The Professional Counselor | Volume 11, Issue 4 389 imputation (Duprey et al., 2018; Ingels & Dalton, 2013). Further, the NCES provides analytic weighted variables and replication weights associated with those main sampling weights. The analytic weights make estimates from the sample data representative of the target population (i.e., ninth grade students in 2009–2010). These analytic weights account not only for differential selection probabilities, but also for differential patterns of response and nonresponse—in other words, nonresponse bias (Duprey et al., 2018). In addition to the analytic weight variables accounting for stratified sampling and nonresponse bias, replication weight variables address standard error concerns. Standard error calculation ensures appropriate standard errors based on the differences between the estimates of the full sample and a series of replicates (Duprey et al., 2018). These replication weights are done with the balanced repeated replication method and help account for the possibility of artificially low standard errors due to clustering in sampling (Duprey et al., 2018). Prior to running the sequential logistic regression, assumptions testing was completed. Logistic regression analyses allow the use of criterion measures on a binary outcome (Meyers et al., 2017). The result of a logistic regression is the impact of each variable on the probability of the observed event of interest (Sperandei, 2014). Sequential logistic regression allows the researcher to specify the entry order of predictor variables into the model (Tabachnick & Fidell, 2013). Model 1, the baseline model, represented person inputs and background environmental influences in SCCT. It included the following variables: FGCS status, race/ethnicity, sex, and SES. Model 2 represented self-efficacy, after controlling for person inputs and background environmental influences. Self-efficacy variables included math self-efficacy, science self-efficacy, and STEM GPA. Model 3 examined school counseling access, after controlling for the variables in the previous two models. School counseling access variables were school counselor caseload and school counselor percentage of time spent on collegereadiness counseling. Table 1 displays the model steps and variables. Table 1 Logistic Regression Model Steps Step Variables SCCT Tenets 1 First-generation student status Race/ethnicity Sex Socioeconomic status Person inputs and background environmental influences 2 Math Self-Efficacy Science Self-Efficacy STEM GPA Self-efficacy 3 School counselor caseload Percentage of time spent on college-readiness counseling Proximal environmental influences Note. SCCT = Social Cognitive Career Theory.