.98 |
| Years per OSA | 3,952.00 | 2.14 | 1.12 | 43,667.00 | 2.57 | 0.72 | 307,251.00 | 2.53 | 0.76 |
| Proportion of OSAs that are sports | 3,952.00 | 0.23 | 0.32 | 43,667.00 | 0.26 | 0.27 | 307,251.00 | 0.27 | 0.29 |
Here we show correlations between computer generated likelihoods, traditional predictors of graduation, and four- and six- year graduation from college. Below the diagonal, we show correlations controlling demographic factors—including gender, parent education and marital status, English as a first language, ethnicity, and type of high school. Zero-order bivariate Pearson correlations are shown above the diagonal.
| Variable | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1. Self-Concordant Motivation | .00 | -.14*** | .03*** | -.01 | .03*** | .02*** | .00 | .20*** | .11*** | .01** | -.09*** | .00 | -.21*** | |
| 2. Goal | .00 | -.04*** | -.10*** | .00 | .00 | .12*** | -.04*** | .07*** | -.13*** | -.18*** | .03*** | .02*** | .12*** | |
| 3. Leadership | -.14*** | .07*** | -.05*** | .01† | .01* | .00 | .01* | .06*** | .06*** | .06*** | -.02** | .02*** | .21*** | |
| 4. Learning | -.04*** | -.08*** | -.03*** | .06*** | .07*** | .02*** | .03*** | .08*** | .00 | .03*** | .10*** | .03*** | -.05*** | |
| 5. Perseverance | .03*** | .02*** | -.02*** | .11*** | -.08*** | .05*** | -.03*** | .04*** | -.02*** | .07*** | -.04*** | .48*** | -.30*** | |
| 6. Self-Transcendence | -.10*** | .12*** | .00 | -.13*** | -.18*** | .09*** | -.02*** | -.02*** | .03*** | .05*** | -.09*** | .25*** | .09*** | |
| 7. Teamwork | -.05*** | .00 | .20*** | .06*** | .06*** | -.09*** | .00 | .03*** | -.01† | .01* | .02*** | .10*** | .16*** | |
| 8. HSGPA | -.01 | .02*** | .07*** | .00 | .03*** | .03*** | .04*** | .05*** | .02*** | .05*** | .03*** | -.18*** | .08*** | |
| 9. Admissions test scores | .00 | .05*** | .06*** | -.02*** | .07*** | -.02** | .05*** | .48*** | .01** | .00 | .04*** | .15*** | .04*** | |
| 10. Number of OSAs | .01† | .09*** | .08*** | .03*** | .05*** | .10*** | .07*** | .25*** | .31*** | .02*** | .05*** | .23*** | .08*** | |
| 11. Years per OSA | .06*** | .00 | .04*** | -.01† | .01* | -.04*** | .00 | .10*** | .16*** | -.05*** | .07*** | .31*** | -.03*** | |
| 12. Proportion of OSAs that are sports | .03*** | -.04*** | -.02*** | .02*** | .05*** | -.09*** | .02*** | -.18*** | -.21*** | -.30*** | .08*** | .16*** | -.05*** | |
| 13. 4-year graduation | .00 | .01* | .03*** | .01** | .00 | .02*** | .02*** | .15*** | .12*** | .09*** | .04*** | -.03*** | .44*** | |
| 14. 6-year graduation | .01* | .03*** | .05*** | .01** | .02*** | .03*** | .03*** | .23*** | .21*** | .16*** | .08*** | -.05*** | .44*** | |
| M | 0.51 | 0.38 | 0.2 | 0.47 | 0.21 | 0.35 | 0.29 | 0.92 | 1,837 | 5.37 | 2.57 | 25.55% | 39.92% | 77.51% |
| SD | 0.44 | 0.35 | 0.36 | 0.45 | 0.34 | 0.44 | 0.35 | 0.13 | 268 | 1.86 | 0.72 | |||
| n | 43,667 | 43,667 | 43,667 | 43,667 | 43,667 | 43,667 | 43,667 | 43,667 | 41,838 | 43,667 | 43,667 | 43,667 | 43,667 | 43,667 |
Here’s a summary of all the models run: controlling or not for institutional graduation rates, including GPA or not, for four- and six-year graduation outcomes, ran with continuous likelihoods and binary predictions. All VIFs were below 1.76. More details can be seen below.
| term | grad4 | grad6 | ||
|---|---|---|---|---|
| Full Sample | GPA Subsample | Full Sample | GPA Subsample | |
| (Intercept) | 0.89*** | 0.71*** | 5.27*** | 4.76*** |
| Self-Concordant Motivation | 1.00 | 1.01 | 1.00 | 1.01 |
| Goal | 1.00 | 1.00 | 1.01* | 1.02 |
| Leadership | 1.03*** | 1.03* | 1.07*** | 1.06*** |
| Learning | 1.03*** | 1.02* | 1.04*** | 1.04** |
| Perseverance | 1.00 | 0.98 | 1.01* | 1.03* |
| Self-Transcendence | 1.04*** | 1.02 | 1.07*** | 1.07*** |
| Teamwork | 1.03*** | 1.01 | 1.04*** | 1.00 |
| stdtest | 1.23*** | 1.09*** | 1.50*** | 1.23*** |
| OSAn | 1.11*** | 1.08*** | 1.24*** | 1.19*** |
| OSAt | 1.02*** | 1.01 | 1.09*** | 1.09*** |
| OSAsport | 1.02*** | 1.03* | 1.05*** | 1.06*** |
| sexM | 0.78*** | 0.83*** | 0.69*** | 0.72*** |
| parentdegreeOne | 1.11*** | 1.16*** | 1.21*** | 1.26*** |
| parentdegreeTwo | 1.06*** | 1.19*** | 1.36*** | 1.45*** |
| parentmarriedOther | 0.86*** | 0.88*** | 0.76*** | 0.74*** |
| firstlanguageOther | 0.94*** | 0.86*** | 0.81*** | 0.70*** |
| raceBlack | 0.73*** | 0.79*** | 0.82*** | 0.79*** |
| raceLatino | 0.79*** | 0.86*** | 0.89*** | 0.95 |
| raceAsian | 0.79*** | 0.84*** | 0.75*** | 0.77*** |
| raceachieve | 0.82*** | 0.88** | 0.77*** | 0.79*** |
| raceMissing | 1.00 | 0.97 | 0.87*** | 0.84*** |
| hstypehs_pubT1 | 0.88*** | 0.94* | 0.87*** | 0.88*** |
| hstypehs_priv | 1.00 | 1.04 | 0.82*** | 0.94* |
