Measuring Institutional Effects on Student Activism John Zilvinskis (bio), Demetri L. Morgan (bio), and Brendan Dugan (bio) To best accentuate individual-level outcomes, some quantitative research on political engagement in higher education includes institutional variables only as controls (i.e., Dodson, 2014; Weerts, Cabrera, & Mejías, 2014 ). This leads to a complementary strand of research that seeks to simultaneously understand both individual and institutional factors that contribute to student political engagement (Baker & Blissett, 2018; Doyle & Skinner, 2017). Often researchers use some configuration of Berger and Milem's (2000) structural features of institutions, since these measures help differentiate institutions and may influence student outcomes. Consequently, Niehaus, Campbell, and Inkelas (2014) encouraged researchers to be transparent about design decisions. In particular, a growing chorus of scholars has called for research to account for the role of the institutional context in the emergence of civic and political outcomes such as activism (Cabrera, Matias, & Montoya, 2017; Kezar, 2010). Since students are nested within groups (e.g., classes, programs, or institutions), researchers often invoke multilevel modeling in education scholarship to simultaneously account for individual-level and group-level variation (Ethington, 1997). This method is vulnerable in regard to how theory and statistics complement each other in research design (Smart, 2005). This lack of specificity hinders institutions' ability to respond to dynamic and complex outcomes such as student activism (Morgan & Davis, 2019). Following insights gleaned from a previous study we conducted, and to encourage greater transparency in the quantitative research design of higher education studies, we tested which institution-level variables found within the activism literature can be retained when modeling student activism behavior at multiple levels (Morgan, Zilvinskis, & Dugan, 2019). This allowed us to statistically examine the value that distinct variables provide, while understanding how the significant institutionlevel variables are related to student-level outcomes. Our research questions included: Which institution-level variables found within activism literature can be retained when modeling student activism behavior at multiple levels? How do the retained institution-level variables relate to individual activism behaviors? INSTITUTIONAL DEMOGRAPHICS An institution's racial composition is an important point of reference for a range of educational outcomes (e.g., sense of belonging, graduation, retention; Museus, Nichols, & Lambert, 2008). We also followed Baker and Blissett's (2018) finding that the Pell percentage was a significant predictor of institutional involvement in a type of social media activism [End Page 372] (e.g., the "I, too, am [College]" campaigns). Research from other disciplines corroborates the connection between socioeconomic status and political activities (Pacheco & Plutzer, 2008). FACULTY INTERACTION AND RESOURCE ALLOCATION The impact on educational outcomes of faculty composition (i.e., ratio of nontenure track to full-time faculty) in higher education is still not well understood (Kezar, 2012 ); yet other research very clearly delineates the importance of student–faculty interactions ( Garvey, BrckaLorenz, Latopolski, & Hurtado, 2018). Hence, we include two variables to better understand whether student–faculty ratios and different compositions of faculty connect to the dependent variables. Finally, research on how resource allocations influence student experience suggests it is critical to consider how institutions use resources to support endeavors (e.g., Cantwell, 2016; Gansemer-Topf, Saunders, Schuh, & Shelly, 2004). METHOD Data Sources The 2017 administration of the National Survey of Student Engagement (NSSE; http://nsse.indiana.edu) provided data from 27 institutions for this study. After the divisive 2016 presidential election, more than 7,000 students responded to a supplemental survey on political activism. Of the 4,168 senior respondents, 3,257 (78.1%) who began the survey completed the additional items of interest. Approximately two thirds of the students identified as women (67.5%), and a similar proportion identified as White (70.2%); first-generation students (neither parent has earned a college degree) constituted 41.0% of respondents. Student demographic covariates were based on survey responses, and institutional characteristics were derived from the Integrated Postsecondary Education Data System (IPEDS). Indicators were developed for students who identified as Black or African American; identified as Lesbian, Gay, Bisexual, or Queer; and those who chose not to share their sexual orientation. Institutional support for activism is the averaged value of responses from students asked about the degree to which their institution emphasized four types of activism (codebook variables...