compared to the drop from 1 to 5. From these data, there appears to be a noticeable difference in
egg numbers between transects in regards to distance from the farm epicenter.
Looking at the data summarized in Table 3-1, we see that the highest standard deviations
correspond with the highest mean egg numbers. These high standard deviations wreak havoc on
statistical tests, lessening any significance that the data may inherently possess. Consequently,
we cannot say that there is a significant difference in egg averages between transect 1 and
transect 11, much less between other more closely-spaced transects. However, when we look at
the 3rd quartile data (Table 3-1), there is a clear increase in frequency of higher egg numbers with
transects more proximal to the epicenter, specifically transects 1 through 4. Additionally,
median values are significantly higher with transects 1 through 3. Perhaps due to the sample
size, we cannot prove significant differences between any of the transects using summary
statistics alone. The next section will detail more complicated statistical analyses and findings.
In order to further analyze butterfly oviposition, the researcher used the model of
repeated measures over time, treating the experiment as a randomized block design with day,
location and transect as the factors. This allows the researcher to put the fixed effects of
location, day and transect into blocks in order to measure how influential each fixed effect is to
the variance in egg number. The raw data were then transformed (i.e., egg numbers) using a
log+1 transformation in SAS to standardize the residuals in order to meet the assumptions of the
model. The transformed data were then analyzed in SAS using a proc mixed procedure, which
generated the significance statistics found in Table 3-2.
Variations in egg number can be attributed to 4 different effects: location, transect, day,
and the interaction between location and day. There were no reported interaction effects
between location and transect, nor day and transect. Table 3-2 shows a list of the effects and