Student-faculty similarity and graduation outcomes for African American students

In our previous blog post, we introduced the concept of using location quotient (LQ) scores, which serve as a measure of demographic similarity of faculty to students, as a predictor of six-year graduation rates within race/ethnicity groups. In general, we saw that a higher representation of Black or African American faculty when compared to student representation on campus is positively correlated with higher graduation rates for Black or African American students. In this post, we further explore data for Black or African American students/faculty to see whether the same patterns found in the overall data exist within institutional types as defined by Carnegie Classification groups. There is also a discussion of other factors which contribute to successful student outcomes in addition to the student-faculty similarity index.

Does institutional type influence the link between demographic similarity and graduation rates?

By way of review, the LQ scores on the x-axis in the two visualizations below show how closely the representation of Black or African American faculty at a campus mirrors that of the student body. An LQ score of 1.0 represents a perfect match, with LQ scores below 1.0 showing underrepresentation of faculty, while scores above 1.0 show overrepresentation of faculty. Almost 90-percent of the public universities in the IPEDS dataset have LQ scores below 1.0. In the two visualizations below, the size of each mark is based on the total undergraduate enrollment in fall 2019. The color-coding scheme is associated with the institutional levels in the Carnegie Classification system. The vertical line at the 1.0 mark on the x-axis shows where there is a perfect match between demographic representation of students and faculty. The lines running left-to-right across the scatterplot shows the relationship between LQ Score and Graduation Rate for African American students. Texas universities are labeled for easier identification.

  • As with the general findings in the previous post, all three lines for the Doctoral Universities show a positive association between the similarity of faculty representation and 6-year graduation rates of African American students. However, we do see that there are differences in the intercepts (where the line crosses 0.0 on the x-axis) and the slopes of the lines. 
  • The darker blue line at the top is for Doctoral: Very High Research Universities. In the original data, the average 6-year graduation rate for African American students who begin as first-time, full-time, degree-seeking students at Very High Research institutions was 63%. The lighter blue line in the middle is for Doctoral: High Research universities, which has an average graduation rate for African American students of 47%. The orange line is for Doctoral/Professional universities, which collectively had an average graduation rate of 41% for African American students.
  • The slopes of the lines for Very High and High Research institutions are similar, indicating that the “improvement” in LQ scores up-to and exceeding 1.0 has a similar trend for these two groups of institutions. Although starting at a similar point as High Research institutions, the Doctoral/Professional universities (orange line) have a much flatter trendline, showing that higher faculty representation has a lower correlation on graduation rates for African American students at these institutions.
  • At public four-year Master’s universities (seen by clicking on “LQ Score & Grad Rate (Masters)” in the visualization below), the positive association between LQ score and graduation rates continue for African American students. The Masters: Larger (grey line) and Masters: Medium (blue line) classification groups have average graduation rates in the original data of 43% and 39%, respectively. 
  • The slopes for the two lines for the Master’s institutions are different. This indicates that the correlation of faculty similarity to students varies based on institutional type within the Master’s level of institutions within the Carnegie Classification system, although it is positive for both groups.

So What?

The use of LQ scores provides a succinct way to compare demographic similarity between faculty and students across institutional types and contexts. However, as with any univariate analysis, there are limitations, especially when viewing an outcome as complex as graduating from college. In order to better understand the influence of demographic similarity in the context of other variables, we conducted a series of regression analyses that allowed us to estimate the effects of input variables on graduation rates for African American students, including institutional selectivity, level of preparation, underrepresented minority representation, percentage of Pell recipients, institutional resources for instruction, along with LQ Score. 

When taking all of these variables into account, including the hierarchical structure of universities being nested within Carnegie Classification groups, demographic similarity has a positive influence on graduation rates for Black or African American students, but the effect was not at a level of significance. Other variables, such as institutional selectivity, student preparedness, and Pell status had greater effects on graduation outcomes for African American students at public universities in the US.

The regression analysis also serves as a way to measure where institutions are “predicted” to be in terms of 6-year graduation rates based on the combination of inputs in the statistical model. For the public universities with graduation rate data for African American students in the IPEDS dataset, the visualization below shows one example of how we could compare the “predicted” (gray box) graduation rate with the actual or “observed” (blue box) graduation rate at each institution. 

  • Where “OBS” (the blue box) is greater than “PRED” (the gray box), we would say that these institutions are out-performing their expected performance based on the statistical model predicted graduation rates.
  • Conversely, when “PRED” is greater than “OBS”, we would say these institutions are under-performing their expected graduation rates for Black or African American students based on statistical model predictions.
  • Overall, approximately one-half of the Doctoral and Masters institutions in the dataset performed “Better” than their predicted graduation rate for African American students, while one-half performed “Worse” than their predicted graduation rate.
  • Using the filters to the right, you can further explore the predicted versus observed data for 6-year graduation rates for African American students at Doctoral and Masters institutions.