Talk:Community Insights/Community Insights 2021 Report/Thriving Movement
@RMaung (WMF):: Thanks for sharing these, I find this information very fascinating! One note: I think something has gone wrong with Figure 8. Summing the percentages of U.S. population in this chart (13+18+0.9+5.7+72), you get 109.6%, which is more than 100% and thus not possible. Similarly, your U.S. editor numbers in the chart (.5+5.2+.1+8.8+89) sum to 103.6%. Can you update this chart with the correct numbers? --LiAnna (Wiki Ed) (talk) 23:38, 22 June 2021 (UTC)
- Hi @LiAnna (Wiki Ed):! The racial and ethnic categories are not mutually exclusive (e.g., someone in the US might identify as both Asian and White), so we wouldn't expect those percentages to total exactly 100% for Figures 8 and 9. Let me know if you have any other questions! RMaung (WMF) (talk) 15:05, 23 June 2021 (UTC)
- I should also add that the category of "mixed or multiple" in figure 9 includes respondents who self-identified as "Mixed" in an open field, referring to this UK-specific census category and/or ethnic category: https://en.wikipedia.org/wiki/Mixed_(United_Kingdom_ethnicity_category). I'll update the notes for each of these figures to make this more clear! RMaung (WMF) (talk) 15:16, 23 June 2021 (UTC)
- Ah, ok, thanks RMaung (WMF) for adding the clarification to the report, that makes a lot more sense! Now I'm more curious what you did to get the US population numbers to compare it to; it says in the notes it's from the 2017 American Community Survey by the Census Bureau. I'm looking at that data; did you just manually add in all of the individual answers in "population of two or more races" section there so it matched how you did your question? I'm asking because I'm trying to compare these numbers to the survey demographic data we have from our program participants (who are mostly U.S. college students editing Wikipedia through our program); instead of having people select multiple, we offered a "biracial/multiracial" option, which makes it easy to contextualize it for us to both the 2019 American Community Survey and the 2016 2016 National Center for Education Statistics undergraduate enrollment data, but is obviously a bit more challenging to compare to the U.S. editor population data from your study. I'd like to understand how you used the 2017 ACS data to get your contextualization so I can most accurately contextualize ours, too. Thank you! --LiAnna (Wiki Ed) (talk) 16:48, 23 June 2021 (UTC)
- LiAnna (Wiki Ed): It's definitely not a perfect comparison-- We made a methodological choice to err on the side of inclusion (the logic being that 'multiracial' is not the primary identity that most folks who are multiracial would select for themselves), which means that when we compare to the US population we basically are excluding that 2-3% of US folks that identified as multiracial ('two or more') on the ACS. We made this choice in part because our basic research question was "At least in these two geographies, do the racial and/or ethnic identities of contributors represent that of the population?", and this comparison is enough to conclude: "nope!". In the future, if we want to explore if and to what extent racial and ethnic representation among contributors is improving or worsening, any statistical test would "recode" respondents who did select more than one race to that "two or more" category to make a more direct comparison with census data. I hope that is helpful! RMaung (WMF) (talk) 17:30, 23 June 2021 (UTC)
- Makes sense! Thanks! --LiAnna (Wiki Ed) (talk) 21:19, 23 June 2021 (UTC)
- LiAnna (Wiki Ed): It's definitely not a perfect comparison-- We made a methodological choice to err on the side of inclusion (the logic being that 'multiracial' is not the primary identity that most folks who are multiracial would select for themselves), which means that when we compare to the US population we basically are excluding that 2-3% of US folks that identified as multiracial ('two or more') on the ACS. We made this choice in part because our basic research question was "At least in these two geographies, do the racial and/or ethnic identities of contributors represent that of the population?", and this comparison is enough to conclude: "nope!". In the future, if we want to explore if and to what extent racial and ethnic representation among contributors is improving or worsening, any statistical test would "recode" respondents who did select more than one race to that "two or more" category to make a more direct comparison with census data. I hope that is helpful! RMaung (WMF) (talk) 17:30, 23 June 2021 (UTC)
- Ah, ok, thanks RMaung (WMF) for adding the clarification to the report, that makes a lot more sense! Now I'm more curious what you did to get the US population numbers to compare it to; it says in the notes it's from the 2017 American Community Survey by the Census Bureau. I'm looking at that data; did you just manually add in all of the individual answers in "population of two or more races" section there so it matched how you did your question? I'm asking because I'm trying to compare these numbers to the survey demographic data we have from our program participants (who are mostly U.S. college students editing Wikipedia through our program); instead of having people select multiple, we offered a "biracial/multiracial" option, which makes it easy to contextualize it for us to both the 2019 American Community Survey and the 2016 2016 National Center for Education Statistics undergraduate enrollment data, but is obviously a bit more challenging to compare to the U.S. editor population data from your study. I'd like to understand how you used the 2017 ACS data to get your contextualization so I can most accurately contextualize ours, too. Thank you! --LiAnna (Wiki Ed) (talk) 16:48, 23 June 2021 (UTC)
Native American/Alaskan underrepresentation
editHi. It looks like out of the racial groups, one of the most severely underrepresented seems to be Native Americans/Alaskan Natives, but this group isn't mentioned at all in the text. It feels like a mention might be a good thing, so that people don't feel even more invisible. Clayoquot (talk) 22:28, 1 October 2021 (UTC)
Can we just see what all the questions and possible responses were?
editI'm finding it is often difficult to understand what exactly the numbers in this report are measurements of. E.g. "the endnotes say "the difference in "unsure" responses to this question between 2019 (9.2%) and 2020 (3.3%) indicate that the year-over-year difference may be a result of an undercounting in 2019, rather than an increase in harassment for this group" but it's really difficult to understand this without transparency in how the questions and possible responses were posed.
Similarly, I've gone through a long process of deducing from the way the report was written that for the questions about gender, respondents could select multiple genders, and that "transgender or nonbinary" was a single option. (BTW it seems odd to me that the survey combines these categories.)
Here's another example: "Over one-third of all contributors surveyed indicated that they had felt unsafe or uncomfortable..." and then the report goes on to talk the proportions of contributors subsets who had felt "unsafe". Were there separate questions for feeling unsafe versus feeling uncomfortable? Or are all these statistics about the number of people who said yes to a question that included the words "unsafe or uncomfortable"? Endnote 22 suggests the latter interpretation but it's not entirely clear.
If the survey questions and responses are already posted somewhere and someone could point me to the page, I would so appreciate it. If they aren't already posted, could you please post them? Clayoquot (talk) 00:07, 16 October 2021 (UTC)
Endnote 22
editEndnote 22 refers to "the proportion of contributors reporting harassment" but it's about a different question from the harassment question. Could this be a copy-paste error? Clayoquot (talk) 00:36, 16 October 2021 (UTC)