Community Insights/Community Insights 2021 Report/Endnotes

From Meta, a Wikimedia project coordination wiki
Jump to navigation Jump to search
2021 Community Insights Report

Below are methodological endnotes corresponding to in-text references.

Brand Awareness[edit]

  1. The following analyses are weighted by editing activity and home project. n = 2263 for all contributors responding to the question in 2020, 90.4% responding "yes" (95% CI: 89.1%, 91.6%). Respondents who finished less than 50% of the survey were less likely (85.6%) to indicate "yes" (n = 578, χ2 = 19.881, p = 0.000), though this effect size was small (d = -0.216). The proportion of contributors indicating awareness of the Foundation increased from 85.3% in 2019 to 90.4% in 2020 (χ2 = 28.768, p = 0.000). Those who were women were more likely to indicate Foundation awareness in 2020 (88.2%, n = 228) than in 2019 (75.9%, n = 108) (χ2 = 8.253, p = 0.004). The following analyses are weighted by editing activity. Contributors living in Western Europe were more likely in 2020 (93.0%, n = 473) than in 2019 (88.8%, n = 383) to be aware of the Wikimedia Foundation (χ2 = 4.726, p = 0.030). Among 2020 respondents, those living in Northern America were more likely (97.0%, n = 234) than others (91.6%, n = 1381) to be aware of the Foundation (χ2 = 8.318, p = 0.004), whereas those in Eastern Europe were less likely to indicate such awareness (85.8%, n = 190) than others (93.3%, n = 1425) (χ2 = 13.309, p = 0.000).
  2. “How well does this vision represent your values…” (n = 2073, average score 3.45 on a 1-4 scale, 95% CI: 3.42, 3.48. “The Wikimedia Foundation has the right programs and strategies…” (n = 1579, average score of 3.66 on a 1-5 scale, 95% CI: 3.62, 3.71). “I am confident in the Wikimedia Foundation’s ability to support the global movement” (n = 1869, average score of 3.65 on a 1-5 scale, 95% CI: 3.61, 3.69. Those who finished less than 50% of the survey had a lower average score of 3.57, n = 378, U = 287345, p = 0.029), though this effect size was small (d = -0.105).
  3. Foundation Support is calculated by averaging responses to the three index questions (or those index questions responded to), standardized to a scale of 1 to 5. A Foundation Support score (FSS) was calculated for 2102 contributors who answered any combination of the three index questions (average FSS score was 3.92, 95% CI: 3.89, 3.95). For the following statistical tests, group means are reported for clarity even though assumptions of normality and outliers are violated. A Mann-Whitney U-test is used in each case to identify group differences. The following analyses are weighted by editing activity. In 2020, contributors living in Africa had higher Foundation support scores (4.39, n = 22) than other contributors (3.91, n = 1530) (mean rank of 1032.52 vs. 744.97, Mann-Whitney U = 21452, p = 0.002). Contributors to the English language Wikipedia had lower Foundation support scores (3.78, n = 404) than other contributors (3.64, n = 1723) (mean rank of 917.56 vs. 1042.71, Mann-Whitney U = 275206.5, p = 0.000). These contributors also had lower Foundation support scores than en.wiki contributors had in 2019 (3.87, n = 286) (mean rank of 293.87 vs. 326.91, Mann-Whitney U = 39450, p = 0.024). Contributors living in South America had lower Foundation support scores (3.74, n = 52) than other contributors (3.93, n = 1500) (mean rank of 614.41 vs. 753.65, Mann-Whitney U = 29445.5, p = 0.025). Contributors living in East Asia had lower Foundation support scores (3.70, n = 112) than other contributors (3.94, n = 1440) (mean rank of 578.86 vs. 761.44, Mann-Whitney U = 53790.5, p = 0.000). Contributors to the Japanese language Wikipedia had lower Foundation support scores (3.60, n = 76) than other contributors (3.93, n = 2051) (mean rank of 723.84 vs. 1029.71, Mann-Whitney U = 46875, p = 0.000). The following analyses are weighted by editing activity and home project.  Newer contributors had higher Foundation support scores (4.08, n = 252) than more tenured contributors (3.89, n = 1766) (mean rank of 1117.37 vs. 974.32, Mann-Whitney U = 231317.5, p = 0.000). Contributors not fluent in English had lower Foundation support scores (3.86, n = 255) than other contributors (3.93, n = 1294) (mean rank of 707.81 vs. 775.58, Mann-Whitney U = 179879.5, p = 0.023). Contributors who are women had higher Foundation support scores in 2020 (3.96, n = 217) than in 2019 (3.70, n = 90) (mean rank of 143.09 vs. 106.83, Mann-Whitney U = 7960, p = 0.001) (weighted by editing activity and home project). Contributors living in Northern Europe had lower Foundation support scores in 2020 (3.95, n = 153) than in 2019 (4.07, n = 120) (mean rank of 108.27 vs. 134.28, Mann-Whitney U = 4873.5, p = 0.005) (weighted by editing activity).

Platform Evolution[edit]

  1. For the following statistical tests, group means are reported for clarity even though assumptions of normality and outliers are violated. A Mann-Whitney U-test is used in each case to identify group differences. The following analyses are weighted by editing activity and home project. In 2020, n = 73 for volunteer developers responding to “MediaWiki and related software currently have a safe, healthy, and sustainable codebase” (3.56, 95% CI: 3.35, 3.77). n = 72 for “The code for MediaWiki and related software can be applied to other Wikimedia projects when needed” (3.99, 95% CI: 3.81, 4.17). n = 75 for "The code for MediaWiki and related software is appropriate for Wikimedia and other projects" (3.87, 95% CI: 3.70, 4.05). n = 78 for "The code development processes for MediaWiki and related software make it easy for me to contribute high-quality code" (3.38, 95% CI: 3.11, 3.64). n = 62 for “The MediaWiki deployment tooling for Wikimedia projects is easy to use” (3.35, 95% CI: 3.11, 3.60). n = 60 for “The MediaWiki deployment tooling for Wikimedia projects is effective” (3.63, 95% CI: 3.41, 3.84. The following analyses are weighted by editing activity. Developers living in emerging geographies were more likely to agree that the codebase is safe and healthy (72.3%, average of 3.79, n = 37) than others (60.0%, average of 3.35, n = 35) (mean rank of 40.92 vs. 32.08, Mann-Whitney U = 807, p = 0.048). Developers living in Asia were more likely to agree that Mediawiki deployment tooling is easy to use (61.9%, average of 3.68, n = 21) than others (36.8%, average of 3.15, n = 38) (mean rank of 35.83 vs. 26.78, Mann-Whitney U = 521.5, p = 0.042); as were those living in emerging geographies (53.1%, average of 3.58, n = 32) compared to others (33.3%, average of 3.04, n = 27) (mean rank of 34.00 vs. 25.26, Mann-Whitney U = 560, p = 0.041).
