Alcohol Guidelines Fail: Self-Underreporting Is Usual

Federal alcohol guidelines fail. That is, they’re too low. That’s because most people greatly under report their alcohol drinking.

Suppose a study shows a medical problem may occur if people report drinking four drinks per day. That really means that the problem might occur if they have five, six, or many more drinks daily.

The more drinkers underreport their consumption, the lower the levels of any apparent risk becomes.

Research has shown widespread underreporting of alcohol drinking. This is not surprising. Answers to questions about drinking are strongly effected by the social desirability bias.


I.   Self-Underreporting of Alcohol Consumption

II.  Summary

III. Resources: Alcohol Guidelines Fail

Alcohol Guidelines Fail

I. Self-Underreporting of Alcohol Consumption

Comparison with sales and taxes
alcohol guidelines fail
Much underreporting is unintentional rather than lying.

Comparisons of alcoholic beverage sales with survey data are often done. In those, 20-70% of actual alcohol consumption was reported by people.1 However, most is probably a result of social desirability bias.

Using several surveys, researchers found that certain groups tended to underreport more. They tended to be those who were younger, had certain social views, or were Hispanic.2

Researchers in a national Canadian survey found underreporting to range from 38% to 66%. They found it more often among underage and younger drinkers. But there were no sex differences.3

Also, researchers in a national Australian survey found that people under reported consumption by 33%. But so-called binge drinkers underreported by 43%. And middle-aged females under eported by 49%.4

Analysts used responses of 2,198,089 adults in the U.S. They then compared self-reported alcohol consumption to sales. They found from 55.7 to 67.7 underreporting.5

Urine testing

alcohol guidelines failUrinary alcohol was measured daily for up to six months in patients with alcoholic liver disease.  Every week researchers asked the patients about their drinking during the past week. Patients with alcohol in their urine denied drinking 52% of the times that they were asked. And 25% of them denied drinking every time they were asked. Only 17% of all patients admitted it at all times.6 That’s why alcohol guidelines fail.

An acute spinal cord injury clinic analyzed self-reported drinking at the time of spinal cord injury. The found that 50% of those with alcohol in their urine reported not having consumed alcohol before their injury.7

Sweat patch

A study of volunteers who kept a record of their alcohol consumption for one week. They wore a sweat patch indicator. Researchers found that 50% underreported their actual alcohol consumption.8

College students required to attend an alcohol program underreported their history of use. And they explained why. They feared the results of being truthful.9


Biomarkers are substances that show something else. For example drinking. In this case the biomarker was PEth. Researchers studied over 200 women. They found that 53% of participants who denied drinking in the prior 30 days. However, they tested positive for PEth. That is, over half were underreporting.10

Researchers asked emergency room patients about their alcohol use. To do so, patients filled in the Michigan Alcoholism Screening Test (MAST). Researchers also tested for PEth in their blood. They found substantial substantial underreporting. Also, that it was greater by heavy drinkers.11

Bogus pipeline

Underreporting of alcohol consumption has been studied using the bogus pipeline method. It’s used to obtain more accurate responses. Respondents are led to believe that a polygraph (“lie detector”) or other device is being used. And that it will permit researchers to know if the answers are false.

A study of pregnant women using the bogus pipeline technique found that 27% of the women in the bogus pipeline group reported the consumption of alcohol during their pregnancy. That’s in comparison with only14% in the group using only self-reports. That is, they underreported about half the time.12

A study used the bogus pipeline with adolescent students. It found that a higher proportion of students in the bogus pipeline condition reported drinking. That is, compared to the non-bogus condition.13

Randomized response

This approach has also been used to get more  accurate answers about alcohol consumption. The methods is used in interviews. It allows  people to answer while maintaining confidentiality. (To learn more, see the Resources section below.)

Researchers asked 133 school students to complete a survey. It asked if they had consumed alcohol in the past three months. Then they used the randomized response technique. It showed that students underreported their use by 44.45%.14

Similarly, this technique drew a higher responses from high school students on alcohol-related questions. That’s compared with a self-report questionnaire.15

The technique was also used with high blood pressure patients. Those who were questioned using it reported a higher consumption. That was in comparison to those for whom the technique was not used.16

Recency of recall

This is asking about drinking yesterday, last week, or last month. It assumes that recency will enable people to be more accurate.