| hstypehs_home | 0.59*** | 0.77 | 0.48*** | 0.66** |
| n | 271358 | 39684 | 271358 | 39684 |
| gpa | NA | 1.24*** | NA | 1.42*** |
| term | grad4 | grad6 | ||
|---|---|---|---|---|
| Full Sample | GPA Subsample | Full Sample | GPA Subsample | |
| (Intercept) | 0.82*** | 0.68*** | 4.55*** | 4.15*** |
| Self-Concordant Motivation | 1.00 | 1.01 | 1.01 | 1.00 |
| Goal | 0.99 | 1.00 | 1.02* | 1.04 |
| Leadership | 1.09*** | 1.06* | 1.17*** | 1.16*** |
| Learning | 1.05*** | 1.05* | 1.08*** | 1.08** |
| Perseverance | 1.01 | 0.96 | 1.02 | 1.07* |
| Self-Transcendence | 1.09*** | 1.04 | 1.14*** | 1.15*** |
| Teamwork | 1.07*** | 1.01 | 1.07*** | 0.99 |
| stdtest | 1.23*** | 1.09*** | 1.50*** | 1.23*** |
| OSAn | 1.11*** | 1.08*** | 1.25*** | 1.20*** |
| OSAt | 1.02*** | 1.01 | 1.09*** | 1.09*** |
| OSAsport | 1.02*** | 1.03* | 1.05*** | 1.06*** |
| sexM | 0.78*** | 0.83*** | 0.69*** | 0.72*** |
| parentdegreeOne | 1.11*** | 1.16*** | 1.21*** | 1.26*** |
| parentdegreeTwo | 1.07*** | 1.19*** | 1.36*** | 1.46*** |
| parentmarriedOther | 0.86*** | 0.87*** | 0.76*** | 0.74*** |
| firstlanguageOther | 0.93*** | 0.86*** | 0.81*** | 0.70*** |
| raceBlack | 0.73*** | 0.79*** | 0.82*** | 0.79*** |
| raceLatino | 0.79*** | 0.86*** | 0.89*** | 0.95 |
| raceAsian | 0.79*** | 0.84*** | 0.75*** | 0.77*** |
| raceachieve | 0.82*** | 0.88** | 0.77*** | 0.79*** |
| raceMissing | 1.00 | 0.97 | 0.87*** | 0.84*** |
| hstypehs_pubT1 | 0.88*** | 0.94* | 0.87*** | 0.88*** |
| hstypehs_priv | 1.00 | 1.03 | 0.82*** | 0.94* |
| hstypehs_home | 0.59*** | 0.77 | 0.48*** | 0.66** |
| n | 271358 | 39684 | 271358 | 39684 |
| gpa | NA | 1.24*** | NA | 1.42*** |
| term | Four-year graduation | Six-year graduation | ||
|---|---|---|---|---|
| Full Sample | GPA Subsample | Full Sample | GPA Subsample | |
| Controlling for institutional graduation rates - Modelled Together - Binarized | ||||
| Self-Concordant Motivation | 1.00 | 1.01 | 1.00 | 1.00 |
| Goal | 0.98* | 0.98 | 1.00 | 1.01 |
| Leadership | 1.06*** | 1.05 | 1.13*** | 1.11** |
| Learning | 1.04*** | 1.04 | 1.06*** | 1.09** |
| Perseverance | 0.99 | 0.93* | 0.98 | 1.04 |
| Self-Transcendence | 1.06*** | 1.01 | 1.11*** | 1.12** |
| Teamwork | 1.05*** | 1.00 | 1.05** | 0.99 |
| Controlling for institutional graduation rates - Modelled Together - Continuous | ||||
| Self-Concordant Motivation | 1.00 | 1.01 | 1.00 | 1.01 |
| Goal | 0.99** | 0.99 | 1.00 | 1.00 |
| Leadership | 1.03*** | 1.03* | 1.05*** | 1.05** |
| Learning | 1.02*** | 1.02 | 1.03*** | 1.04** |
| Perseverance | 0.99 | 0.97** | 0.99 | 1.01 |
| Self-Transcendence | 1.03*** | 1.00 | 1.06*** | 1.06*** |
| Teamwork | 1.02*** | 1.00 | 1.02*** | 1.00 |
| Not controlling for institutional graduation rates - Modelled Together - Binarized | ||||
| Self-Concordant Motivation | 1.00 | 1.01 | 1.01 | 1.00 |
| Goal | 0.99 | 1.00 | 1.02* | 1.04 |
| Leadership | 1.09*** | 1.06* | 1.17*** | 1.16*** |
| Learning | 1.05*** | 1.05* | 1.08*** | 1.08** |
| Perseverance | 1.01 | 0.96 | 1.02 | 1.07* |
| Self-Transcendence | 1.09*** | 1.04 | 1.14*** | 1.15*** |
| Teamwork | 1.07*** | 1.01 | 1.07*** | 0.99 |
| Not controlling for institutional graduation rates - Modelled Together - Continuous | ||||
| Self-Concordant Motivation | 1.00 | 1.01 | 1.00 | 1.01 |
| Goal | 1.00 | 1.00 | 1.01* | 1.02 |
| Leadership | 1.03*** | 1.03* | 1.07*** | 1.06*** |
| Learning | 1.03*** | 1.02* | 1.04*** | 1.04** |
| Perseverance | 1.00 | 0.98 | 1.01* | 1.03* |
| Self-Transcendence | 1.04*** | 1.02 | 1.07*** | 1.07*** |
| Teamwork | 1.03*** | 1.01 | 1.04*** | 1.00 |
| Controlling for institutional graduation rates - Modelled Separately - Continuous | ||||
| Self-Concordant Motivation | 0.98* | 1.00 | 0.98 | 0.99 |
| Goal | 0.98* | 0.97 | 1.01 | 1.03 |
| Leadership | 1.09*** | 1.07* | 1.16*** | 1.12** |
| Learning | 1.04*** | 1.04 | 1.06*** | 1.08* |
| Perseverance | 0.97* | 0.91** | 0.96* | 1.02 |
| Self-Transcendence | 1.06*** | 1.01 | 1.11*** | 1.12** |
| Teamwork | 1.08*** | 1.02 | 1.08*** | 1.00 |
| Not controlling for institutional graduation rates - Modelled Separately - Continuous | ||||
| Self-Concordant Motivation | 0.98* | 1.00 | 0.98* | 0.98 |
| Goal | 1.01 | 1.00 | 1.06*** | 1.09* |
| Leadership | 1.11*** | 1.08* | 1.21*** | 1.18*** |
| Learning | 1.05*** | 1.05 | 1.08*** | 1.06* |
| Perseverance | 1.00 | 0.95 | 1.01 | 1.06 |
| Self-Transcendence | 1.08*** | 1.04 | 1.13*** | 1.15*** |
| Teamwork | 1.11*** | 1.04 | 1.13*** | 1.01 |
There is a consistent and significant effect for leadership and learning; with mixed evidence for goal orientation, perseverance (for 6-year graduation), self-transcendence, and Teamwork.
