Research:Knowledge Gaps Index/Measurement/Readers Survey 2024
This project aims to understand the demographics and motivations of Wikipedia Readers across language editions. It is part of the Knowledge Gaps Index focus on Readers of Wikipedias, and continues the work of the 2019 and 2023 Readers surveys.
Progress on this project can be followed at T369087.
Key Takeaways
editThis section offers a high-level summary of 2024 Global Reader Survey findings. For the full results please check the Results section.
Survey Administration
editThe survey reached a simple random sample of desktop and mobile web readers of 11 Wikipedia projects (arwiki, dewiki, enwiki, eswiki, frwiki, idwiki, jawiki, kowiki, ptwiki, ruwiki, zhwiki). Surveys were translated to all 11 languages (with both simplified and traditional Chinese editions) and were fielded from November 25—December 19, 2024. A total of 25,929 complete survey responses were collected.
Demographics
editAge
edit- Consistent with findings from 2023, Wikipedia readers skew young overall, although this can vary substantially by project. German Wikipedia readers in particular tend to skew older. In addition, when we compare the share of readers age 18-24 for projects surveyed in 2019, 2023, and 2024, we observe significant declines in French, German, and Spanish Wikipedia.
Gender
edit- Wikipedia readers across all projects identify disproportionately as solely men. By project, readership of kowiki, frwiki, ruwiki, and enwiki are closest to gender parity, but readership on all surveyed projects is over 60% composed of those identifying exclusively as men. In addition, the share of readers identifying solely as men increased significantly from 2023—2024 (both overall and in 7 of 11 projects surveyed in both years).
Education
edit- Wikipedia readers are highly-educated. Readership skews heavily towards the university-educated (or those currently enrolled in Bachelor's or postgraduate degree programs).
Language
edit- Wikipedia readers are highly multilingual: about 35% of readers claim fluency in two or more languages. However, readers of any given Wikipedia are overwhelmingly reading in a primary language. Readers of enwiki are most likely to be non-native language speakers. At the same time, English-language fluency is common among readers of projects other than enwiki, ranging from 22.5% on zhwiki to 50.0% on frwiki.
Reader Motivations and Information Needs
edit- When readers are asked why they are reading the Wikipedia article they were sampled from, they are most likely to say its topic is personally important to them, followed by reading out of boredom or exploration for fun. Readers were also likely to say the article topic was referenced in a piece of media they consumed. Readers were comparatively less likely to say they were reading Wikipedia for a work or school-related assignment or to inform a personal decision.
- Wikipedia readers are more likely to say they are are reading to "get an overview of the topic" rather than to "look up a specific fact or to get a quick answer" or to "get an in-depth understanding of the topic". However, each use case is relatively common.
- Wikipedia readers are more likely to say they are already familiar with the topic of the article they are reading than that they're learning about it for the first time.
- Over 80% of Wikipedia readers said they were able to find the information they were looking for on Wikipedia. However, readers of kowiki and especially arwiki, were less likely to say they were successful in finding the information they sought.
Accessing Knowledge Online
editOther platforms for learning and accessing knowledge online
editWhen Wikipedia readers are asked where else they go online to learn or access knowledge:
- Readers are overwhelmingly most likely to visit YouTube and Google (or another search engine)
- Reddit and ChatGPT are the next most-cited online destinations for learning and knowledge, particularly for readers aged 18-24.
- Even youth readers are substantially less likely to cite social media, whether text- (X/Twitter), image- (Instagram), or video-based (TikTok) as destinations for learning and knowledge.
- While "local" apps and platforms have varying levels of popularity for learning and knowledge across the surveyed Wikipedias, YouTube and Google are always much more likely to be named by readers, regardless of the Wikipedia.
Platform favorability ratings
editWhen Wikipedia readers are asked to give favorability ratings (on a scale of 0 to 10) for platforms they say they use to learn and access knowledge:
- Overall, Wikipedia is rated more favorably than any other platform.
- Wikipedia is the most-favorably rated platform for every age group of readers.
- There is substantial variation in favorability ratings by project, however. Notably, arwiki readers rate Wikipedia less favorably in absolute terms (mean rating) as well as in comparison to other platforms like Google and ChatGPT.
Wikipedia Use Frequency
editWhen asked about their use of Wikipedia:
- Almost two-thirds of readers say how much they use Wikipedia hasn't changed much in the past year.
- But readers are more than twice as likely to say they've increased rather than decreased their use of Wikipedia over the past year.
- More than three-in-four readers say they visit Wikipedia at least "a few times a week", while more than one-third say they visit at least daily.
Wikipedia Experience and Knowledge
editWhen we asked readers about their experience with editing and donating to Wikipedia:
- A supermajority (about 70%) of readers say they have never made a Wikipedia edit.
- Just 8% say they have edited more than "one or two times".
- About two-in-five of dewiki, enwiki, eswiki, frwiki, jawiki, kowiki, and ptwiki readers say they have donated at least once to Wikipedia or the Wikimedia Foundation.
We measured readers' factual knowledge about Wikipedia with a ten-item true/false battery.
- A substantial minority (about 30%) of readers were highly knowledgeable, answering correctly on nine or more of the questions.
- Conversely, a similar proportion (about 25%) of readers did no better than chance (5 or fewer correct).
- Readers who visited Wikipedia more frequently, donors, and editors were more knowledgeable about Wikipedia on average.
- More-educated readers were more knowledgeable on average.
- Wikipedia knowledge varied considerably by project. Readers of dewiki were most knowledgeable on average.
Data Collection
editThis project employed simple random sampling of Wikipedia readers using the QuickSurveys extension. The QuickSurveys opt-in was displayed to non-logged in users and asked them whether they would like to participate in a survey to help improve Wikipedia. Survey responses were collected using LimeSurvey, an external survey tool.
The goals of the survey were to make demographic estimates of Wikipedia readers across different language projects within the scope of the Knowledge Gaps Index, to understand motivations for reading Wikipedia, and to analyze whether there are differences by motivation and demographics in who reads which type of content.
Timeline
editDate | Milestone |
---|---|
November 25, 2024 | enwiki full sample survey launched |
November 28, 2024 | enwiki full sample survey closed |
December 10, 2024 | arwiki, jawiki, ptwiki, ruwiki, zhwiki full sample surveys launched |
December 12, 2024 | dewiki, eswiki, frwiki full sample surveys launched |
December 17, 2024 | arwiki, dewiki, frwiki, jawiki, ptwiki, ruwiki, zhwiki full sample surveys closed |
December 19, 2024 | eswiki, idwiki, kowiki full sample surveys closed |
- January & February 2025: Data cleaning and analysis
- March 2025: Internal shareout of initial results
- May 2025: Internal shareout of report
- June 2025: Report published
Policy, Ethics and Human Subjects Research
editThis survey is governed by the Global Readers Survey privacy statement.
Survey Administration
editSurveys were fielded across 11 projects from November 25, 2024—December 19, 2024. A total of 25,929 complete survey responses were collected.
Project | Fielding Dates (Month/Day) | QuickSurvey Sampling Ratio | Total LimeSurvey Initiations | Total Completes |
---|---|---|---|---|
arwiki (Arabic) | 12/10—12/17 | 25% | 26829 | 2864 |
dewiki (German) | 12/12—12/17 | 5% | 10497 | 3613 |
enwiki (English) | 11/25—12/2 | 4% | 27290 | 6146 |
eswiki (Spanish) | 12/12—12/19 | 6% | 13399 | 2642 |
frwiki (French) | 12/12—12/17 | 10% | 14687 | 2531 |
idwiki (Indonesian) | 12/10—12/19 | 30% | 19192 | 1026 |
jawiki (Japanese | 12/10—12/17 | 5% | 6975 | 1780 |
kowiki (Korean) | 12/10—12/19 | 30% | 5339 | 419 |
ptwiki (Portuguese) | 12/10—12/17 | 15% | 16624 | 1610 |
ruwiki (Russian) | 12/10—12/17 | 2% | 6445 | 1855 |
zhwiki (Simplified and Traditional Chinese) | 12/10—12/17 | 10% | 6551 | 1443 |
Survey Attrition
editOut of a total 120,698 users with unique QuickSurvey tokens who initiated the survey on Limesurvey (navigated to the survey landing page), more than one-in-four (27%) viewed every substantive question (completed the survey). The vast majority of survey attrition occurred immediately, as an outright majority of unique Limesurvey initiations ended without anyone opening the survey. In contrast, comparatively few respondents quit during any of the substantive modules (9% overall).
Survey Results
editScreeners and Preliminaries
editAge Screener
editOnly readers aged 18 years and older were considered eligible for the survey. As a result, all readers who opted into the survey were first shown an age-based screener question. Those who indicated they were under 18 had their survey sessions terminated.
G0Q01. Are you at least 18 years of age? ○ Yes ○ No Unfortunately, legal protections for people under 18 mean we cannot survey you. Thank you for your interest!
zhwiki Language Type
editReaders sampled from zhwiki were asked to choose their preferred form of written Chinese for the survey. While we cannot directly estimate the proportion of zhwiki readers within China, due to the well-known ban on Wikipedia in the People's Republic of China, the proportion of respondents choosing to take the survey in simplified Chinese (predominantly used in the People's Republic of China) can provide a very rough estimate (although simplified Chinese is also used in Malaysia and Singapore).
G0QLANG. Please choose whether you would like to take this survey in Simplified or Traditional Chinese. 请选择您想用简体中文还是繁体中文参加此调查。 請選擇您想用簡體中文還是繁體中文參加此調查。
Demographics
editWe measure demographics in each wave of the Readers Survey to provide regular measurements as part of the Knowledge Gap Index. Note that (as described in the [Survey Attrition] section, above) while the demographics survey items were administered in the final module of the survey (meaning that these items were seen by respondents after the other survey modules), results from these items are presented first here as demographic data will frequently be used in the analysis of other survey items.
Geography
editGeography was coded based on IP address recorded by Limesurvey (the survey host) and coded using the ip2location
R package and the IP2LOCATION LITE country database. Geographic data is weighted only by project share of total traffic. Because geography is coded "administratively" rather through survey response, it is recorded for essentially all 120,698 unique users arriving at the survey, including the 62,683 (51.9%) who did not proceed past the landing page, opening the survey.
When ___location is coded based on current Wikimedia Foundation regions, Wikipedia readers are most likely to be located in North America (24.1%) and Northern & Western Europe (22.8%) and least likely to be located in the Middle East & North Africa (3.3%) or Sub-Saharan Africa (2.7%).
Overall at the country level, readers from 208 separate countries are represented in the sample. However, these readers are highly concentrated in a relatively small number of countries. Only sixteen individual countries each have 1% or more of total readers in the sample, but collectively account for 71.3% of readers in the sample. In particular, readers are much more likely to be located in the United States (22.1%) than in any other single country.
Age
editWe employ the following item (also employed in the 2023 Global Readers Survey and the Community Insights surveys) to measure age (only readers 18 and older qualify for the survey). Note that 3.4% of respondents to the age screener indicated that they are under 18 years old (4.4% total did not advance past the age screener).
G3Q01. What is your age? ○ 18-24 ○ 25-29 ○ 30-39 ○ 40-49 ○ 50-59 ○ 60-69 ○ 70+ ○ I prefer not to say
Consistent with findings from the 2019 and 2023 Global Readers Surveys, we find that readers skew young overall, with readers aged 18-24 constituting the single largest age group in our sample. Overall, readers age 18-29 make up over one third of respondents (36.1%).
However, the age distribution of readers varies considerably across the 11 studied projects: readers 18-29 make up half or more of respondents on ruwiki (49.6%), idwiki (51.7%), arwiki (53.7%)), and zhwiki (53.8%). Conversely, readers aged 18-29 make up fewer than 30% of respondents on eswiki (29.8%), jawiki (27.6%), frwiki (23.9%), and dewiki (14.0%).
What's more, when we compare the proportion of readers age 18-29 across the seven projects consistently surveyed in the 2019, 2023, and 2024 Global Readers Surveys (arwiki, dewiki, enwiki, eswiki, frwiki, ruwiki, and zhwiki), we observe substantial and sustained declines in two projects: French and German Wikipedia.
Examining just the youngest age cohort surveyed (readers age 18-24) across the same seven projects from 2019--2024 suggests that Spanish Wikipedia may be experiencing an analogous decline in youth readership as well.
Finally, we see similar trends when it comes to readers who say they are under the age of 18. When we compare the proportion of readers who are disqualified from continuing in the survey by saying they are under 18 (across all eleven projects surveyed in both 2023 and 2024) most projects do not see statistically significant changes between 2023 and 2024. The four exceptions are ruwiki, eswiki, frwiki, and dewiki, all of which show significant declines in the proportion of readers saying they are under 18. In addition, dewiki is a clear outlier in both 2023 and 2024 as the project with the lowest proportion of surveyed readers saying they are under 18.
Gender
editIn order to facilitate comparisons between surveys of Wikipedia readers and contributors to Wikimedia projects, the 2023 and 2024 Global Readers Surveys employ a gender identity survey item aligned with that used in the 2024 Community Insights survey. Note that respondents to the arwiki, and idwiki surveys were not presented with the "transgender", "non-binary", and "genderfluid" response options.
G3Q02. Which of these categories describe your gender identity? Select all that apply. □ Man □ Woman □ Transgender □ Non-binary □ Genderfluid □ Other: _________________ ○ I prefer not to say
Readers were given the opportunity to select as many gender identities as applied. Nearly seven-in-ten (69.5%) identified as Men (alone or in combination), 20.3% identified as women (alone or in combination), 2.2% identify as transgender (alone or in combination), 3.0% identify as non-binary (alone or in combination), 1.3% identify as genderfluid (alone or in combination), and 2.9% identify with another identity (alone or in combination).
When gender identity is recoded into mutually-exclusive and mutually comprehensive categories, 67.8% of respondents identify solely as men, 19.0% identify solely as women, 7.6% identify as genderdiverse (identifying as transgender, non-binary, genderfluid, another identity, or any of two or more identities).
At the project level, readers are most likely to identify solely as men on idwiki (75.5%), jawiki (74.4%), arwiki (73.3%), zhwiki (71.7%), and dewiki (70.5%). However, readers identifying solely as men make up more than six-in-ten respondents on every surveyed project, even on relatively more-gender-balanced projects such as kowiki (61.6%) and frwiki (66.0%).
In comparison to 2023, we observe a significant increase in the share of readers identifying solely as men (63.3% to 67.8%) and a significant decrease in those identifying solely as women (25.1% to 19.0%). In addition, 7 of 11 projects surveyed in both years show significant increases in the share of readers identifying solely as men. The exceptions are dewiki, idwiki, kowiki, and ptwiki (although we measure a non-significant increase in readers identifying solely as men on idwiki). Future research is needed to evaluate whether this represents a temporary blip or a persistent change in the gender composition of Wikipedia readership.
Urbanity (Community Density)
editIn order to measure the type of communities in which Wikipedia readers live, we asked respondents to describe their community using one of five density categories.
G3Q05. How would you describe the community where you currently live? ○ Rural (a farm or home in the country outside of any town or village) ○ Rural (a small country village) ○ Suburban (a small city or town located near a larger city) ○ Suburban (the outskirts or suburbs of a city) ○ Urban (in or near the center of a city) ○ Something else not listed above (please describe here):_____________
Readers were most likely to describe their local community as "Urban (in or near the center of a city)" (43.8%), followed by suburban—"a small city or town located near a larger city" or "the outskirts or suburbs of a city"—(37.0%). Only 12.3% of readers described their community as rural ("a farm or home in the country outside of any town or village" or "a small country village").
Education
editMeasuring educational attainment cross-nationally is a longstanding methodological challenge in survey research[1]. This is further complicated in our case by the fact that Global Readers surveys are designed and sampled by language project rather than by geography (e.g., enwiki respondents alone are educated under a wide variety of very different educational systems). We also sought to balance survey item simplicity with cross-system comparability. Together, these constraints made it difficult for us to substantially localize our measures of educational attainment.