Alcohol Guidelines failA researcher observed drinking by 58 men in a simulated restaurant. She did so for six nights. The next day after each night she asked each about his drinking. She found that underreporting was greatly reduced. Also she found that heavy drinkers did so more than light drinkers.17

A study asked several thousand adults to keep diaries about their drinking. Diary entries led to less underreporting than did later interviews. Also, those who drank more often and more heavily underreported more.18

Researchers asked women to report their drinking throughout their pregnancies. Five years later they asked the same women about their consumption during that pregnancy. At that later period, there was much more underreporting.19

Researchers asked respondents to think about their drinking the past week. Doing so reduced their underreporting by 17%.20

Computer interface

Researchers made a study of the population of Edinburgh, Scotland. They found two different average self-reported consumption rates. One was 33% higher than the other. The higher rate was when respondents were interviewed by a computer. The lower rate was when respondents were in a face-to-face interviews with a person.21

A researchers made a study of clients of an alcohol agency.  He found that they reported greater amounts of alcohol consumption to a computer than to a human interviewer.22

Another study in Scotland found that alcohol consumption was widely underreported when speaking with an interviewer. Especially so among males, the young, and being employed.23


Researchers asked college students to complete an alcohol use survey. Then they asked the students to do three tasks. They were to pour liquid into different size cups. In doing so, first, they were to judge the volume of a beer. Second, they were to judge the volume of a shot of spirits. Finally, they were to judge the amount of spirits in a mixed drink. In each case, they overestimated the size of a standard drink. Overall, students overpowered drinks by 43.67%.24 So when they honestly answered questions about drinking, they massively underreported.

II. Summary: Alcohol Guidelines Fail

There is overwhelming evidence that widespread underreporting of alcohol consumption in self-reports. The more drinkers underreport their consumption, the lower the levels of any apparent harm becomes. So it’s clear that alcohol guidelines fail. In short, they’re too low.

III. Resources

A. Randomized response technique

Chance decides if the question is to be answered truthfully or not. Or if it should be answered “yes” regardless of the truth. For example, a person can be asked if they drank alcohol yesterday. Before they answer they are asked to flip a coin not visible to the questioner.

Alcohol Guidelines failThey are told to answer “yes” if the coin comes up tails. But they are told to answer truthfully, if it comes up heads. Only they know if their answer reflects the toss of the coin or the truth. 

Half the people who did not drink yesterday get tails and the other half get heads when they flip the coin.

Therefore, half of those who didn’t have a drink yesterday will answer “yes” even though they didn’t do so. So whatever proportion of the group said “no,” the true number who didn’t have a drink yesterday is double that. For example, if 30% people questioned said “no,” then the true number who didn’t have a drink yesterday is 60%.25       

B. Web
C. Footnotes: Alcohol Guidelines Fail
    1. Perm, K. Validity of survey data on alcohol use. In Gibbins et al. (Eds) Advances in Alcohol Problems. NY: Wiley.
    2. Rogers, J. et al. Estimates of alcohol consumption. Can Causes Cont, 25(6), 393-99.
    3. Stockwell, T. et al. Who under-reports their alcohol consumption in surveys?  Addict, 109(10), 1657-66.
    4. Livingston, M. and Callinan, S. Underreporting in alcohol surveys. J Stud Alco Drugs, 76(1),158-64.
    5. Esser, M. et al. Adjusting for the underreporting of alcohol consumption. J Stud Alc Drugs, 83(1), 134-44.
    6. Orrego, H. et al. Alcohol intake based on interviews. Lancet, 2, 1354-6.
    7. Hein, A. et al. Toxicology screening in acute spinal cord injury. Alco Clin Exper Res, 12(6), 815-819.
    8. Phillips, M. An improved patch. Biomat Med Dev Art OrgB, 13-21.
    9. Walker, S., and Cosden, M. Reliability of self­ reported drinking. J Sub Abuse Treat, 33(4), 405-409.
    10. Littlefield, A. et al. PEth as a biomarker of alcohol consumption. AIDS Behav, 21, 1938-49.
    11. Ver, C. et al. Screening for alcohol use. Alc Clin Exper Res.
    12. Lowe, J. et al. Accuracy of self-reported alcohol consumptionJ Stud Alco, 47, 173-5.
    13. Botvin, E. et al. Self-reports of alcohol use. Psych Rep, 55(2), 379-86.
    14. Fisher, M. et al. Substance use in a school-based clinic. J Adol Health, 13(4), 281-285.
    15. Barth, J. and Sand, H. Evaluation in a drinking survey. J Alco Drugs, 37(5), 690-
    16. Volcer, B. and Volcer, L. Estimating alcohol use. J Stud Alco, 43(7), 739-50
    17. Poik, K. Underestimation of recalled alcohol intake. Brit J Addict, 80(2), 215–6.
    18. Boniface, S. et al. Under-reporting of alcohol consumption. BMC Pub Health, 14, #1297.
    19. Morrow, M. et al. Underreporting of alcohol use in pregnancy. Alc Clin Exper Res.
    20. Staudt, A. et al. Does prior recall of past week alcohol use affect screening results? PLoS One, 14(6).
    21. Duffy, J. and Water, J. Under-reporting of alcohol consumption. Brit J Addict, 79, 303-308.
    22. Lucas, R. et al. Patients with alcohol-related illnesses. Brit J Psy, 131, 160-7.
    23. Cooke, D. and Allan, C. Self-reported alcohol drinking. J Stud Alco, 44, 617-29.
    24. White, A. et al. Do college students drink more than they think? Alc Clin Exp Res, 11, 1750-6. Why alcohol guidelines fail.
    25. Cook, J. Randomized Response.

D. Note: Alcohol Guidelines Fail

This site gives no advice. Please see your doctor.