Menu Close

Statistics Are Great! Except When They Aren’t

By Millie Plotkin, MLS

A few years ago, a colleague asked for the source of the statistic “Eating disorders are the third most common chronic illness among adolescents, after obesity and asthma.” It seems like an easy question to answer because that statement is included in many journal articles. All I had to do was look at their citations, right? Ha!

Article 1: This one cited overview article A, which did not actually mention that stat. Rewind! Article 2: Cited article B that cited article C that didn’t have the stat. I wish this was unusual, but it isn’t. Rewind! Article 3: Finally, this cited the actual study! I had the source.

This seemed like great news, except for a few problems. The study* was published in 1991, using data collected between 1935 and 1984. Also, it was done in a city in Minnesota with a population that was 98% white. The authors noted that there were a greater proportion of females in the age 15-25 group compared to the national average. In other words, the group being studied was in no way representative of the United States. Lastly, the authors state that obesity can be measured in different ways, which affects the prevalence data.

To sum up, this is a lousy study to extrapolate to the entire national (and ever worse, global) population. I’m not saying this to attack authors who drop this stat into the introduction of their articles. The intention is to state that eating disorders are serious before moving on to explain their own work in detail. My point is that this gives us a couple things to think about when looking at research articles and picking statistics or other information to highlight.

  1. When was the data generated? Research on treatment generally uses current patients, though sometimes it can be older. The study may include a chart review in which the authors go back through past patient records to find data. When it comes to prevalence data, it is common to use large surveys such as the National Comorbidity Survey. These capture much larger numbers, but may not be repeated for several years, leaving us to depend on older data. Is old information bad? No, not necessarily. But take a few minutes to check if there is anything more recent you can use.
  1. Who is being studied? As mentioned in my first point, some research depends on large national surveys. These try to capture a representative group, meaning the percentages of Blacks or Latinx is similar to the overall national percentage and is seen as a good stand-in for the population. Community surveys, such as the Rochester study, may be skewed if that area is mostly White. When it comes to treatment data, most studies are of white females in a particular age group. There is usually a table of demographics that shows this data. Should we throw all of this research away? Again no, but do your homework so you can discuss the weaknesses.

Finally, I want to talk about use of these types of stats on websites, often as a list of bullet points. Sometimes the site doesn’t even bother giving any citations at all. If you see that, run away! They are just parroting information found on other sites and have not bothered to verify any of it. If the “citations” are just organizations names (NEDA, ANAD, or yes, FEAST), I don’t consider that sufficient. Any reference should be given to an actual study or report, such as the recent Deloitte Economics report from AED and STRIPED. Just citing “AED” isn’t enough, because the reader can’t look at the source themselves to learn more.

I agree with the great American writer Mark Twain  when he said, “Facts are stubborn things, but statistics are pliable.” Stats are very important for supporting advocacy efforts, whether in the halls of Congress or a clinician’s office. But don’t just grab onto a statement because it sounds good and supports your argument. Take a little time to examine the source and ask questions. When in doubt, you can reach out and ask me, your eating disorders librarian, for assistance.

Millie can be reached at

*Lucas, A. R., Beard, C. M., O’Fallon, W. M., & Kurland, L. T. (1991). 50-year trends in the incidence of anorexia nervosa in Rochester, Minn.: a population-based study. American Journal of Psychiatry, 148(7), 917-922.


  1. Danielle

    Great analysis of the problems with statistics. But now I want to know is there current data about the prevalence of EDs among adolescents? It seems like it wouldn’t be that hard to gather those kinds of numbers now that data is shared electronically. We must have something better than 1935-1984, right!?

    • Millie Plotkin

      I am part of a group that is working on a large systematic review of the literature to try and get a better handle on prevalence numbers broken down by age groups, ethnicities, genders, etc. In the meantime, this review showed a point prevalence (measured at a single point in time rather than over the lifetime) of 5.7% for adolescents.

      Galmiche, M., Dechelotte, P., Lambert, G., & Tavolacci, M. P. (2019). Prevalence of eating disorders over the 2000-2018 period: a systematic literature review. American Journal of Clinical Nutrition, 109(5), 1402-1413.

Leave a Reply

Your email address will not be published. Required fields are marked *