I'm analysing a dataset and comparing three zones at two points in time. The data in question is:

Hourly sightings of wild fish (CPUE)

In one of three adjacent zones (A, B, and C)

During the Spring of 2009 and Spring of 2017

I'm trying to see if there was any significant change in CPUE between 2009 and 2017, for each zone, as well as for the entire region (A+B+C). Null hypothesis is that there was no observed change in CPUE over time. I'm not necessarily interested in finding differences between the zones, just the same zone across the two time periods. Simplys put i'm comparing CPUE between:

A2009 vs A2017

B2009 vs B2017

C2009 vs C2017

(A+B+C)2009 vs (A+B+C)2017

The data is not normally distributed, includes many 0s, and not all zones have the same amount of data.

So far I have conducted Mann-Whitney U tests comparing each zone between years, as well as for the entire region in both years. Is the Mann-Whitney U the right choice? I've looked into kruskall wallace and similar tests, but I'm ony interested in comparisons between time period, not zone. Is there a more appropriate test? Would a linear / time series model be more appropriate? I've attached an analogue of my data for clarity.


All data simplified

  • $\begingroup$ Are the zones the same in 2009 and 2017? Why do you have a varying amount of measurements in each zone? Are these repeated measures? $\endgroup$ – user2974951 Feb 6 at 10:12
  • $\begingroup$ Yes they are the same zones, but the exact locations within are different. Since they are the same zones ant different times I think they're repeated measures but am not sure. Can you clarify? $\endgroup$ – Christian Daniell Feb 6 at 10:19
  • $\begingroup$ What are these multiple measurements in each zone? Are these measurements taken from different locations inside a zone for a given year? That is, are these measurements independent? $\endgroup$ – user2974951 Feb 6 at 10:45
  • $\begingroup$ Each zone is a transect (approx 0.5km by 8km) and 90 minute observations were made at random sites within those zones. I think this qualifies as a repeated measures design yes. $\endgroup$ – Christian Daniell Feb 6 at 12:42
  • $\begingroup$ That is definitely a repeated measures design then. The tests you mentioned above Mann-Whitney / Kruskal-Wallis are not appropriate here. You could solve this with a repeated-measures ANOVA or a mixed model. Can you post your raw data in your question? $\endgroup$ – user2974951 Feb 6 at 12:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.