I used the 3 days of data from my point count surveys of spotted towhees across sites within Beacon Hill park, categorized by shrub cover category (sparse, moderate, dense), to make a bar graph comparing the mean proportion of days towhees were present to the corresponding shrub cover category of the site.
It was a relatively long process of finding a way to organize my data logically. Originally I was not planning to aggregate my counts across the 3 survey days for each site to get a proportion, but I realized that if I were to run an analysis like ANOVA from my experiment data (which ended up being my best option), I’d need to make my response variable continuous.
I didn’t have much difficulty with organizing the final data I collected into a complete table, but it was hard to produce a figure from it that made sense and communicated the patterns my data clearly. I eventually decided that a bar graph was the clearest way to show the trend from my data of towhees being present on more occasions in sites with more dense shrub cover. I expected my data to show this trend, but I didn’t expect it to have as many outliers which increased the standard deviation quite a lot as you can see with my error bars. This makes me think that if I had the opportunity to do several more survey days (e.g. 10-15 instead of just 3) I might see a much stronger relationship in my data because I wouldn’t be dealing with so much variability and having proportion values only being 0.33 (1/3), 0.67 (2/3), or 1 (3/3). If I were to explore my research further, I would do this.