A dataset of key bullying and cyberbullying statistics in the U.S., including prevalence by age, sex, identity, and school environment.

In the 2021-22 school year, about **** percent of female students in the United States between the ages of 12 and 18 reported that they were bullied either online or by text. This is compared to **** percent of male students who were cyberbullied in that year.

Discover key social media bullying statistics, including victim rates, platform trends, age groups affected, and mental health impact!

Context This dataset is a collection of datasets from different sources related to the automatic detection of cyber-bullying. The data is from different social media platforms like Kaggle, Twitter, Wikipedia Talk pages and YouTube. The data contain text and labeled as bullying or not. The data contains different types of cyber-bullying like hate speech, aggression, insults and toxicity.

Content The data is from different social media platforms like Kaggle, Twitter, Wikipedia Talk pages and YouTube. The data contain text and labeled as bullying or not. The data contains different types of cyber-bullying like hate speech, aggression, insults and toxicity.

Acknowledgements Elsafoury, Fatma (2020), “Cyberbullying datasets”, Mendeley Data, V1, doi: 10.17632/jf4pzyvnpj.1

Decrease the number of high school youth (grades 9-12) who report they were bullied on school property from 18.6% in 2013 to 17.5% by 2020.

During the second half of 2024, the largest number of content and account reports submitted to Snapchat involved alleged violations for harassment and bullying, with over 6.3 million reports being due to this type of content. Sexual content was the second highest reason for users to report content and accounts, with 5.2 million reports, followed by spam.

Data on the direct results of bullying and the proportion of bullying victims who experienced these results in the United Kingdom (UK) as of 2020 shows that the most common reported result was depression and anxiety, with 36 and 44 percent of respondents stating so, respectively. These impacts were followed closely by suicidal thoughts, with 33 percent of respondents having stated to develop such thoughts because of bullying.

In the third quarter of 2025, Facebook took action on *** million pieces of bullying- and harassment-related content, down from *** million in the previous quarter. Overall, the first quarter of 2022 saw the highest ever number of bullying- and harassment related pieces of content removed by the platform. Between the first quarter of 2020 and the first quarter of 2021, the removal of bullying and harassment content increased significantly.

This dataset contains annotated social media comments specifically curated for the analysis and detection of cyberbullying. The data is labeled across multiple categories including race, religion, gender, sexual orientation, and miscellaneous attributes. Each comment has been carefully reviewed and annotated by multiple annotators to ensure accuracy and reliability.

Original Dataset in hateXplain was in JSON Format

Sample Entry ->

{
"1179055004553900032_twitter": {
"post_id": "1179055004553900032_twitter",
"annotators": [ {"label": "normal", "annotator_id": 1, "target": ["None"]}, {"label": "normal", "annotator_id": 2, "target": ["None"]}, {"label": "normal", "annotator_id": 3, "target": ["None"]} ],
"rationales": [],
"post_tokens": ["i", "dont", "think", "im", "getting", "my", "baby", "them", "white", "9", "he", "has", "two", "white", "j", "and", "nikes", "not", "even", "touched"]
}

This Dataset is a Scraped and Well-Processed Version of the above data, specially designed for Meta Data Extraction

hateXplain.csv

post_id : Unique id for each post
annotators : The list of annotations from each annotator
annotators[label] : The label assigned by the annotator to this post. Possible values: [Hatespeech, Offensive, Normal]
annotators[annotator_id] : The unique Id assigned to each annotator
annotators[target] : A list of target community present in the post
rationales: A list of rationales selected by annotators. Each rationale represents a list with values 0 or 1. A value of 1 means that the token is part of the rationale selected by the annotator. To get the particular token, we can use the same index position in "post_tokens"
post_tokens: The list of tokens representing the post which was annotated

final_hateXplain.csv

Different Label Categories and Their Respective Values

Race: no_race, african, arab, asian, caucasian, hispanic, indian, indigenous
Religion: nonreligious, buddhism, christian, hindu, islam, jewish
Gender: no_gender, men, women
Sexual Orientation: no_orientation, asexual, bisexual, heterosexual, homosexual
Miscellaneous: none, disability, economic, minority, other, refugee

ACKNOWLEDGEMENT ->

I am here citing the original work and the creators

@inproceedings{mathew2021hatexplain, title={HateXplain: A Benchmark Dataset for Explainable Hate Speech Detection}, author={Mathew, Binny and Saha, Punyajoy and Yimam, Seid Muhie and Biemann, Chris and Goyal, Pawan and Mukherjee, Animesh}, booktitle={Proceedings of the AAAI Conference on Artificial Intelligence}, volume={35}, number={17}, pages={14867--14875}, year={2021} }

Bullying reports the Total number of bullying incidents and the number of students with at least 1 bullying incident at the school district and state level.

