Publishing Quality Statistics Videos on YouTube: Analysis of Sources and Publishing Formats Authors: Aizhan Oralbek, Olena Zimba, Ainur B Qumar, Nurlan Aukenov, Ramazan Azim Okyay, Burhan Fatih Kocyigit Published: 2026 ABSTRACT AbstractBackgroundThis study aims to assess the content quality, presentation characteristics, and viewer engagement of YouTube videos on statistics, and to compare video quality categories based on fundamental video parameters, interaction metrics, presentation formats, and content sources.MethodsYouTube videos were searched for on November 17, 2025, using the search terms “statistical analysis,” “data analysis,” “statistics,” and “statistical methods.” The top 50 most relevant videos for each search term were assessed; after applying exclusion criteria, 114 videos were reviewed. The Global Quality Scale was used to measure video quality, and reliability was determined using a modified DISCERN tool. Videos were divided into low-, intermediate-, and high-quality categories. Engagement metrics and presentation methods were compared among quality groups.ResultsEighty-seven (76.3%) of the videos were categorized as high quality. Statistically significant differences were observed among video quality groups in daily views, likes, and comments (P < 0.001 for all comparisons); these differences were attributed to the disparity between the low- and high-quality groups. Videos produced by academic institutions, nonprofit associations, and professional experts had higher percentages classified in the high-quality category (n = 1; 100%, n = 5; 83.3%, and n = 32; 80%, respectively). Presentation formats differed significantly across quality categories, with greater diversity observed in high-quality videos. Within this group, slide-based presentations accounted for 33.3%, animated content for 32.2%, and demonstration-based (software-centric) videos for 18.4% (P = 0.020).ConclusionA substantial proportion of YouTube videos on statistics were rated high quality, and significant differences were observed in engagement metrics across quality groups. The quality distribution varied by video source and presentation format. These findings emphasize the need for educators and content developers to integrate accurate statistical content with engaging presentation formats to improve the educational value of YouTube videos.Keywords: Social Media, Online Social Networking, YouTube, Statistics, Data Analysis INTRODUCTION Statistics is a crucial domain within the scientific realm, involving the systematic collection, organization, examination, and interpretation of data obtained through observation and measurement.1 The Medical Subject Headings (MeSH) define data analysis as a process that encompasses the systematic description, visualization, summarization, and evaluation of data through the application of statistical and/or logical techniques.2 In this context, statistics facilitates the description of numerical information while also allowing for the interpretation of complex data structures and the extraction of trustworthy inferences from the data.3 Therefore, statistics is a vital component of modern scientific inquiry and is regarded as a key instrument for assessing the reliability, validity, and scientific merit of the findings.4 Technological advancements and the rapid progression of digitalization have significantly transformed access to information and the ways of knowledge acquisition. Currently, customers and scholars can access information more rapidly and extensively via online platforms.5 Online information resources are becoming pivotal for disseminating academic knowledge, supporting learning, and enabling research processes due to their accessibility, speed, and diversity.6 This circumstance requires assessing the reliability and accuracy of scientific information available online. YouTube, a prominent online platform, has transcended its role as a mere entertainment platform and has emerged as a significant repository of information, facilitating the dissemination of educational and scientific content.7 YouTube facilitates the communication of complex topics to a broad audience by integrating visual and audio components and featuring videos crafted by professionals, academics, and creators from various fields. Nonetheless, the considerable variety in provider profiles, presentation formats, and the scientific foundations of the content available on the platform demands a thorough assessment of the quality and reliability of the information provided.8,9 This study aims to evaluate the content and presentation characteristics of YouTube videos related to statistics. Within this scope, videos were divided into quality groups, sources were examined, and basic video parameters and daily interaction metrics were compared between quality groups. In addition, the presentation techniques used in the videos were determined, and their relationships with quality groups were analyzed. METHODS This study was designed as a cross-sectional study. The videos were listed on November 17, 2025. Within the context of the study, the search terms “statistical analysis,” “data analysis,” “statistics,” and “statistical methods” were used to identify YouTube videos. Given that YouTube provides personalized results based on user behavior, searches were conducted in incognito mode after deleting browser cookies and browsing history to minimize personalization bias and ensure consistent search results. This strategy aimed to ensure that the videos included in the research were selected independently of user-specific algorithmic guidance. Search results were obtained with YouTube’s default ‘Relevance’ sorting option, which represents the platform’s standard ranking setting and aligns with a typical user experience.10,11 According to prior research, the average internet user examines only a limited number of search results and regularly accesses items toward the top of the list. Consequently, the first 50 videos for each search phrase were assessed to reflect user behavior.12,13 On the stated date, 200 video URLs were documented, and the key video parameters for all videos were collected on the same day. The exclusion criteria for video evaluation were predetermined. Accordingly, videos in languages other than English, duplicate content, videos unrelated to statistics, and videos exhibiting substantial technical deficiencies in audio or visual quality were excluded from the study. During the video review procedure, two independent researchers, both professors experienced in the evaluation of scientific and educational content, first assessed the videos independently, unaware of each other's scores. After the independent evaluations, the scores were juxtaposed to identify discrepancies among the researchers. A third researcher re-evaluated videos having inconsistencies in their assessments, and the final judgment was based on this examination. The concordance of the researchers' assessments was assessed using Cohen’s kappa coefficient, and the reliability of the evaluation procedure was quantitatively recorded.14 Video parameters In the context of video metrics, the total counts of views, likes, and comments for each video were documented, along with the video’s duration in seconds, the interval between the upload date and the evaluation date in days, and daily counts of views, likes, and comments. Daily engagement metrics (views per day, likes per day, and comments per day) were calculated by dividing the total number of views, likes, and comments by the number of days elapsed since the video upload date. Furthermore, videos were categorized into four main groups based on the presentation methods used: 1) Narrative-focused videos feature a single educator presenting statistical topics primarily through vocal explanations, with limited visual elements. 2) Demonstration-based (software-focused) videos feature details of statistical analyses conducted with software such as SPSS, R, Python, Stata, or Excel, prominently displaying screen recordings and sequential workflows. 3) Animated content comprises videos that elucidate statistical concepts, distributions, and test logic through animations, motion graphics, or visual metaphors, thereby reducing complex knowledge through visual representation. 4) Slide-based presentations incorporate videos that provide information through static or slightly animated slides, which may feature bullet points, formulas, diagrams, or flowcharts, accompanied by narration or text. Sources Videos are classified by the channel source from which they are uploaded. This categorization aims to represent the variations across content producer profiles on YouTube. Seven distinct categories of sources were determined within this area. The academic institutions category encompasses organized information provided by universities, faculties, medical schools, or research institutes, often produced by academic personnel or institutional training departments. The professional experts category encompasses content generated by individuals possessing acknowledged expertise in statistics, biostatistics, data analysis, or research methods. This group comprises personal channels from academics, practical statisticians, and data scientists, as well as instructional videos from field specialists unaffiliated with any institution. The category of healthcare-related organizations includes materials created by hospitals, clinics, healthcare systems, public health authorities, or clinical research facilities, often emphasizing statistical methodologies for health data analysis or clinical applications. The category of nonprofit organizations and professional associations encompasses videos uploaded by scientific societies, professional organizations, and research foundations for educational, professional development, and methodological guiding objectives. The news or media outlets category comprises videos produced by news agencies, media networks, or science communication channels that generally deliver statistical material related to current events, scientific discoveries, or data-driven news. The educational or professional platforms category includes organized instructional content created by online training companies, learning platforms, and analytics-oriented training websites. Although these platforms lack formal affiliation with academic institutions, they provide structured instructional materials. The independent users category comprises statistics videos provided by individual content providers lacking academic, professional, or institutional affiliation, and produced by students, amateur educators, or individual users. Content evaluation In assessing the quality of video content, the Global Quality Scale (GQS) is widely used to evaluate the usefulness and effectiveness of online educational resources. The GQS uses a five-point scale (1–5) to evaluate content quality. A score of 1 signifies that the content lacks coherence, exhibits considerable informational deficiencies, and offers minimal benefits to the user; conversely, a score of 5 denotes that the content is exceptionally coherent, thorough, and profoundly beneficial from an educational standpoint. This study categorizes videos based on their scores: 4 or 5 indicate high quality, 3 indicates intermediate quality, and 1 or 2 indicate low quality. This approach to employing the GQS has been favored in prior studies for objectively and comparably assessing content quality.15,16 To evaluate the reliability of video content, a modified DISCERN tool was employed, which has been widely used in research on online health and educational materials. This tool offers a systematic evaluation that includes characteristics such as content clarity and comprehensibility, absence of bias, degree of neutrality, and availability of sources and supplemental information. The modified DISCERN evaluation comprises binary (yes/no) questions, with affirmative responses earning 1 point and negative responses earning 0 points. The maximum achievable total score on the scale is 5, with elevated values signifying greater content reliability.17 Statistical analysis Statistical analyses were conducted utilizing the Statistical Package for the Social Sciences (SPSS) software (version 20.0; IBM Corp., Armonk, NY, USA). Prior to conducting the analysis, the distribution of continuous variables was assessed using the Shapiro-Wilk test, which indicated that the data did not meet the assumption of normality. Consequently, the findings are presented as median (minimum-maximum), frequency (n), and percentage (%) values, with nonparametric methods used for analysis. Videos were categorized into three groups: low, intermediate, and high quality. The Kruskal-Wallis test was utilized to compare daily views, daily likes, and daily comments across video quality groups. For variables exhibiting statistically significant differences in the Kruskal-Wallis test, pairwise comparisons (post-hoc analysis) were conducted utilizing the Mann-Whitney U test with Bonferroni correction to identify the groups contributing to the difference. In post hoc analyses involving multiple comparisons, the adjusted significance threshold was set at P < 0.0167 (0.05/3). The χ2 test was used to compare categorical variables among quality groups. Spearman’s rho test was employed for correlation analysis. The agreement between researchers during the video evaluation procedure was assessed using Cohen’s kappa coefficient. Unless specified otherwise, a P < 0.05 value was deemed statistically significant in all statistical analyses. Ethics statement Approval from an ethics committee was not required for this study, as it involved analyzing publicly accessible online content. RESULTS A total of 200 video URLs were initially identified across all search queries. After eliminating duplicate videos identified across various search terms (n = 41), 159 distinct videos were evaluated for eligibility. Consequently, 45 videos were eliminated according to established criteria, comprising 22 in languages other than English and 23 unrelated to statistics. No videos were omitted due to technical flaws in audio or visual quality. Following this screening procedure, 114 videos were included in the final analysis. Fig. 1 shows details on the sample selection procedure. The median duration of the videos was 811 (115–55,103) seconds. The median number of views was 625,438.50 (810–15,171,596), the median number of likes was 13,642.50 (0–575,290), and the median number of comments was 284.50 (0–12,000). When examining the daily engagement metrics, the median daily number of views was 529.36 (0.37–4,697.09), the median daily number of likes was 9.51 (0–178.11), and the median daily number of comments was 0.24 (0–5.26). Upon assessment of presentation formats, it was noted that 39.4% (n = 45) of the videos were constructed as slide-based presentations, 21.1% (n = 24) were narrative-centric, 25.4% (n = 29) featured animated content, and 14.1% (n = 16) comprised demonstration-oriented (software-focused) presentations. Table 1 provides detailed information on the key properties of the videos. Flow diagram of the video selection process. Table 1Primary features of the videosVideo featuresValuesDuration, sec811 (115–55,103)No. of views625,438.50 (810–15,171,596)No. of likes13,642.50 (0–575,290)No. of comments284.50 (0–12,000)Days since upload1,626 (53–6,164)Views per day529.36 (0.37–4,697.09)Likes per day9.51 (0–178.11)Comments per day0.24 (0–5.26)Presentation methodSlide-based presentations45 (39.4)Narration-centric24 (21.1)Animated content29 (25.4)Demonstration-based16 (14.1)Data are expressed as median (minimum–maximum) or number (%). Primary features of the videos Data are expressed as median (minimum–maximum) or number (%). Videos were categorized into three groups based on GQS scoring: low, intermediate, and high quality. Twelve videos (10.5%) were categorized as low quality, 15 videos (13.2%) as intermediate quality, and 87 (76.3%) videos as high quality. Upon assessing video sources by quality categories, one (n = 1) (100%) video from academic institutions was classified as high quality, 83.3% (n = 5) of videos from nonprofit organizations, 80% (n = 32) from professional experts, 74.2% (n = 23) from independent users, and 72.2% (n = 26) from educational or professional platforms were deemed high quality. No videos were available from healthcare-related organizations or news sources (Table 2). Table 2Categorization of the videos according to sourcesSourceLow qualityIntermediate qualityHigh qualityTotalAcademic institutions0 (0.0)0 (0.0)1 (100.0)1Professional experts2 (5.0)6 (15.0)32 (80.0)40Healthcare-related organizations0 (0.0)0 (0.0)0 (0.0)0Nonprofit associations1 (16.7)0 (0.0)5 (83.3)6News0 (0.0)0 (0.0)0 (0.0)0Educational or professional platforms5 (13.9)5 (13.9)26 (72.2)36Independent user4 (12.9)4 (12.9)23 (74.2)31Data are expressed as number (%). Categorization of the videos according to sources Data are expressed as number (%). Statistically significant differences were found among video quality groups in terms of daily views, likes, and comments (P < 0.001 for all). The median daily view count was 158.14 (0.37–332.18) in the low-quality group, while it was 694.62 (4.84–1847.14) and 633.03 (9.17–4,697.09) in the intermediate- and high-quality groups, respectively. The median daily like count was 2.63 (0–6.02), 8.23 (0–45.4), and 11.22 (0.11–178.11) in the low-, intermediate-, and high-quality groups, respectively. The median daily comment count was 0.03 (0–0.13), 0.21 (0–1.77), and 0.30 (0–5.26) in the quality groups, respectively (Table 3). Post-hoc analysis indicated statistically significant differences between high and low quality groups for daily views (U = 136.5), likes (U = 131.5), and comment counts (U = 138.5) (P < 0.001 for all). No significant difference was observed between the low-intermediate and intermediate-high quality groups regarding daily interaction metrics following Bonferroni correction (P > 0.0167 for all). The results indicate that the overall significance was mainly due to the difference between the low- and high-quality groups. Table 3Comparison of the video parameters between the low quality, intermediate, and high quality groupsParametersLow qualityIntermediate qualityHigh qualityP valueViews per day158.14 (0.37–332.18)694.62 (4.84–1,847.14)633.03 (9.17–4,697.09)< 0.001Likes per day2.63 (0–6.02)8.23 (0–45.4)11.22 (0.11–178.11)< 0.001Comments per day0.03 (0–0.13)0.21 (0–1.77)0.3 (0–5.26)< 0.001 Comparison of the video parameters between the low quality, intermediate, and high quality groups Upon examining the distribution of presentation formats across video quality groups, it was noted that in the low-quality group, the predominant format was slide-based presentations (66.7%), followed by narration-centric videos (25.0%). In the intermediate quality category, slide-based presentations constituted 53.3%, while narration-centric presentations accounted for 46.7%. In the high-quality group, presentation styles exhibited diversity, with slide-based presentations (33.3%), animated content (32.2%), and demonstration-based (software-focused) videos (18.4%) (P = 0.020) (Table 4). Table 4Distribution of presentation methods within video quality groupsPresentation methodLow qualityIntermediate qualityHigh qualitySlide-based presentations8 (66.7)8 (53.3)29 (33.3)Narration-centric3 (25.0)7 (46.7)14 (16.1)Animated content1 (8.3)0 (0.0)28 (32.2)Demonstration-based0 (0.0)0 (0.0)16 (18.4)Values are expressed as the number of videos (% within each quality category). Percentages are calculated within quality categories owing to the unequal size of the high-quality group. Distribution of presentation methods within video quality groups Values are expressed as the number of videos (% within each quality category). Percentages are calculated within quality categories owing to the unequal size of the high-quality group. Correlation analyses revealed a statistically significant positive correlation between GQS scores and modified DISCERN scores (rho = 0.674, P < 0.001). Additionally, a statistically significant correlation was found between video duration and quality assessment scores (rho = 0.353, P < 0.001). Conversely, no statistically significant correlation was found between days since upload and quality scores (P = 0.805). Agreement between researchers was assessed using Cohen’s kappa coefficient, which was calculated as 0.805. DISCUSSION This study shows that over three-quarters of YouTube statistics videos are classified as high-quality, with significant differences in daily views, likes, and comments across video quality groups. High-quality videos had significantly higher daily engagement metrics than low-quality videos. Upon evaluating video sources, those from academic institutions, nonprofit organizations, and professional experts were more likely to be categorized as high quality. Moreover, presentation approaches showed significant variation across quality groups; the high-quality group exhibited greater diversity. The categorization of over three-quarters of the videos into the high-quality category suggests that YouTube potentially has a solid, rich content pool in the statistics field. Previous research on YouTube concluded that the proportion of high-quality videos varied. These discrepancies are presumed to stem from factors involving the topic matter, sampling procedure, inclusion and exclusion criteria, and evaluation approaches.18,19,20 The superior quality of videos from academic institutions, nonprofit organizations, and professional experts shows that these sources follow more consistent methodologies for scientific accuracy, structured narratives, and credibility. This study identified statistically significant differences in daily views, likes, and comments across video quality groups, suggesting that content quality is a critical factor in viewer engagement. The emergence of significance solely between the low- and high-quality groups in post-hoc analysis suggests that the increase in engagement metrics may be associated with surpassing a specific quality threshold rather than a progressive rise in quality. The significantly higher scores for high-quality videos across all engagement indicators, compared with low-quality videos, indicate that content quality improves the ability to attract, retain, and engage viewers.21,22 These findings affirm that content quality, rather than mere quantitative reach metrics, is a crucial factor in understanding user behavior when assessing online educational content. The substantial disparities in presentation formats between video quality groups indicate that content quality is profoundly affected by presentation approach. The dominance of slide-based presentations in the low-quality group suggests that the content in such videos is mostly presented with a static, limited interactive narrative, resulting in relatively poor cognitive engagement with the viewer. In contrast, the greater variety of presentation formats in the high-quality group, particularly the higher proportion of animated and demonstration-based (software-oriented) videos, suggests that conveying statistical concepts through visual, practical, and dynamic methods can play an important role in improving content quality.23,24 This outcome corroborates the notion that conveying intricate statistical data through animations and interactive demonstrations can enhance both clarity and educational efficacy. The positive correlation between GQS and modified DISCERN scores suggests that quality and reliability in video content align, with high- quality videos providing more reliable content.25 The significant correlation between video duration and quality ratings suggests that extended videos may more thoroughly and intricately cover statistical concepts, thereby enhancing quality. This study has several limitations. Due to its cross-sectional design, the study does not establish a causal relationship between video quality and viewer engagement. The examined videos, derived from search results on a specific date, do not reflect the content and engagement levels that may fluctuate over time owing to YouTube’s dynamic, algorithmic framework. Moreover, assessing the top 50 videos for each search phrase, although intended to reflect user behavior, may fail to provide a comprehensive picture of all relevant content on the platform. Restricting the analysis to top-ranked search results and a specified set of English-language search phrases may have favoured higher-quality content, thus overestimating the number of high-quality videos. As a result, the findings may not be entirely generalizable to all statistics-related videos available on YouTube. Regional variations in YouTube search results may have influenced the findings. Although the video evaluation included acknowledged instruments such as GQS and a modified DISCERN, these scales are likely to incorporate subjective evaluations. Nonetheless, this was countered mainly by a high inter-researcher agreement coefficient. Moreover, as the study encompassed only English-language videos, it was not possible to assess the quality and presentation attributes of content in other languages. Ultimately, viewer engagement measures (views, likes, and comments) provide only a numerical representation of user activity; this study did not evaluate viewers’ learning outcomes, the perceived accuracy of the information, or the material’s actual educational value. The present study shows that a substantial proportion of the videos was classified as high quality, with statistically significant differences favoring the high-quality group in terms of daily views, likes, and comments. Video sources and presentation formats were related to quality levels, with more varied and dynamic presentation methods being employed frequently in high-quality content. The elevated percentage of high-quality videos produced by academic institutions, nonprofit organizations, and professional experts suggests that these sources likely prioritize scientific accuracy, structured narratives, and content reliability. These findings can contribute to a more informed evaluation of online educational materials for internet consumers. Footnotes Disclosure: The authors have no potential conflicts of interest to disclose. Data Sharing Statement: Raw data can be provided to researchers upon request. Author Contributions: Conceptualization: Oralbek A, Zimba O, Qumar AB, Aukenov N, Okyay RA, Kocyigit BF.Data curation: Oralbek A, Okyay RA, Kocyigit BF.Formal analysis: Okyay RA Kocyigit BF.Investigation: Oralbek A, Zimba O, Qumar AB, Aukenov N, Okyay RA, Kocyigit BF.Supervision: Zimba O, Qumar AB, and Aukenov N.Methodology: Oralbek A, Zimba O, Qumar AB, Aukenov N, Okyay RA, Kocyigit BF.Writing - review & editing: Oralbek A, Zimba O, Qumar AB, Aukenov N, Okyay RA, Kocyigit BF. Conceptualization: Oralbek A, Zimba O, Qumar AB, Aukenov N, Okyay RA, Kocyigit BF. Conceptualization: Oralbek A, Zimba O, Qumar AB, Aukenov N, Okyay RA, Kocyigit BF. Data curation: Oralbek A, Okyay RA, Kocyigit BF. Data curation: Oralbek A, Okyay RA, Kocyigit BF. Formal analysis: Okyay RA Kocyigit BF. Formal analysis: Okyay RA Kocyigit BF. Investigation: Oralbek A, Zimba O, Qumar AB, Aukenov N, Okyay RA, Kocyigit BF. Investigation: Oralbek A, Zimba O, Qumar AB, Aukenov N, Okyay RA, Kocyigit BF. Supervision: Zimba O, Qumar AB, and Aukenov N. Supervision: Zimba O, Qumar AB, and Aukenov N. Methodology: Oralbek A, Zimba O, Qumar AB, Aukenov N, Okyay RA, Kocyigit BF. Methodology: Oralbek A, Zimba O, Qumar AB, Aukenov N, Okyay RA, Kocyigit BF. Writing - review & editing: Oralbek A, Zimba O, Qumar AB, Aukenov N, Okyay RA, Kocyigit BF. Writing - review & editing: Oralbek A, Zimba O, Qumar AB, Aukenov N, Okyay RA, Kocyigit BF. References