| binaries | term | grad4 | grad6 | ||
|---|---|---|---|---|---|
| Full Sample | GPA Subsample | Full Sample | GPA Subsample | ||
| yes | (Intercept) | 0.82*** | 0.68*** | 4.55*** | 4.15*** |
| yes | Self-Concordant Motivation | 1.00 | 1.01 | 1.01 | 1.00 |
| yes | Goal | 0.99 | 1.00 | 1.02* | 1.04 |
| yes | Leadership | 1.09*** | 1.06* | 1.17*** | 1.16*** |
| yes | Learning | 1.05*** | 1.05* | 1.08*** | 1.08** |
| yes | Perseverance | 1.01 | 0.96 | 1.02 | 1.07* |
| yes | Self-Transcendence | 1.09*** | 1.04 | 1.14*** | 1.15*** |
| yes | Teamwork | 1.07*** | 1.01 | 1.07*** | 0.99 |
| yes | stdtest | 1.23*** | 1.09*** | 1.50*** | 1.23*** |
| yes | OSAn | 1.11*** | 1.08*** | 1.25*** | 1.20*** |
| yes | OSAt | 1.02*** | 1.01 | 1.09*** | 1.09*** |
| yes | OSAsport | 1.02*** | 1.03* | 1.05*** | 1.06*** |
| yes | sexM | 0.78*** | 0.83*** | 0.69*** | 0.72*** |
| yes | parentdegreeOne | 1.11*** | 1.16*** | 1.21*** | 1.26*** |
| yes | parentdegreeTwo | 1.07*** | 1.19*** | 1.36*** | 1.46*** |
| yes | parentmarriedOther | 0.86*** | 0.87*** | 0.76*** | 0.74*** |
| yes | firstlanguageOther | 0.93*** | 0.86*** | 0.81*** | 0.70*** |
| yes | raceBlack | 0.73*** | 0.79*** | 0.82*** | 0.79*** |
| yes | raceLatino | 0.79*** | 0.86*** | 0.89*** | 0.95 |
| yes | raceAsian | 0.79*** | 0.84*** | 0.75*** | 0.77*** |
| yes | raceachieve | 0.82*** | 0.88** | 0.77*** | 0.79*** |
| yes | raceMissing | 1.00 | 0.97 | 0.87*** | 0.84*** |
| yes | hstypehs_pubT1 | 0.88*** | 0.94* | 0.87*** | 0.88*** |
| yes | hstypehs_priv | 1.00 | 1.03 | 0.82*** | 0.94* |
| yes | hstypehs_home | 0.59*** | 0.77 | 0.48*** | 0.66** |
| yes | n | 271358 | 39684 | 271358 | 39684 |
| no | (Intercept) | 0.89*** | 0.71*** | 5.27*** | 4.76*** |
| no | Self-Concordant Motivation | 1.00 | 1.01 | 1.00 | 1.01 |
| no | Goal | 1.00 | 1.00 | 1.01* | 1.02 |
| no | Leadership | 1.03*** | 1.03* | 1.07*** | 1.06*** |
| no | Learning | 1.03*** | 1.02* | 1.04*** | 1.04** |
| no | Perseverance | 1.00 | 0.98 | 1.01* | 1.03* |
| no | Self-Transcendence | 1.04*** | 1.02 | 1.07*** | 1.07*** |
| no | Teamwork | 1.03*** | 1.01 | 1.04*** | 1.00 |
| no | stdtest | 1.23*** | 1.09*** | 1.50*** | 1.23*** |
| no | OSAn | 1.11*** | 1.08*** | 1.24*** | 1.19*** |
| no | OSAt | 1.02*** | 1.01 | 1.09*** | 1.09*** |
| no | OSAsport | 1.02*** | 1.03* | 1.05*** | 1.06*** |
| no | sexM | 0.78*** | 0.83*** | 0.69*** | 0.72*** |
| no | parentdegreeOne | 1.11*** | 1.16*** | 1.21*** | 1.26*** |
| no | parentdegreeTwo | 1.06*** | 1.19*** | 1.36*** | 1.45*** |
| no | parentmarriedOther | 0.86*** | 0.88*** | 0.76*** | 0.74*** |
| no | firstlanguageOther | 0.94*** | 0.86*** | 0.81*** | 0.70*** |
| no | raceBlack | 0.73*** | 0.79*** | 0.82*** | 0.79*** |
| no | raceLatino | 0.79*** | 0.86*** | 0.89*** | 0.95 |
| no | raceAsian | 0.79*** | 0.84*** | 0.75*** | 0.77*** |
| no | raceachieve | 0.82*** | 0.88** | 0.77*** | 0.79*** |
| no | raceMissing | 1.00 | 0.97 | 0.87*** | 0.84*** |
| no | hstypehs_pubT1 | 0.88*** | 0.94* | 0.87*** | 0.88*** |
| no | hstypehs_priv | 1.00 | 1.04 | 0.82*** | 0.94* |
| no | hstypehs_home | 0.59*** | 0.77 | 0.48*** | 0.66** |
| no | n | 271358 | 39684 | 271358 | 39684 |
| yes | gpa | NA | 1.24*** | NA | 1.42*** |
| no | gpa | NA | 1.24*** | NA | 1.42*** |
Things look similar if we control for institutional graduation rates.