  2. Developers indicated whether they experienced each barrier never (1), some of the time (2) most of the time (3), or all of the time (4). n = 65 for volunteer developers responding to "Difficulty setting up development environment" (average score 2.24, 95% CI: 2.05, 2.43). n = 70 responding to “Unclear or missing documentation about how to contribute software” (2.38, 95% CI: 2.19, 2.58).  n = 51 responding to “Slow code review process” (2.62, 95% CI: 2.35, 2.89). n = 70 for "Difficulty finding a person or documentation to get questions answered" (2.31, 95% CI: 2.14, 2.48). n = 66 for “Members of the technical community are not helpful or not friendly" (1.47, 95% CI: 1.31, 1.64). Weighted by editing activity and home project.
  3. All analyses weighted by editing activity and home project.  For the following statistical tests, group means are reported for clarity even though assumptions of normality and outliers are violated. A Mann-Whitney U-test is used in each case to identify group differences. n = 1629 for contributors responding to “I have the right buttons or tools to do what I need to do” with an average score of 3.75 (95% CI: 3.71, 3.79). n = 1619 for “the software works the way I expect it to” with an average score of 3.88, (95% CI: 3.72, 3.80). n = 1629 for “The software is easy to use” with an average score of 3.63 (95% CI: 3.59, 3.68). n = 1494 for “I know how and where to report a "bug" that affects my editing” with an average score of 3.07 (95% CI: 3.01, 3.14). Between 2019 and 2020, agreement with this statement rose from 40.7% to 45.2% (an average score of 3.00, n = 804, to 3.07, n = 1494, mean rank of 949.93 vs. 1051.58, Mann-Whitney U = 464642, p = 0.000).
  4. All analyses weighted by editing activity.  For the following statistical tests, group means are reported for clarity even though assumptions of normality and outliers are violated. A Mann-Whitney U-test is used in each case to identify group differences. Contributors to ru.wikipedia had higher agreement with "I know how and where to report a "bug"" in 2020 (63.0%, average of 3.55, n = 82) compared to 2019 (42.6%, average of 3.07, n = 54) (mean rank of 67.98 vs. 45.59, Mann-Whitney U = 2210.5, p = 0.004); as did contributors to Central and Eastern European language projects (57.1% in 2020, average of 3.30, n = 84; 44.9% in 2019, average of 3.09, n = 112) (mean rank of 89.6 vs. 71.38, Mann-Whitney U = 3961, p = 0.010); and contributors to ja.wikipedia (24.6%, average of 2.55, n = 57; 12.8% in 2019, average of 1.86, n = 39) (mean rank of 51.53 vs. 32.81, Mann-Whitney U = 1301.5, p = 0.000). Japanese Wikipedians were less likely than others to agree they knew how to report a technical problem compared to others (46.5%, average of 3.11, n = 1463) (mean rank of 548.15 vs. 740.83, Mann-Whitney U = 27126, p = 0.000); as are editors of the French Wikipedia (35.1%, average of 2.76, n = 97) compared to others (46.3%, average of 3.11, n = 1423) (mean rank of 628.25 vs. 741.16, Mann-Whitney U = 54056, p = 0.010), Dutch Wikipedia (34.5%, average of 2.51, n = 28) compared to others (45.8%, average of 3.10, n = 1492) (mean rank of 563.39 vs. 737.20, Mann-Whitney U = 14833.5, p = 0.028); and English Wikipedia (36.3%, average of 2.92, n = 289) compared to others (47.8%, average of 3.13, n = 1231) (mean rank of 673.79 vs. 747.77, Mann-Whitney U = 146511, p = 0.007).
  5. All analyses weighted by editing activity.  For the following statistical tests, group means are reported for clarity even though assumptions of normality and outliers are violated. A Mann-Whitney U-test is used in each case to identify group differences. Contributors living in East Asia are less likely to say that the software is easy to use (54.2%, average of 3.43, n = 119) compared to others (69.1%, average of 3.66, n = 1471) (mean rank of 652.09 vs. 773.58, Mann-Whitney U = 64540, p = 0.002), that it works as expected (62.8%, average of 3.62, n = 120) compared to others (72.9%, average of 3.78, n = 1459) (mean rank of 686.48 vs. 764.02, Mann-Whitney U = 68253.5, n = 0.047), and that they have the right buttons or tools to do what they need to do (63.1%, average of 3.58, n = 122) compared to others (73.9%, average of 3.76, n = 1460) (mean rank of 688.63 vs. 765.57, Mann-Whitney U = 69644.5, p = 0.047). Youth are more likely to say that the software is easy to use (74.2%, average of 3.83, n = 224) compared to others (66.7%, average of 3.61, n = 1357) (mean rank of 840.31 vs. 747.87, Mann-Whitney U = 158070, p = 0.002); as are contributors living in Latin America or the Caribbean (80.0%, average of 3.95, n = 75) compared to others (67.4%, average of 3.62, n = 1514) (mean rank of 915.15 vs. 757.69, Mann-Whitney U = 62420, p = 0.001).
