Is it appropriate to make a poisson regression with life time number of suicidal attemps obtained by survey? or do we need to follow them for a particular time period?
I think it's better to have a specific time period. Like "number of suicidal attempts of 25-year old adults". So that the time period is from the time they were born until they reach 25 years of age. In that sense, you may also express the counts in terms of annual attempts (morbid as it may sound).
Moreover, use Poisson regression when there is an equidispersion. Observe the Deviance measures. Otherwise, use other count regression models (Negative binomial, truncated poisson/negative binomial, hurdle, or zero-inflated models (ZINB and ZIP)).
Thank's a lot for your answer. But we don't have enough participants to chosse a spesific age. But we have duration of illness times of them, so we plan to add duration of illness times to Poisson regression as an independet variable. Do you thing it would be appropriate then?
then i guess it’s better to express Y as number of attempts divided by age. it may be a very small value but it still keeps the Y as a discrete random variable. i hope this makes sense 😀
- Badges
- Science topic
- Similar topics
- Statistics
- Survey