Analysis of ICU Data

  1. Run a logistic regression using all 19 predictor variables -- thru (and the intercept). Determine the coefficient with the largest P-value.
  2. Remove the predictor variable determined in II.1 and rerun the logistic regression with the remaining predictors. Determine the coefficient with the largest P-value.
  3. Continue removing predictor variables one at a time in this manner until no predictor has a P-value greater than . List the remaining predictor variables and the output from the logistic regression.

    NOTE: This process is called backward deletion. You can continue the model selection process by trying to add terms back ( stepwise addition). Once you find the significant variables, you can add their quadraticis and interactions to improve the model. For the purpose of this lab we will stop here.

  4. Without further analsis (or biological background) does the model seem reasonable ? Which terms did you expected to see ?