  6. n = 998 for contributors responding to “When the Wikimedia Foundation creates software, they collaborate well with people in my project or community” (3.34, 95% CI: 3.28, 3.40). n = 1098 for those responding to "The Wikimedia Foundation prioritizes the software that is most needed for Wikipedia and Wikimedia projects" (3.54, 95% CI: 3.48, 3.59). n = 1274 for those responding to "The Wikimedia Foundation produces high-quality software" (3.68, 95% CI: 3.63, 3.73). For the following statistical tests, results of a t-test are reported for clarity even though assumptions of normality and outliers are violated. A Mann-Whitney U-test is used in each case to confirm findings. The following analyses are weighted by editing activity. Contributors living in Latin America or the Caribbean are more likely to agree that the Foundation produces high-quality software (77.0%, average of 4.08, n = 65) compared to others (63.4%, average of 3.66, n = 1174) (mean rank of 743.81 vs. 594.83, Mann-Whitney U = 44163, p = 0.000), with similar trends across subcontinental regions. Contributors living in Eastern Europe are less likely to agree that the Foundation produces high-quality software (50.0%, average of 3.42, n = 142) compared to others (66.0%, average of 3.72, n = 1096) (mean rank of 507.55 vs. 614.99, Mann-Whitney U = 61187.5, p = 0.000). Contributors living in South Asia are more likely to agree that the Foundation prioritizes software well (84.4%, average of 4.02, n = 32) compared to others (56.1%, average of 3.52, n = 1046) (mean rank of 699.18 vs. 524.03, Mann-Whitney U = 20510.5, p = 0.000), whereas women are less likely to agree (49.6%, average of 3.38, n = 137) compared to others (58.8%, average of 3.57, n = 922) (mean rank of 463.64 vs. 528.34, Mann-Whitney U = 53083, p = 0.013); as are en.wikipedia contributors (47.3%, average of 3.34, n = 188) compared to others (58.5%, average of 3.56, n = 937) (mean rank of 484.68 vs. 567.14, Mann-Whitney U = 71861, p = 0.000). Contributors living in Western Europe are less likely to agree that the Foundation collaborates well when producing software (41.6%, average of 3.25, n = 265) compared to others (47.7%, average of 3.36, n = 712) (mean rank of 455.03 vs. 492.97, Mann-Whitney U = 85527, p = 0.045), as were those in Eastern Europe (33.4%, average of 3.18, n = 120) compared to others (47.9%, average of 3.36, n = 857) (mean rank of 437.74 vs. 488.80, Mann-Whitney U = 44951, p = 0.047) and Northern Europe (33.0%, average of 3.06, n = 82) compared to others (47.4%, average of 3.36, n = 895) (mean rank of 413.85 vs. 489.05, Mann-Whitney U = 31193, p = 0.012). Contributors living in South Asia were more likely to agree that the Foundation collaborates well when producing software (80.7%, average of 3.82, n = 31) compared to others (44.9%, average of 3.32, n = 945) (mean rank of 633.10 vs. 477.66, Mann-Whitney U = 18528, p = 0.001), as were those in Latin America and the Caribbean (63.0%, average of 3.85, n = 53) compared to others (45.0%, average of 3.31, n = 924) (mean rank of 614.89 vs. 475.10, Mann-Whitney U = 3033.5, p = 0.000), with similar trends across subcontinental regions.
  7. For the following statistical tests, results of a t-test are reported for clarity even though assumptions of normality and outliers are violated. A Mann-Whitney U-test is used in each case to confirm findings. The following analyses are weighted by editing activity. On-wiki administrators were less likely to agree that the Foundation produces high-quality software (48.5%, average of 3.41, n = 190) compared to others (58.2%, average of 3.71, n = 1107) (mean rank of 546.59 vs. 648.06, Mann-Whitney U = 91277.5, p = 0.000) They were also less likely than on-wiki administrators in 2019 (56.8%, average of 3.48, n = 168) to say so (mean rank of 128.96 vs. 152.92, Mann-Whitney U = 5244, p = 0.025). On-wiki administrators were less likely to agree that the Foundation prioritizes the most needed software (51.8%, average of 3.20, n = 176) compared to others (66.0%, average of 3.59, n = 950) (mean rank of 465.85 vs. 571.91, Mann-Whitney U = 70915.5, p = 0.000) They were also less likely than on-wiki administrators in 2019 (54.2%, average of 3.36, n = 154) to say so (mean rank of 116.98 vs. 153.69, Mann-Whitney U = 3833, p = 0.000). On-wiki administrators were less likely to agree that the Foundation collaborates well when producing software (37.5%, average of 2.95, n = 175) compared to others (47.6%, average of 3.71, n = 1107) (mean rank of 416.04 vs. 528.72, Mann-Whitney U = 61128.5, p = 0.000) They were also less likely than on-wiki administrators in 2019 (38.2%, average of 3.01, n = 157) to say so (mean rank of 119.32 vs. 145.51, Mann-Whitney U = 4454, p = 0.012).

Thriving Movement[edit]

  1. n = 1581 for all indicating both tenure (year when began contributing) and country or territory of residence. A Fisher’s exact test found new contributors more likely to live in Africa (3.0%) than tenured contributors (1.2%) (p = 0.044). A Pearson χ2 test also found new contributors more likely to live in Asia (28.0% vs. 14.9%, χ2 = 24.607, p = 0.000) and Latin America or the Caribbean (12.3% vs. 3.6%, χ2 = 31.992, p = 0.000), and less likely to live in Europe (45.3% vs. 60.7%, χ2 = 19.499, p = 0.000) and Northern America (9.7% vs. 17.2%, χ2 = 8.194, p = 0.004) than more tenured contributors. Even though Oceania includes Melanesia, Micronesia, and Polynesia, the only respondents from this region lived in Australia and New Zealand. These analyses are weighted by editing activity and home project.
  2. For newcomers, 153 reported their country or territory of residence in 2019, and 236 did so in 2020. Changes in proportion residing on each continent between 2019 and 2020 are reported here, though none of these changes was statistically significant. The proportion of those living in Africa increased from 1.3% to 3.0% (Fisher's exact p = 0.242), in Asia fell from 36.6% to 28.0% (χ2 = 3.215, p = 0.073), in Europe increased from 41.2% to 45.3% (χ2 = 0.654, p = 0.419), in Latin America and the Caribbean increased from 7.8% to 12.3% (χ2 = 1.945, p = 0.163), in Northern America fell from 13.1% to 9.7% (χ2 = 1.044, p = 0.307), and in Oceania rose from 0.0% to 2.1% (Fisher's exact p = 0.082). For tenured contributors, 769 reported their country or territory of residence in 2019, and 1345 did so in 2020. The proportion of those living in Africa increased from 0.7% to 1.2% (this was not statistically significant at χ2 = 1.447, p = 0.229), in Asia rose from 14.3% to 14.9% (this was not statistically significant at χ2 = 0.117, p = 0.732), in Europe increased from 53.6% to 60.7% (this was statistically significant at χ2 = 9.993, p = 0.002), in Latin America and the Caribbean fell from 4.8% to 3.6% (this was not statistically significant at χ2 = 1.711, p = 0.191), in Northern America fell from 24.3% to 17.2% (this was statistically significant at χ2 = 15.801, p = 0.000), and in Oceania rose from 2.2% to 2.5% (χ2 = 0.125, p = 0.724). These analyses are weighted by editing activity and home project.
  3. n = 1634 for contributors responding to a question about rural/urban location, and 36.4% said they live in the country or in a small city or town (95%CI: 34.1%, 38.8%). In 2020, newcomers were more likely to be under the age of 25 (27.2%, n = 232) than tenured contributors (12.0%, n = 1344) (χ2 = 37.371, p = 0.000), but newcomers in 2020 were less likely to be under 25 (27.2%, n = 232)  than they were in 2019 (37.3%, n = 134) (χ2 = 4.107, p = 0.043). n = 1559 for contributors indicating their level of education. Newer contributors were less likely (75.8%, n = 227) to have completed a college or post-secondary degree than tenured contributors (88.7%, n = 1332) (χ2 = 28.559, p = 0.000).
