Behavioral Addiction Statistics
Behavioral addiction statistics measure different things, from screening symptoms to diagnosed disorders. Compare US findings, study populations and the limits behind the numbers.
Battling addiction & ready for help?
How Many People Have a Behavioral Addiction?
There is no reliable single total for how many Americans have a behavioral addiction. Studies count different things: diagnosed disorders, symptoms on a questionnaire, or people who describe themselves as addicted. Combining those results would mix populations and could count the same person more than once.[1][2]
That does not make the problems less real. Behavioral addiction describes harmful patterns involving activities rather than substances. Gambling disorder and gaming disorder are recognized diagnoses in the World Health Organization’s classification. Experts have not settled on diagnostic definitions for several other behaviors. That makes it harder to compare how common they are.[3][4][5]
- US gambling estimates: The National Council on Problem Gambling estimates that 2.5 million adults have a severe gambling problem in a given year.[6]
- A screening result is not a clinical diagnosis. Exercise-addiction researchers warn that a high questionnaire score does not necessarily mean exercise is seriously disrupting someone’s life.[5]
- Time spent is a different measure. In a 2024 US teen survey, 45% said they spent too much time on social media.[7] That was a question about their own perception, not an addiction assessment.[7]
Behavioral Addiction Statistics in the United States
These US findings show why the population and question belong beside the percentage. They describe different concerns and cannot be added together.
| Finding | Population and date | What the number means |
|---|---|---|
| 2.5 million adults, about 1%[6] | NCPG estimate for US adults in a given year. The FAQ does not identify a survey year. | Severe gambling problems. NCPG separately estimates 5–8 million adults, or 2–3%, have mild or moderate problems.[6] |
| 13% met a food-addiction screening threshold[8] | A July 2022 national poll of 2,163 US adults ages 50–80.[8] | Symptoms plus significant distress or serious problems in daily life on the modified Yale Food Addiction Scale 2.0 questionnaire. It is not a clinician-assessed rate for all US adults.[8] |
| 11% of men and 3% of women[2] | A study published in 2019 with 2,075 US adult internet users.[2] | Some agreement with “I am addicted to pornography.” These are self-perceptions, not diagnosed cases.[2] |
| 11% of men and 17% of women reported compulsive exercise[9] | 8,251 college students in the US Healthy Minds Study, across academic survey years 2016–2020.[9] | Self-reported compulsive exercise during the previous four weeks. The sample does not represent all adults.[9] |
How to Read a Behavioral Addiction Statistic
Prevalence means how common a condition or measured problem is in a particular group and time period. A study’s sample is the group of people it actually studied. Those people may or may not represent everyone you want to understand.
Ask what a study counted before deciding what its percentage means. A question about feeling addicted, a high questionnaire score and a professional assessment can produce different answers. The pornography and exercise studies above illustrate that distinction.[2][5]
A screening threshold is the score a questionnaire uses to flag a possible concern. A clinical assessment is a closer evaluation of the person and their circumstances. A high screening score can suggest something worth discussing, but it does not establish a diagnosis by itself.[5]
In their 2023 review, researchers Aviv Weinstein and Attila Szabo explain that high exercise-questionnaire scores “may or may not reflect dysfunctional exercise.” They call for interviews and other evidence to complement questionnaire results. Their point helps explain why a positive screen should lead to a closer assessment rather than a label by itself.[5]
Put the Question Beside the Number
When sharing a statistic, include who was asked and what counted as a problem. Those few words can keep a useful finding from becoming a misleading headline.
Use these questions when comparing findings or discussing a number with a clinician:
- Who was studied? Look for age, country, sample size and whether people were recruited from the general population, a school, a clinic or an online volunteer group.
- What counted? Was it time spent, self-identification, a screening threshold, or a diagnosis?
- When? A publication date and a survey date can be years apart. Lifetime experiences also differ from problems in the past month or year.
- How uncertain is the estimate? A confidence interval gives a range around the estimate. A wide interval is a reason to avoid treating one percentage as precise.
- Is the comparison like for like? Check that studies used comparable questions and populations before describing an increase or ranking conditions.
An Example of Keeping the Meaning
“Teens who felt they spent too much time on social media” preserves what Pew asked. Rewriting that as “teens addicted to social media” would turn a question about time into a diagnosis.
Worldwide Estimates During the COVID-19 Pandemic
Read the Whole Range
When a table gives both an estimate and an uncertainty range, keep them together. Choosing only the headline percentage can make a finding sound more certain than it is.
A meta-analysis is a review that combines results from several studies. One published in 2022 brought together 94 studies with 237,657 participants in 40 countries.[1] Its first combined estimate was 33%.[1] After a statistical adjustment for potentially missing research, the estimate fell to 11.1%, with a 95% confidence interval of 5.4–16.8%.[1]
The adjustment addresses possible publication bias. This happens when the research available to a review gives an uneven picture, for example because some results are more likely to be published than others. The large change here is a reason to be cautious about treating one percentage as a settled answer.[1]
The studies were conducted during the pandemic and used different groups of people and ways of identifying problems. Their combined results do not tell us the current percentage of people worldwide with a professionally diagnosed disorder. The broad uncertainty ranges also make comparisons difficult.[1]
| Behavior | Estimate | 95% confidence interval |
|---|---|---|
| Internet | 10.6%[1] | 6.2–15.1%[1] |
| Gaming | 5.3%[1] | 0–15.3%[1] |
| Gambling | 7.2%[1] | 0–15.4%[1] |
| Smartphone | 30.7%[1] | 16.3–45.2%[1] |
| Social media | 15.1%[1] | 0–36.5%[1] |
| Sex | 9.4%[1] | 0–24.6%[1] |
| Shopping | 7.2%[1] | 0–54.3%[1] |
| Food | 21%[1] | 10–32%[1] |
| Exercise | 7%[1] | 3–12%[1] |
Source: Alimoradi and colleagues. The first seven estimates include a statistical adjustment for potentially missing research. Food and exercise do not, because the researchers did not detect publication bias in those categories. The rows therefore do not all use the same adjustment.[1]
Statistics by Type of Behavioral Concern
Gambling Disorder
Gambling disorder has clearer diagnostic recognition than many other behaviors discussed here. However, harm can occur below the diagnostic threshold. WHO describes financial stress, damaged relationships and mental-health consequences among gambling’s harms.[3]
NCPG’s US figures distinguish severe problems from mild or moderate problems. Its FAQ does not provide a dated national sample behind those estimates, so they should not be presented as a newly measured 2026 rate. Its helpline can connect people with support in their state.[6]
Gaming Disorder
Playing video games is not the same as having gaming disorder. WHO says the disorder affects a small proportion of people who game. An assessment asks whether gaming is difficult to control, takes priority over other activities and continues despite harm. It must seriously disrupt daily life, with a pattern normally lasting at least 12 months.[4]
A review published in 2021 combined 53 studies from 2009–2019, covering 226,247 participants in 17 countries.[10] Its overall estimate was 3.05%, with an uncertainty range (95% confidence interval) of 2.38–3.91%.[10]
When researchers used only studies with stricter rules for choosing participants, the estimate fell to 1.96%, with a much wider uncertainty range (95% confidence interval) of 0.19–17.12%.[10] The results depended partly on who was included and which questionnaire was used. That wide range is a reason not to treat the smaller percentage as an exact answer.[10]
Those findings cannot establish that a fixed percentage of American children are addicted. For a family concerned about gaming, missed responsibilities and loss of control provide more useful starting points for an assessment than whether a child enjoys games.[4]
Internet, Phone and Social Media Use
A phone can provide access to gambling, gaming, shopping and social media. Start by naming the activity when talking about a concern. The relevant question is what the person is doing online and how it affects their life.
Pew Research Center surveyed US teens ages 13–17 in September–October 2024 and published the results in 2025. It found that 45% felt they spent too much time on social media, up from 27% in 2023.[7][11]
The 2024 survey included 1,391 teens.[7][11] Researchers adjusted the answers to better represent US teens living with parents. The reported margin of error was ±3.3 percentage points.[7][11] That is one measure of uncertainty around the survey estimates, rather than a guarantee that each percentage is exact.[7][11]
This is a trend in teenagers’ assessments of their time use, not proof of a rise in diagnosed addiction. Pew also found benefits: 74% said social media helped them feel more connected to friends.[7] Both experiences can matter when discussing a particular teen’s use.[7]
For concerns about phone use or social media, the specific activity matters. When seeking help, describe the activity and its consequences rather than relying on a screen-time total alone.
Exercise and Compulsive Exercise
The Healthy Minds finding above concerns compulsive exercise reported by college students. The estimates were 11% for men (95% confidence interval 9.60–12.42%) and 17% for women (15.86–18.24%).[9] The study also found links with mental health symptoms and substance use.[9]
The researchers measured these experiences at the same point in time. This is called a cross-sectional study. It cannot tell us whether exercise problems came before the other symptoms, followed them, or shared another cause.[9]
A separate review included 13 studies with 3,635 participants without indicated eating disorders.[12] Its combined estimate using two exercise-screening questionnaires was 6.2%, based on 2,560 of the participants.[12] These samples and measures differ from Healthy Minds. The review also stresses the need to distinguish exercise pursued for its own sake from exercise connected with eating or body-image problems.[5]
Read more about exercise addiction if you are concerned about loss of control or consequences. A demanding training schedule alone does not answer those questions.[5]
Food-Addiction Screening
In the 2022 National Poll on Healthy Aging, 13% of adults ages 50–80 met the modified Yale Food Addiction Scale 2.0 threshold.[8] That means they reported at least two symptoms plus significant distress or serious problems in daily life during the past year. The poll did not clinically diagnose respondents.[8]
The poll drew from a national group recruited through probability sampling, which gives people a known chance of being selected. Researchers then adjusted the results to better reflect the population. Those methods improve a survey’s usefulness, but the results still describe questionnaire responses within the stated age group.[8]
An older 2011 review illustrates why samples matter. It reported an 11.4% rate using the original Yale scale in a predominantly normal-weight student sample.[13][14] It also presented these results from a separate German sample, with very different numbers of people in each weight category:[14]
| Study’s weight category | Sample size | Met the Yale scale threshold |
|---|---|---|
| Underweight | 60 | 10.0%[15][14] |
| Normal weight | 572 | 6.3%[15][14] |
| Overweight | 86 | 14.0%[15][14] |
| Obese | 32 | 37.5%[15][14] |
These are the study’s category labels, not a way to diagnose someone from body size. The smallest group had only 32 participants.[14] People were included because they were available to the researchers, rather than chosen to represent the whole population. This older study does not establish current US rates. Researchers also continue to debate how food addiction should be defined.[14]
The newer US poll also found screening symptoms across age and gender groups: 18% of women and 8% of men met the threshold.[8] Its findings apply to the surveyed age range. The poll report recommends talking with a healthcare provider about concerns.[8] Food-addiction concerns and eating disorders need careful assessment rather than interchangeable labels.
Compulsive Buying and Shopping
A 2016 review combined findings on compulsive buying. For adult groups chosen to represent a wider population, it estimated 4.9%, with a 95% confidence interval of 3.4–6.9%.[16] That calculation used eight estimates involving 10,102 participants.[16] The full review included 40 studies across 16 countries.[16]
Its college-student groups produced an 8.3% estimate, compared with 12.3% in adult groups that did not represent the wider population and 16.2% in groups selected specifically for shopping-related research.[16] Different recruitment methods and measures help explain the range. Those higher estimates should not be used as the rate for the whole population.[16]
The review found associations with younger age and being female. These are group findings, not a rule about who can experience compulsive buying.[16]
Pornography Use and Compulsive Sexual Behavior
The 2019 US pornography study used an online sample matched to census characteristics. About 11% of men and 3% of women expressed at least some agreement with being addicted.[2] Responses were associated with use frequency and with religiousness and moral conflict about pornography. The authors explicitly distinguish those self-reports from diagnostic standards.[2]
Neither those percentages nor frequent sexual activity establish compulsive sexual behavior disorder. Cleveland Clinic describes an assessment based on difficulty controlling behavior and its effect on daily life. It also notes that the condition is not defined by sex drive alone.[17]
If pornography use is causing concern, tell a clinician what feels out of control and what harm or distress you are experiencing. That gives them more to work with than a comparison with another person’s frequency of use.
