Analyses

Before analyses were conducted, we removed all direct identifiers (such as respondents’ names and email addresses). The data were further protected by restricting access via password-protected computers and restricted cloud-based storage folders.

>>> To explore survey findings use the navigation menu, which may appear at top left as ☰ on smaller screens. 

Quantitative Data

Quantitative measures, such as frequency of use and ratings of experience, as well as statistical differences between groups of respondents, were analyzed using the statistical software package SPSS.

Please note that the sample sizes in our analyses vary, in some cases widely, due to the fact that not all participants saw all questions (i.e., faculty and staff were not asked about their coursework). Sample sizes also vary because some people skipped questions as they participated.

>>> Go to the Survey Codebook (PDF) for all survey questions and response percentages, as well as interconnected data from other sources.

Qualitative Data

Qualitative measures with free-text responses were manually coded (i.e., categorized) by theme in Dedoose or OpenRefine, with categories further analyzed in SPSS to make the results easier to summarize and understand. Representative or compelling quotes from participants were also identified for inclusion in the findings. 

To further explore key themes and identify quotes across open-responses questions, our team used ChatGPT-4o and U-MGPT to perform thematic and sentiment analysis on each question. We then uploaded the thematic summaries and sentiment analysis results to ChatGPT-4o to identify common themes across all questions. 

Comparisons of Group Differences

During analysis, special emphasis was placed on understanding how the library is experienced by groups that have been marginalized and/or underrepresented at U-M in the past. Addressing potential places where people feel underserved, mistreated, or confused was an important goal of the survey and is in alignment with the library’s values. As such, analyses of relevant study variables have been conducted to look for differences influenced by participant gender, race, first-generation status (for students), international status, and disability status. 

We are reporting only significant findings where differences across groups are significant; if no significant findings are noted, it is safe to assume there weren’t any.