  4. Gender was reported by 995 contributors in 2019 and 1629 contributors in 2020. The proportion identifying as women increased from 11.8% to 14.9% (χ2 = 5.205, p = 0.023). For newcomers (162 reporting in 2019, 234 in 2020), the proportion increased from 22.2% to 28.6%, though this change was not statistically significant (χ2 = 2.044, p = 0.153). For tenured contributors (747 in 2019, 1329 in 2020), the proportion of women increased from 9.1% to 12.4% (χ2 = 5.266, p = 0.022). Among 2020 contributors, 1563 of whom reported both their gender and tenure, newcomers (28.6%) were more likely to be women than tenured editors (12.4%) (χ2 = 41.398, p = 0.000). Results are weighted by editing activity and home project.
  5. Among those living in Africa, gender was reported by 20 contributors in 2019 and 22 contributors in 2020. The proportion identifying as women increased from 5.0% to 31.8% (Fisher's Exact p = 0.032). Among those living in Latin America, gender was reported by 193 contributors in 2019 and 75 contributors in 2020, and the proportion identifying as women increased from 9.3% to 22.7% (χ2 = 8.466, p = 0.004). Among those living in Northern America, gender was reported by 164 contributors in 2019 and 235 contributors in 2020, and the proportion identifying as women increased from 12.8% to 22.1% (χ2 = 5.616, p = 0.018). Among contributors under the age of 45, 861 reported their gender in 2019 and 832 did so in 2020, and the proportion identifying as women increased from 11.1% to 15.5% (χ2 = 6.964, p = 0.008). Results are weighted by editing activity.
  6. Gender was reported by 995 contributors in 2019 and 1629 contributors in 2020. The proportion identifying as transgender or non-binary increased from 1.9% to 2.6% (χ2 = 1.356, p = 0.244). The proportion identifying as men decreased from 86.5% to 83.7% (χ2 = 3.757, p = 0.053). Results are weighted by editing activity and home project.
  7. US Population proportions are from the most recent American Community Survey (2017) by the US Census Bureau. Race and ethnicity was reported by 195 contributors living in the United States at the time of response and is weighted by editing activity and home project, inclusive of unweighted movement organizers for whom editing activity is unknown. 95% Confidence intervals are as follows: Black or African American (0.0%, 1.5%); Hispanic or Latinx (2.1%, 9.2%); American Indian or Alaskan Native (0.0%, 0.7%); Asian (4.8%, 12.8%); White or Caucasian (84.1%, 93.1%). British population proportions are from the 2011 UK Census. Ethnicity was reported by 67 contributors living in Britain at the time of response and is weighted by editing activity and home project, inclusive of unweighted movement organizers for whom editing activity is unknown. 95% confidence intervals are as follows: Asian or Asian British (0.0%, 6.1%); Black or Black British (0.0%, 0.0%); White or White British (74.3%, 92.6%); mixed or multiple (0.0%, 6.1%), other (0.0%, 6.1%).
  8. n = 226 for new contributors responding to “easy to learn editing policies” with an average score of 3.63 (95% CI: 3.49, 3.76). n = 232 for “easy to learn the editing tools” with an average score of 3.88, (95% CI: 3.76, 4.01). n = 227 for “I feel like I understand the rules” with an average score of 3.93 (95% CI: 3.81, 4.04). n = 232 for “learning to edit has been easy” with an average score of 3.91 (95% CI: 3.79, 4.02). n = 214 for “other editors have been kind” with an average score of 3.82 (95% CI: 3.69, 3.94). n = 218 for “other editors have been helpful” with an average score of 3.79 (95% CI: 3.65, 3.93). The Newcomer Empowerment Score is calculated by averaging responses to the six index questions (or those index questions responded to), standardized to a scale of 1 to 5. A score was calculated for 233 new contributors who answered any combination of the six index questions (average score was 3.83, 95% CI: 3.73, 3.92). For the following statistical tests, group means are reported for clarity even though assumptions of normality and outliers are violated. A Mann-Whitney U-test is used in each case to identify group differences. Agreement with "It has been easy to learn the editing tools" increased from an average score of 3.65 in 2019 (n = 157) to 3.88 in 2020 (n = 232) (mean rank of 177.55 vs. 153.42, Mann-Whitney U = 15129.5, p = 0.016). Agreement with "It has been easy to learn editing policies" increased from an average score of 3.33 in 2019 (n = 149) to 3.63 in 2020 (n = 226) (mean rank of 171.98 vs. 146.38, Mann-Whitney U = 14174.5, p = 0.012). These analyses are weighted by editing activity and home project.
  9. For the following statistical tests, group means are reported for clarity even though assumptions of normality and outliers are violated. A Mann-Whitney U-test is used in each case to identify group differences. Newcomer empowerment (an index of all six measures on a 1-5 scale, where 5 is the most favorable) increased from an average of 3.63 in 2019 (n = 154) to 3.82 in 2020 (n = 237) (mean rank of 152.04 vs. 182.85, Mann-Whitney U = 16060, p = 0.005). Analysis weighted by editing activity and home project. For newcomers living in emerging geographies (103 in 2019, 91 in 2020) the average newcomer empowerment score increased from 3.78 to 3.99 (mean rank of 96.81 vs. 81.42, Mann-Whitney U = 4520.5, p = 0.044). Analysis weighted by editing activity. For newcomers who are women (56 in 2019, 65 in 2020), the average newcomer empowerment score increased from 3.50 to 3.83 (mean rank of 58.39 vs. 45.63, Mann-Whitney U = 1674, p = 0.031). Analysis weighted by editing activity. For newcomers living in East Asia, (35 in 2019, 21 in 2020), the average newcomer empowerment score increased from 3.36 to 3.78 (mean rank of 32.44 vs. 21.92, Mann-Whitney U = 398.5, p = 0.015). Analysis weighted by editing activity.
