Dr. Beeper, an Educational Psychologists who studies issues related to higher education, is interested in studying key factors that impact year to year persistence among college students. His review of the literature identifies several factors that appear to be causally related to persistence. Specifically, academic aptitude, goal commitment, institutional commitment, and the number of work hours.
To test the importance of these factors, Dr. Beeper administers a set of questionnaires to 100 randomly selected first-time, full-time freshmen college students (50 male and 50 female) that attended the Freshmen Orientation in the Fall of 2016, at Newton Young University (NYU) in Nebraska.
Measures:
Institutional Commitment (IC) represents the importance that students place on graduating from the college they are currently attending. Institutional Commitment was measured with five-item questionnaire. Each item was rated on a 0, 1, or 2 scale. The possible range of scale scores are zero to 10, where values close to zero indicate little to no importance, and values close to 10 indicate high importance.
Goal Commitment (GC) represents the importance that students place on obtaining a college degree. Goal Commitment was also measured with five-item questionnaire. Each item was rated on a 0, 1, or 2 scale. The possible range of scale scores are zero to 10; where values close to zero indicated little to no importance to obtaining a college degree, and values close to 10 indicated a high importance to graduating from college.
Academic Aptitude was represented as scores on both the SAT-Math and the SAT-Verbal tests. SAT scores for all participants were obtained from high school transcripts.
Hours works, represents the anticipated number of hours the student expected to work throughout the semester.
Finally, Year-to-year persistence was determined by examining the enrollment records for the sample of 100 students. A student that was registered for registered for the Fall 2017 classes was classified as a “Persister”, and given a code of 1, a student that did not re-enroll for classes at NYU, or any other college/university (based on follow-up phone interviews) was considered a “Non-persister”, and was given a code of 0. Therefore, the SPSS variable Persist has two levels, 0 and 1.
The assignment is, using the attached SPSS data file, conduct a binary logistical regression analysis in which IC, GC, SAT-Math, SAT-Verbal, and Hours Worked are the predictor. variables (covariates in SPSS), and the variable Persist is the outcome (DV in SPSS). Use my sample summary as a model for your summary.
The specific elements of the assignment are:
1) Create a Null and Alternative Hypotheses for the Logistical Regression Analysis
2) State the Goals of the analysis
3) Summarize the results and interpret findings the overall model (for example the Chi Square results, Nagelkerke R-Square or Cox Snell R-Square).
4) Summarize and interpret the results for each predictor; and present, summarize and interpret the results for each significant predictor (i.e., B, Wald’s test, df, p and OR (ExpB). Interpret the significant OR using the effect size conventions I posted in last week’s (8) discussion board.
5) Include and refer to the appropriate tables within the summary.
Please read my sample summary see what statistics to report, and how to report and interpret them in correct APA style, as well as the tables to include.
You’ll see that in my sample summary I also include t-tests. You may want to conduct t-tests that compare “persisters” and non-persisters, on the predictor variables (covariates). Please note that the t-test are optional, and will have no impact on your grade whether you include them or not. The t-test are very informative about the bivariate relationship between the predictor variables (covariates in SPSS) and the binomial outcome (DV in SPSS) .
Please note that you are not required to conduct the t-tests, or to compute and report Cohen’s d.
Here’s the syntax for my sample summary.
T-TEST GROUPS=BO(0 1)
/MISSING=ANALYSIS
/VARIABLES=teachsat ressat wkoverld
/CRITERIA=CI(.95).
LOGISTIC REGRESSION VARIABLES BO
/METHOD=ENTER teachsat ressat wkoverld
/PRINT=GOODFIT CI(95)
/CRITERIA=PIN(0.05) POUT(0.10) ITERATE(20) CUT(0.5).
1
The Effects of Job Satisfaction and Work Overload on Burnout Among College
Professors: A Sample APA Summary for Logistical Regression Analysis
Anthony Napoli
State University of New York
2
Overview of study
The present study was designed to examine the effects of teaching and research
satisfaction and work overload on “burnout” among college professors. The sample consisted of
80 fulltime college professors randomly selected from four large public universities in New York
State. Male (n = 40) and female (n = 40) faculty were equally represented in the sample. The
mean age of the participants equaled 51.2 years (SD = 10.5). The mean number of years
teaching equaled 13.5 (SD = 4.2), with a range of 7 to 20 years. All participants completed a
questionnaire that included standard demographic items and measures of teaching satisfaction,
research satisfaction, and work overload. Burnout classifications were dichotomized into two
groups based on instructors’ reports that they were leaving academia due to a dislike of the
teaching and research responsibilities (burnt-out)1 , or that they would be remain in their present
teaching position for the next academic year (non-burnout). Among the 80 respondents (n= 20)
25% reported that were leaving their position due to burnout. Preliminary analyses indicated that
respondents’ age, gender, and years of teaching were not statistically related to burnout.