Work Addiction
An often-repeated estimate that 10% of US adults may be workaholic came from a tentative synthesis of a few studies, most conducted outside the United States.[18] In his 2012 review, Steven Sussman warned that even that estimate could be high for people most likely to need treatment. It was not a current national diagnostic survey.[18]
A 2023 review combined 53 studies with 71,625 participants from 23 countries.[19] It estimated work addiction in 15.2%, with a 95% confidence interval of 12.4–18.5%.[19] An adjustment for potentially missing research lowered the estimate to 14.1%.[19]
Using only groups chosen to represent a country’s general population or workers lowered the estimate further, to 8.0%.[19] This is another example of why the people included in a study can change its answer.[19]
The authors note that definitions and questionnaires differ and that workaholism lacks the formal diagnostic status of gambling disorder. Long working hours alone should not be treated as a diagnosis.[19][18]
Age, Gender, Harms and Gaps in the Data
There is no single demographic profile that fits every behavioral concern. College-student exercise findings, older-adult food-screening findings and teen social-media surveys describe different populations. Comparing their percentages without their age ranges or measures would obscure that difference.[9][8][7]
Sex or gender differences also vary by behavior and study design. Exercise researchers warn that differences can depend on whether studies compare people who do similar amounts of exercise. The shopping review reports associations with age and gender, but substantial variation across samples.[5][16]
Data about treatment for alcohol or drugs should not be relabeled as evidence about treatment for every behavioral concern. For gambling specifically, WHO identifies stigma and shame as barriers to seeking help and describes harms that extend to family members and household finances. It also identifies poverty, discrimination and other social disadvantage as risk factors.[3]
NCPG estimates an annual US social cost of problem gambling of $14 billion.[6] This is a modeled estimate using a 1999 formula updated for inflation and the adult population as of December 2023, not a measured cost for all behavioral addictions.[6]
Comparable US totals for treatment need, access and outcomes across all these behaviors remain a research gap. The studies here do not establish those totals.
An estimate of how common a problem is does not measure all of that harm. It cannot tell you how much debt, missed sleep, relationship conflict or lost time a particular person is experiencing. A useful next step is to describe the consequence you want to change and ask for an assessment.
Finding Help for a Behavioral Concern
You do not need to know whether your experience matches a national percentage before asking for help. For gambling concerns, call 1-800-MY-RESET or text the National Problem Gambling Helpline to connect with support in your state.[6]
When searching for a therapist near you, ask whether they assess and treat the particular behavior you are concerned about. For online therapy, HHS recommends checking whether telehealth fits the health concern and confirming insurance and out-of-pocket costs with the provider.[20]
Before booking, ask:
- What would an assessment involve, beyond a screening questionnaire?
- Do you have experience with this behavior and related concerns, such as anxiety, debt or eating difficulties?
- Can you provide care in the state where I will attend appointments?
- How would we agree on goals and check whether treatment is helping?
- What are the costs, privacy arrangements and options if I need more support?
Learn about behavioral addiction warning signs and treatment, or explore treatment help to prepare for a conversation about the concern affecting your life.
If you are in emotional distress or a suicidal crisis in the United States, call 988 or text 988. Call 911 for immediate danger.[21]
Can I Compare a Survey Percentage With My Own Screening Result?
They answer different questions. A survey percentage describes how common a result was among the people studied. Your score needs to be considered alongside what is happening in your life. Exercise researchers warn that a high questionnaire score does not always mean exercise is causing serious problems.[5]
Does a Newer Publication Mean the Survey Is Newer?
No. Check the data-collection dates as well as the publication date. Pew’s teen report was published in April 2025, but its survey ran in September and October 2024. A new publication can also analyze an older dataset.[7]
Get Treatment Help
If you or someone you love is struggling with addiction, getting help is just a phone call away, or consider trying therapy online with BetterHelp.
Exclusive offer: 20% Off BetterHelp*Following links to the BetterHelp website may earn us a commission that helps us manage and maintain AddictionHelp.com. *Get 20% off your first month of BetterHelp. Offer valid for new BetterHelp users only. Offer cannot be combined with insurance.
- Fact-Checked
- Editor