  10. For the following statistical tests, group means are reported for clarity even though assumptions of normality and outliers are violated. A Mann-Whitney U-test is used in each case to identify group differences. Newcomers on the Spanish Wikipedia had higher empowerment scores (4.23, n = 26) than other newcomers (3.77, n = 204) (mean rank of 131.89 vs. 98.37, Mann-Whitney U = 2648.5, p = 0.011). Newcomers living in emerging geographic areas had higher empowerment scores (3.99, n = 91) than other newcomers (3.69, n = 132) (mean rank of 113.01 vs. 89.22, Mann-Whitney U = 5833, p = 0.004). Those living in Latin America or the Caribbean had higher empowerment scores (4.14, n = 25) than other contributors (3.77, n = 198) (mean rank of 124.33 vs. 95.98, Mann-Whitney U = 2380, p = 0.031). Those under the age of 25 had higher empowerment scores (3.96, n = 61) than other contributors (3.75, n = 158) (mean rank of 109.84 vs. 91.88, Mann-Whitney U = 4501, p = 0.043). Those living in Western Europe had lower empowerment scores (3.60, n = 44) than other contributors (3.86, n = 179) (mean rank of 78.94 vs. 103.95, Mann-Whitney U = 2298.5, p = 0.014). Those who contribute to Wikimedia Commons had lower empowerment scores (3.38, n = 29) than other contributors (3.88, n = 201) (mean rank of 77.14 vs. 105.49, Mann-Whitney U = 1603.5, p = 0.023).
  11. The Belonging index is an averaged score of disagreement or agreement with two statements: (1) I feel like I belong in the Wikimedia movement; and (2) I feel respected in the Wikimedia movement. Those who finished less than 50% of the survey (n = 36) had a lower average belonging score of 3.38, vs. 3.70 for those completing 50% or more (n = 1672) (mean rank of 602.87 vs. 842.36, Mann-Whitney U = 35857.5, p = 0.003), though this effect size was small (d = -0.376). In 2019, a Belonging index score was calculated for 1150 contributors answering these questions, and for 1708 contributors in 2020. For the following statistical tests, group means are reported for clarity even though assumptions of normality and outliers are violated. A Mann-Whitney U-test is used in each case to identify group differences. Between 2019 and 2020, the average belonging score rose from 3.50 to 3.70, and this difference was statistically significant (mean rank of 1323.53 vs. 1122.59, Mann-Whitney U = 813618.5, p = 0.000). This analysis is weighted by editing activity and home project. In 2020, contributors living in South Asia had higher belonging scores (4.14, n = 34) than other contributors (3.71, n = 1571) (mean rank of 994.14 vs. 767.15, Mann-Whitney U = 32245.5, p = 0.003). This analysis is weighted by editing activity. In 2020, movement organizers had higher belonging scores (4.10, n = 281) than other contributors (3.66, n = 1545) (mean rank of 1111.26 vs. 842.51, Mann-Whitney U = 285790, p = 0.000). This analysis is weighted by editing activity and home project inclusive of unweighted organizers for whom editing activity is unknown. In 2020, contributors living in West Asia had higher belonging scores (4.06, n = 53) than other contributors (3.70, n = 1553) (mean rank of 961.19 vs. 765.40, Mann-Whitney U = 48604, p = 0.001). This analysis is weighted by editing activity. In 2020, on-wiki administrators had higher belonging scores (3.98, n = 235) than other contributors (4.14, n = 1774) (mean rank of 978.02 vs. 813.90, Mann-Whitney U = 194628.5, p = 0.000). This analysis is weighted by editing activity and home project. In 2020, contributors who had ever participated in a movement event had higher belonging scores (3.94, n = 510) than other contributors (3.60, n = 1202) (mean rank of 955.34 vs. 760.79, Mann-Whitney U = 363908, p = 0.000). These contributors also had higher belonging scores than event attendees indicated in 2019 (3.85, n = 395) (mean rank of 389.69 vs. 356.39, Mann-Whitney U = 66378, p = 0.047). These analyses are weighted by editing activity. In 2020, contributors who lived in East Asia had lower belonging scores (3.53, n = 124) than other contributors (3.73, n = 1482) (mean rank of 643.58 vs. 782.15, Mann-Whitney U = 66284, p = 0.001). However, these contributors also had higher belonging scores than those in East Asia indicated in 2019 (3.29, n = 132) (mean rank of 120.88 vs. 102.88, Mann-Whitney U = 7218.5, p = 0.034). These analyses are weighted by editing activity. In 2020, contributors to the Japanese Wikipedia had lower belonging scores (3.36, n = 70) than other contributors (3.72, n = 1664) (mean rank of 616.29 vs. 842.18, Mann-Whitney U = 37362.5, p = 0.000). However, these contributors also had higher belonging scores than ja.wiki contributors indicated in 2019 (2.86, n = 56) (mean rank of 66.34 vs. 50.14, Mann-Whitney U = 2165.5, p = 0.009). These analyses are weighted by editing activity.
  12. The Engagement index is an averaged score of disagreement or agreement with five statements: (1) I would recommend Wikimedia projects as a great place to contribute; (2) The vision of the Wikimedia movement is to freely share in the sum of all knowledge. This vision inspires me to contribute more to Wikimedia projects than I would somewhere else; (3) I take pride in contributing to the Wikimedia projects; (4) I often think about quitting as a contributor to Wikimedia projects (scale reversed); and (5) I expect to be contributing to Wikimedia projects two years from now. Those who finished less than 50% of the survey (n = 307) had a lower average score of 3.98, vs. 4.17 for those completing 50% or more (n = 1752) (mean rank of 867.56 vs. 1024.21, Mann-Whitney U = 282278, p = 0.000), though this effect size was small (d = -0.277). For the following statistical tests, group means are reported for clarity even though assumptions of normality and outliers are violated. A Mann-Whitney U-test is used in each case to identify group differences. An Engagement Index score was calculated for 1425 people responding to these questions in 2019 and for 2059 respondents in 2020. Between 2019 and 2020, the average engagement score rose from 4.00 to 4.14 (mean rank of 1390.29 vs. 1601.99, Mann-Whitney U = 1201780, p = 0.000). This analysis is weighted by editing activity and home project. In 2020, contributors living in Africa had higher engagement scores (4.59, n = 22) than other contributors (4.16, n = 1654) (mean rank of 1092.88 vs. 799.67, Mann-Whitney U = 22719.5, p = 0.004). Trends were similar among all subcontinental regions. This analysis is weighted by editing activity. In 2020, movement organizers had higher engagement scores (4.41, n = 301) than other contributors (4.13, n = 1806) (mean rank of 1242.04 vs. 989.72, Mann-Whitney U = 342399.5, p = 0.000). This analysis is weighted by editing activity and home project inclusive of unweighted organizers for whom editing activity is unknown. In 2020, contributors living in Latin America and the Caribbean had higher engagement scores (4.36, n = 79) than other contributors (4.16, n = 1596) (mean rank of 946.78 vs. 796.58, Mann-Whitney U = 67286.5, p = 0.006). These contributors also had higher engagement scores than contributors in Latin America indicated in 2019 (4.04, n = 202) (mean rank of 147.76 vs. 112.29, Mann-Whitney U = 8159, p = 0.000). Trends were similar among all subcontinental regions. These analyses are weighted by editing activity. In 2020, contributors living in Southern Europe had higher engagement scores (4.35, n = 165) than other contributors (4.15, n = 1510) (mean rank of 928.61 vs. 789.85, Mann-Whitney U = 134159, p = 0.000). These contributors also had higher engagement scores than contributors in Southern Europe indicated in 2019 (4.13, n = 184) (mean rank of 162.94 vs. 128.34, Mann-Whitney U = 13184, p = 0.000). These analyses are weighted by editing activity. In 2020, contributors to the Spanish Wikipedia had higher engagement scores (4.34, n = 84) than other contributors (4.13, n = 1996) (mean rank of 1137.18 vs. 982.55, Mann-Whitney U = 83500, p = 0.020). This analysis is weighted by editing activity. In 2020, contributors to the French Wikipedia had higher engagement scores (4.28, n = 140) than other contributors (4.13, n = 1941) (mean rank of 1157.32 vs. 976.32, Mann-Whitney U = 145012.5, p = 0.000). These contributors also had higher engagement scores than contributors to the French Wikipedia indicated in 2019 (4.18, n = 106) (mean rank of 118.03 vs. 97.58, Mann-Whitney U = 6787.5, p = 0.019). These analyses are weighted by editing activity. In 2020, contributors who had attended movement events had higher engagement scores (4.26, n = 561) than other contributors (4.10, n = 1441) (mean rank of 1057.24 vs. 909.32, Mann-Whitney U = 442797, p = 0.000). This analysis is weighted by editing activity. In 2020, contributors living in emerging geographies had higher engagement scores (4.25, n = 413) than other contributors (4.14, n = 1263) (mean rank of 851.78 vs. 787.86, Mann-Whitney U = 257327.5, p = 0.017). These contributors also had higher engagement scores than contributors in emerging geographies indicated in 2019 (4.07, n = 533) (mean rank of 441.23 vs. 371.83, Mann-Whitney U = 95982, p = 0.000). These analyses are weighted by editing activity.
  13. All analyses weighted by editing activity. For the following statistical tests, group means are reported for clarity even though assumptions of normality and outliers are violated. A Mann-Whitney U-test is used in each case to identify group differences. In 2020, contributors living in Eastern Europe had lower engagement scores (4.09, n = 205) than other contributors (4.18, n = 1470) (mean rank of 730.69 vs. 813.8, Mann-Whitney U = 125506.5, p = 0.017). However, these contributors had higher engagement scores than contributors in Eastern Europe indicated in 2019 (3.98, n = 242) (mean rank of 197.83 vs. 174.61, Mann-Whitney U = 19468, p = 0.037). In 2020, contributors who were not fluent in English had lower engagement scores (4.07, n = 310) than other contributors (4.18, n = 1395) (mean rank of 746.29 vs. 834.45, Mann-Whitney U = 219693.5, p = 0.003). In 2020, contributors to the Russian Wikipedia had lower engagement scores (4.02, n = 109) than other contributors (4.14, n = 1972) (mean rank of 873.29 vs. 994.90, Mann-Whitney U = 85362.5, p = 0.034). However, these contributors had higher engagement scores than contributors to ru.wikipedia indicated in 2019 (3.83, n = 102) (mean rank of 101.07 vs. 85.14, Mann-Whitney U = 5051.5, p = 0.044). In 2020, contributors living in East Asia had lower engagement scores (4.01, n = 129) than other contributors (4.18, n = 1546) (mean rank of 647.60 vs. 815.75, Mann-Whitney U = 68866, p = 0.000). In 2020, contributors to the German Wikipedia had lower engagement scores (3.96, n = 292) than other contributors (4.17, n = 1789) (mean rank of 1011.54 vs. 844.75, Mann-Whitney U = 193216.5, p = 0.000). However, these contributors had higher engagement scores than contributors to de.wikipedia indicated in 2019 (3.76, n = 136) (mean rank of 198.68 vs. 171.58, Mann-Whitney U = 16839.5, p = 0.030). In 2020, contributors to the Japanese Wikipedia had lower engagement scores (3.86, n = 76) than other contributors (4.15, n = 2005) (mean rank of 679.56 vs. 999.51, Mann-Whitney U = 43864, p = 0.000).
  14. The following analyses are weighted by editing activity and home project. n = 1623 for all respondents indicating both role and gender, n = 183 for movement organizers sharing this information. 30.1% of movement organizers are women, compared to 13.0% of non-organizing contributors (χ2 = 37.283, p = 0.000), and 4.8% are transgender or non-binary, compared with 2.3% of non-organizing contributors (Fisher's exact p = 0.044). n = 1637 for all indicating both role and age, n = 186 for movement organizers sharing this information. 9.1% of movement organizers are under 25, compared to 14.8% of non-organizing contributors (χ2 = 4.369, p = 0.037). n = 1641 for all indicating both role and country or territory of residence, n = 178 for movement organizers sharing this information. Organizers are more likely to live in Africa (6.2%) than other contributors (0.8%) (χ2 = 35.348, p = 0.000), as well as in South Asia (4.5% vs. 1.9%, Fisher's exact p = 0.034). They are less likely to live in Europe (50.6%) than other contributors (59.2%) (χ2 = 4.895, p = 0.027).
  15. n = 178 for administrators indicating where they live. A Pearson χ2 test found administrators more likely to live in Southern Asia (4.5% compared to 1.9%, χ2 = 4.925, p = 0.026) than non-administrators.
  16. All analyses are weighted by editing activity and home project. In 2020, n = 193 for those responding to "When I am presented with a challenge in my on-wiki role, I usually know how to address it" (4.07, 95% CI: 3.98, 4.16). n = 192 for organizers responding to "...I am usually able to address it in the way I think is most appropriate" (4.16, 95% CI: 4.08, 4.24). n = 191 for those responding to "I have access to the right tools or software to implement my..." (3.96, 95% CI: 3.85, 4.07). The Admin Empowerment Index is an averaged score of disagreement or agreement with these three statements. The average Index score, computed for 193 respondents with on-wiki administrative roles, was 4.07 (95% CI: 3.99, 4.14). The average admin empowerment index score was 4.00 in 2019 (n = 84), 4.07 in 2020 (n = 193). This difference was not statistically significant (Mann-Whitney U = 4617.5, p = 0.438). No statistically significant differences were found in administrator empowerment between demographic groups in 2020, nor for demographic groups over time from 2019 to 2020.