Results
Mean comparisons on the three predictor variables using t-tests indicated that “burnt-out”
participants (M = 52.3, SD = 32.4, N = 20) experienced significantly lower levels of teaching
satisfaction (t(78) = 3.67, p < .001, d = .95) than “non-burnt-out” respondents (M = 106.8, SD = 63.5). Burnt-out participants (M = 50.1 SD = 27.5), also experienced significantly lower research satisfaction (t(78) = 3.27 p < .001, d = .84) than non-burnt-out participants (M = 95.6, 1 In the present study, burnout is operationalized as leaving teaching/research due to the long-term exhaustion, depersonalization, cynicism, and diminished interest in teaching. http://en.wikipedia.org/wiki/Depersonalization http://en.wikipedia.org/wiki/Depersonalization 3 SD = 60). Burnt-out participants (M = 139.9, SD = 27.8) reported significantly higher levels of work overload stress (t(78) = 4.18,. p < .001, d = 1.08), in comparison to their non-burnt-out peers (M = 88.6, SD = 52.2) To examine the direct effect of each of the predictor variables on burnout, a logistical regression analysis was conducted in which teaching satisfaction, research satisfaction, and work overload were the predictor variables, and burnout group, coded 1 for burnout and 0 for non- burnout, was the outcome measure. The results for the full model were statistically significant, χ2(3, N = 80) = 33.2, p < .001, indicating that the set of predictor variables reliably distinguished burnt-out and non-burnt-out faculty. The variance in burnout accounted for by the model was moderately large and (Nagelkerke R2) = .503. Classifications were somewhat impressive but asymmetrical, with 90% of non-burnt-out and 50% of the burnt-out faculty correctly predicted, for an overall success rate of 80%. Appearing in Table 2 are the regression coefficients (β), Wald statistics (W), odds ratios (OR), and 95% confidence intervals for odds ratios for each of the predictors. The results indicated that teaching satisfaction (β = -.018, W = 6.70, p < .01, OR = .98), and research satisfaction (β = -.020, W = 5.46, p < .05, OR = .98) were both significantly and negatively related to burnout - Indicating that as levels of teaching and research satisfaction increased the risk for burnout decreased. Work overload was significantly and positively related to burn-out (β = .023, W = 7.13, p < .01, OR = .1.02) – as levels of work overload increased the risk for burn- out increased. Although all three predictors of burnout were statistically significant the effect size (ORs) for each as small.2 2 The effect size conventions for Odds Ratios are: ORs of 1.44 = Small Effect, 2.47 = Medium Effect, and 4.25 = Large Effect 4 Summary of findings In the present study, a sample of (n = 20) college professors who reported that they were leaving academia due to the long-term exhaustion , depersonalization, cynicism, and diminished interest in teaching (burn-out), reported significantly lower levels of teaching and research satisfactions, and higher levels of work overload stress than a sample (n = 60) of peers who reported that they will remain in academia. Results from a logistical regression analysis indicated that increased levels of teaching and research satisfaction are related to decrease likelihoods of burnout; whereas higher levels of work overload increase the risk for burn-out. Table 1 Summary statistics by group and t-test results Non-Burnt Out Burnt Out t-test Measure M SD N M SD N t df d Teaching satisfaction 106.8 63.5 60 52.3 32.4 20 3.67** 78 .95 Research satisfaction 95.6 60 60 50.1 27.5 20 3.27** 78 .84 Work load 88.6 52.2 60 139.9 27.8 20 4.18** 78 1.08 **p < .01 http://en.wikipedia.org/wiki/Depersonalization 5 Table 2 Logistical Regression Results3 Measure B S.E. Wald df p OR 95% CI for OR Teaching Satisfaction -.18 .007 6.700 1 .01 .982 .969 .996 Research Satisfaction -.02 .009 5.460 1 .029 .980 .963 .998 Work Overload .23 .009 7.130 1 .008 1.023 1.005 1.041 Constant -1.15 1.256 0.839 1 .36 .317 3 Note that the logistical regression (LR) equation for the probability of burnout, P(Y), equals: )( 3322111 1 )( XbXbXbboe YP +++− + = Where e is the base of the natural system of logarithms, or approximately 2.718282. b0 is the Y intercept (constant from spss LR equation). b0 = -1.15 b1 is the regression coeffictent for teaching satisfaction (B value from spss LR equation). b1 = -.018 b2 is the regression coeffictent for research satisfaction (B value from spss LR equation). b2 = -.020 b3 is the regression coeffictent for work overload stress (B value from spss LR equation). b3 = .023 X1 = a score on the teaching satisfaction measure X2 = a score on the research satisfaction measure X3 = a score on the work overload stress measure