| binaries | term | grad4 | grad6 | ||
|---|---|---|---|---|---|
| Full Sample | GPA Subsample | Full Sample | GPA Subsample | ||
| yes | (Intercept) | 1.03* | 0.86*** | 8.49*** | 7.66*** |
| yes | Self-Concordant Motivation | 1.00 | 1.01 | 1.00 | 1.00 |
| yes | Goal | 0.98* | 0.98 | 1.00 | 1.01 |
| yes | Leadership | 1.06*** | 1.05 | 1.13*** | 1.11** |
| yes | Learning | 1.04*** | 1.04 | 1.06*** | 1.09** |
| yes | Perseverance | 0.99 | 0.93* | 0.98 | 1.04 |
| yes | Self-Transcendence | 1.06*** | 1.01 | 1.11*** | 1.12** |
| yes | Teamwork | 1.05*** | 1.00 | 1.05** | 0.99 |
| yes | stdtest | 0.99 | 0.94*** | 1.11*** | 1.00 |
| yes | rates | 1.63*** | 1.58*** | 1.93*** | 1.84*** |
| yes | OSAn | 1.01** | 1.00 | 1.12*** | 1.09*** |
| yes | OSAt | 0.98*** | 0.97* | 1.07*** | 1.07*** |
| yes | OSAsport | 1.00 | 1.01 | 1.03*** | 1.05** |
| yes | sexM | 0.80*** | 0.84*** | 0.68*** | 0.68*** |
| yes | parentdegreeOne | 1.06*** | 1.11** | 1.12*** | 1.17*** |
| yes | parentdegreeTwo | 0.95*** | 1.07* | 1.15*** | 1.25*** |
| yes | parentmarriedOther | 0.90*** | 0.89*** | 0.79*** | 0.74*** |
| yes | firstlanguageOther | 0.86*** | 0.80*** | 0.65*** | 0.57*** |
| yes | raceBlack | 0.68*** | 0.74*** | 0.79*** | 0.76*** |
| yes | raceLatino | 0.79*** | 0.85** | 0.93** | 0.98 |
| yes | raceAsian | 0.80*** | 0.85*** | 0.69*** | 0.72*** |
| yes | raceachieve | 0.83*** | 0.90** | 0.78*** | 0.84** |
| yes | raceMissing | 0.98 | 0.94* | 0.84*** | 0.81*** |
| yes | hstypehs_pubT1 | 0.88*** | 0.96 | 0.89*** | 0.93 |
| yes | hstypehs_priv | 0.90*** | 0.90*** | 0.68*** | 0.75*** |
| yes | hstypehs_home | 0.64*** | 0.82 | 0.54*** | 0.78 |
| yes | n | 239100 | 34988 | 239100 | 34988 |
| no | (Intercept) | 1.09*** | 0.87*** | 9.39*** | 8.49*** |
| no | Self-Concordant Motivation | 1.00 | 1.01 | 1.00 | 1.01 |
| no | Goal | 0.99** | 0.99 | 1.00 | 1.00 |
| no | Leadership | 1.03*** | 1.03* | 1.05*** | 1.05** |
| no | Learning | 1.02*** | 1.02 | 1.03*** | 1.04** |
| no | Perseverance | 0.99 | 0.97** | 0.99 | 1.01 |
| no | Self-Transcendence | 1.03*** | 1.00 | 1.06*** | 1.06*** |
| no | Teamwork | 1.02*** | 1.00 | 1.02*** | 1.00 |
| no | stdtest | 0.99 | 0.94*** | 1.11*** | 1.00 |
| no | rates | 1.63*** | 1.58*** | 1.93*** | 1.84*** |
| no | OSAn | 1.01** | 1.00 | 1.12*** | 1.09*** |
| no | OSAt | 0.98*** | 0.97* | 1.07*** | 1.07*** |
| no | OSAsport | 1.00 | 1.01 | 1.03*** | 1.05** |
| no | sexM | 0.80*** | 0.84*** | 0.68*** | 0.69*** |
| no | parentdegreeOne | 1.06*** | 1.11** | 1.12*** | 1.17*** |
| no | parentdegreeTwo | 0.95*** | 1.07* | 1.15*** | 1.25*** |
| no | parentmarriedOther | 0.90*** | 0.89*** | 0.79*** | 0.74*** |
| no | firstlanguageOther | 0.86*** | 0.80*** | 0.65*** | 0.57*** |
| no | raceBlack | 0.68*** | 0.74*** | 0.79*** | 0.76*** |
| no | raceLatino | 0.79*** | 0.85** | 0.93** | 0.98 |
| no | raceAsian | 0.80*** | 0.85*** | 0.69*** | 0.72*** |
| no | raceachieve | 0.83*** | 0.90** | 0.78*** | 0.84** |
| no | raceMissing | 0.98 | 0.94* | 0.84*** | 0.81*** |
| no | hstypehs_pubT1 | 0.88*** | 0.96 | 0.89*** | 0.93 |
| no | hstypehs_priv | 0.90*** | 0.90*** | 0.68*** | 0.75*** |
| no | hstypehs_home | 0.64*** | 0.82 | 0.54*** | 0.78 |
| no | n | 239100 | 34988 | 239100 | 34988 |
| yes | gpa | NA | 1.18*** | NA | 1.30*** |
| no | gpa | NA | 1.18*** | NA | 1.30*** |
The following appendix tables show how each of the codes predict college graduation.
| term | grad4 | grad6 | ||
|---|---|---|---|---|
| Full Sample | GPA Subsample | Full Sample | GPA Subsample | |
| (Intercept) | 0.15*** | 0.06*** | 0.13*** | 0.04*** |
| Self-Concordant Motivation | 0.98* | 1.00 | 0.98* | 0.98 |
| stdtest | 1.00*** | 1.00*** | 1.00*** | 1.00*** |
| OSAn | 1.06*** | 1.05*** | 1.12*** | 1.11*** |
| OSAt | 1.02*** | 1.02 | 1.12*** | 1.13*** |
| OSAsport | 1.09*** | 1.09* | 1.18*** | 1.22*** |
| sexM | 0.77*** | 0.82*** | 0.68*** | 0.71*** |
| parentdegreeOne | 1.11*** | 1.16*** | 1.21*** | 1.27*** |
| parentdegreeTwo | 1.07*** | 1.19*** | 1.37*** | 1.46*** |
| parentmarriedOther | 0.86*** | 0.87*** | 0.75*** | 0.74*** |
| firstlanguageOther | 0.93*** | 0.86*** | 0.81*** | 0.70*** |
| raceBlack | 0.73*** | 0.79*** | 0.82*** | 0.79*** |
| raceLatino | 0.79*** | 0.86** | 0.89*** | 0.96 |
| raceAsian | 0.79*** | 0.84*** | 0.75*** | 0.78*** |
| raceachieve | 0.81*** | 0.88** | 0.77*** | 0.79*** |
| raceMissing | 0.99 | 0.96 | 0.87*** | 0.84*** |
| hstypehs_pubT1 | 0.88*** | 0.94* | 0.87*** | 0.88*** |
| hstypehs_priv | 1.00 | 1.03 | 0.82*** | 0.94 |
| hstypehs_home | 0.58*** | 0.76* | 0.47*** | 0.64** |
| n | 271358 | 39684 | 271358 | 39684 |
| gpa | NA | 5.31*** | NA | 14.48*** |
| term | grad4 | grad6 | ||
|---|---|---|---|---|
| Full Sample | GPA Subsample | Full Sample | GPA Subsample | |
| (Intercept) | 0.15*** | 0.06*** | 0.13*** | 0.04*** |
| Goal | 1.01 | 1.00 | 1.06*** | 1.09* |
| stdtest | 1.00*** | 1.00*** | 1.00*** | 1.00*** |
| OSAn | 1.06*** | 1.05*** | 1.12*** | 1.11*** |
| OSAt | 1.02*** | 1.02 | 1.12*** | 1.13*** |
| OSAsport | 1.08*** | 1.09* | 1.17*** | 1.22*** |
| sexM | 0.77*** | 0.82*** | 0.68*** | 0.71*** |
| parentdegreeOne | 1.11*** | 1.16*** | 1.21*** | 1.27*** |
| parentdegreeTwo | 1.07*** | 1.19*** | 1.37*** | 1.46*** |
| parentmarriedOther | 0.86*** | 0.87*** | 0.75*** | 0.74*** |
| firstlanguageOther | 0.93*** | 0.86*** | 0.81*** | 0.70*** |
| raceBlack | 0.73*** | 0.79*** | 0.82*** | 0.79*** |
| raceLatino | 0.79*** | 0.86** | 0.89*** | 0.96 |
| raceAsian | 0.80*** | 0.84*** | 0.75*** | 0.78*** |
| raceachieve | 0.81*** | 0.88** | 0.77*** | 0.79*** |
| raceMissing | 0.99 | 0.96 | 0.87*** | 0.84*** |
| hstypehs_pubT1 | 0.88*** | 0.94* | 0.87*** | 0.88*** |
| hstypehs_priv | 1.00 | 1.03 | 0.82*** | 0.94* |
| hstypehs_home | 0.58*** | 0.76* | 0.47*** | 0.64** |
| n | 271358 | 39684 | 271358 | 39684 |
| gpa | NA | 5.31*** | NA | 14.55*** |
| term | grad4 | grad6 | ||
|---|---|---|---|---|
| Full Sample | GPA Subsample | Full Sample | GPA Subsample | |
| (Intercept) | 0.15*** | 0.06*** | 0.13*** | 0.04*** |
| Leadership | 1.11*** | 1.08* | 1.21*** | 1.18*** |
| stdtest | 1.00*** | 1.00*** | 1.00*** | 1.00*** |
| OSAn | 1.06*** | 1.05*** | 1.12*** | 1.10*** |
| OSAt | 1.02*** | 1.02 | 1.11*** | 1.12*** |
| OSAsport | 1.08*** | 1.09* | 1.17*** | 1.22*** |
| sexM | 0.77*** | 0.82*** | 0.68*** | 0.71*** |
| parentdegreeOne | 1.11*** | 1.16*** | 1.21*** | 1.27*** |
| parentdegreeTwo | 1.07*** | 1.19*** | 1.37*** | 1.46*** |
| parentmarriedOther | 0.86*** | 0.87*** | 0.76*** | 0.74*** |
| firstlanguageOther | 0.94*** | 0.86*** | 0.81*** | 0.70*** |
| raceBlack | 0.73*** | 0.79*** | 0.82*** | 0.79*** |
| raceLatino | 0.79*** | 0.86*** | 0.89*** | 0.96 |
| raceAsian | 0.80*** | 0.84*** | 0.75*** | 0.78*** |
| raceachieve | 0.81*** | 0.88** | 0.77*** | 0.79*** |
| raceMissing | 1.00 | 0.97 | 0.87*** | 0.84*** |
| hstypehs_pubT1 | 0.87*** | 0.94* | 0.87*** | 0.88*** |
| hstypehs_priv | 1.00 | 1.04 | 0.82*** | 0.95 |
| hstypehs_home | 0.58*** | 0.77 | 0.47*** | 0.65** |
| n | 271358 | 39684 | 271358 | 39684 |
| gpa | NA | 5.28*** | NA | 14.32*** |
| term | grad4 | grad6 | ||
|---|---|---|---|---|
| Full Sample | GPA Subsample | Full Sample | GPA Subsample | |
| (Intercept) | 0.15*** | 0.06*** | 0.13*** | 0.04*** |
| Learning | 1.05*** | 1.05 | 1.08*** | 1.06* |
| stdtest | 1.00*** | 1.00*** | 1.00*** | 1.00*** |
| OSAn | 1.06*** | 1.05*** | 1.12*** | 1.11*** |
| OSAt | 1.02*** | 1.02 | 1.12*** | 1.13*** |
| OSAsport | 1.08*** | 1.09* | 1.17*** | 1.21*** |
| sexM | 0.77*** | 0.82*** | 0.68*** | 0.71*** |
| parentdegreeOne | 1.11*** | 1.16*** | 1.21*** | 1.27*** |
| parentdegreeTwo | 1.07*** | 1.19*** | 1.36*** | 1.46*** |
| parentmarriedOther | 0.86*** | 0.87*** | 0.75*** | 0.74*** |
| firstlanguageOther | 0.93*** | 0.86*** | 0.81*** | 0.70*** |
| raceBlack | 0.73*** | 0.79*** | 0.82*** | 0.79*** |
| raceLatino | 0.79*** | 0.86*** | 0.89*** | 0.96 |
| raceAsian | 0.79*** | 0.84*** | 0.75*** | 0.78*** |
| raceachieve | 0.82*** | 0.88** | 0.77*** | 0.79*** |
| raceMissing | 0.99 | 0.96 | 0.87*** | 0.84*** |
| hstypehs_pubT1 | 0.88*** | 0.94* | 0.87*** | 0.88*** |
| hstypehs_priv | 1.00 | 1.03 | 0.82*** | 0.94 |
| hstypehs_home | 0.58*** | 0.77* | 0.47*** | 0.65** |
| n | 271358 | 39684 | 271358 | 39684 |
| gpa | NA | 5.31*** | NA | 14.49*** |
| term | grad4 | grad6 | ||
|---|---|---|---|---|
| Full Sample | GPA Subsample | Full Sample | GPA Subsample | |
| (Intercept) | 0.15*** | 0.06*** | 0.13*** | 0.04*** |
| Perseverance | 1.00 | 0.95 | 1.01 | 1.06 |
| stdtest | 1.00*** | 1.00*** | 1.00*** | 1.00*** |
| OSAn | 1.06*** | 1.05*** | 1.12*** | 1.11*** |
| OSAt | 1.02*** | 1.02 | 1.12*** | 1.13*** |
| OSAsport | 1.08*** | 1.10* | 1.17*** | 1.21*** |
| sexM | 0.77*** | 0.82*** | 0.68*** | 0.71*** |
| parentdegreeOne | 1.11*** | 1.16*** | 1.21*** | 1.27*** |
| parentdegreeTwo | 1.07*** | 1.19*** | 1.37*** | 1.46*** |
| parentmarriedOther | 0.86*** | 0.87*** | 0.75*** | 0.74*** |
| firstlanguageOther | 0.93*** | 0.86*** | 0.81*** | 0.70*** |
| raceBlack | 0.73*** | 0.79*** | 0.82*** | 0.79*** |
| raceLatino | 0.79*** | 0.86** | 0.89*** | 0.96 |
