3
$\begingroup$

I am assessing the presence or absence of a number of developmental milestones in mice in their first $21$ days of life. Each day, from the day they are born until they have $21$ days, I test if pups have or not specific developmental milestone. This analysis involves $2$ groups, a control and an experimental.

I want to test whether the experimental group are delayed in accomplishing these milestones compared to controls. For each test, I end up with a range of days which the presence was observed. For controls, my $n=17$, for experimental my $n=15$. So, for a particular test (for example, cliff aversion), I have observed the presence as early as day $2$ and latest as day $9$. Not sure if a Chi-square would be appropriate and if so, do I have to do a chi-square test for each day comparing both groups? Could I also perform any type of "survival" analysis that allows me to estimate the probability (what day) for each group to have the presence of a specific milestone?

Improve this question
$\endgroup$
4
  • 1
    $\begingroup$ Can a milestone that's been achieved ever be lost, or is it a permanent change of state (like death in survival analysis)? $\endgroup$
    – Glen_b
    Commented Jun 13, 2014 at 4:39
  • $\begingroup$ Do you have any covariates (other than group membership?) If not, then survival analysis (even if appropriate) is maybe more than needed! $\endgroup$
    – kjetil b halvorsen ♦
    Commented Jun 13, 2014 at 8:19
  • 1
    $\begingroup$ Thanks for the answers. Once the milestone is achieved is a permanent change. Regarding other covariates, I would say no, I am testing 2 groups over 21 days on the acquisition of these milestones. Luis $\endgroup$
    – user48296
    Commented Jun 15, 2014 at 16:07
  • 1
    $\begingroup$ @user48296 If you are the original poster and you forgot your login credentials, please consult our Help center to merge your accounts. $\endgroup$
    – chl
    Commented Jun 15, 2014 at 16:34

1 Answer 1

Reset to default
1
$\begingroup$

You seem to have two groups observed at many different variables (milestones for different tasks, with values in 1...21). So this could be seen as a problem of multivariate regression, or maybe manova.

But a simpler solution is to use logistic regression, with group (experimental or control) as outcome variable. The logic is that if there is some differences between the groups, they should be able to predict the group.

This is proposed at T-tests, manova or logistic regression - how to compare two groups? and otherwise! This should be accompanied with some data visualizations.

Improve this answer
$\endgroup$

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.