OSU Freshman Retention Analysis Fall 2007 to Fall 2008
OSU Freshman Retention Analysis Fall 2007 to Fall 2008 K Celeste Campbell, K. Campbell Ph.D., Ph D University Registrar Registrar, Oklahoma State University April 2009
Purpose of this Study „ To identify factors that were potential predictors of freshman
retention „ To evaluate the relative predictive importance of these factors „ To use these factors to describe how retained students differ from those who were not retained (departed). (departed)
Subjects 3500
3000
2500
2000
1500
3110
2835 2400
1000
500
435 0 Fall 07
Spr 08
Sum 08
Fall 08
Univariate Analyses „ Performed to determine the univariate relationship of 22
initial variables with retention group. These are primarily pre-college and college academic variables, financial variables, and demographic variables. (Table 1, p. 5) „ Tables 2 and 3 (pp. 6-7) 6 7) provide results. „ The only variables that showed no statistically significant relationship with retention group were OK resident status and Residential Life status.
Multivariate Discriminant Analysis „ The 20 variables that showed significance during the
univariate analyses were reduced (redundancy tests) to a set of 12 predictors for the discriminant analysis (Table 4, p. 8). „ This set of variables was moderately successful in predicting to which retention group students belonged (82.4% accuracy; 65% improvement over chance).
Most Important Predictors The four most important predictor variables provide the following distinction between retained and departed students. Students who were not retained: • had significantly lower college GPAs; • were significantly g y more likelyy to have an unpaid p OSU bill;; • withdrew from significantly more credit hours of coursework
during the first semester; and • had significantly lower high school GPAs.
Relatively Small Contributions Other variables that made relatively small contributions to the retention prediction: • ACT score (departed students had lower scores); • number of credit hours attempted during the first semester (departed • • • • •
students attempted fewer hours); family income (departed students had lower family income); amountt off scholarships, h l hi grants, t andd tuition t iti waivers i (departed (d t d students had less assistance that did not require repayment); unmet financial need during the first semester (departed students had more unmet financial need); gender (male students were more likely to depart); and race (minority students were more likely to depart).
Note: College variable was treated uniquely in the analysis.
Further Examination of Departed Students „ The model predicted as retained 39% of those students who
departed. „ Additional analyses were performed to determine characteristics of these students. „ These students were generally very similar to retained students academically, financially, etc. (Table 6, p. 11) „ Clearly, these students departed for reasons that are not measured by this study.
Prior Research Telephone surveys conducted with students who left OSU between 2005 and 2007 identified the following factors as most important in their departure decision: „ financial problems, „academic difficulty, difficulty „personal problems, and „the location of OSU.
Questions or Comments
Oklahoma State University Retention Analysis Report Focusing on Fall 2007 Freshmen K. Celeste Campbell, Ph.D., University Registrar, Oklahoma State University January 2009
Bulleted Summary •
• •
•
•
The retention rate at Oklahoma State University for the Fall 2007 freshman cohort was the lowest in 10 years (77.1%). The highest freshman retention rate during the past 10 years was 84.6% for the 1999 cohort. Using the Fall 2007 freshman cohort, retention status was predicted moderately well (82.4% accuracy) from a set of academic, financial, and demographic variables. The most important variables that predicted retention status were the college GPA, unpaid OSU balance (bursar hold), the number of credit hours dropped during the first semester, and the high school GPA. Students who were not retained: o had significantly lower college GPAs; o were significantly more likely to have an unpaid OSU bill; o withdrew from significantly more credit hours of coursework during the first semester; and o had significantly lower high school GPAs. Other variables that made a relatively small contribution to the prediction of retention were: ACT score (departed students had lower scores); number of credit hours attempted during the first semester (departed students attempted fewer hours); family income (departed students had lower family income); amount of scholarships, grants, and tuition waivers (departed students had less assistance that did not require repayment); unmet financial need during the first semester (departed students had more unmet financial need); gender (male students were more likely to depart); and race (minority students were more likely to depart). A group of students has been identified who were predicted as retained but who in fact have left OSU. These students, who resemble retained students on most of the variables examined in this study, are of particular interest. Recommendations include interviewing a sample of these students to determine the reasons for their departure. Narrative Summary