| raceAsian | 0.80*** | 0.84*** | 0.75*** | 0.78*** |
| raceachieve | 0.81*** | 0.88** | 0.77*** | 0.79*** |
| raceMissing | 0.99 | 0.97 | 0.87*** | 0.84*** |
| hstypehs_pubT1 | 0.88*** | 0.94 | 0.87*** | 0.88*** |
| hstypehs_priv | 1.00 | 1.03 | 0.82*** | 0.94* |
| hstypehs_home | 0.58*** | 0.76* | 0.47*** | 0.64** |
| n | 271358 | 39684 | 271358 | 39684 |
| gpa | NA | 5.31*** | NA | 14.50*** |
| term | grad4 | grad6 | ||
|---|---|---|---|---|
| Full Sample | GPA Subsample | Full Sample | GPA Subsample | |
| (Intercept) | 0.15*** | 0.06*** | 0.13*** | 0.04*** |
| Self-Transcendence | 1.08*** | 1.04 | 1.13*** | 1.15*** |
| stdtest | 1.00*** | 1.00*** | 1.00*** | 1.00*** |
| OSAn | 1.06*** | 1.05*** | 1.12*** | 1.10*** |
| OSAt | 1.02*** | 1.02 | 1.12*** | 1.13*** |
| OSAsport | 1.09*** | 1.10* | 1.19*** | 1.24*** |
| sexM | 0.78*** | 0.83*** | 0.69*** | 0.72*** |
| parentdegreeOne | 1.11*** | 1.16*** | 1.21*** | 1.27*** |
| parentdegreeTwo | 1.07*** | 1.19*** | 1.37*** | 1.46*** |
| parentmarriedOther | 0.86*** | 0.87*** | 0.76*** | 0.74*** |
| firstlanguageOther | 0.93*** | 0.86*** | 0.81*** | 0.70*** |
| raceBlack | 0.73*** | 0.79*** | 0.82*** | 0.79*** |
| raceLatino | 0.79*** | 0.86*** | 0.89*** | 0.95 |
| raceAsian | 0.79*** | 0.84*** | 0.75*** | 0.78*** |
| raceachieve | 0.82*** | 0.88** | 0.77*** | 0.79*** |
| raceMissing | 0.99 | 0.96 | 0.87*** | 0.84*** |
| hstypehs_pubT1 | 0.88*** | 0.94* | 0.87*** | 0.88*** |
| hstypehs_priv | 1.00 | 1.03 | 0.82*** | 0.94* |
| hstypehs_home | 0.58*** | 0.76* | 0.47*** | 0.64** |
| n | 271358 | 39684 | 271358 | 39684 |
| gpa | NA | 5.30*** | NA | 14.39*** |
| term | grad4 | grad6 | ||
|---|---|---|---|---|
| Full Sample | GPA Subsample | Full Sample | GPA Subsample | |
| (Intercept) | 0.15*** | 0.06*** | 0.13*** | 0.04*** |
| Teamwork | 1.11*** | 1.04 | 1.13*** | 1.01 |
| stdtest | 1.00*** | 1.00*** | 1.00*** | 1.00*** |
| OSAn | 1.06*** | 1.05*** | 1.12*** | 1.11*** |
| OSAt | 1.02*** | 1.02 | 1.12*** | 1.13*** |
| OSAsport | 1.08*** | 1.09* | 1.16*** | 1.22*** |
| sexM | 0.77*** | 0.82*** | 0.68*** | 0.71*** |
| parentdegreeOne | 1.11*** | 1.16*** | 1.21*** | 1.27*** |
| parentdegreeTwo | 1.07*** | 1.19*** | 1.37*** | 1.46*** |
| parentmarriedOther | 0.86*** | 0.87*** | 0.75*** | 0.74*** |
| firstlanguageOther | 0.94*** | 0.86*** | 0.81*** | 0.70*** |
| raceBlack | 0.73*** | 0.79*** | 0.82*** | 0.79*** |
| raceLatino | 0.79*** | 0.86** | 0.89*** | 0.96 |
| raceAsian | 0.80*** | 0.84*** | 0.75*** | 0.78*** |
| raceachieve | 0.82*** | 0.88** | 0.77*** | 0.79*** |
| raceMissing | 0.99 | 0.96 | 0.87*** | 0.84*** |
| hstypehs_pubT1 | 0.88*** | 0.94* | 0.87*** | 0.88*** |
| hstypehs_priv | 1.00 | 1.03 | 0.82*** | 0.94 |
| hstypehs_home | 0.58*** | 0.76* | 0.47*** | 0.64** |
| n | 271358 | 39684 | 271358 | 39684 |
| gpa | NA | 5.31*** | NA | 14.49*** |
| term | grad4 | grad6 | ||
|---|---|---|---|---|
| Full Sample | GPA Subsample | Full Sample | GPA Subsample | |
| (Intercept) | 0.32*** | 0.15*** | 0.15*** | 0.06*** |
| Self-Concordant Motivation | 0.98* | 1.00 | 0.98 | 0.99 |
| stdtest | 1.00 | 1.00*** | 1.00*** | 1.00 |
| grad4rates | 9.58*** | 7.77*** | NA | NA |
| OSAn | 1.01*** | 1.00 | 1.06*** | 1.05*** |
| OSAt | 0.98*** | 0.96* | 1.09*** | 1.10*** |
| OSAsport | 1.00 | 1.05 | 1.11*** | 1.20** |
| sexM | 0.80*** | 0.84*** | 0.67*** | 0.68*** |
| parentdegreeOne | 1.07*** | 1.11** | 1.12*** | 1.17*** |
| parentdegreeTwo | 0.95*** | 1.07* | 1.15*** | 1.25*** |
| parentmarriedOther | 0.89*** | 0.89*** | 0.78*** | 0.74*** |
| firstlanguageOther | 0.86*** | 0.80*** | 0.65*** | 0.57*** |
| raceBlack | 0.68*** | 0.74*** | 0.79*** | 0.76*** |
| raceLatino | 0.79*** | 0.85** | 0.93** | 0.99 |
| raceAsian | 0.80*** | 0.85*** | 0.69*** | 0.72*** |
| raceachieve | 0.83*** | 0.90** | 0.78*** | 0.84** |
| raceMissing | 0.98 | 0.94* | 0.84*** | 0.81*** |
| hstypehs_pubT1 | 0.88*** | 0.96 | 0.89*** | 0.93 |
| hstypehs_priv | 0.90*** | 0.90*** | 0.67*** | 0.75*** |
| hstypehs_home | 0.63*** | 0.81 | 0.53*** | 0.76 |
| n | 239100 | 34988 | 239100 | 34988 |
| gpa | NA | 3.46*** | NA | 7.54*** |
| grad6rates | NA | NA | 53.66*** | 38.48*** |
| term | grad4 | grad6 | ||
|---|---|---|---|---|
| Full Sample | GPA Subsample | Full Sample | GPA Subsample | |
| (Intercept) | 0.32*** | 0.15*** | 0.14*** | 0.06*** |
| Goal | 0.98* | 0.97 | 1.01 | 1.03 |
| stdtest | 1.00 | 1.00*** | 1.00*** | 1.00 |
| grad4rates | 9.59*** | 7.79*** | NA | NA |
| OSAn | 1.01*** | 1.00 | 1.06*** | 1.05*** |
| OSAt | 0.98*** | 0.96* | 1.09*** | 1.10*** |
| OSAsport | 1.00 | 1.05 | 1.11*** | 1.20** |
