The Effects of the Freshman Triad System on Course Persistence and Grades

A Statistical Analysis

Dr. Blair Sterba-Boatwright

Texas A&M University-Corpus Christi

January 18, 1999

Abstract: Grades from eight courses for three semesters were analyzed to see if there was a significant difference between the results for freshmen who participated in learning communities at Texas A&M University-Corpus Christi (locally called "Triads") and those who didn't. The results showed that freshmen not participating in a triad were almost twice as likely to withdraw from a course. After removing the withdrawals from the dataset, the average grade of freshmen triad participants was usually 0.2 to 0.3 higher (on a 4-point scale) than the grade of non participating freshmen. Both these results are significant statistically (p < 0.05). Some results indicate a stronger positive effect among women and hispanics in the freshmen class.

1. Introduction

Since the first admission of freshmen in the Fall 1994 semester, Texas A&M University Corpus Christi has required full-time freshmen to enroll in learning communities called triads each of their first two regular semesters. A triad consists of a block of three disciplinary courses with required concurrent registration. A fourth required course, Freshman Seminar, is used to teach college survival skills as well as to draw together the material learned in the block of three courses. Of the three courses in a triad, two are in a large lecture format (150-250 students) and the third is an English Composition class into which a student is placed by English faculty. Enrollment in the English Composition course is limited to 25 students, and corresponds precisely to the enrollment in a section of Freshman Seminar. 

Previous studies at TAMU-CC have demonstrated that Freshman Seminars are effective in creating several important precursors for freshman retention, including creating nascent student communities, teaching study skills, and bringing freshman into close contact with faculty. However, no study before this one has been performed to determine what the effect of the Triad system is on student academic outcomes.

2. Methodology

Grades were collected from the Fall 1997, Spring 1998, and Fall 1998 semesters for the following eight courses: HIST 1370 and 1371, POLS 2305 and 2306, ESCI 1470, MUSI 1306, PSYC 2301 and SOCI 1301. These are the eight possible large lecture courses a freshman may enroll in in a triad. These semesters were chosen because a revision in format and method in Freshman Seminar was instituted in Fall 1997, and so these semesters represent the current "start of the art". We collected grades from all sections of these courses, not just those participating in the triads. Our goal was to compare the performance of freshmen in the triad sections of these courses with their performance in non-triad sections.

These eight courses were separated into two groups of four each: the HIST and POLS courses in one group, the other four in the other. In the HIST and POLS courses, sections which were part of a triad were closed to students not participating in that triad. Thus, in this first group, freshmen could be divided into "TR", students who enrolled in these courses as part of a triad students and "NT", students who took a section which was not part of a triad. In the other four courses, it was possible for students to enroll in the triad sections without enrollment in the rest of the courses of that triad. Thus, in addition to "TR" and "NT", there is a third group, dubbed "LL", which was students who were enrolled in the triad section of a course, but not the rest of the triad.

In addition to the main effect variable, triad participation, we also collected two other variables about the students in the study: sex (M or F) and ethnic background (Anglo or Hispanic; other ethnic groups were too small to study independently, and were excluded from any consideration of ethnic effect).

3. Results and Discussion

Our results for the HIST and POLS groups are presented in Table 1:

Table 1

Results from HIST and POLS courses (n = 2627 grades analyzed)

In Table 1, the rows labelled "W's t-test" are the results of a t-test with the following hypotheses:

H₀: There is no difference between the proportion of withdrawals in the two groups.

H₁: There is a difference between the proportion of withdrawals in the two groups.

The corresponding numbers in the TR columns are the percentages of W's among the two groups, followed by the p-values for each test. As can be seen, both overall and in the different sex and ethnic subsamples, NT students are about twice as likely to withdraw from a class as TR students.

The rows labelled "Grds ANOVA" are the results of an analysis of variance done on mean grades in the two groups. After "W" grades were excluded, the remaining grades were assigned numbers on the usual four-point scale ("A" = 4, "B" = 3, etc.) and means computed. The hypotheses were:

H₀: There is no difference between mean grades in the two groups.

H₁: There is a difference between mean grades in the two groups.

The corresponding numbers in the TR columns are mean grades for the two groups, followed by the p-values for each test. As can be seen, both overall and in the different sex and ethnic subsamples, TR students perform about 0.2 GPA points better than NT students, with the exception of Hispanics, where the difference is about 0.3 points.

In each case above, there was no significant cross effect between section type (TR/NT) and either sex or ethnic background. Course by course analysis showed consistent trends, although reduced sample sizes also reduced statistical significance. In general, the effects were weaker in HIST 1370 and POLS 2305, stronger in HIST 1371 and POLS 2306.

The results from the ESCI, MUSI, PSYC and SOCI courses are presented in Table 2:

Table 2

Results from ESCI, MUSI, PSYC, and SOCI courses (n = 1808 grades analyzed)

The format of Table 2 is the same as that of Table 1. Note that there were fewer significant results in this table, but in all cases, the "TR" group outperformed the "NT" group. A significantly lower withdrawal rate prevailed among women and hispanics, but not men and anglos. In the "LL" group, note both a higher withdrawal rate and a higher GPA. There was a significant cross effect between ethnic background and section type because of the LL group.

In keeping with the less significant differences found in these courses, course by course analysis showed inconsistent trends. In general, the effects were stronger in SOCI 1301, insignificant in ESCI 1470 and PSYC 2301, and actually reversed in MUSI 1306. In all cases however, the size of the LL and NT groups were so small that the results are of questionable validity, particularly in the sex and ethnic subsamples.

4. Conclusions

In most cases studied, freshman who took a large lecture course as part of a triad were more likely to complete the course, and those completing the course received a higher average grade, than did those freshmen who took the same course without triad participation. The positive effects of triad participation were slightly stronger among women and hispanics. Further study is needed to determine which pedagogical components of the triads are responsible for these benefits.