Quote:
Originally Posted by BILL H
Bug Guy,
...What we have is a set of data with absolutely no controls on it, and no real way to determine the direction of bias or whether there is bias. I have many problems with the data set, and size is not the only or the greatest one....
|
No doubt nobody can make the kinds of conclusions that can from a replicated study with controls, experimental variables, dependent variables, etc. And making decisions about "effect" from this study would be wrong.
I guess from the data I see pattern. Really, it is a small and poorly done creel survey. This links to a page that I found on a quick google search that does a decent job describing a good survey:
http://www.michigan.gov/dnr/0,1607,7...7829--,00.html
For example, the bias comes from the people polled - an unbiased survey would randomize where people fished, what days fished, etc. and take into account other things. In addition, the data would include skunks (as stated previously in this thread). So, this data should take into account these biases and limit any conclusions about pattern to the types of fisherman, location of fisherman, and other variables that describe the respondents. Conclusions on pattern should not be applied to the entire bay. No conclusions on effect to the fish should be made. But, properly done creel surveys (which is what I think the registry will do a better job of) has successfully helped to manage fisheries in other places.
But I will admit - I don't have the greatest confidence that this data is use in the limited form that it should be. So I'm not necessarily disagreeing with everyone here.