| sexM | 0.80*** | 0.84*** | 0.67*** | 0.68*** |
| parentdegreeOne | 1.07*** | 1.11** | 1.12*** | 1.17*** |
| parentdegreeTwo | 0.95*** | 1.07* | 1.15*** | 1.25*** |
| parentmarriedOther | 0.89*** | 0.89*** | 0.78*** | 0.74*** |
| firstlanguageOther | 0.86*** | 0.80*** | 0.65*** | 0.57*** |
| raceBlack | 0.68*** | 0.74*** | 0.79*** | 0.76*** |
| raceLatino | 0.79*** | 0.85** | 0.93** | 0.99 |
| raceAsian | 0.80*** | 0.85*** | 0.69*** | 0.72*** |
| raceachieve | 0.83*** | 0.90** | 0.78*** | 0.84** |
| raceMissing | 0.98 | 0.94* | 0.84*** | 0.81*** |
| hstypehs_pubT1 | 0.88*** | 0.95 | 0.89*** | 0.93 |
| hstypehs_priv | 0.90*** | 0.90*** | 0.67*** | 0.75*** |
| hstypehs_home | 0.63*** | 0.81 | 0.53*** | 0.76 |
| n | 239100 | 34988 | 239100 | 34988 |
| gpa | NA | 3.46*** | NA | 7.55*** |
| grad6rates | NA | NA | 53.64*** | 38.44*** |
| term | grad4 | grad6 | ||
|---|---|---|---|---|
| Full Sample | GPA Subsample | Full Sample | GPA Subsample | |
| (Intercept) | 0.31*** | 0.15*** | 0.15*** | 0.06*** |
| Leadership | 1.09*** | 1.07* | 1.16*** | 1.12** |
| stdtest | 1.00 | 1.00*** | 1.00*** | 1.00 |
| grad4rates | 9.56*** | 7.77*** | NA | NA |
| OSAn | 1.01*** | 1.00 | 1.06*** | 1.05*** |
| OSAt | 0.98*** | 0.96* | 1.08*** | 1.09*** |
| OSAsport | 1.00 | 1.05 | 1.11*** | 1.20** |
| sexM | 0.80*** | 0.84*** | 0.67*** | 0.68*** |
| parentdegreeOne | 1.07*** | 1.11** | 1.12*** | 1.17*** |
| parentdegreeTwo | 0.95*** | 1.07* | 1.15*** | 1.25*** |
| parentmarriedOther | 0.89*** | 0.89*** | 0.78*** | 0.74*** |
| firstlanguageOther | 0.86*** | 0.80*** | 0.65*** | 0.57*** |
| raceBlack | 0.68*** | 0.74*** | 0.79*** | 0.76*** |
| raceLatino | 0.79*** | 0.85** | 0.93** | 0.99 |
| raceAsian | 0.80*** | 0.85*** | 0.69*** | 0.73*** |
| raceachieve | 0.83*** | 0.90** | 0.78*** | 0.84** |
| raceMissing | 0.98 | 0.94* | 0.84*** | 0.82*** |
| hstypehs_pubT1 | 0.88*** | 0.95 | 0.89*** | 0.93 |
| hstypehs_priv | 0.90*** | 0.90*** | 0.68*** | 0.75*** |
| hstypehs_home | 0.63*** | 0.82 | 0.53*** | 0.77 |
| n | 239100 | 34988 | 239100 | 34988 |
| gpa | NA | 3.44*** | NA | 7.47*** |
| grad6rates | NA | NA | 53.34*** | 38.36*** |
| term | grad4 | grad6 | ||
|---|---|---|---|---|
| Full Sample | GPA Subsample | Full Sample | GPA Subsample | |
| (Intercept) | 0.31*** | 0.15*** | 0.14*** | 0.05*** |
| Learning | 1.04*** | 1.04 | 1.06*** | 1.08* |
| stdtest | 1.00 | 1.00*** | 1.00*** | 1.00 |
| grad4rates | 9.58*** | 7.78*** | NA | NA |
| OSAn | 1.01*** | 1.00 | 1.06*** | 1.05*** |
| OSAt | 0.98*** | 0.96* | 1.09*** | 1.10*** |
| OSAsport | 1.00 | 1.05 | 1.11*** | 1.20** |
| sexM | 0.80*** | 0.84*** | 0.67*** | 0.68*** |
| parentdegreeOne | 1.06*** | 1.11** | 1.12*** | 1.17*** |
| parentdegreeTwo | 0.95*** | 1.07* | 1.15*** | 1.25*** |
| parentmarriedOther | 0.89*** | 0.89*** | 0.78*** | 0.74*** |
| firstlanguageOther | 0.86*** | 0.80*** | 0.65*** | 0.57*** |
| raceBlack | 0.68*** | 0.74*** | 0.80*** | 0.76*** |
| raceLatino | 0.79*** | 0.85** | 0.93** | 0.99 |
| raceAsian | 0.80*** | 0.85*** | 0.69*** | 0.72*** |
| raceachieve | 0.83*** | 0.90** | 0.78*** | 0.84** |
| raceMissing | 0.98 | 0.94* | 0.84*** | 0.81*** |
| hstypehs_pubT1 | 0.88*** | 0.95 | 0.89*** | 0.93 |
| hstypehs_priv | 0.90*** | 0.90*** | 0.67*** | 0.75*** |
| hstypehs_home | 0.63*** | 0.82 | 0.53*** | 0.77 |
| n | 239100 | 34988 | 239100 | 34988 |
| gpa | NA | 3.46*** | NA | 7.53*** |
| grad6rates | NA | NA | 53.58*** | 38.52*** |
| term | grad4 | grad6 | ||
|---|---|---|---|---|
| Full Sample | GPA Subsample | Full Sample | GPA Subsample | |
| (Intercept) | 0.31*** | 0.15*** | 0.14*** | 0.06*** |
| Perseverance | 0.97* | 0.91** | 0.96* | 1.02 |
| stdtest | 1.00 | 1.00*** | 1.00*** | 1.00 |
| grad4rates | 9.59*** | 7.80*** | NA | NA |
| OSAn | 1.01*** | 1.00 | 1.06*** | 1.05*** |
| OSAt | 0.98*** | 0.96* | 1.09*** | 1.10*** |
| OSAsport | 1.01 | 1.06 | 1.12*** | 1.20** |
| sexM | 0.80*** | 0.84*** | 0.67*** | 0.68*** |
| parentdegreeOne | 1.07*** | 1.11** | 1.12*** | 1.17*** |
| parentdegreeTwo | 0.95*** | 1.07* | 1.15*** | 1.25*** |
| parentmarriedOther | 0.89*** | 0.89*** | 0.78*** | 0.74*** |
| firstlanguageOther | 0.86*** | 0.80*** | 0.65*** | 0.57*** |
| raceBlack | 0.68*** | 0.74*** | 0.79*** | 0.76*** |
| raceLatino | 0.79*** | 0.85** | 0.93** | 0.99 |
| raceAsian | 0.80*** | 0.85*** | 0.69*** | 0.72*** |
| raceachieve | 0.83*** | 0.90** | 0.78*** | 0.84** |
| raceMissing | 0.98 | 0.94* | 0.84*** | 0.81*** |
| hstypehs_pubT1 | 0.88*** | 0.96 | 0.89*** | 0.93 |
| hstypehs_priv | 0.90*** | 0.90*** | 0.67*** | 0.75*** |
| hstypehs_home | 0.63*** | 0.81 | 0.53*** | 0.76 |
| n | 239100 | 34988 | 239100 | 34988 |
| gpa | NA | 3.46*** | NA | 7.54*** |
| grad6rates | NA | NA | 53.76*** | 38.46*** |
| term | grad4 | grad6 | ||
|---|---|---|---|---|
| Full Sample | GPA Subsample | Full Sample | GPA Subsample | |
| (Intercept) | 0.30*** | 0.15*** | 0.14*** | 0.05*** |
| Self-Transcendence | 1.06*** | 1.01 | 1.11*** | 1.12** |
| stdtest | 1.00 | 1.00*** | 1.00*** | 1.00 |
| grad4rates | 9.56*** | 7.77*** | NA | NA |
| OSAn | 1.01*** | 1.00 | 1.06*** | 1.05*** |
| OSAt | 0.98*** | 0.96* | 1.09*** | 1.10*** |
| OSAsport | 1.01 | 1.05 | 1.13*** | 1.21** |
| sexM | 0.80*** | 0.84*** | 0.67*** | 0.68*** |
| parentdegreeOne | 1.07*** | 1.11** | 1.12*** | 1.17*** |
| parentdegreeTwo | 0.95*** | 1.07* | 1.15*** | 1.25*** |
| parentmarriedOther | 0.89*** | 0.89*** | 0.79*** | 0.74*** |
| firstlanguageOther | 0.86*** | 0.80*** | 0.65*** | 0.56*** |
| raceBlack | 0.68*** | 0.74*** | 0.79*** | 0.76*** |
| raceLatino | 0.79*** | 0.85** | 0.93** | 0.98 |
| raceAsian | 0.80*** | 0.85*** | 0.69*** | 0.72*** |
| raceachieve | 0.83*** | 0.90** | 0.78*** | 0.84** |
| raceMissing | 0.98 | 0.94* | 0.84*** | 0.81*** |
| hstypehs_pubT1 | 0.88*** | 0.96 | 0.89*** | 0.93 |
| hstypehs_priv | 0.90*** | 0.90*** | 0.67*** | 0.75*** |
| hstypehs_home | 0.63*** | 0.81 | 0.53*** | 0.76 |
| n | 239100 | 34988 | 239100 | 34988 |
| gpa | NA | 3.46*** | NA | 7.50*** |
| grad6rates | NA | NA | 53.50*** | 38.25*** |
| term | grad4 | grad6 | ||
|---|---|---|---|---|
| Full Sample | GPA Subsample | Full Sample | GPA Subsample | |
| (Intercept) | 0.31*** | 0.15*** | 0.14*** | 0.06*** |
| Teamwork | 1.08*** | 1.02 | 1.08*** | 1.00 |
| stdtest | 1.00 | 1.00*** | 1.00*** | 1.00 |
| grad4rates | 9.56*** | 7.77*** | NA | NA |
| OSAn | 1.01*** | 1.00 | 1.06*** | 1.05*** |
| OSAt | 0.98*** | 0.96* | 1.09*** | 1.10*** |
| OSAsport | 1.00 | 1.05 | 1.11*** | 1.20** |
| sexM | 0.80*** | 0.84*** | 0.67*** | 0.68*** |
| parentdegreeOne | 1.06*** | 1.11** | 1.12*** | 1.17*** |
| parentdegreeTwo | 0.95*** | 1.07* | 1.15*** | 1.25*** |
| parentmarriedOther | 0.89*** | 0.89*** | 0.78*** | 0.74*** |
| firstlanguageOther | 0.86*** | 0.80*** | 0.65*** | 0.57*** |
| raceBlack | 0.68*** | 0.74*** | 0.79*** | 0.76*** |
| raceLatino | 0.79*** | 0.85** | 0.93** | 0.99 |
| raceAsian | 0.80*** | 0.85*** | 0.69*** | 0.73*** |
| raceachieve | 0.83*** | 0.90** | 0.78*** | 0.84** |
| raceMissing | 0.98 | 0.94* | 0.84*** | 0.81*** |
| hstypehs_pubT1 | 0.88*** | 0.96 | 0.89*** | 0.93 |
| hstypehs_priv | 0.90*** | 0.90*** | 0.68*** | 0.75*** |
| hstypehs_home | 0.63*** | 0.81 | 0.53*** | 0.76 |
| n | 239100 | 34988 | 239100 | 34988 |
| gpa | NA | 3.46*** | NA | 7.54*** |
| grad6rates | NA | NA | 53.48*** | 38.48*** |
There is no issues of multicollinearity in our models. Independent on how we run the models, all VIFs are below 1.76. Below we present VIFs for models not controlling for institutional graduation rates and ran with continuous likelyhoods. Back to models.
| Term | grad4 | grad6 | ||
|---|---|---|---|---|
| full | gpa | full | gpa | |
| `Self-Concordant Motivation` | 1.04 | 1.05 | 1.04 | 1.05 |
| Goal | 1.03 | 1.04 | 1.03 | 1.03 |
| Leadership | 1.08 | 1.08 | 1.07 | 1.07 |
| Learning | 1.04 | 1.04 | 1.04 | 1.04 |
| Perseverance | 1.06 | 1.06 | 1.06 | 1.06 |
| `Self-Transcendence` | 1.11 | 1.12 | 1.10 | 1.11 |
| Teamwork | 1.07 | 1.07 | 1.06 | 1.06 |
| stdtest | 1.37 | 1.68 | 1.37 | 1.65 |
| OSAn | 1.23 | 1.25 | 1.23 | 1.25 |
| OSAt | 1.06 | 1.08 | 1.06 | 1.08 |
| OSAsport | 1.21 | 1.23 | 1.22 | 1.25 |
| sex | 1.10 | 1.11 | 1.11 | 1.12 |
| parentdegree | 1.25 | 1.27 | 1.26 | 1.29 |
| parentmarried | 1.07 | 1.07 | 1.08 | 1.08 |
| firstlanguage | 1.35 | 1.39 | 1.39 | 1.45 |
| race | 1.52 | 1.60 | 1.61 | 1.72 |
| hstype | 1.05 | 1.15 | 1.07 | 1.17 |
| gpa | NA | 1.40 | NA | 1.35 |
It seems like factorizing the codes is not the optimal solution. The KMO statistic is low (KMO = 0.52), and while Bartlett Test of Sphericity is significant, that doesn’t mean much with such a large dataset (chisq(21, [N = 307254])= 49300.16, p = < .001).
If we were to extract factors anyway, parallel analysis suggests the extraction of 3 factors. The solution of such model shown below.
| Item | F1 | F2 | F3 |
|---|---|---|---|
| Self-Concordant Motivation | .03 | -.17 | .09 |
| Goal | .99 | .01 | -.09 |
| Leadership | .06 | .71 | -.00 |
| Learning | -.05 | -.01 | .25 |
| Perseverance | .06 | -.03 | .36 |
| Self-Transcendence | .07 | -.02 | -.50 |
| Teamwork | .01 | .28 | .20 |