Over the past 10 years, the freshman retention rate at Oklahoma State University has ranged from 77.1% for the 2007 cohort to 84.6% for the 1999 cohort (OSU Student Profile, 2008). Freshman retention is defined as the percent of full time freshman students enrolled during a fall semester who return for the following fall semester. The purpose of this study was to identify factors that were potential predictors of freshman retention, to evaluate the relative predictive importance of these factors, and to use them to describe how the retained students differ from those who departed. Records were examined from 3,110 freshmen who were enrolled in at least 12 credit hours at OSU during the Fall 2007 semester. After one year, during the Fall 2008 semester, 2,400 of these students were still enrolled at OSU and 710 were not enrolled. Using an initial set of 22 variables, preliminary univariate and subsequent multivariate discriminant analyses resulted in a set of 11 variables that were moderately successful in predicting whether or not students were retained (82.4% accuracy; 65% improvement over chance as measured by the proportional reduction in error). The initial set of variables represented pre college skills and abilities, college academic performance, academic and social integration, financial resources, family and community background, and personal attributes. The most important discriminating variables were the college cumulative graduation/retention GPA, the presence or absence of a bursar hold one year after 1
beginning college, the number of credit hours dropped during the first semester of college, and the high school GPA. Students who were not retained had significantly lower graduation/retention GPAs, had significantly more bursar holds, withdrew from significantly more credit hours of coursework during the first semester, and had significantly lower high school GPAs. The set of predictor variables was quite successful predicting students who were retained (88.8% accuracy). However, 39% of the students who departed were predicted to be retained. That is, 39% of the students who did not enroll at OSU for the Fall 2008 semester appeared to be successful academically and did not appear to have severe financial difficulties. Additional analysis was performed to compare these students with those who were retained and with the other students who were not retained, and the results suggest that the students in this group are generally very similar to the retained group on the academic and financial variables used in this analysis. Further research is needed to determine reasons for the departure of these students. Telephone surveys conducted with students who left OSU between 2005 and 2007 identified the following factors as most important in their departure decision: financial problems, academic difficulty, personal problems, and the location of OSU (OSU First Semester Retention, 2007; Retention, 2007). A similar survey, targeting the students who departed but were predicted to be retained, would be beneficial. Because it is already known that these students did not experience academic or severe financial difficulties, the survey should focus on a few brief, open ended questions to determine the reasons for departure and whether OSU could have done anything to prevent it.
2
Methodology and Results Introduction and Purpose of the Study Over the past 10 years, the freshman retention rate at Oklahoma State University has ranged from 77.1% for the 2007 cohort to 84.6% for the 1999 cohort (OSU Student Profile, 2008). Freshman retention is defined as the percent of full time freshman students enrolled during a fall semester who return for the following fall semester. The purpose of this study was to identify factors that were potential predictors of retention, to evaluate the relative predictive importance of these factors, and to use them to describe how the retained students differ from those who departed. Subjects The subjects for this study are the students in the Fall 2007 Oklahoma State University freshman cohort: 3,110 students who were enrolled in at least 12 credit hours as new freshmen during the Fall 2007 semester. Two students were omitted from the original sample of 3,112 because they were deceased sometime during the Spring 2008 semester. Cohort Enrollment Status The enrollment status of the Fall 2007 freshman cohort is shown in Figure 1. Spring 2008 enrollment showed a loss of 275 of the original students (8.8%). One year after initial enrollment, the Fall 2008 enrollment showed a loss of 710 of the original cohort students (22.8%). This constitutes a freshman retention rate of 77.1%, which is the lowest in 10 years (Fall 2008 OSU Student Profile). A total of 28 students withdrew from all courses before completing the Fall 2007 semester. Three of these students returned for Spring 2008, but none of the 28 were enrolled for Fall 2008. Of the students who were not enrolled for Spring 2008, 15 came back to be enrolled for Fall 2008. In Summer 2008, 435 of the cohort students (14.0%) were enrolled. Of the summer students, 433 were continuing from Spring and 2 were returning from Fall 2007. Thirty five students left at the end of summer. Figures 2 and 3 provide one year retention information by the College of initial enrollment. Figure 1. Fall 2007 Freshman Cohort Enrollment 3500 3000 2500 2000 1500
3110
2835 2400
1000 500 435 0 Fall 07
Spr 08
Sum 08
3
Fall 08
Figure 2. Cohort Enrollment Status after One Year by College of Initial Enrollment 1000 900 800 700 600 500 400 300 200 100 0
661
Retained 437
Departed
423
306 87 AG
208 AS
186
162
225
118
60
106
79
52
BU
ED
EN
GU
HES
Figure 3. Cohort Enrollment Status after One Year: Percentage by College of Initial Enrollment 100% 90%
77.90% 76.10% 78.70% 75.60% 80.00% 67.20% 81.20%
80% 70% 60% 50%
Retained
40%
Departed
30% 20% 10%
32.80% 22.10% 23.90% 21.30% 24.40% 20.00%
18.80%
0% AG
AS
BU
ED
EN
GU
HES
Definitions of Predictor Variables For this study, an initial set of 22 variables was examined to identify factors that were potential predictors of student retention status after one year—factors that would be able to predict whether students were still enrolled at OSU (members of the retained group) or were not enrolled (members of the departed group). Table 1 provides a listing of the initial predictor variables, along with their descriptions and the corresponding category from Tinto’s model of student departure (1993, chap. 4).
4
Table 1. Initial Predictor Variables Predictor
Definition/Explanation
ACT Highest Composite
Highest composite ACT score
HS Class Rank
High school class rank as a percentage, where first in class = 100%
HS Core GPA
High school core GPA
HS GPA
High school cumulative GPA
076 Hours of W
Number of credit hours with grades of W during Fall 07 (courses dropped between week 2 and week 12) Ratio of OSU earned credit hours to OSU attempted credit hours
Total OSU % Completed Hours 076 Earned Hours
Institutional GPA
Number of credit hours earned Fall 07 (excludes courses dropped, courses with grades of F or I, and remedial courses) Satisfactory academic progress for financial aid purposes assessed at the end of the Freshman year (Summer 08); based on good academic standing and successful completion of at least 75% of total cumulative hours attempted; Cleared, probation, suspended. Binary variable indicating academic standing, where good academic standing is a 1.7 grad/ret GPA with 30 earned hours or less and a 2.0 grad/ret GPA with more than 30 earned hours Cumulative graduation/retention GPA, including courses from other institutions. Zero GPAs that resulted from withdrawal from all courses were replaced with “missing.” GPA calculated from all OSU courses
076 Attempted Hours 076 College
Number of credit hours attempted Fall 07 College of initial enrollment: AG, AS, BU, ED, EN, GU, HES
Residential Life during Fall 07
Binary variable indicating whether or not student lived in on campus housing during Fall 07 Binary variable indicating whether or not student is classified as an Oklahoma resident Binary variable indicating student’s gender Binary variable indicating whether or not student is Caucasian Total dollar amount of Fall 07 scholarships, grants, and tuition waivers Fall 07 cost of education minus expected family contribution and all “free” assistance (scholarships, grants, and tuition waivers). Need to borrow assumed to be zero if student did not file FAFSA. Fall 07 cost of education minus expected family contribution and all assistance (scholarships, tuition waivers, grants, student loans, and parent loans). Unmet need assumed to be zero if student did not file FAFSA. Family adjusted gross income; available for students who filed FASFA (65% of sample); missing value for other students. Binary variable indicating whether or not student had a bursar hold November 2008. Generally a bursar hold is placed on a student’s record if there is an overdue balance of at least $100.
SAPFinAidStatus084
Good Academic Standing
Grad/Ret GPA
OK Resident Status Gender Race 076 assistance no repayment 076 need to borrow
076 unmet need
Adjusted Gross Income Current Bursar Hold
5
Category Precollege skills and abilities Precollege skills and abilities Precollege skills and abilities Precollege skills and abilities College academic performance College academic performance College academic performance College academic performance
College academic performance College academic performance College academic performance Academic integration Academic/social integration Social integration Family and community background Personal attributes Personal attributes Financial resources Financial resources
Financial resources
Financial resources Financial resources
Univariate Analysis Preliminary screening analyses (independent samples t tests) were performed to help understand significant univariate differences between the students who were retained and those who departed and to determine a more parsimonious set of viable discriminating variables. Significant group differences were found on all of the variables listed in Table 2. The variables in the table are arranged by decreasing effect size. The larger the effect size (in absolute value), the more different the two groups are on the particular variable. Table 2. Comparisons by Retention Group for Continuous Variables Arranged by Decreasing Effect Size Predictor Departed Retained Cohen’s N M SD N M SD t p d1 Large effect size Total OSU % Completed Hrs** 710 59.20 34.00 2400 91.07 13.52 24.4
Purpose of this Study „ To identify factors that were potential predictors of freshman
retention „ To evaluate the relative predictive importance of these factors „ To use these factors to describe how retained students differ from those who were not retained (departed). (departed)
Subjects 3500
3000
2500
2000
1500
3110
2835 2400
1000
500
435 0 Fall 07
Spr 08
Sum 08
Fall 08
Univariate Analyses „ Performed to determine the univariate relationship of 22
initial variables with retention group. These are primarily pre-college and college academic variables, financial variables, and demographic variables. (Table 1, p. 5) „ Tables 2 and 3 (pp. 6-7) 6 7) provide results. „ The only variables that showed no statistically significant relationship with retention group were OK resident status and Residential Life status.
Multivariate Discriminant Analysis „ The 20 variables that showed significance during the
univariate analyses were reduced (redundancy tests) to a set of 12 predictors for the discriminant analysis (Table 4, p. 8). „ This set of variables was moderately successful in predicting to which retention group students belonged (82.4% accuracy; 65% improvement over chance).
Most Important Predictors The four most important predictor variables provide the following distinction between retained and departed students. Students who were not retained: • had significantly lower college GPAs; • were significantly g y more likelyy to have an unpaid p OSU bill;; • withdrew from significantly more credit hours of coursework
during the first semester; and • had significantly lower high school GPAs.
Relatively Small Contributions Other variables that made relatively small contributions to the retention prediction: • ACT score (departed students had lower scores); • number of credit hours attempted during the first semester (departed • • • • •
students attempted fewer hours); family income (departed students had lower family income); amountt off scholarships, h l hi grants, t andd tuition t iti waivers i (departed (d t d students had less assistance that did not require repayment); unmet financial need during the first semester (departed students had more unmet financial need); gender (male students were more likely to depart); and race (minority students were more likely to depart).
Note: College variable was treated uniquely in the analysis.
Further Examination of Departed Students „ The model predicted as retained 39% of those students who
departed. „ Additional analyses were performed to determine characteristics of these students. „ These students were generally very similar to retained students academically, financially, etc. (Table 6, p. 11) „ Clearly, these students departed for reasons that are not measured by this study.
Prior Research Telephone surveys conducted with students who left OSU between 2005 and 2007 identified the following factors as most important in their departure decision: „ financial problems, „academic difficulty, difficulty „personal problems, and „the location of OSU.
Questions or Comments
Oklahoma State University Retention Analysis Report Focusing on Fall 2007 Freshmen K. Celeste Campbell, Ph.D., University Registrar, Oklahoma State University January 2009
Bulleted Summary •
• •
•
•
The retention rate at Oklahoma State University for the Fall 2007 freshman cohort was the lowest in 10 years (77.1%). The highest freshman retention rate during the past 10 years was 84.6% for the 1999 cohort. Using the Fall 2007 freshman cohort, retention status was predicted moderately well (82.4% accuracy) from a set of academic, financial, and demographic variables. The most important variables that predicted retention status were the college GPA, unpaid OSU balance (bursar hold), the number of credit hours dropped during the first semester, and the high school GPA. Students who were not retained: o had significantly lower college GPAs; o were significantly more likely to have an unpaid OSU bill; o withdrew from significantly more credit hours of coursework during the first semester; and o had significantly lower high school GPAs. Other variables that made a relatively small contribution to the prediction of retention were: ACT score (departed students had lower scores); number of credit hours attempted during the first semester (departed students attempted fewer hours); family income (departed students had lower family income); amount of scholarships, grants, and tuition waivers (departed students had less assistance that did not require repayment); unmet financial need during the first semester (departed students had more unmet financial need); gender (male students were more likely to depart); and race (minority students were more likely to depart). A group of students has been identified who were predicted as retained but who in fact have left OSU. These students, who resemble retained students on most of the variables examined in this study, are of particular interest. Recommendations include interviewing a sample of these students to determine the reasons for their departure. Narrative Summary
Over the past 10 years, the freshman retention rate at Oklahoma State University has ranged from 77.1% for the 2007 cohort to 84.6% for the 1999 cohort (OSU Student Profile, 2008). Freshman retention is defined as the percent of full time freshman students enrolled during a fall semester who return for the following fall semester. The purpose of this study was to identify factors that were potential predictors of freshman retention, to evaluate the relative predictive importance of these factors, and to use them to describe how the retained students differ from those who departed. Records were examined from 3,110 freshmen who were enrolled in at least 12 credit hours at OSU during the Fall 2007 semester. After one year, during the Fall 2008 semester, 2,400 of these students were still enrolled at OSU and 710 were not enrolled. Using an initial set of 22 variables, preliminary univariate and subsequent multivariate discriminant analyses resulted in a set of 11 variables that were moderately successful in predicting whether or not students were retained (82.4% accuracy; 65% improvement over chance as measured by the proportional reduction in error). The initial set of variables represented pre college skills and abilities, college academic performance, academic and social integration, financial resources, family and community background, and personal attributes. The most important discriminating variables were the college cumulative graduation/retention GPA, the presence or absence of a bursar hold one year after 1
beginning college, the number of credit hours dropped during the first semester of college, and the high school GPA. Students who were not retained had significantly lower graduation/retention GPAs, had significantly more bursar holds, withdrew from significantly more credit hours of coursework during the first semester, and had significantly lower high school GPAs. The set of predictor variables was quite successful predicting students who were retained (88.8% accuracy). However, 39% of the students who departed were predicted to be retained. That is, 39% of the students who did not enroll at OSU for the Fall 2008 semester appeared to be successful academically and did not appear to have severe financial difficulties. Additional analysis was performed to compare these students with those who were retained and with the other students who were not retained, and the results suggest that the students in this group are generally very similar to the retained group on the academic and financial variables used in this analysis. Further research is needed to determine reasons for the departure of these students. Telephone surveys conducted with students who left OSU between 2005 and 2007 identified the following factors as most important in their departure decision: financial problems, academic difficulty, personal problems, and the location of OSU (OSU First Semester Retention, 2007; Retention, 2007). A similar survey, targeting the students who departed but were predicted to be retained, would be beneficial. Because it is already known that these students did not experience academic or severe financial difficulties, the survey should focus on a few brief, open ended questions to determine the reasons for departure and whether OSU could have done anything to prevent it.
2
Methodology and Results Introduction and Purpose of the Study Over the past 10 years, the freshman retention rate at Oklahoma State University has ranged from 77.1% for the 2007 cohort to 84.6% for the 1999 cohort (OSU Student Profile, 2008). Freshman retention is defined as the percent of full time freshman students enrolled during a fall semester who return for the following fall semester. The purpose of this study was to identify factors that were potential predictors of retention, to evaluate the relative predictive importance of these factors, and to use them to describe how the retained students differ from those who departed. Subjects The subjects for this study are the students in the Fall 2007 Oklahoma State University freshman cohort: 3,110 students who were enrolled in at least 12 credit hours as new freshmen during the Fall 2007 semester. Two students were omitted from the original sample of 3,112 because they were deceased sometime during the Spring 2008 semester. Cohort Enrollment Status The enrollment status of the Fall 2007 freshman cohort is shown in Figure 1. Spring 2008 enrollment showed a loss of 275 of the original students (8.8%). One year after initial enrollment, the Fall 2008 enrollment showed a loss of 710 of the original cohort students (22.8%). This constitutes a freshman retention rate of 77.1%, which is the lowest in 10 years (Fall 2008 OSU Student Profile). A total of 28 students withdrew from all courses before completing the Fall 2007 semester. Three of these students returned for Spring 2008, but none of the 28 were enrolled for Fall 2008. Of the students who were not enrolled for Spring 2008, 15 came back to be enrolled for Fall 2008. In Summer 2008, 435 of the cohort students (14.0%) were enrolled. Of the summer students, 433 were continuing from Spring and 2 were returning from Fall 2007. Thirty five students left at the end of summer. Figures 2 and 3 provide one year retention information by the College of initial enrollment. Figure 1. Fall 2007 Freshman Cohort Enrollment 3500 3000 2500 2000 1500
3110
2835 2400
1000 500 435 0 Fall 07
Spr 08
Sum 08
3
Fall 08
Figure 2. Cohort Enrollment Status after One Year by College of Initial Enrollment 1000 900 800 700 600 500 400 300 200 100 0
661
Retained 437
Departed
423
306 87 AG
208 AS
186
162
225
118
60
106
79
52
BU
ED
EN
GU
HES
Figure 3. Cohort Enrollment Status after One Year: Percentage by College of Initial Enrollment 100% 90%
77.90% 76.10% 78.70% 75.60% 80.00% 67.20% 81.20%
80% 70% 60% 50%
Retained
40%
Departed
30% 20% 10%
32.80% 22.10% 23.90% 21.30% 24.40% 20.00%
18.80%
0% AG
AS
BU
ED
EN
GU
HES
Definitions of Predictor Variables For this study, an initial set of 22 variables was examined to identify factors that were potential predictors of student retention status after one year—factors that would be able to predict whether students were still enrolled at OSU (members of the retained group) or were not enrolled (members of the departed group). Table 1 provides a listing of the initial predictor variables, along with their descriptions and the corresponding category from Tinto’s model of student departure (1993, chap. 4).
