BBS2007: “Linear and Logistic regression and Repeated Measures Analysis”
Lecture 2
Logistic regression
analysis
Bjorn Winkens
Methodology and Statistics
Maastricht University
Methodology and Statistics | University of Maastricht © Bjorn Winkens 2021
, Content
• Simple logistic regression analysis
– Risk, odds, relative risk (RR), odds ratio (OR)
– Wald test + relation with chi-square test
• Multiple logistic regression analysis I
– Correction for other variables:
• dummy variables
• Interpretation and 95%CI
• Multiple logistic regression analysis II
– Effect-modification:
• interpretation and tests (Wald, Likelihood ratio)
• top-down procedure
– Assumptions
– Prediction (cohort study) 2
, Applications (1)
Study goal:
• Compare two probabilities with correction for other
factors
– E.g. Effect smoking during pregnancy (yes/no) on small-for-
gestational age baby (SGA: yes/no),
with correction for alcohol use, drug use, age of mother
1. smoking (yes/no),
2. alcohol use (yes/no), SGA
3. drug use (yes/no), (yes/no)
4. age of mother (years)
3
, Applications (2)
Study goal:
• Determine which prognostic/risk factors are
important/not important
– E.g. which variables are independently related to small-
for-gestational age baby (SGA: yes/no)?
1. smoking (yes/no),
2. alcohol use (yes/no), SGA
3. drug use (yes/no), (yes/no)
4. age of mother (years)
4
Lecture 2
Logistic regression
analysis
Bjorn Winkens
Methodology and Statistics
Maastricht University
Methodology and Statistics | University of Maastricht © Bjorn Winkens 2021
, Content
• Simple logistic regression analysis
– Risk, odds, relative risk (RR), odds ratio (OR)
– Wald test + relation with chi-square test
• Multiple logistic regression analysis I
– Correction for other variables:
• dummy variables
• Interpretation and 95%CI
• Multiple logistic regression analysis II
– Effect-modification:
• interpretation and tests (Wald, Likelihood ratio)
• top-down procedure
– Assumptions
– Prediction (cohort study) 2
, Applications (1)
Study goal:
• Compare two probabilities with correction for other
factors
– E.g. Effect smoking during pregnancy (yes/no) on small-for-
gestational age baby (SGA: yes/no),
with correction for alcohol use, drug use, age of mother
1. smoking (yes/no),
2. alcohol use (yes/no), SGA
3. drug use (yes/no), (yes/no)
4. age of mother (years)
3
, Applications (2)
Study goal:
• Determine which prognostic/risk factors are
important/not important
– E.g. which variables are independently related to small-
for-gestational age baby (SGA: yes/no)?
1. smoking (yes/no),
2. alcohol use (yes/no), SGA
3. drug use (yes/no), (yes/no)
4. age of mother (years)
4
