https://communities.sas.com/t5/Statistical-Procedures/Oversample-of-covariates-what-effects-would-be-expected/m-p/683814#M32929 <P>Suppose we'd like to see some interactions with certain diseases and SES. So, we select all the individuals with the diseases from the entire population data, say 5%. Then we select other 5% from the rest of the population with the stratified random sampling by age-sex-income. Then, we'll see time to death through survival analyses by including both diseases, SES, and other covariates. Here are my questions.&nbsp;</P><P>-Is it possible to assume that the 5% of the population without the disease is a representative sample of the rest 90%?</P><P>-If we don't have the weight variable, what procedures can be used to test that distributions of predictors?</P><P>-are the estimates from Cox proportional analyses biased in this case?</P><P>&nbsp;</P> Tue, 15 Sep 2020 01:13:24 GMT Lily07 2020-09-15T01:13:24Z https://communities.sas.com/t5/Statistical-Procedures/Oversample-of-covariates-what-effects-would-be-expected/m-p/683814#M32929 <P>Suppose we'd like to see some interactions with certain diseases and SES. So, we select all the individuals with the diseases from the entire population data, say 5%. Then we select other 5% from the rest of the population with the stratified random sampling by age-sex-income. Then, we'll see time to death through survival analyses by including both diseases, SES, and other covariates. Here are my questions.&nbsp;</P><P>-Is it possible to assume that the 5% of the population without the disease is a representative sample of the rest 90%?</P><P>-If we don't have the weight variable, what procedures can be used to test that distributions of predictors?</P><P>-are the estimates from Cox proportional analyses biased in this case?</P><P>&nbsp;</P> Tue, 15 Sep 2020 01:13:24 GMT https://communities.sas.com/t5/Statistical-Procedures/Oversample-of-covariates-what-effects-would-be-expected/m-p/683814#M32929 Lily07 2020-09-15T01:13:24Z