I am trying to determine if certain contaminants are decreasing in birds over time (1960-Present). Data was pulled from the literature and I've grouped birds based on their dietary habits. When I analyze all the birds together I get a significant negative correlation. I get the same results for "songbirds" (n=80), "fish eating" (n=91), and "birds of prey" (n=28) groups, but not birds that feed on "aquatic inverts" (n=163) group which has the largest sample size. Is there a way to determine if results are actually significant and not a result of smaller sample sizes for the other groups?
We as scientists may try to consider and use available data. It makes sense that years of environmental regulation is apt to and should reduce contaminants in sensitive species as birds. However, the time span of nearly 60 years, potential for data from various research efforts, locations, species, date, etc., suggest circumstantial evidence, not necessarily significant results that you may seek. Obviously, there is no opportunity to go back in time, unless you find some dated frozen birds. Also I would be surprised if 200 plus birds are all that have been tested through time, so is there a bird data base? Nonetheless the effort, properly qualified and reported suggests the effort may have circumstantial evidentual worth.
Aquatic invertebrates are insects, so not necessarily directly comparible with birds. Whether different species of bird types can be combined and compared is beyond my expertise level. Personally, I would probably plot data for example song birds, date on x axis, concentration of contaminant on y, and use a different figure symbol for each species of song bird. Then do same for fish eating, then same for birds of prey. But lets say 8 of the birds of prey were bald eagles, taken various years, some in remote Rocky Mountains, and rest in more populated areas with higher air and pesticide pollutants. The habitat conditions are not equal relative to pollution, so that may be a problem with just grabbing available data. You may need to disect the data, qualify it, etc. before using to support an idea or theory. If you intend to run ststistics to prove a point for publication, I would recommend consulting a statistician to get advice. There is a difference in running data through a statistical package, and statistically designing a study to address a hypothesis.
1. I agree with Dr. Willium F you should correlate each group individually.
2. Aquatic inverts are not birds
3. Confounding environmental effects should be adressed
4. Bit confusing study what is your hypothesis
Yes, I am very well aware that aquatic invertebrates are not birds. I grouped the birds based on their dietary habits... this group feeds on aquatic inverts (which are not all insects). I am focusing only on Latin America so the data is very limited... hence the small sample sizes and large variability and difficulty in any sort of data analysis.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size