Which test? - Significance of Sample Size

#1
Hi all, (First-time poster)

I'm working with tagging data for 35 tracked fish. Each fish was tracked of 3 seasons, with detections (3873 overall) corresponding to a fish entering the detection area of one of the many placed monitors within the study area.

I would like to perform a stats test to determine how significant/relevant my overall data set is, given the variance in detections among tagged fish. The values below give an example of such variance.


# S1 S2 S3
1 0 29 118
2 1 0 5
3 439 244 110
4 0 4 935
.
.

I'm wondering what test would be best fit for this sort of data?
Any assistance would be much appreciated.
 

rogojel

TS Contributor
#2
hi,
I think, that there is no such thing as a "significant data set " . What are the research questions you want to answer with this data?

reagards
rogojel
 
#3
You may check whether the data is normal. The Shapiro test could help you to do it. Beyond this, you have not mentioned what's your research question, objective etc.
 
#4
The research objective is to confirm that the movement of these fish are in fact significantly linked to the season and location in which they were tagged. Tagging locations were split between the upper and lower area of the study site (related to deep and shallow regions).

Could a comparison of detections between upper and lower regions be performed and have any sort of benefit to this objective?
 

rogojel

TS Contributor
#5
hi,
I do not know much about your research, so just assuming for the sake of example:
you should have a data set that contains some measure of the fish movement at differente seasons and different locations (like a two dimensional table with the rows being the locations, say, and the columns the seasons. And conceptually, a third dimension for each location containing the upper and the lower regions. Then you could try an ANOVA for instance: that would sort out the influence of the location, region and season.

regards
rogojel
 
#6
Thanks all!
I have performed an ANOVA on the data set. My Fcrit > F (3.15 > 2.2), therefore does that mean I accept the null hypothesis? (meaning that the influence of the three seasons as an example, were all the same?)

Additionally I am also in the process of trying to use the Shapiro-Wilk test, performing the test on one season's worth of data at a time. I'm following the instructions from this site (http://www.real-statistics.com/test...l-tests-normality-symmetry/shapiro-wilk-test/) and am now stuck on how to calculate b. Any suggestions/assistance would be much appreciated :)