3
$\begingroup$

Given a dataset as below

day      id  time_with_new_teach time_in_prog dur_train dur_appr age graduated
2020-01  02  20                  35           4         9        20  no
2020-01  19  10                  20           9         19       20  no
2020-02  02  21                  36           5         10       21  yes
2020-01  03   1                   4           3          7       29  no
....
2020-01  09  24                  45           20        19       20  yes

Using graduated as the outcome, which kind of model/packages can i use to determine impact of switching teachers (time_with_new_teach) and time in program (time_in_prog) on graduation probability? Ideally I would like to know if you change your teacher within 6 months, your graduation likelihood increases or decreases by x%. Also for people with time in program over 20 months (time_in_prog > 20), changing your teacher in 6 months (time_in_prog <= 6) leads to x% decrease in graduation probability.

$\endgroup$

1 Answer 1

2
$\begingroup$

The outcome graduated is binary, so you need a model for binary data such as a logistic model.

There appears to be repeated measures within id so you need to account for correlations within id. You could use a model with random effects for id to do so.

To answer your research questions you can fit a model with fixed effects for time_with_new_teach and time_in_prog, and possibly the interaction between them.

Since you seem to be interested in a cut-off point for these variables, then you may want to dichotomize them - however this always results in a loss of statistical power and it might be the case that there are interesting associations among the variables that will be lost if you dichotomize, so I would strongly suggest trying to work with the original variables first.

Such a model would be a generalise linear mixed model (glmm) and using the logit link function it will model the odds of graduation, so you can obtain odds ratios for the fixed effects - that is, the % change in odds of graduation between for a 1 unit change in the variables (if using the the original variables), or between the groups (if using dichotomized variables)

$\endgroup$
1
  • $\begingroup$ @Mia does this answer your question ? If so, please consider marking it as the accepted answer. If not then please let us know why $\endgroup$ Sep 26, 2020 at 18:21

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.