Students might actually overestimate rather than underestimate their levels of drinking

Sep 20 2005

Numerous studies have shown that misuse of alcohol by college students in the U.S. is a pervasive problem. Most of what is known about consumption levels is based on self-report survey data, which assumes that students are aware of and can accurately calculate how much alcohol they consume.

A study in the September issue of Alcoholism: Clinical & Experimental Research compares students' actual blood alcohol concentrations (BACs) against their self-reported consumption and estimated BACs. The results indicate that students might actually overestimate rather than underestimate their levels of drinking.

"In two of our previous studies, we examined how students define standard, single servings of alcohol," said Aaron White, assistant professor in the department of psychiatry at Duke University Medical Center and corresponding author for the study. In that research, students were asked to pour drinks of different types into containers of different sizes in a laboratory setting. "The findings from those studies indicated that students pour drinks that are way too big, which suggested to us that students probably underestimate how much alcohol they actually drink when they go out."

A tendency to underestimate alcohol consumption, said White, would mean that the students' actual BACs would be higher than their expected BACs.

To test this, researchers collected breathalyzer readings and self-reported drinking data from 152 college students (104 males, 48 females) walking on campus between the hours of 11 p.m. and 2 a.m. during the fall of 2004. BACs were calculated using a standard formula, and the relationship between actual and estimated BACs was examined closely.

"We wanted to measure both actual and estimated BACs and compare them," said White. "If we had been right that students drink more than they think, then their actual BAC levels should have been much higher than expected. In other words, we expected that students would be a lot drunker than they thought. They weren't. Their actual BAC levels were quite similar to the BAC levels estimated based on their self-reported drinking levels. This is actually good news for those of us that like to collect data via surveys. It means that students' self-reported drinking levels on surveys are probably fairly accurate; much more accurate than we expected based on our laboratory studies of drink size."

White added that some researchers may be disappointed that these findings do not fit perfectly with the previous work that has been done. "That's just not how science works," he observes. "We generate hypotheses and then design studies to test these hypotheses. Sometimes you just don't get what you anticipate. That doesn't make the process any less exciting. In this case, our previous laboratory studies suggested that students probably underestimate how much they actually drink. The current study, which involved bringing the researchers to the students instead of vice versa, suggests that we were wrong – self-reported drinking levels are probably more similar to actual drinking levels than we believed."

What these results call into question, said White, is the extension of laboratory studies on college drinking to real-world settings, not the existence of the problem itself.

"The current study says that students are fairly accurate when they report how much alcohol they drink, which means that the drinking problem is … every bit as big as we think it is," he said. "The data from this study just suggest that the problem isn't bigger than we thought."