1/16
Looks like no tags are added yet.
Name | Mastery | Learn | Test | Matching | Spaced |
|---|
No study sessions yet.
What tests are good to see if independent observations are correlated?
Relative Risks, Chi-square test, logistic regression
What tests are good to see if correlated observations are correlated?
McNemar’s chi-square test, conditional logistic regression, GEE modeling
What are the alternatives to the chi-square test if there are sparse cells?
Fisher’s exact test, McNemar’s exact test
Chi-square test
compares proportion between groups with independent observations
McNemar’s Chi-square
for correlated data
What chi-square tests are for independent data?
Mantel Haenzel, Pearsons
Logistic regression
multivariate technique, independent data
conditional logistic regression
multivariate technique, correlated data (matched)
generalized estimating equations (GEE)
for repeated measures, correlated data
Simple Linear Regression
assessing association between two continuous variables
uses Least Square Method
gives regression coefficients
Multiple Linear Regression
gives linear association between a continuous outcome after the adjustment of covariates
predictor variables can be categorical or continuous
uses Maximum Likelihood estimate
Logistic Regression
looks at assoication between exposure and birnar outcome after adjusting for covariates
based on max likihood method
used in case control and cross-sectional studies
can be used in cohort studies but assumes complete follow-up
log(odds); OR= e^b1
Cox Proportional Hazards Model
follows proportional hazards assumption
uses time-to-event data (person-time)
used in cohort studies
log(h(t/x')/(ho(t)
use for survival analysis
gives hazard ratio (HR)
Possion Regression
use for person-time
gives RR (log(Rate))
matched case-control
use conditional logistic regression (analogous to logistic model)
nested case-control studies
use conditional logistic regression
matching var is length of followup
analogous to cox regression
case-cohort study
use cox regression
handles possibility that control becomes case