  17. All analyses are weighted by editing activity and home project, inclusive of unweighted organizers for whom editing activity is unknown. For the following statistical tests, group means are reported for clarity even though assumptions of normality and outliers are violated. A Mann-Whitney U-test is used in each case to identify group differences. In 2020, n = 270 for those responding to "I have people who can help me with my..." (4.14, 95% CI: 4.04, 4.24). n = 263 for organizers responding to "I know whom to ask for help with my Wikimedia program or partnership" (4.03, 95% CI: 3.92, 4.13). n = 258 for those responding to "I have access to the right tools or software to implement my..." (3.74, 95% CI: 3.63, 3.86). n = 255 for those responding to "I have access to documentation and training materials to help me implement my..." (3.77, 95% CI: 3.66, 3.88). n = 256 for those responding to "I have access to communications materials and tools that help me reach the people I want to engage with my..." (3.72, 95% CI: 3.59, 3.84). n = 250 for those responding to "I have access to funding that I need to implement my..." (3.49, 95% CI: 3.34, 3.65). n = 255 for those responding to "I have access to the non-monetary resources (e.g., event space) that I need for implementing my..." (3.58, 95% CI: 3.44, 3.71). The Organizer Empowerment Index is an averaged score of disagreement or agreement with these six statements. The average Index score, computed for 271 organizers, was 3.78 (95% CI: 3.69, 3.87). Overall, organizer empowerment increased from an average of 3.64 in 2019 (n = 200) to 3.78 in 2020 (n = 271) (mean rank of 208.57 vs. 238.62, Mann-Whitney U = 27204.5, p = 0.018). Agreement with "I have access to funding that I need to implement my events, campaigns, groups, or partnerships" increased from an average of 3.09 in 2019 (n = 182) to 3.49 in 2020 (n = 250) (mean rank of 184.59 vs. 223.15, Mann-Whitney U = 23930.5, p = 0.001).
  18. All analyses are weighted by editing activity, inclusive of unweighted organizers for whom editing activity is unknown. For the following statistical tests, group means are reported for clarity even though assumptions of normality and outliers are violated. A Mann-Whitney U-test is used in each case to identify group differences. In 2020, organizers living in Africa had a higher organizer empowerment score (4.21, n = 37) than other organizers (3.75, n = 232) (mean rank of 183.08 vs. 132.81, Mann-Whitney U = 3071, p = 0.000). These organizers also had higher empowerment in 2020 than they did in 2019 (3.82, n = 27) (mean rank of 36.69 vs. 25.33, Mann-Whitney U = 654.5, p = 0.015). Organizers living in Western Europe had higher empowerment in 2020 than in 2019 (3.82, n = 65 in 2020; 3.54, n = 31 in 2019; mean rank of 44.93 vs. 31.30, Mann-Whitney U = 685.5, p = 0.049). In 2020, organizers living in emerging geographies had a higher organizer empowerment score (3.87, n = 121) than they did in 2019 (3.50, n = 134) (mean rank of 135.20 vs. 108.18, Mann-Whitney U = 8866, p = 0.004). Movement organizers in Asia had lower empowerment scores (3.67, n = 53) than other organizers (3.85, n = 216) (mean rank of 119.18 vs. 144.4, Mann-Whitney U = 4950.5, p = 0.038).
  19. The following analyses are weighted by editing activity and home project, inclusive of unweighted organizers for whom editing activity is unknown. For the following statistical tests, group means are reported for clarity even though assumptions of normality and outliers are violated. n = 249 for organizers responding to "The Wikimedia Foundation provides the help and support I need to plan and execute my projects(s)" (3.57, 95% CI: 3.43, 3.70). n = 236 for those responding to "My Wikimedia community or affiliate provides the help and support I need to plan and execute my projects(s)" (3.97, 95% CI: 3.86, 4.09). While there was no change over time for organizers regarding the Foundation's provision of support, organizers were more likely in 2020 to say that their community or affiliate did so than they were in 2019 (n = 194, 3.74) (mean rank of 223.77 vs. 198.33, Mann-Whitney U = 24136.5, p = 0.026). The following analyses are weighted by editing activity and home project. n = 133 for admins responding to "If I contact them, the Wikimedia Foundation usually provides the help and support I need to execute my on-wiki role" (3.12, 95% CI: 2.92, 3.32). n = 189 for admins responding to "Other volunteers usually provide the help and support I need to execute my on-wiki role" (3.95, 95% CI: 3.83, 4.07).
  20. The following analyses are weighted by editing activity, inclusive of unweighted organizers for whom editing activity is unknown. For the following statistical tests, group means are reported for clarity even though assumptions of normality and outliers are violated. Organizers living in emerging geographies (73.1% favorable, average of 3.87, n = 115) were more likely than others (50.0% favorable, average of 3.30, n = 119) to say that the Wikimedia Foundation provides the support they need (mean rank of 141.63 vs. 102.97, Mann-Whitney U = 9643, p = 0.000). This was also the case for those living in Africa (83.7% favorable, average of 4.28, n = 36) compared to others (52.2% favorable, average of 3.45, n = 198) (mean rank of 170.21 vs. 112.99, Mann-Whitney U = 5461.5, p = 0.000); whereas those living in Western Europe (47.9% favorable, average of 3.25, n = 48) were less than others (64.8% favorable, average of 3.66, n = 186) to indicate Foundation support (mean rank of 100.35 vs. 127.15, Mann-Whitney U = 3792, p = 0.010). Among admins, those living in Asia were more likely to say that the Foundation provides the support they need (73% favorable, average of 3.59, n = 25) than those elsewhere (35.2% favorable, average of 3.08, n = 105) (mean rank of 88.59 vs. 66.83, Mann-Whitney U = 2014, p = 0.010), as were admins under the age of 25 (65.0% favorable, average of 3.60, n = 20) compared to others (38.6% favorable, average of 3.11, n = 114) (mean rank of 90.45 vs. 70.49, Mann-Whitney U = 1737, p = 0.034).
