Is unadjusted odds ratio same as bivariate regression analysis for categorical variables ?
If your dependent variable is 0-1, logistic regressin can be told to produce two forms of results for the indpendent variable: a logistic regression coefficient (which is hard to interpret) and an unadjusted odds ratio (which is slightly less har to interpret). A bivariat regression for such a dep var is usually the use of OLS regression :-)
The bivariate regression analysis for categorical variables would be performed by logistic regression. First one has to define dependent and independent variable in bivariate logistic regression. It would provide the Odds ratio. As I have mentioned that it was bivariate so it is unadjusted odds ratio. If other independent variables are also present then one can get the adjusted odds ratio through logistic regression.
For Bivariate analysis, unadjusted odds ratio can also be calculated through cross tabulation of categorical data. Chi square and Fischer exact test can be applied. In this case the unadjusted odds ratio calculations and results would be different.
Dependent and independent variables are very important to define to get similar results in regression models.
Yes, unadjusted odds ratio same as bivariate regression analysis for categorical variables , only when you include one categorical independent variable in the bivariate logistic regression model. If more than one independent variable are included in the bivariate logistic regression, you will get the adjusted odds ratio for that specific variable. For example if you are studying association between CHD (Yes/no) and smoking (Yes/no), you will get unadjusted odds ratio. In this model if you add more variables like gender, alcohol consumption, you will get adjusted odds ratio of CHD for smoking . The adjusted odds ratio shows strength of association between CHD and smoking, where effects of other variables in the logistic regression model are held constant.
You can find unadjusted odds ratio by manual calculation (OR=ad/bc) or through any software for logistic regression by including only one independent variable.
Yes, you can obtain unadjusted odd ratios by mere cross-tabulation of say, two categorical variables, but in order to obtain adjusted ratios you need to control/adjust or include more than one independent variable in your model. The interpretation of the adjusted odd ratios is such that the effect of one variable on the dependent variable can only be discerned if the others in the model are held constant (other things being equal).
Great answers there presented by several colleagues: Omar B, MS Kulkarni, Saira, and so forth. In epidemiological studies, analysis should where possible, flow from univariate (descriptive statistics mainly) analysis, to bivariate level of analysis, and lastly to multivariate analysis level.
Yes, the same. If you run bivariate analysis for two categorical variables using cross-tabulation, the P value obtained from the Chi-square test will be the same as that obtained from the logistic regression model for those same two categorical variables. The Chi-square test will inform whether there is a difference between those two categorical variables while logistic regression model will quantify the magnitude of that difference in the form of odds ratio.
When you run the logistic model make sure to enter those variables as categorical not as contentious variables.
Hope this would help
Take only one independent variable and your dependent variable in logistic regression, you'll get unadjusted odd ratio.
Take two or more independent variables and your dependent variable in logistic regression, you'll get adjusted odd ratio.
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication