3
$\begingroup$

This is a repeated measure analysis. The dependent variable is a binary Tuberculosis (TB), Yes/No. I have several independent variables(X) like age, BMI, number of family members, income and others, mix of continuous and discrete.

Once the patient develops TB (Y=1), there is no going back. The patient has reached the final absorbing state. My question is, does collecting more observations, independent variables(X) like age, BMI, number of family members, income and others, matter after the patient has reached the final stage, TB=Yes (Y=1), while analyzing the association between Y and X ? Thanks in advance.

$\endgroup$
2
  • $\begingroup$ What exactly is your data? Could you describe it in greater detail? You have data collected over time for patients and want to predict when the particular patient would develop the disease? $\endgroup$
    – Tim
    Nov 3, 2021 at 8:48
  • $\begingroup$ @Tim, binary Y, continuous or discrete X. $\endgroup$
    – Science11
    Nov 4, 2021 at 3:41

1 Answer 1

1
$\begingroup$

So this is regression with a categorical dependent variable, or binary logistic regression. We want to know how changes in X are associated with the probability of Y. For example, we want to know how changes in age (X) are related to the probability of developing TB (Y). Thus, any observations made after data collected to examine the relationship between age (X) and TB (Y), BMI (X) and TB (Y), income (X) and TB (Y) (etc.) changes the intent of the initial research question. Rather, conceptually we are now looking at how changes in X after developing TB (or not developing TB), contribute to the probability of Y (TB health).

Beyond this, Bayesian approaches allow us to update the regression model with new data (bayesian prior), however I don't think this is applicable here.

$\endgroup$
1
  • 1
    $\begingroup$ you are absolutely right. I realized the mistake in my thinking. Thanks. $\endgroup$
    – Science11
    Nov 4, 2021 at 5:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.