  21. The following analyses are weighted by editing activity, inclusive of unweighted organizers for whom editing activity is unknown. For the following statistical tests, group means are reported for clarity even though assumptions of normality and outliers are violated. Organizers living in Africa (91.7% favorable, average of 4.35, n = 36) were more likely than others (69.4% favorable, average of 3.96, n = 148) to say that their community or affiliate provides the support they need (mean rank of 154.03 vs. 123.72, Mann-Whitney U = 4879, p = 0.013); whereas organizers living in Asia (68.7% favorable, average of 3.82, n = 50) were less likely than others (81.3% favorable, average of 4.07, n = 196) to say that their community or affiliate provides the support they need (mean rank of 109.86 vs. 132.53, Mann-Whitney U = 4277, p = 0.032). Organizers who are women were more likely to say their community or affiliate provides this support in 2020 (84.9% favorable, 4.07 average, n = 72) than they were in 2019 (67.1% favorable, average of 3.68, n = 78) (mean rank of 82.84 vs. 66.38, Mann-Whitney U = 3346.5, p = 0.011). The same was true for those in emerging geographies in 2020 (81.5% favorable, 3.99 average, n = 129 compared to 2019 (69.0% favorable, average of 3.72, n = 126) (mean rank of 119.85 vs. 103.91, Mann-Whitney U = 6983.5, p = 0.049). The following analyses are weighted by editing activity. For the following statistical tests, group means are reported for clarity even though assumptions of normality and outliers are violated. Admins not fluent in English (50.0% favorable, average of 3.36, n = 20) were less likely than others (82.5% favorable, average of 4.03, n = 177) to say that other volunteers provide the support they need (mean rank of 70.02 vs. 111.26, Mann-Whitney U = 2914.5, p = 0.000). Admins living in Asia were more likely in 2020 (93.2% favorable, average of 4.16, n = 31) than in 2019 (71.9% favorable, average of 3.69, n = 32) to say that other volunteers provided the support they needed (mean rank of 28.82 vs. 20.83, Mann-Whitney U = 390, p = 0.028).
  22. The following analyses are weighted by editing activity and home project. n = 1705 for all contributors responding to "In the last 12 months, how often have you felt unsafe or uncomfortable contributing to Wikimedia projects online?" 34.2% said they had this experience at least once (95% CI: 31.9%, 36.4%). A Pearson χ2 test found no difference in the proportion of contributors reporting harassment between 2019 (18.5%) and 2020 (n = 1133 for 2019, χ2 = 2.347, p = 0.126). The following analyses are weighted by editing activity unless otherwise noted. Contributors to South Asian language projects were more likely to report feeling unsafe (60.9%, n = 23) than other contributors (34.5%, n = 1711) (χ2 = 6.961, p = 0.008); as were contributors to the Chinese language Wikipedia (52.5%, n = 40 vs. 34.4%, n = 1693, χ2 = 5.657, p = 0.017), contributors who are gender minorities (51.2%, n = 43 vs. 33.7%, n = 1563, χ2 = 5.661, p = 0.017), contributors to the Russian language Wikipedia (45.5%, n = 88 vs. 34.2%, n = 1644, χ2 = 4.624, p = 0.032), the German language Wikipedia (43.6%, n = 234 vs. 33.5%, n = 1500, χ2 = 9.138, p = 0.003), and (weighted by editing activity inclusive of unweighted organizers for whom editing activity is unknown) movement organizers (43.4%, n = 272 vs. 32.6%, n = 1522, χ2 = 11.942, p = 0.000). Contributors to the English language Wikipedia were less likely to report feeling unsafe (30.0%, n = 323) than other contributors (35.9%, n = 1411) (χ2 = 4.032, p = 0.045); as were contributors living in Northern America (28.2%, n = 234 vs. 35.3%, n = 1366, χ2 = 4.447, p = 0.035), contributors living in Northern Europe (27.1%, n = 155 vs. 35.0%, n = 1445, χ2 = 3.900, p = 0.048) and contributors to the Japanese language Wikipedia (20.9%, n = 67 vs. 35.4%, n = 1666, χ2 = 5.980, p = 0.014). Contributors to ja.wikipedia were also less likely to report having felt unsafe than they were in 2019 (40.0%, n = 55, χ2 = 5.300, p = 0.021). While women were no more likely to feel unsafe overall, they were among contributors to the English language Wikipedia (42.3%, n = 52 vs. 27.2%, n = 243, χ2 = 4.695, p = 0.030) and among those living in Northern America (39.6%, n = 48 vs. 24.7%, n = 178, χ2 = 4.155, p = 0.042).
  23. The following analyses are weighted by editing activity and home project. n = 1701 for all contributors responding to "In the last 12 months, how often have you been harassed, bullied, or attacked in any Wikimedia online or offline space?" 20.9% said they had this experience at least once (95% CI: 19.0%, 22.8%). A Pearson χ2 test found no difference in the proportion of contributors reporting having felt unsafe between 2019 (36.5%) and 2020 (n = 1088 for 2019, χ2 = 1.617, p = 0.204). The measure used in 2020 was slightly different from the one used in 2019 (respondents were asked if they had experienced harassment, not how often), so where year-over-year differences are suspected to be a result of measurement change it will be noted. Contributors in Oceania were more likely to say they had been harassed than others (37.8%, n = 37 vs. 21.0%, n = 1572, χ2 = 6.103, p = 0.013), as were movement organizers (weighted by editing activity inclusive of unweighted organizers for whom editing activity is unknown) (29.7%, n = 269 vs. 19.5%, n = 1526, χ2 = 14.345, p = 0.000) and contributors to the English Wikipedia  (28.0%, n = 321 vs. 20.0%, n = 1408, χ2 = 9.930, p = 0.002). Contributors in Southern Europe were less likely to say they had experienced harassment  (15.3%, n = 163 vs. 22.1%, n = 1446, χ2 = 3.940, p = 0.047).
  24. The measure of harassment used in 2020 was slightly different from the one used in 2019 (respondents were asked if they had experienced harassment, not how often), so where year-over-year differences are suspected to be a result of measurement change it will be noted. The following analyses are weighted by editing activity. Women contributors were more likely to report having experienced harassment in 2020 (24.2%, n = 236) than they were in 2019 (13.0%, n = 177, χ2 = 8.063, p = 0.005) though 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. Contributors to Wikidata were also more likely to report having experienced harassment in 2020 (23.9%, n = 46) than they were in 2019 (8.5%, n = 59, χ2 = 4.770, p = 0.029) though the difference in "unsure" responses to this question between 2019 (9.2%) and 2020 (4.2%) 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. Contributors in Asia were more likely to report having experienced harassment in 2020 (22.0%, n = 268) than they were in 2019 (12.4%, n = 299, χ2 = 9.339, p = 0.002). Contributors in Europe were more likely to report having experienced harassment in 2020 (19.8%, n = 978) than they were in 2019 (11.8%, n = 905, χ2 = 22.474, p = 0.001).