# Which test is suitable?

#### beth534

##### New Member
I have collected data on the frequency of invertebrates on 13 paired plants (paired based on similar sizes). 1 plant from each pair was covered and the other was uncovered (i.e. 2 conditions). I'm confused as to what statistical test is appropriate for such data. Since I am not measuring the same plants for both conditions, I assume paired tests would not work. Chi-square would not work as the plants are paired, and the DV (frequency) is not continuous so ANOVA would be unsuitable. Is there any such test that exists for this kind of data?

#### Karabiner

##### TS Contributor
Your dependent variable is "number of invertebrates" (interval scaled). Your unit of observation is a plant,
but observations are also paired. Therefore, a dependent sample t-test with n=13 pairs would perhaps be
suitable, but sample size is small, and it is doubtful whether the test's assumptions are met. A Wilcoxon
signed rank test could be used here.

With kind regards

Karabiner

#### beth534

##### New Member
Your dependent variable is "number of invertebrates" (interval scaled). Your unit of observation is a plant,
but observations are also paired. Therefore, a dependent sample t-test with n=13 pairs would perhaps be
suitable, but sample size is small, and it is doubtful whether the test's assumptions are met. A Wilcoxon
signed rank test could be used here.

With kind regards

Karabiner
Thank you for your answer. I have tried using a Wilcoxon signed rank test, but it seems that there are too many "ties", meaning the p-value given is probably not that accurate. This is because I am working with discrete frequency data, where there are multiple 0's and the maximum number is 3. Not sure where to go from here - is there any way a paired t-test or a variation of this test could be used on this kind of data?

Last edited:

#### Karabiner

##### TS Contributor
Then there's the sign test.

With kind regards

Karabiner