Forum Replies Created
-
Posts
-
I’ve encountered a few box and whisker plots only when meeting with statisticians, but I’ve seen them kind of hit a blank wall when presented to non-statisticians on the team. Histograms are more commonly presented in my experience to show a basic distribution of the data. But even then, histograms aren’t immediately recognizable in the way that my non-statistician PI’s usually want to see.
Since I joined a team of social scientists who have little background in quantitative data, I’ve started to present what little quant data we have in relation to other data. Presenting just values of data don’t seem to mean much, at least not to my team members. Since box and whisker plots and histograms are extremely useful for me to translate data though, they’ll probably be common on my analysis pages. But not the final dashboard.
The one exception I can think of would be a histogram (which is easier to understand immediately than box and whiskers) with a very explicit title describing what piece of data they’re supposed to get from the plot.
According to Stephen Few, bar charts are often the best option to display ranking. The most frequent service category of claims by far are curative care visits, almost double the number of claims compared to the next most frequent category. And this is true across both Eastern and Northern zones.
The Northern and Eastern zone begin to differ with the 5th ranked service category, and the least frequent is different depending on zone as well. In the Eastern zone, both inpatient obstetrics and inpatient medical claims were the least frequent, with each service category only having 2 claims. In the Northern zone, speech therapy was the least frequent with only 1 claim.
That’s an interesting question because the values of the different variables (employee/spouse/dependent, male and female, etc.) that could be included on the membership trends graph very widely. It seems that because the breakouts of enrollments vary between the two zones in ways that are significant for health plan decision makers, we should be sure to include more variables than just simple overall enrollment trend over time. I’m thinking that perhaps a line graph is still the best way to show the trends over the course of the year, even though you’ll be including employee/spouse/dependent breakouts and/or Eastern and Northern zones simultaneously. But the line graph should be accompanied by a simple two bar graph showing the average proportion of females and males enrolled in each zone for 2011. Since the proportion of females in the Northern zone is so much higher than males in that zone AND ALSO higher than both males and females in the Eastern zone, that information will affect health plan choices by Flores del Mundo.
