Data Analysis
Introduction
Wayne State University Diversity Campus Climate Study Data Analysis Approach
The Wayne State University Diversity Campus Climate Study group (DCCS) is the group that planned and implemented the climate survey through a partnership with the Survey Research Lab (SRL) at the University of Illinois, Chicago. The SRL's report of the Faculty survey follows this introduction. The DCCS's in-depth review of the reports provided by SRL motivated the DCCS to carry out independent analyses of the survey data. This document explains the motivation for those independent analyses, explains how the DCCS approach differs from the SRL approach, and alerts readers to important considerations in interpreting the SRL reports.
Wayne State University Diversity Campus Climate Study
- Faculty Report with narrative (PDF)
- Staff report with narrative (PDF)
- Student report with narrative (PDF)
Important information on data analysis approach
The approach taken by SRL was to identify a core outcome variable for campus climate and to utilize other data from the survey to identify significant predictors of responses on the core outcome variable. The SRL constructed a "climate" variable based on empirical analyses of the survey data (see Computing the Outcome Variable: Overall Climate on page 2 of the report). The "climate" variable was composed of three items (see Table 1):
Table 1. Items contributing to the SRL "climate" variable
| Faculty | Staff | Students |
|---|---|---|
| Climate at WSU | Climate at WSU | Climate at WSU |
| Climate in the primary department or unit | Climate in the primary department or unit | Climate in the primary department or unit |
| Job satisfaction | Job satisfaction | Happiness as a student at WSU |
Conceptually, however, job satisfaction and happiness are notably different types of indicators than the other two climate variables and should be treated separately. Having them together as part of the SRL "climate" variable therefore makes it difficult to interpret results that identify important predictors of climate. Thus, we cannot assume that the items SRL identified as key predictors of their "climate" variable would be the same key predictors for a "climate" variable that separated out job satisfaction and happiness. This difference in strictly statistical versus conceptual definitions of the "climate" variable motivated the DCCS to carry out independent analyses of the survey data using only (1) climate at WSU and (2) climate in the primary department or unit as key components of overall climate.
Further, SRL did not examine the composite climate variable and the components in terms of identity group membership. Thus, the SRL report does not discuss how members of different identity groups responded on these predictive indicators. They do provide demographic analyses for each item in the survey, which are in the appendices.
Preliminary analysis of climate
While climate at WSU and climate in the department/unit are likely to be related, they also reflected different aspects of climate. Thus, we also examined these two climate variables and job satisfaction separately. Further, from discussions with various groups we identified other variables ("barometers of connection") that are important outcome indicators, specifically belongingness, happiness, respect and intent to leave. We also highlight specific experiences of hostility and unfairness as well as self-expression. We examined each indicator in terms of role (students, faculty, staff) as well as by three key identity group membership gender identity; race/ethnicity; and disability. These have formed the basis for the preliminary findings that have been shared with the campus (https://climatestudy.wayne.edu/report).
Further information for data interpretation of this report
There are two other features of this report that we want to highlight:
- All analyses in this report are based on weighted samples. Tables 3a and 3b show the demographic profile for respondents in each sample. To some extent, these sample profiles differ from the population of students, faculty and staff as identified from the core demographic data the institution has. A statistical approach utilized when the desire is to generalize the results of a sample to a population is to weight the sample to match the demographic distribution of the population (see Sample Weights, pg 6 of the report for further explanation). Another approach is to use the samples as they exist (unweighted). Because the SRL utilized a weighted sample, for consistency and to prevent confusion we utilized weighted samples in our analysis as well.
- The items having to do with hostile behaviors were originally scored with 1 indicating never and 5 indicating very often. SRL chose to reverse code these so that 5 indicates never and 1 indicates very often (see Interpretation of the Means, pg. 4-5). This is confusing as higher numbers are typically associated with "more of" something. Thus, in looking at the data provided in this report, means of 4 and 5 represent very infrequent occurrences. In our own analyses and presentations, we have maintained the original structure of the data so that higher numbers mean more frequently occurring behavior.