Improvement on the aggregated rapid reviews dataviz

[dasaderi]

Brainstorm ideas to make the visualization of aggregated rapid reviews more intuitive to interpret.

[mortonanalytics]

Hi there! I'm understanding you want feedback and ideas for the data viz. My feedback is relatively simple as there aren't too many options with yes/no data.

Option 1) The fixed length plot will be easier to read if it's ordered by yes responses - that way the relative differences will be easier to discern.

Option 2) I'd consider taking the fixed length off and try a more butterfly chart where yes and no are each a wing that grows in size relative to the number of responses. You'll get peaks and troughs that draw attention to lopsided responses on various areas.

Option 3) I think changing chart type to a gauge or pie chart would not improve things considering there are several areas of interest that will have responses unless you can change the UI around a bit. Below works ok for 4-5 areas. Overflowing to a second row (for a total of 8-10) is probably as deep as you'd want to go.

[dasaderi]

@mortonanalytics Thank you so much for your feedback. The possible answers to each question right now are "yes", "no", "unsure", and "N/A". I like option 2 and 3. But what should we do with the "unsure" and "N/A" options?

[mortonanalytics]

I'm not sure option 2 works with more than two responses. I'd stick with the existing chart and play around with the sizing. Try something not as long and perhaps wider to increase the readability and visual impact. If we could break out the NA and unsure into a sub category, we could maybe keep the yes/no butterfly idea (you'll need your imagination here where the yes/no replaces the likert-like responses for the first category and NA/unsure only get one bar):

My only concern is that it's getting a bit busy in the wrong kind of way that may be difficult to scan quickly.

I think option 3 could be modified to multiple categories (3 or 4 is ok for a gauge or donut chart):

I like that you can play around with the UI using different center options - either icons or perhaps the number of respondents.

We'll call this option four. It's really just about rearranging the text with regards to the horizontal bar chart. The pivot makes it a little easier to scan up and down than the existing chart, and perhaps you can get a sense of why widening the bar vertically is visually helpful.

[dasaderi]

@mortonanalytics thank you so much. These are great suggestions. I'm a fan of donut charts but I fear it may get too busy to have one of those for every answer. However, if we can come up with a summary score, that could be a way to show aggregated summaries.

Option 4 seems very similar to existing plot but with more definition, wider bar, and % instead of absolute number, or am I missing something? I do like the percentage. We were thinking about doing that instead of the absolute number, but I guess % makes a lot sense when the n is high. for now our numbers per preprint are low. I also like the widening of the bar. I agree that is helpful.

I'll discuss these options with the developers. Thank You!

[mortonanalytics]

You’re not missing anything on Option 4. Just wanted to demo how even minor layout changes can make some easier to scan and digest easily.

I agree multiple donuts can be a bit much. Another option is to have a 2-3 donut summary at top, with detailed bar chart breakdowns below.

On Wed, Mar 25, 2020 at 3:28 PM Daniela Saderi notifications@github.com wrote:

@mortonanalytics https://github.com/mortonanalytics thank you so much. These are great suggestions. I'm a fan of donut charts but I fear it may get too busy to have one of those for every answer. However, if we can come up with a summary score, that could be a way to show aggregated summaries.

Option 4 seems very similar to existing plot but with more definition, wider bar, and % instead of absolute number, or am I missing something? I do like the percentage. We were thinking about doing that instead of the absolute number, but I guess % makes a lot sense when the n is high. for now our numbers per preprint are low. I also like the widening of the bar. I agree that is helpful.

I'll discuss these options with the developers. Thank You!

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/PREreview/rapid-prereview/issues/79#issuecomment-604097443, or unsubscribe https://github.com/notifications/unsubscribe-auth/AC3JVCTGZ4L7M2UO344GRILRJJZRLANCNFSM4LS3PWHA .

-- Ryan Morton, Principal Analyst Morton Analytics www.morton-analytics.com

"39...This appears to be the first uninteresting number, which of course makes it an especially interesting number, because it is the smallest number to have the property of being uninteresting. It is therefore also the first number to be simultaneously interesting and uninteresting." ---David Wells. The Penguin Dictionary of Curious and Interesting Numbers.

[murkatr]

This could be helpful for #80