New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Confusing visualization of score poll when allowing negative ratings #8395
Comments
Hi @despens thanks for taking the time to explain this. I've not done much testing with negative value score polls to date. It's quite broken eh. Thanks for all these actionable suggestions too. I'll find some time to fix the results table with negative values. I'm intending on making the chart type used for a poll selectable in the future. But I agree were going to need a stacked bar chart that can handle having both negative and positive values. My workload is full right now, but I'll come back to this in a few weeks. |
hi @despens - the next release of Loomio will have bar graph next to each option in the table and "% of votes" is removed for score poll. I'm not totally sure what the right way to represent positive and negative scores is with bars. Do you think negative goes left and positive goes right, with 0 in the center.. or does the lowest value (possibly negative) become the left value (previously 0) and the highest on the right? |
I thought the best visualization would be a bar chart in which the negative score is expressed below the x axis: This is able to express "controversy" for options that receive both positive and negative votes that cancel out, but can show where further discussion needs to happen. I know this looks difficult to implement as it would require some kind of legend on the chart as labels would be oriented orthogonal to the bars, and horizontal scrolling for polls with more than a few options. To stack the bars on top of each other would seem to make sense, but there is a weird effect in which the classic coordinate system logic of "left below 0, right above 0" looks wrong, because the lowest entry will probably have the largest amount of negative votes: Positive votes on the left and negative on the right looks more logical, at least to me: It might be confusing to users, would need to be tested. Perhaps the reduced color saturation in the negative direction helps to express the "mirrored x-axis" here, because in reading direction first the "full color" is introduced, and then a "lesser" version of it comes along. |
I really, really appreciate this response. It's unlocked my thinking about the problem - and energized me a little too. I think I can make this work, even in our email templates... It's really not that complex once you have calculated the values you're after. |
Right. Damn.
Thanks so much for checking up on this.
The "how to visualize a score poll" thing has been interesting to think
about, while I've been rewriting our emails and chatbots to all support the
latest results table format. I've also had a lot of fun making ascii art
graphs and tables to display results within Discord etc.
What my smooth brain has finally realised is: As soon as you bring negative
values into the mix, you're not really talking about the sum of scores.
You're really interested in the frequency of scores. Lots of 10s and lots
of -10's are not the same as 0.
It's actually easier to visualize too, provided there are some additional
constraints. So if the maximum range between min and max score is limited
to (say) 10, then you can show a "bar" which is actually a table of 10
cells, and control the darkness/intesity of each cell depending on how
often it was selected as a score.
I guess if it was a grey scale value, you'd see a cloud where darker values
indicate more frequent.
I think this is actually a cool representation to consider for a few of the
voting methods.
…On Wed, May 4, 2022 at 12:10 AM Dragan Espenschied ***@***.***> wrote:
In the latest version 2.13.1, score polls with negative rating don't show
any visualization at all:
[image: Screenshot-20220503140505-778x527]
<https://user-images.githubusercontent.com/571494/166449984-3ff07320-27dc-4099-8b12-ed7ef638064a.png>
(Screenshot from the same poll as in the original bug report.)
—
Reply to this email directly, view it on GitHub
<#8395 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AADWXX2ESZ2AVK6B2EY2JYTVIEJTBANCNFSM5ODMJPKA>
.
You are receiving this because you commented.Message ID:
***@***.***>
|
Hi @despens - this is a proof of concept I've just made. It's using randomly generated data, so there isn't trends like you might see with real vote data.. but.. what do you think? |
I expect real vote data will make a big improvement |
Maybe we need the scale at the top, too. -5 to 5 |
I would suggest to
Indeed the email display looks more legible without the gaps in between the scores. I think that already might be a great improvement! Perhaps using a higher brightness contrast for the rectangles versus the background would be good. |
Thanks for the feedback. But to see who voted what, we may as well make an option to change the chart type to what we use for meeting poll. Which is users across the top and options down the side, and what value they gave each option. Infact, in the next big revision I hope to make it so that you can choose the chart type for your poll. So you can choose between bar chart, frequency chart (this), and individual weightings per user (meeting grid). |
When running a score poll and allowing for negative numbers, the chart drawn as a result is showing the option with the most negative rating as the largest block, according to "% of points".
My sense is that users would assume a negative rating to express that something is undesirable and that this option would be drawn as a smaller block rather than a larger one. The "% of points" assigning a negative value to options that have received positive score is surprising. Additionally, the order of options rendered appears like looking at the tail of results.
The thumbnail visualization of the same poll highlights the option that received the most positive ranking as the largest, but at the bottom of the stack.
In general that type of horizontal bar graph representation seems more adequate than the full-size single horizontal stacked bar graph drawn in full view, at least when negative scores are allowed in the poll.
I would suggest to
The text was updated successfully, but these errors were encountered: