Failure, withdrawals, and retakes in intermediate microeconomics

Abstract

Using data from 11 institutions, the authors investigate enrollments in intermediate microeconomics to determine characteristics of successful and unsuccessful students and follow the retake behavior of unsuccessful students. Successful students are significantly different from unsuccessful ones, and unsuccessful students differ by type (unsuccessful completers vs. withdrawers). Aptitude is a strong predictor of success but not of retake. Having taken or concurrently taking intermediate macro reduces the likelihood of withdrawal and increases that of retake. Proficiency in calculus reduces the likelihood of unsuccessful completion but is uncorrelated with withdrawal and retake. Predictors of unsuccessful completion, withdrawal, and retake are similar in direction and significance for males and females, although magnitudes differ. URM students are more likely than white and Asian students to earn less than a C−.

Keywords:

• Class attrition
• DFW rates
• economics education
• intermediate microeconomics
• student achievement
• withdrawal

JEL CODE:

• A2

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Aptitude refers to a student’s overall academic aptitude, which we proxy for with their cumulative GPA.

2 While our sample, like that of others, is also drawn from the MIDFIELD dataset, our study uses a different MIDFIELD version. Our sample, described in greater detail below, is drawn from MIDFIELD version 9.

3 The MIDFIELD database also includes Utah State, Elizabethtown College, Florida A&M, and North Carolina A&T. After dropping observations with missing values for measures of interest, the remaining samples from Utah State and Elizabethtown College are too small to support analysis, and thus these institutions are omitted from our study. Florida A&M and North Carolina A&T are historically Black universities (HBCUs). Because our analysis focuses, in part, on differences across racial and ethnic groups, including the HBCUs would potentially have confounded our findings. Thus, observations from these institutions are also dropped from our analysis.

4 Students may complete their prerequisites in a manner that we do not observe (e.g., transferring in credit), or they may have the prerequisites waived for some reason. Although we do have student records where we do not observe students completing traditional prerequisites (e.g., principles of microeconomics, calculus, etc.) for intermediate micro theory, we elect to retain these students in our sample.

5 Our data provide an indicator of “W” for students who withdrew from their intermediate micro theory course. Note that the designation on a student’s transcript of a “W” occurs only after the student has been in the course beyond the add/drop period. Students who “drop” a course within the “add/drop” window have no indicator on their transcript of ever having been enrolled in the course. Withdrawal policies vary across institutions and time and the withdrawal window may extend through most of a semester. Thus, students in our sample who have a “W” for intermediate microeconomics have likely been in the course for at least several weeks but possibly most of the term. While the timing, circumstances, and grade at the time of withdrawal are of interest, our transcript data do not contain any of this additional information.

6 While investigation of predictors of performance on retake would also be of interest, our sample of students retaking intermediate micro theory after a failed attempt is too small for rigorous statistical analysis to yield meaningful results.

7 Existing studies also often used SAT scores to measure aptitude. As Von Allmen (1996) discusses, while both the SAT and GPA are measures of aptitude and will likely be correlated, they may also measure somewhat different things. In particular, the SAT tends to be a good predictor of performance for freshmen but less so for students later in their academic careers. Because intermediate micro theory is generally considered a junior-level course, one might expect that the SAT may have less predictive power for this course than a lower-level course. Thus, we use GPA. However, we also estimate our model using the SAT in place of GPA, and any material differences are footnoted where relevant. Further, a student’s GPA is calculated using their performance in all courses that we observe them completing prior to enrollment in intermediate microeconomics. We estimate the student’s likelihood of an unsuccessful outcome both including and excluding a separate control for their micro principles course grade. The exclusion of a student’s micro principles course grade does not materially alter our estimates. Thus, we choose to include it in the results reported here. A final note regarding the use of GPA and micro principles grades. Even though intermediate micro theory is considered a junior-level course and micro principles is usually a prerequisite, we have no record of some students in our sample having taken (1) micro principles, and/or (2) any courses prior to enrolling in intermediate micro with which to calculate a GPA. At this stage of the analysis, students missing values for one or both of these measures are omitted from our analysis.

