# A power analysis question - am I making a big mistake?

#### Graham

##### New Member
Help! It's a very long time since I have had to tackle Power Analysis but I have recently been handed a couple of problems, one of which I hope someone can help me with here.

An area of sea bed was surveyed by divers for a particular type of coral that grew as discrete individuals. The method was to lay out a 10m tape measure and two divers each counted the coral individuals within a two metre band either side of the tape measure. Once completed, a further 10 m was laid taking care to remain within a set depth range (effectively following a depth contour). This was repeated until a contiguous 100m length transect was completed. A second transect was completed in the same way in the same area. Now, depending on how you look at it, I either have forty 10mx2m, four 100mx2m or two 100mx4m sample areas.

I would like to know the method's ability to detect change from year to year (probably more interested in it's ability to detect a decline in reality).

This is where I probably get myself into trouble. I decided to treat the data as four 100mx1m sample areas. The data are as follows:

Sample 1 = 27 corals
Sample 2 = 38 corals
Sample 3 = 9 corals
sample 4 = 20 corals

I have Minitab. Is it appropriate to use the "Power and sample size two sample t" option for these data? In which case I'm assuming I would input:

Sample size = 4
Difference = (left blank)
Power values = 0.8 (or a range)
Standard deviation = how should this be calculated?

Since sample 1 and sample 2 are transects side by side, and 3 & 4 likewise am I committing a major statistical error?

Is the test appropriate for what I'm trying to do? Should I be doing something else?

Really sorry if this if I'm perhaps asking to many or too basic questions. Like I say it's been a very long time...

Last edited: