The Centers for Disease Control and Prevention’s new finding that 38 million Americans engage in binge drinking is, quite appropriately, causing widespread alarm. But below that headline is an important secondary point: We have been underestimating the rate of binge drinking for a long time because researchers haven’t surveyed cell phone users until now. This year’s survey was the first to include cell phones, and the number of binge drinkers jumped as a result.
There's an intriguing history here.
The arrival of landlines in virtually every American home was a godsend to survey researchers. A survey researcher could draw a random sample of phone numbers and be confident that it was representative of the U.S. population.
To preserve this representativeness, researchers wouldn’t necessarily survey the person who answered the phone. If you're old enough to remember growing up in a home with a landline, you'll know why: who answered the family phone was rarely random. For example, in some homes mom always did it even when dad was home, in other homes dad did it only if he had a teenage daughter and wanted to screen potential suitors, and so on.
Telephone survey researchers would handle this by asking the person who answered the landline phone a randomizing question within the household (e.g., the would ask to speak to the person in the household whose birthday was closest to a randomly chosen date). This technique combined with the ubiquity of landlines made household landlines a fabulous way to survey random, representative samples of Americans about political attitudes, dietary habits, product purchasing patterns and a million others things, including, of course, drinking patterns.
The emergence of cell phones ruined all that. Particularly when they first became available, people who owned them were a non-representative sample of the population. Since they were usually not shared, you couldn’t “re-randomize” within a group when someone answered as you could on a shared household landline; your only choice was typically the phone owner. In addition, the phone numbers weren’t usually listed so it was hard to get the sample of phones to call in the first place.
For years I've been attending meetings of alcohol survey researchers during which colleagues lamented the rise of cell phones as a threat to survey research but couldn’t come up with a solution. Each year their response rates were getting lower, and their samples less representative as younger people opted not to have landlines.
The CDC managed to crack into the cell phone survey game with its latest effort - something for which they should be applauded. (How they got the numbers I don’t know.) The changed picture reflected in this year’s results show the advantage of their method, and show that cell phone surveys are the wave of the future not just for drinking problems but for all the other health and social phenomena that are the focus of survey research.
Addiction expert Keith Humphreys, PhD, is a professor of psychiatry and behavioral sciences at Stanford and a career research scientist at the Palo Alto VA. He recently completed a one-year stint as a senior advisor in the Office of National Drug Control Policy in Washington.