Forum Replies Created
-
Posts
-
I’ve mainly encountered box-and-whisker plots being used to display results from scientific experiments in reports and papers. I have not seen them used on other dashboards. The histograms I have seen have mainly been in the setting of displaying grade distributions on examinations.
I think histograms are easy to understand at a glance. Box-and-whisker plots, however, are less immediately clear and require an explanation of what the box and “whiskers” represent. I think that once you know how to read them, they are useful for showing more information about the statistical distribution of the data.
The best type of graph to show rankings with are bar charts, with the categories organized from greatest or most frequent to least. One could possibly consider using a pie chart, as well, depending on one’s goals.
Based on claims in the Northern zone, the most frequent would be Curative Care Visits, and the least would be Speech Therapy.
Based on claims in the Eastern zone, the most frequent would still be Curative Care Visits, but here the least frequent would be tied between Inpatient Medical and Inpatient Obstetrics visits.
For the purposes of displaying trends in Flores del Mundo’s total membership (Eastern & Northern regions combined), and for showing Eastern and Northern regional memberships separately (but on the same graph), I would use a line graph. For the former, I would use a line without dots. For the latter, I would opt to include the dots to help highlight the differences between the two at each monthly interval.
If we wanted to subdivide the data and show graphs depicting Employee, Spouse, and Dependent enrollment all on the same graph, I might be inclined to suggest that a bar graph would be more effective. The difference between the range of numbers for each subdivision is great enough that, in my opinion, a line graph doesn’t effectively compare the groups to one another. For the Northern region, where there is such a large gap between the number of employee enrollees and the number of either spouse or dependent enrollees, I might even consider using a stacked bar chart. Otherwise, I would consider using a break in the Y-axis in order to effectively demonstrate the trends between the much larger and much smaller values (without having the smaller values appear like an indiscriminate flat line).
Hi all,
Going by Alexander’s and Few’s definitions, I would say the WHO local maternity dashboard is not a “true” dashboard, mainly because the data is not presented in an especially visual way. The only visual component is the color coding of data relative to a target — there is no incorporation of bar graphs, trendlines, pie charts, etc. In a way, it almost seems closer to a report with color coding rather than a strict dashboard.
On the other hand, the data presented is high level data relevant to a goal (namely, meeting targets for better maternal/neonatal outcomes) presented on a single screen, and the colors do help to create a visual effect that can be interpreted at a glance. However, the presentation is not as visually clear as, for example, a trendline. Given the sheer number of metrics they were tracking, this is an understandable choice, since having that many trendlines would be confusing in its own right.
– Masis
