Detecting Change in not Normally Distributed Count Data

AJHM

New Member
#1
Hello,

I was wondering if anyone could shed some light on how to interrogate count data whichholds a lot of 0'sand therefore not normally distributed. I have pasted in an example of the dataset i'm using. It gives 4 species counts across 27 sites (replicates) across intermittent years.



Site Year Sp.1 Sp.2 Sp.3 Sp.4
1 1993 1 0 5 0
2 1993 89 3 8 0
3 1993 57 0 15 0
4 1993 204 7 0 0
5 1993 104 15 7 0
6 1993 79 4 0 0
7 1993 3 0 0 0
8 1993 0 0 1 0
9 1993 1 0 0 0
10 1993 2 0 0 0
11 1993 53 8 0 0
12 1993 48 11 0 0
13 1993 62 19 0 0
14 1993 1 0 0 0
15 1993 28 7 1 0
16 1993 67 0 0 0
17 1993 56 3 0 0
18 1993 62 7 0 0
19 1993 157 6 2 0
20 1993 4 0 7 0
21 1993 16 7 0 0
22 1993 5 0 0 0
23 1993 0 0 0 0
24 1993 3 0 1 0
25 1993 1 0 0 0
26 1993 28 0 0 0
27 1993 4 0 3 0
1 1994 7 0 28 0
2 1994 65 10 6 0
3
ect.
ect.

Something I would be interested in looking into would be if there was a difference in Sp counts pre and post a year e.g. 2000. If I code pre and post with a 1 and 2 and add the data to SPSS would it then, in theory, be possible ask this question even if the data is still not normally distributed and contains many 0 values?

Thanks for your time,

Andrew