Blog Post 4: Sampling Strategies

Compare the percentage error of the different strategies for the two most common and two rarest species.: both were right skewed, however haphazard sampling resulted in less error dispersion, where as random sampling showed a clearer dispersion of error. Seen as how they both skew in roughly the same way, I would conclude that random sampling is accurately representing possible error, where as haphazard may be showing a tendency to show false positive.

Further thoughts: It may also be that with a right hand skew, plus less diffuse error distribution in the haphazard model may actually be a fat tail. If this is the case, then some very important outliers could be hiding in that tail. On second thought, I don’t think I’d want to use this sampling technique where missing the effects of outlying, or phenomenon could have a major impact on stake holders involved in decision making. I wouldn’t use this in helping with ecological assessments around environmental safety, conservation issues regarding extremely endangered species, or economically and culturally vital species, such as salmon or herring populations. If there is error, the randomized model seems represents it more effectively, with a more dispersed deviation around the mean, there by prevent the right skew kurtosis.

Did the accuracy change with species abundance?: No, it didn’t seem to. Although, this may reflect the fact that when I sampled in haphazard sampling, I followed a pattern, and did not attempt to emulate a random pattern. I may have to try sampling again.