Homework 06

Logistic Regression

For Homework 06, you will be using the HELP dataset, learn more at:

• https://melindahiggins2000.github.io/N736Fall2017_HELPdataset/ &
• https://github.com/melindahiggins2000/N736Fall2017_HELPdataset

Refer to the logistic regression analysis example and codes we ran during lesson 18 and 19 - see https://github.com/melindahiggins2000/N736Fall2017_lesson1819

For the HELP dataset:

• OUTCOME VARIABLE: consider the variable g1b “Experienced serious thoughts of suicide (last 30 days) - Baseline”
• PREDICTOR VARIABLE: consider these variables as potential predictors for g1b:
  • age, female, pss_fr, homeless, pcs, mcs, cesd, indtot

Complete the following:

1. Consider the continuous variable cesd as a predictor for g1b
   1. run a logistic regression of the probability of suicidal thoughts (g1b) given their depressive symptoms scores (cesd)
   2. make a plot of the the predicted probability of suicidal thoughts (g1b) by the depressive symptoms scores (cesd)
   3. what value of the cesd leads to a probability of suicidal thoughts => 0.5? (hint: use the plot you just made)
2. Using variable selection methods, develop a logistic regression model for the probability of suicidal thoughts (g1b) considering all of these variables for possible inclusion: age, female, pss_fr, homeless, pcs, mcs, cesd, indtot
   1. present the final model results
   2. write a few sentences describing your results including:
      1. model fit
      2. model classification table results - remember to report the threshold used for the classification table - you can change it from 0.5 if you think a different threshold might work better
      3. odds ratios for each significant predictor in the model