Timeline for Linear spline and 'interaction' p value
Current License: CC BY-SA 4.0
8 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Mar 24, 2023 at 10:35 | history | edited | dipetkov | CC BY-SA 4.0 |
added 723 characters in body
|
| Mar 12, 2023 at 23:25 | history | edited | dipetkov | CC BY-SA 4.0 |
deleted 590 characters in body
|
| Mar 12, 2023 at 21:08 | comment | added | dipetkov | Interesting. Thank you for providing the extra information which makes it clear that all individuals (cis women) will experience the life event (but not at the same age?). Since the age of event was ~25 and the groups men & women, I was thinking more along the lines of having children. In general, I find abstraction makes it harder rather than easier to write suitable answers. | |
| Mar 12, 2023 at 20:58 | comment | added | LucaS | Thank you - you could be right. In my case I am looking at a population of women who have undergone menopause (as the 'life event'). And I was essentially replicating other analyses - e.g. pubmed.ncbi.nlm.nih.gov/33609298; pubmed.ncbi.nlm.nih.gov/26447063 | |
| Mar 12, 2023 at 19:53 | comment | added | dipetkov | I was interested in the counterfactual interpretation implied by "The aim of this analysis was to see what 'effect' a particular life event had on an outcome"; not in how to implement change of slope. To me these are very two different problems: (a) compare two groups with age as covariate but don't assume that the effect of age is linear througout; and (b) model a single time series with a change point. To estimate the effect of an event you need to do (a). Don't this require overlap in ages as well as to allow for a change of slope in both groups as Y might increase in older people anyway. | |
| Mar 11, 2023 at 21:29 | comment | added | LucaS | Thanks for that. You're absolutely right about the extrapolation issue. I initially simulated data that overlapped for males and females but then thought it would be better to use the same data in illustrating my issue with the change in slope analysis. And yes, there are easier ways to calculate emmeans and I should have remembered that. But (with respect) I think you are missing my point somewhat in what I was failing to grasp in the change in slope analysis. I don't think this is any different, in principle, to something like an interrupted time series anlaysis. | |
| Mar 11, 2023 at 20:09 | history | edited | dipetkov | CC BY-SA 4.0 |
added 1295 characters in body
|
| Mar 11, 2023 at 17:28 | history | answered | dipetkov | CC BY-SA 4.0 |