The 2024 Global Readers Survey builds on the survey items used to measure education in the 2023 Global Readers Survey, but modifies them slightly to improve cross-national comparability. In particular, education is measured with a series of survey items: first, one asking whether respondents were currently enrolled as students and subsequent items asking current students to indicate the level of education they are pursuing and for non-students to indicate their level of educational attainment (categories for both questions are based like the 2023 survey item on the ISCED-1997 classifications.
G3Q03. Are you currently enrolled as a student in school (for example, high school, vocational or trade school, a college or university)? ○ Yes ○ No ○ I'm not sure ○ I prefer not to say
Only shown to respondents who selected "Yes" in G3Q03
G3Q04A. Are you currently… ○ An upper secondary (high) school student ○ A vocational or trade school student ○ A college or university student (Bachelor’s degree program) ○ A graduate college or university student (post-graduate degree program) ○ I am not currently a student ○ I prefer not to say
Only shown to respondents who selected "No" in G3Q03
G3Q04B. What is the highest level of formal education you have completed? ○ I have no formal schooling ○ Some primary or elementary school ○ Primary or elementary school ○ Lower secondary or middle school ○ Upper secondary or high school ○ A post-secondary technical or vocational degree or certificate ○ Some university education, but less than a Bachelor’s degree ○ A Bachelor’s degree or equivalent ○ A post-graduate degree (e.g., master's, doctorate, or professional degree) ○ I'm not sure ○ I prefer not to say
Student Status
editOverall, about one-in-four readers (27.1%) say they are currently enrolled in formal education (upper secondary or higher). The share of current student readers by project ranges from around two-in-five (40.0% on arwiki, 38.2% of zhwiki) to just 13.6% on dewiki.
Educational Attainment / Student Level
editAmong current students (age 18 and older), a clear plurality (44.3%) are currently enrolled as undergraduates (in a bachelor's degree program) with a further 16% currently enrolled in a post-graduate degree program. However, the 24.7% of current student respondents who say they are currently high school or upper secondary students almost certainly understates the true proportion as only those over age 18 qualified for the survey.
Among those who are no longer enrolled in formal education, the overwhelming majority of respondents (86.2%) have completed at least upper secondary or high school (93.6% of those providing an answer). In addition, a majority of respondents (53.7%) have a Bachelor's degree or more, with over one-in-four (25.4%) having attained a postgraduate degree. These findings are consistent with those from the 2023 Global Readers Survey, with the proportion of postgraduate degree holders declining marginally compared to 2023 (27.2% ±0.5%).
When the education items are combined, readers appear to skew even more towards the university-educated, with a total of 58.9% of respondents reporting being enrolled in or having completed a university degree program (37.5% Bachelor's degree/student; 21.4% postgraduate degree/student).
Language
editResults from the 2023 Global Readers Survey show that Wikipedia readers are highly multilingual, overall. For the 2024 Global Readers Survey, we employed the same survey items used in the 2023 edition. Readers were first asked whether the project language is their primary language. Those who indicate it is their primary language are then shown a question (G3Q07A) listing 20—21 languages (depending on the project surveyed) and asked which, if any, they are also fluent in. Respondents can select as many as apply and can also add other languages they are fluent in as open text entries. Those whose primary languages are not the project language are instead first shown the same list of languages as in G3Q07A and asked to indicate their primary language(s) in question G3Q07B. Afterwards, these respondents are shown a question (G3Q08) with the same structure as G3Q07A and asked which, if any, they are also fluent in (selecting as many as relevant)
G3Q06. Is [sampled language] your first or native language or the language you are most fluent in? ○ Yes ○ No ○ I prefer not to say
Only shown to respondents who selected "Yes" in G3Q06. Languages shown are those shown to enwiki respondents. Language options alphabetized in project language.
G3Q07A. What other languages are you fluent in? ○ I’m not fluent in any other language □ عربى(Arabic) □ 汉语 (Chinese) □ Čeština (Czech) □ Nederlands (Dutch) □ فارسی (Farsi / Persian) □ Français (French) □ Deutsch (German) □ עברית (Hebrew) □ हिन्दी (Hindi) □ Bahasa Indonesia (Indonesian) □ Italiano (Italian) □ 日本語 (Japanese) □ 한국어 (Korean) □ Polski (Polish) □ Português (Portuguese) □ Pусский (Russian) □ Español (Spanish) □ Svenska (Swedish) □ Українська (Ukrainian) □ Tiếng Việt (Vietnamese) □ Other language(s) not listed above (please enter here): _________________ ○ I prefer not to say
Respondents who selected "No" in G3Q06 are shown the same language list as in G3Q07A with the addition of the project language and are asked (question shown as administered to enwiki respondents):
G3Q07B. What is your first, or native language, or what language are you most fluent in? □ English ... □ Other language(s) not listed above (please enter here): _________________ ○ I prefer not to say
Respondents who selected "No" in G3Q06 are next shown the same language list as in G3Q07B and asked (question shown as administered to enwiki respondents):
G3Q08. What other languages are you fluent in? ○ I’m not fluent in any other language □ English ... □ Other language(s) not listed above (please enter here): _________________ ○ I prefer not to say
In each surveyed project, a clear majority of respondents indicate that they are reading in their primary language. In all but one project (enwiki), about nine-in-ten or more readers say they are reading in their primary language. This is broadly consistent with findings from the 2023 Global Readers Survey, although we measure somewhat fewer English Wikipedia readers who say English is their primary language (65.4%±1.2%) in 2024 compared to 2023 (73.0%±1.4%).
When we recoded responses to G3Q06, G3Q07A, G3Q07B, and G3Q08 to count the number of languages readers say they are fluent in, 35.9% total say they are fluent in two or more languages. Notably, this is a substantial decrease from the 54.9% of readers who claimed multilingual fluency in the 2023 Global Readers Survey. This decrease appears to be primarily due to a substantial increase in monolinguality among English Wikipedia readers (from 52.1% ± 1.4% in 2023 to 70.7% ± 1.2% in 2024).
Examining claimed English language fluency specifically, we find that English fluency appears widespread among Wikipedia readers in all surveyed projects. Because English Wikipedia traffic represents such a large proportion of total Wikipedia traffic, the overall (weighted) level of English fluency closely resembles that of English Wikipedia readers. However, even on the projects with the lowest claimed levels of English fluency (zhwiki, 22.5%; arwiki, 24.1%) about one-in-four readers say they are fluent in English. At the same time, about half of readers on idwiki (48.3%) and frwiki (50.0%) say they are fluent in English. This likely reflects not only the fact that English can claim more speakers than any other language, but also the high overall levels of formal education we observe among Wikipedia readers.
Reader Motivations and Information Needs
editReader Motivation
editConsistent with previous survey research conducted by the Wikimedia Foundation [1] [2] , we asked readers about their motivations for reading Wikipedia. However, in this survey, we allowed respondents to select multiple motivations and to write-in other motivations that were not listed as answer options.
G1Q01. I am reading this article because ... Please select all answers that apply □ I have a work or school-related assignment □ I need to make a personal decision based on this topic (e.g., buy a book, choose a travel destination) □ I want to know more about a current event (e.g., a soccer game, a recent earthquake, somebody’s death) □ the topic was referenced in a piece of media (e.g., TV, radio, article, film, book) □ the topic came up in a conversation □ I am bored or randomly exploring Wikipedia for fun □ this topic is important to me and I want to learn more about it (e.g., to learn about a culture) □ Other:__________________________
Consistent with results from the 2023 Global Readers Survey, 2024 respondents were overall most likely to say the article topic was "personally important" to them. (Respondents were able to select multiple motivations).
Reader Information Needs
editAgain, following previous survey research, we asked readers about the specific information needs that motivated them to read the article from which they were sampled.
G1Q02. I am reading this article to … ○ look up a specific fact or to get a quick answer ○ get an overview of the topic ○ get an in-depth understanding of the topic
Overall, we observe similar results to 2023. Wikipedia readers are most likely to say they are are reading to "get an overview of the topic" (44.3%). However, readers also frequently say they visited an article to "look up a specific fact or to get a quick answer" (29.0%) or to "get an in-depth understanding of the topic" (26.0%).
Reader Topic Prior Knowledge
editWe presented readers with the same survey question measuring their prior knowledge of the topic of the article they were reading that was used in previous readers surveys.
G1Q03. Prior to visiting this article … ○ I was already familiar with the topic ○ I was not familiar with the topic, and I am learning about it for the first time
Overall, Wikipedia readers are more likely to say that they are already familiar with the topic they are reading about (58.0%) than not (41.5%). This represents a marginal increase for "already familiar" in comparison to 2023.
Success in Finding Information
editIn the 2024 survey, we introduced an item measuring readers' self-reported success in finding the information they were seeking on Wikipedia.
G1Q04. Were you able to find the information you were looking for on Wikipedia? ○ Yes ○ No ○ I'm not sure
Respondents overwhelmingly said that they were successful in locating the information they were looking for on Wikipedia (80.7%), while very few (3.9%) said they were unsuccessful. A further 15.2% indicated that they were not sure.
Surprisingly, we observed no variation in self-reported success by either the type of information sought or by readers' prior familiarity with the topic.
However, we do observe some substantial variation in successful information retrieval at the project level. Notably, readers of arwiki (63.6%) and kowiki (70.5%) are less likely to say they were able to find the information they were looking for on Wikipedia.
Accessing Knowledge Online
editFor the 2024 Global Readers Survey, we introduced a new module across all 11 surveyed projects designed to understand how Wikipedia readers learn and access knowledge on the internet—both on and off Wikipedia. This includes for example, asking respondents where else they access knowledge online, as well as their affective feelings towards those other apps, platforms, or websites, their experiences of engagement with Wikipedia as readers, editors, and donors, as well as a battery testing their factual knowledge about Wikipedia.
Other Online Platforms
editG2Q01. What other online apps, platforms or websites have you visited in the past 30 days in order to learn or access knowledge? Please select all answers that apply □ YouTube □ Quora □ Reddit □ TikTok □ Instagram □ ChatGPT □ Facebook □ X (formerly known as Twitter) □ WhatsApp □ Coursera □ Linkedin □ Google / another search engine □ Another app, platform or website:__________________________ □ None of the above □ I prefer not to say
Survey respondents were asked "What other online apps, platforms or websites have you \nvisited in the past 30 days in order to learn or access knowledge?" and given a list from which they could select as many options as applied. Respondents were also able to select "another app, platform or website" and provide additional responses as free text. The sample question displayed above shows the choices shown to English Wikipedia readers, but respondents sampled from other projects were also shown other, localized choices based on market share in relevant markets. Respondents who selected "none of the above" or "I prefer not to say" were not able to select any other options.
Overall, readers were most likely to name YouTube (81.0%) and Google / Other Search Engines (78.1%) as online platforms they have used recently to learn or access knowledge. Reddit (36.1%) and ChatGPT (29.2%) were next most-likely to be named. Popular messaging and social media platforms such as X (Twitter) (20.3%), Instagram (20.0%), WhatsApp (13.2%), Tiktok (11.7%), and Facebook (18.7%) were comparatively less likely to be cited as places to learn or access knowledge.
What's more, the youngest respondents (age 18-24) generally did not differ dramatically in their responses from respondents overall. Young readers were somewhat more likely to say they visited YouTube (88.4% vs. 81.0%) and marginally more likely to say they used Google or another search engine (80.7% vs. 78.1%). However, young readers are substantially more likely to say they use Reddit (48.5% vs. 36.1%) and ChatGPT (45.0% vs. 29.2%) and substantially less likely to say they use Facebook (7.4% vs. 18.7%) to learn or access knowledge compared to readers as a whole.
In most cases, we did not observe substantial differences in information sources cited by readers across projects. However, readers of Japanese, Korean, Russian, and Chinese Wikipedia were relatively more likely to say they used some other platforms, including regionally-popular apps or platforms, to learn or access knowledge in the past 30 days.
Japanese Wikipedia readers were, like readers overall, most likely to say they used YouTube (81.2%) and Google or another search engine (77.3%), but were also much more likely to say they used X (Twitter) (48.8%) and Yahoo (39.5%) compared to readers overall. Messaging app Line was also cited by 25.9% of Japanese Wikipedia readers, more than Instagram (24.9%), Facebook (13.7%), and TikTok (9.8%). In addition, comparatively few Japanese Wikipedia readers named Reddit as a place to learn or access knowledge (6.9%).
While Korean Wikipedia readers were also most likely to cite YouTube (72.3%) and Google or other search engines (66.0%) as sources of knowledge, they were notably less likely to do so than readers overall. A majority of Korean Wikipedia readers said they used South Korean platform Naver (53.9%) to learn or access knowledge, while sizeable groups said they used the messaging app Kakaotalk (20.2%) and web portal Daum (14.1%).
Russian Wikipedia readers are marginally more likely to say they visited YouTube to learn or access knowledge (84.2%), but are substantially less likely to name Google (71.0%) compared to readers overall, possibly because 44.1% of them cite Yandex. Social media app VK is named by 29.8% of Russian Wikipedia readers.
Chinese Wikipedia readers were distinctively likely to say they used ChatGPT to learn or access knowledge (49.3%) when compared to readers overall. "Super app" WeChat was named by 23.7% of Chinese Wikipedia readers, while Line (popular in Taiwan was named by 13.5%, and Chinese messaging platform TencentQQ was named by 10.3%.
Affective Ratings of Online Platforms
editIn order to provide a single measure of readers' subjective feelings towards both Wikipedia and other online platforms they use to learn or access knowledge, we constructed an adapted feeling thermometer item. Respondents were asked to rate their feelings towards Wikipedia as well as any platform they had previously indicated having visited in the past 30 days to learn or access knowledge on a scale of 0 to 10, with higher numbers meaning more positive or favorable feelings. A sample question for a reader who had previously selected YouTube, TikTok, and ChatGPT is shown below.
G2Q03. Please rate how you feel about each of the following apps, platforms or websites that you have used to learn or access knowledge on a scale between 0 and 10. * The higher the number, the more positive or favorable you feel about that app, platform or website. 10 would indicate an extremely positive or favorable rating, 5 would indicate a neutral rating and 0 would indicate an extremely negative or unfavorable rating. 0 1 2 3 4 5 6 7 8 9 10 I'm not sure Wikipedia ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ YouTube ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ TikTok ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ChatGPT ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○
Overall, Wikipedia (shown in purple) is rated more favorably than any other platform, receiving a mean rating of 8.5 out of 10. The next-highest rated platforms (of the "global" list presented to all respondents) are Google (mean rating of 7.6), YouTube (mean rating of 7.3), ChatGPT (mean rating of 7.4), and Coursera (mean rating of 7.6—although Coursera was rated by many fewer respondents). Notably, "social networking" platforms generally receive lower favorability ratings overall: X (Twitter) is the lowest-rated overall (5.8 mean rating), followed by Facebook (5.9 mean rating), Instagram (6.0 mean rating), TikTok (6.3 mean rating), Linkedin (6.3 mean rating), and Reddit (6.5 mean rating).
When platform favorability ratings are compared across age groups, Wikipedia ratings appear relatively stable across age groups. While mean Wikipedia ratings do increase marginally with age, these differences are only statistically significant for comparisons of the youngest (18-24) and oldest (70+) age groups. In contrast, ratings of TikTok and Reddit show much clearer variation in favorability by users' age, with older age cohorts generally rating these platforms less favorably.
