Video 10 - SAS Studio for Logistic Regression + ROC Curve & Hosmer-Lemeshow Goodness-of-Fit Analysis

SAS Studio permits the simple development of logistic regression models (and probit models too). Here, the emphasis is the construction of a multivariable logistic regression model for predicting E-cigarette Use in the previous 30 days. The focus is for public health statistical output with an emphasis on obtaining odds ratios (or adjusted odds ratios since the model is multivariable). Logistic regression is not able to be fully taught in 16 minutes, and it's use, especially for consequential research and/or report purposes should be guided by a statistician or a person with experience using logistic regression (as it can be taught as one semester of graduate education or through intensive workshops). For any logistic regression, it is important, upon proper model construction, to assess model fit and model discrimination. Here, model fit is assessed using the Hosmer-Lemeshow Goodness of Fit Test, and the model's discrimination is assessed using the area under the ROC curve. SAS Studio provides these post-test model assessment tools through a series of simple options via the drop-downs.

• #education
• #logistic_regression
• #multivariable_logistic_regression
• #adjusted_odds_ratios
• #odds_ratio
• #sas_studio
• #sas
• #lroc
• #roc
• #roc_curve
• #area_under_roc_curve
• #goodness_of_fit
• #fit
• #hosmer_lemeshow
• #statistics
• #public_health
• #biostatistics