4
Table 1. Initial Predictor Variables Predictor
Definition/Explanation
ACT Highest Composite
Highest composite ACT score
HS Class Rank
High school class rank as a percentage, where first in class = 100%
HS Core GPA
High school core GPA
HS GPA
High school cumulative GPA
076 Hours of W
Number of credit hours with grades of W during Fall 07 (courses dropped between week 2 and week 12) Ratio of OSU earned credit hours to OSU attempted credit hours
Total OSU % Completed Hours 076 Earned Hours
Institutional GPA
Number of credit hours earned Fall 07 (excludes courses dropped, courses with grades of F or I, and remedial courses) Satisfactory academic progress for financial aid purposes assessed at the end of the Freshman year (Summer 08); based on good academic standing and successful completion of at least 75% of total cumulative hours attempted; Cleared, probation, suspended. Binary variable indicating academic standing, where good academic standing is a 1.7 grad/ret GPA with 30 earned hours or less and a 2.0 grad/ret GPA with more than 30 earned hours Cumulative graduation/retention GPA, including courses from other institutions. Zero GPAs that resulted from withdrawal from all courses were replaced with “missing.” GPA calculated from all OSU courses
076 Attempted Hours 076 College
Number of credit hours attempted Fall 07 College of initial enrollment: AG, AS, BU, ED, EN, GU, HES
Residential Life during Fall 07
Binary variable indicating whether or not student lived in on campus housing during Fall 07 Binary variable indicating whether or not student is classified as an Oklahoma resident Binary variable indicating student’s gender Binary variable indicating whether or not student is Caucasian Total dollar amount of Fall 07 scholarships, grants, and tuition waivers Fall 07 cost of education minus expected family contribution and all “free” assistance (scholarships, grants, and tuition waivers). Need to borrow assumed to be zero if student did not file FAFSA. Fall 07 cost of education minus expected family contribution and all assistance (scholarships, tuition waivers, grants, student loans, and parent loans). Unmet need assumed to be zero if student did not file FAFSA. Family adjusted gross income; available for students who filed FASFA (65% of sample); missing value for other students. Binary variable indicating whether or not student had a bursar hold November 2008. Generally a bursar hold is placed on a student’s record if there is an overdue balance of at least $100.
SAPFinAidStatus084
Good Academic Standing
Grad/Ret GPA
OK Resident Status Gender Race 076 assistance no repayment 076 need to borrow
076 unmet need
Adjusted Gross Income Current Bursar Hold
5
Category Precollege skills and abilities Precollege skills and abilities Precollege skills and abilities Precollege skills and abilities College academic performance College academic performance College academic performance College academic performance
College academic performance College academic performance College academic performance Academic integration Academic/social integration Social integration Family and community background Personal attributes Personal attributes Financial resources Financial resources
Financial resources
Financial resources Financial resources
Univariate Analysis Preliminary screening analyses (independent samples t tests) were performed to help understand significant univariate differences between the students who were retained and those who departed and to determine a more parsimonious set of viable discriminating variables. Significant group differences were found on all of the variables listed in Table 2. The variables in the table are arranged by decreasing effect size. The larger the effect size (in absolute value), the more different the two groups are on the particular variable. Table 2. Comparisons by Retention Group for Continuous Variables Arranged by Decreasing Effect Size Predictor Departed Retained Cohen’s N M SD N M SD t p d1 Large effect size Total OSU % Completed Hrs** 710 59.20 34.00 2400 91.07 13.52 24.4