At the project level, we observe much greater variation in favorability ratings, overall. Specifically, Wikipedia is viewed less favorably in both absolute and relative terms among readers of arwiki, idwiki, and kowiki. Arabic Wikipedia readers rate Wikipedia (7.7 mean rating) significantly less favorably than Google (8.2 mean rating), and ChatGPT (8.0 mean rating). Similarly, while favorability estimates are noisier on Korean Wikipedia, Wikipedia receives a similar mean rating in absolute terms (7.6). Indonesian Wikipedia readers regard Wikipedia more favorably in absolute terms (8.2 mean rating), but rate Google significantly more favorably on average (8.6 mean rating) and give ChatGPT (8.2 mean rating) and YouTube (8.0 mean rating) statistically indistinguishable ratings.
Use of Wikipedia Compared to Last Year
editTo measure readers' perceptions of trends in their use of Wikipedia, the 2024 Global Readers Survey includes an item measuring how respondents' frequency of use of Wikipedia has changed in the past year.
G2Q04. Thinking about your use of Wikipedia now compared to one year ago... Do you use Wikipedia: ○ Less now than a year ago ○ About the same now as a year ago ○ More now than a year ago ○ I'm not sure ○ I prefer not to say
Most respondents (64.1%) report that they use Wikipedia about the same now as a year ago. However, respondents are about twice as likely to say they have increased their use of Wikipedia over the past year (20.3%) as to say they have decreased their use of Wikipedia (9.1%).
Wikipedia Visit Frequency
editThe 2024 Global Readers Survey also introduced a new item measuring how frequently respondents say they visit Wikipedia. This measure can complement existing metrics of reader retention as well as providing context for how Global Readers Survey respondents compare to more casual readers.
G2Q05. How often do you visit Wikipedia? ○ Never ○ A few times per year ○ A few times per month ○ A few times per week ○ Every day ○ I'm not sure
Overall, about three-in-four respondents (78.2%) say they visit Wikipedia a "few times per week" or more with 36.3% saying they visit "every day".
Wikipedia Experience and Knowledge
editWikipedia Editing Experience
editReaders were asked about their experience with editing Wikipedia for the first time in the 2024 Global Readers Survey.
G2Q06. Have you ever added, changed, or edited anything in a Wikipedia article? ○ No, I didn't know you could edit Wikipedia ○ No, but I know you can edit Wikipedia ○ Yes, once or twice ○ Yes, many times ○ I'm not sure
About seven-in-ten (69.7%) of surveyed readers indicated that they have never made any Wikipedia edits, while 21.6% say they've made one or two edits, with just 8.1% saying they have edited "many times". While this likely overstates to some extent the proportion of readers who may become active editors (due for example, to acquiescence and social desirability bias), it represents a plausible upper bound of how many frequent readers may be open to becoming editors.
Donating to Wikipedia
editAnother new item in the 2024 Global Readers Survey asks readers whether they have previously donated to Wikipedia. Because on-platform fundraising is limited to certain markets, this question was not included in the arwiki, idwiki, ruwiki, or zhwiki surveys.
G2Q07. Have you ever donated money to Wikipedia or the Wikimedia Foundation? ○ No ○ No, but I've considered it ○ Yes, once or twice ○ Yes, multiple times ○ I'm not sure ○ I prefer not to say
Overall, almost two-in-five (38.8%) respondents to the dewiki, enwiki, eswiki, frwiki, jawiki, kowiki, and ptwiki surveys say they have donated once or twice (17.2%) or multiple times (21.6%) to Wikipedia or the Wikimedia Foundation. However, readers were most likely to say they have never considered donating (32.9%) or have considered donating, but ultimately decided not to (24.1%).
At the project level, dewiki, enwiki, and frwiki readers are most likely to say they have donated, while kowiki readers are least likely to have donated.
Wikipedia Knowledge Battery
editThe 2024 Global Readers Survey also measured respondents' knowledge about Wikipedia through a knowledge battery developed for the recent Volunteer Archetypes research project. Readers were presented with a series of ten statements (reproduced below) and asked to indicate whether each one is True or False. Readers could also respond "I'm not sure" or skip the statement.
G2Q08 | True | False | I'm not sure |
---|---|---|---|
Wikipedia articles are written by paid experts. | |||
Wikipedia articles must have pictures. | |||
Wikipedia articles can have many authors. | |||
Wikipedia articles have advertisements. | |||
Wikipedia articles can be changed by anyone. | |||
Wikipedia articles change over time. | |||
Wikipedia articles are only changed when mistakes are found. | |||
Changes to Wikipedia articles must be approved by content experts. | |||
Wikipedia relies on donations to support its operations. | |||
The people who write Wikipedia articles are unpaid volunteers. |
Overall, we observe considerable variation in readers' factual knowledge about Wikipedia. While about three-in-ten (31.6%) were able to answer at nine or ten correctly, a similar proportion (26.3%) performed no better than chance (five or fewer correct answers out of ten).
At the individual statement level, readers were most likely to answer correctly that Wikipedia articles can have many authors (89.9%), that Wikipedia articles can change over time (85.8%), that Wikipedia articles do not have ads (79.3%) and least likely to answer correctly that anybody can edit a Wikipedia article (57.2%) and that changes to articles do not need to be approved by content experts (28.9%).
What Explains Wikipedia Knowledge Levels?
editAs noted above, we measure considerable variation in readers' levels of knowledge about Wikipedia. But what explains this variation? In the following, we examine some plausible explanations: visit frequency, editing behavior, donation behavior, and education. In addition, we examine some plausible explanations for the single item about which readers were least knowledgeable: "Changes to Wikipedia articles must be approved by content experts".
