I am running Stata 13 for the statistical analysis & results of my dissertation. I am comparing the effect of a light source upon numbers of Anopheles vectors sampled per night in tent traps. Paired collections were carried out for a span of a few days (n=5). For each day, there are 5 separate tent traps being run. This gives me multiple entries for the same date. My data is over-dispersed (mean significantly greater than the mean) and am attempting to run a negative binomial regression model that will take into consideration the effect of night-to-night variation on my catch numbers. Based on what I have read on the forum, I would think that I need to set an id to go along with the date. In my model, the number of vectors caught is dependent (count data) and the independent variables are light, no light, location, human bait, and date.
Any advice or signposting would be much appreciated.
Lastly, given that my data is over-dispersed and I have large number of "zero" catches I would prefer to calculate the Williams mean to compare average density between trap and mosquito species rather than the geometric and/or arithmetic mean. Is there any advice on how this may best be accomplished?
Thank you
Any advice or signposting would be much appreciated.
Lastly, given that my data is over-dispersed and I have large number of "zero" catches I would prefer to calculate the Williams mean to compare average density between trap and mosquito species rather than the geometric and/or arithmetic mean. Is there any advice on how this may best be accomplished?
Thank you