8 We consider a student to have retaken the course if they do so in the three years following their initial attempt. Further, we restrict the sample so that students must be observed at least one semester after their initial attempt in order for them to have at least some opportunity to retake the course.

9 Recall that our calculus indicator combines students earning Cs, Ds, or Fs in calculus with those who withdraw, complete the course but are not assigned a grade, and whom we do not observe taking calculus prior to intermediate micro theory. That is, we combine those with low proficiency with those without exposure. We note that over 71.7 percent of our sample took calculus prior to enrolling in their first intermediate micro theory course. Further, the fraction of those completing a calculus course prior to intermediate micro was significantly larger for those who were unsuccessful in intermediate theory (successful completers at 71.3%, unsuccessful completers at 74.6%, and withdrawers at 73.5%). Considering only those students whom we observe taking calculus, 43.3 percent of those successfully completing intermediate micro earned an A or B in their calculus course. That value is 27.2 percent for unsuccessful intermediate micro completers and 30 percent for withdrawers (with those fractions also being significantly different).

10 Because odds ratios are less frequently used in economics; we take a moment to remind the reader of their interpretation. With odds ratios, the direction and magnitude of the effects are relative to a coefficient of 1 (as opposed to 0). Specifically, coefficients with values less than 1 indicate a reduced likelihood of the event occurring for a one-unit increase in the control variable, while coefficients greater than 1 suggest an increased likelihood of occurrence. To further illustrate this point, we provide an expanded interpretation of the coefficients for underrepresented minorities (URM) for the first two specifications in table 2. The coefficient for URM in the first specification is 1.238. This suggests that URM males are 23.8 percent more likely to complete intermediate micro unsuccessfully than white males. For females, the coefficient is 1.423. Thus, URM females are 42.3 percent more likely to complete their intermediate micro course unsuccessfully than white females.

11 In sensitivity testing, using the SAT instead of cumulative GPA to proxy for aptitude, the SAT is a significant predictor of an unsuccessful outcome in only one specification where a 100-point increase in a male student’s combined SAT score (i.e., math plus verbal) is associated with a 10 percent reduction in the likelihood of his earning less than a C−. Estimates using the SAT instead of cumulative GPA are available from the authors upon request.

12 We performed additional sensitivity analysis for our calculus measure. For the much blunter indicator measure of whether a student has taken a calculus course or not, we find that males who took calculus (with any outcome) prior to enrolling in intermediate micro were 16.8 percent less likely to complete intermediate micro unsuccessfully. In no other specification was this simple exposure measure a significant predictor. In terms of intermediate micro course outcomes, we find that students whom we do not observe taking calculus are no more or less likely to be successful than those for whom we do not have a calculus course grade, suggesting that exposure without proficiency is no different than no exposure. Collectively, these results suggest that proficiency (as measured by earning a “good grade” in calculus) is a more important predictor of intermediate micro success than simple exposure or no exposure at all.

13 We note that using the SAT to proxy for aptitude, significance of calculus proficiency and grade in micro principles as predictors of unsuccessful outcomes in intermediate micro are consistent with estimates using cumulative GPA. However, where significant, estimated decreases in the likelihood of unsuccessful outcomes are larger using the SAT to measure aptitude.

14 Since the SAT and cumulative GPA likely capture somewhat different aspects of a student’s aptitude, it is not entirely surprising that estimates using the SAT instead of cumulative GPA yield somewhat different results. The SAT is not a significant predictor of the likelihood of withdrawal as compared to unsuccessful completion.

15 Using the SAT as our proxy for aptitude, both calculus proficiency and performance in micro principles are significant predictors of withdrawal as compared to earning less than a C−. For both males and females, the likelihood of withdrawing is significantly increased for those earning an A or B in calculus and those earning higher course grades in their introductory micro course.