Forum Replies Created
-
Posts
-
I have not encountered many box-and-whisker plots or histograms in the course of my work, nor on other dashboards. I find them easy to understand and useful as a visual representation of distribution data. I think that histograms are probably easy for broad audiences to understand but box-and-whisker plots may be less intuitive for lay audiences. Box-and-whisker plots are probably more useful in academic settings and for audiences with some background in statistics and/or research.
Bar graphs are best for showing rankings, as it is visually easy to compare the sizes of the bars across categories, particularly when bars are arranged in descending order of value. If we looked at all claims in both the Northern and Eastern Zones and ranked service categories from most to least frequent, Curative Care would be at the top of the list (2097 total claims) and Inpatient Surgery would be at the bottom (6 total claims). For both the Northern and Eastern Zones, the highest ranking service category is Curative Care. For the Northern Zone, the lowest ranking category is Speech Therapy (1 claim), while for the Eastern Zone it is Inpatient Medical and Inpatient Obstetrics (2 claims each).
I think that a line graph would be the best option for showing Flores del Mundo’s membership trends over the course of the year. This is a time series relationship, and so creating a line graph with time in months on the x-axis and number of members on the y-axis would enable viewers to easily see the trends in membership over the course of the year. I would also use a line graph to depict the Eastern and Northern regions on the same graph, with a separate line for each region.
As others have noted, the best option for graphing total membership versus employee/spouse/dependent depends on the intervals of time used. If I wanted to show the overall breakdown for the entire year, I would create a bar graph with a categorical x-axis for each member type, and the number of members on the y-axis. This allows for a quick comparison across member types. If I wanted to show the trends in membership over time for each of these categories, I would use a stacked bar graph, but would likely use larger time intervals like quarters instead of months to reduce visual clutter (as others have suggested).
I agree with others who have suggested that the WHO maternity dashboard at the Mpilo Central Hospital meets Stephen Few’s definition of a dashboard but could be substantially improved to support decision-making in the hospital. The WHO dashboard does display important information on a single screen to achieve the objective of improving maternal and perinatal clinical outcomes. The information displayed could be considered a “high-level summary” in that it shows change over time for a number of indicators without providing additional details about how and why any changes might have occurred. The table does concisely display the data, but this display format is not optimal for quick decision-making, as others have pointed out. Compared to, for example, widgets or metric boxes that display key indicators individually, the table used in this dashboard takes longer to visually process, because it requires that the user visually move between the left side of the screen (where the performance measures are listed) and the right side of the screen, which displays the most recent month of data. Additionally, the authors noted that the dashboard was printed, but that no color printer was available at the hospital; the use of gray-scale would potentially make it even more challenging for the user to make quick assessments at a glance. The tool therefore might not meet Few’s “customized” criterion in that the visual elements were designed without consideration of limitations of the local environment, such as the lack of color printing.