First, it seems plausible that readers' knowledge about Wikipedia should be correlated with their level of engagement with Wikipedia—those who use Wikipedia more frequently and engage more deeply with it (eg, as editors or donors) are likely to be better-informed about it. In the following, we show the relationship between readers' overall knowledge about Wikipedia (as measured by their total correct answers on the Wikipedia knowledge battery) and three measures of Wikipedia engagement: visit frequency, editing behavior, and donation behavior. Each violin plot compares the smoothed probability density of total correct answers for different levels of reader engagement. Subgroup means (along with corresponding 95% confidence intervals) are depicted in red.
The more frequently a reader reports visiting Wikipedia, the more knowledgeable they are about Wikipedia, on average.
Likewise, the greater a readers' self-reported engagement with Wikipedia editing, the higher their average score on the Wikipedia knowledge battery.
Finally, knowledge about Wikipedia is also clearly correlated with readers' self-reported donation behavior: the more "serious" they are about donating to Wikipedia, the more knowledgeable they are about Wikipedia on average.
Readers' level of education is also consistently correlated with Wikipedia knowledge: those with higher levels of educational attainment (or, for those who are currently students, more advanced levels of current schooling) tend to score higher on the Wikipedia knowledge battery across each surveyed project. Notably however, we also observe considerable variation in Wikipedia knowledge at the project level such that, for example, the most-educated arwiki readers are less-knowledgeable on average than the least-educated dewiki readers. This project-level variation is similarly in evidence at the level of individual Wikipedia knowledge battery items.
"Changes to Wikipedia articles must be approved by content experts"
editOne possible explanation for cross-project variation in correct responses to "Changes to Wikipedia articles must be approved by content experts" is the use of the Flagged Revisions tool on the project a respondent was reading. Specifically, readers of projects where Flagged Revisions is enabled could conceivably interpret the review by experienced editors as review by "content experts". Flagged Revisions is currently implemented on arwiki, dewiki and ruwiki (and was implemented on idwiki between 2010 and 2011).
However, it is not immediately clear that the project-level results are consistent with this explanation. For example, dewiki readers are more likely than average to answer the "content experts" question correctly. In addition, while arwiki and idwiki readers are least likely to answer this question correctly, this is consistent with the overall pattern of responses across the whole Wikipedia knowledge battery.
Moreover, it seems likely that only readers who also have editing experience would likely be familiar with how edits are moderated on a given Wikipedia. Thus, if Flagged Revisions enablement explains project-level variation, we might expect editing experience to predict a lower likelihood of correct answers on this statement for readers on projects where Flagged Revisions is enabled. Instead, we observe the opposite relationship: the likelihood of correct answers on this statement increases monotonically with editing experience for all surveyed projects other than kowiki.
Methodology
editSampling
editThis project employed simple random sampling of Wikipedia readers using the QuickSurveys extension. Sampling rates vary by project and are described in the Survey Administration section above. The QuickSurveys opt-in was displayed to non-logged-in readers only and asked whether they would consent to "Take a short survey and help us improve Wikipedia". We chose to employ the QuickSurvey tool to sample readers (rather than e.g., a Central Notice Banner) both for consistency with previous research conducted by the Wikimedia Foundation and to avoid sampling readers from non-article pages (e.g., talk pages, community pages, Wikipedia home pages).
Readers who consented to the survey were then linked out to a survey hosted on LimeSurvey, an open-source survey platform.
Weighting
editIn order to account for sampling design and to better match the global population of Wikipedia readers, we apply weights based on global population parameters following the method described in DeBell and Krosnick (2009)[2] implemented using the 'anesrake'
software package written for R.
Survey responses were weighted at the project level by geography (weighting categories vary by project). For analyses at the global level, responses were also weighted by project shares of overall traffic during the time when the surveys were in the field. Survey weights applied differ from the 2023 Global Readers Survey due to data collection issues experienced with reading session data for English Wikipedia readers participating in the survey, making weighting on OS family, referrer class, and session length impossible for English Wikipedia readers. We chose to limit weights to geography at the project level and project shres of total traffic at the global level for the following reasons:
- These variables were available for all surveyed projects and employing the same weighting scheme for all projects facilitated cross-project comparisons
- Due to the substantive importance of English Wikipedia, it seemed most appropriate to weight all projects using the same scheme utilized for English Wikipedia
- These variables were the two most-influential variables in determining final weights applied to the 2023 Global Readers Survey
Similarly, while the 2023 Global Readers Survey utilized a geography variable (country) coded (from IP addresses) at the point of QuickSurvey interaction, the 2024 Global Readers Survey utilizes country codings determined at the point of Limesurvey interaction. We chose to do this because this was the only way we could recover geography data for English Wikipedia respondents for this survey, but also because—for as-yet-unknown reasons—Limesurvey records IP addresses (and therefore geography at the country level) for a larger proportion of readers in the survey sample. As described above, IP address data recorded by Limesurvey was coded at the country level using the ip2location
R package and the IP2LOCATION LITE country database.
Resources
editReferences
edit- ↑ Connelly, Roxanne; Gayle, Vernon; Lambert, Paul S. (2016). "A review of educational attainment measures for social survey research". Methodological Innovations 9: 1–11. ISSN 2059-7991. doi:10.1177/2059799116638001.
- ↑ DeBell, Matthew; Krosnick, Jon A. (2009). "Computing Weights for American National Election Study Survey Data" (PDF). ANES Technical Report series (nes012427): 1–14.