Introduction

Social Media Bullying Statistics: Social media is now a regular part of day-to-day life, particularly for young users. These platforms enable people to communicate and exchange ideas easily, but they have also facilitated online bullying. Social media bullying, also termed cyberbullying, mostly takes place when individuals use digital platforms to intimidate, insult, or harm others. Because online content spreads quickly and reaches many people at once, its effects can be more damaging than face-to-face bullying.

Studying this issue is essential to safeguard mental well-being, promote responsible online conduct, and create safer digital environments.

Financial overview and grant giving statistics of Bullying Recovery Resource Center

Financial overview and grant giving statistics of Bullying-Were Kickin It Inc.

The 2015-16 tables are based on data collected from all of the nation’s school districts and schools—approximately 17,300 school districts and 96,300 schools. This set of Excel files contains data on students reported as harassed or bullied or disciplined for harassment or bullying on the basis of sex, race, or disability category for all states. Each file contains three spreadsheets: total students, male students, and female students. The Excel file contains data on allegations of harassment or bullying by type of allegations for all states.

During the second half of 2024, harassment and bullying content accounted for 42.6 percent of total enforcements of reported content on video platform Snapchat. Sexual content accounted for 32.2 percent of content violations, and spam accounted for 2.8 percent.

Financial overview and grant giving statistics of Anti Bullying Teen Rough Ing

This dataset is a collection of over 100,000 Bengali-language social media comments, annotated for various types of cyberbullying behavior. It was created to support research and development in NLP-based toxicity detection, especially in underrepresented languages like Bangla.

The data has been collected, cleaned, and labeled with great care to cover a wide range of offensive, chauvinistic, boorish, and affinity-based online interactions.

Explore alarming cyberbullying statistics, uncover the trends, victim demographics, online behavior patterns, and the growing impact!

Название: Percentage of students experiencing bullying in the last 12 months, low socio-economic status, both sexes (%) Тема: School Safety/Violence Описание: Percentage of students who, during a school year, were physically attacked, participated in a physical fight, experiencing bullying, corporal punishment, harassment, sexual discrimination or abuse. Bullying, includes verbal and relational abuse. This indicator intends to measure experiences related to bullying such as being called by an offensive nickname, being threatened to be hurt, or other students posting offensive pictures or texts about them. Bullying has been linked to reduce academic and health outcomes for victims and for perpetrators. The figure is calculated as the number of students in a given level of education reporting that they have experienced any of the different types of violence or abuse in the past year expressed as a percentage of all students at the same level of education. Data for this indicator may come from two different school based surveys coordinated by the World Health Organization (WHO): a) The Global School-based Student Health Survey (GSHS) developed by the World Health Organization (WHO) and the US Center for Disease Control and Prevention (CDC) in collaboration with UNICEF, UNESCO, and UNAIDS. The GSHS is conducted primarily among students aged 13-17 years and has a global coverage; b) The Health Behaviour in School-aged Children (HBSC) study is a WHO collaborative cross-national study of adolescents' health and well-being administered in schools every four years and using a questionnaire for 11-, 13- and 15-year-olds. In addition, data points could be reported using student background data from international student assessments. For more information, consult the UNESCO Institute for Statistics: http://uis.unesco.org/ Источник: UNESCO Institute for Statistics Код индикатора: UIS.PER.11T15.BULLIED.LOWSES Датасет содержит следующие поля: Код индикатора (indicator_id) — Уникальный идентификатор индикатора Всемирного банка Название индикатора (indicator_name) — Полное название индикатора на английском языке Код страны (country_id) — Уникальный идентификатор страны (код Всемирного банка) Название страны (country_name) — Полное название страны или региона на английском языке