My data has a nested structure, thus temporal trends were calculated using hierarchical modelling.

sex ~ year + (1 | town)

Males proportion increased by 5% [95% CI: 2; 7], reaching 60% [95% CI: 50; 70] in the last year of the study.

Calculations on crude data showed that male proportion was 50% (750/1500) in the last year of the study.

As you see, there is 10% discrepancy between these analyses.

  1. Do you agree that I should publish the both results? Modelling gives significance and crude calculations shows the "actual" situation.

  2. Should I address the discrepancy between the analyses or this is self-evident? How would you address the discrepancy?

  3. And is it correct to say calculations on crude data or are there better terminology or wording? Raw data, initial data, original data, unadjusted analysis?


1 Answer 1


IMHO it's a good idea to publish both results. This is because you should do you uttermost best to explain the situation to the reader. I think that a significant result is not the full story.

To that point, this model isn't beyond criticism. Here you assume a fixed linear trend for the proportion of males (presumably with a logistic link).

Fixed, the alternative would be sex ~ (1 + year | town) Linear, a bigger problem than the former I think, but harder to solve. Depending on the number of years you are fitting, it may be overly strict to assume a linear trend in time. This doesn't fit "hockeysticks", "bumps" and other things that are very likely a priori in this data. I say "a priori" because I don't know about your data and the number of years. If you are modelling survival over a single year, this is fine.


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.