Excellent plain language Philadelphia Inquirer article on the impact of innumeracy on health

Health literacy, reading, English language, science, and other colleagues,

JoAnn Weinberger, former Executive Director of the Center For Literacy in Philadelphia and longtime advocate of adult basic skills education, just posted the article Fear of Math is Harmful to Your Health. This woman wants to help to the AAACE-NLA adult basic skills advocacy Google group. It is an engaging, easy-to-read article about the importance of comfort and competence with numeracy for better health. The author, Sandy Bauers, points out that more than twice as many people have problems with numeracy than literacy, and gives examples of what that looks like in their health decisions and behaviors. The article is about  Ellen Peters, a visiting scholar at the Annenberg Public Policy Center at the University of Pennsylvania. She is writing a book, Innumeracy in the Wild, and is a distinguished professor of psychology at Ohio State. At the end of the article are simple suggestions that patients who are not comfortable with numeracy can ask medical practitioners to help them make better health decisions. I like that the article is written in plain language, and includes easy-to-understand examples. Although it is about health, not how to teach numeracy as such, the author makes numeracy in the context of important health decisions feel manageable to learn. The article might be useful for you, and for your students. It might be used, for example, in an ESL/ESOL, reading, health, numeracy, or science class. When the book is available, I expect it will include the research citations, which the article does not.

How might, or might not, you use this article with adult learners, in teacher professional development, or in adult numeracy advocacy?

David J. Rosen




The relationship between numeracy and health is much more nuanced than this article implies. Most prior research on numeracy & health has NOT controlled for background characteristics that are related both to better health and better numeracy scores (e.g., formal education). This is a big problem because then you’re not comparing apples with apples. Instead, you’re comparing the numeracy scores of, say, someone with a college degree versus someone who did not graduate from high school. It’s not necessarily the numeracy that’s influencing their health; it may be their educational attainment or some other characteristic like employment status, age, or access to health insurance.

See my colleagues’ and my articles on this topic, using PIAAC data. We found that once we controlled for these background characteristics, a higher numeracy score was NOT associated with better odds of reporting better health. The same held true for the problem solving in technology-rich environments score. Literacy, on the other hand, was still significant (although it only raised the chances of reporting better health by 3%). 

Prins, E., Monnat, S., Clymer, C., & Toso, B. W. (2015). How is health related to literacy, numeracy, and technological problem-solving skills among U.S. adults? Evidence from the Program for the International Assessment of Adult Competencies (PIAAC). Journal of Research and Practice for Adult Literacy, Secondary, and Basic Education, 4(3), 22 42. https://static1.squarespace.com/static/55a158b4e4b0796a90f7c371/t/57995fe7725e2582f34ddd00/1469669391938/COABE+Journal+Winter+2015.pdf

Prins, E., & Monnat, S. (2015). Examining associations between self-rated health and proficiency in literacy and numeracy among immigrants and U.S.-born adults: Evidence from the Program for the International Assessment of Adult Competencies (PIAAC). PLOS ONE 10(7), e0130257. http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0130257

Prins E., Monnat S., Clymer C., and Toso B. (2015). Examining associations between adult health and literacy, numeracy, technological problem-solving skills, and post-initial learning in the U.S. Washington, DC: American Institutes for Research and the National Center for Education Statistics.  http://static1.squarespace.com/static/51bb74b8e4b0139570ddf020/t/54da78a6e4b0f5214f04f907/1423603878589/Prins_Monnat_Clymer_Toso_PIAAC.pdf

Here is an excerpt from the discussion section of the first article:

"This paper explored proficiencies in literacy, numeracy, and technological problem solving as potential social determinants of adult health in the U.S. Our study is the first to use PIAAC data to analyze this topic. The results indicate that although all three proficiencies are positively related to self-rated health, these skills largely operate through socioeconomic status indicators to influence adult health. Only literacy remained significant after accounting for other respondent characteristics. Consistent with previous research, we found a positive association between numeracy and self-rated health (Lipkus & Peters, 2009; Reyna et al., 2009; Rothman et al., 2006; Rothman et al., 2008; Waldrop-Valverde et al., 2010) and technological problem-solving skills and health (Birru & Steinman, 2004; Gilmour, 2007; Zarcadoolas et al., 2002). However, unlike previous research—including official PIAAC presentations (Schleier, 2013, slide 8)—we tested whether that relationship remained after controlling for various individual characteristics, and found that these other characteristics were driving the association between numeracy and health. Additional research is needed to determine the mechanisms by which literacy enhances health, after accounting for other individual characteristics, and whether our findings apply to adults in other PIAAC countries."


Dr. Ellen Peters emailed to ask me to post her reply to Dr. Esther Prins which you will find below.

David J. Rosen


First, I wanted to thank Dr. Prins for sharing some of her research results.  I have a student looking up her papers for me as they seem quite relevant to what we do. 

With respect to her comments on the numeracy literature, I think more research exists than she realizes. 

First, many numeracy studies have been done with college students for whom education differences are quite minor. These studies tend to be focused on decision-making competence, however, rather than health behaviors or outcomes. They support the role of numeracy in decision making in the effective absence of education differences. Similarly, studies with physicians (who have about equal education levels) nonetheless show numeracy differences (e.g., in shared decision making, communication preferences).

Second and similar to a point made by another commentator, numeracy produces education and so we usually do not control for it in my lab although some people do and find health differences controlling for education and a number of other demographics, risky habits, BMI, trust in physicians, etc. Garcia-Retamero, Andrade, Sharit, & Ruiz (2015), for example, examined numeracy differences in patient health records controlling for these variables. Nonetheless Dr. Prins is correct that studies need to do a better job of controlling for relevant variables.

Third, education is not the best variable in my opinion. Education teaches numeracy and it teaches other non-numeric intelligence. In our studies, we generally control for non-numeric intelligence instead, and effects remain.  Where we do not control for non-numeric intelligence (it's time consuming to measure), we usually will control for education.  And effects remain.

Fourth, what we really need are more studies that experimentally manipulate adult numeracy and examine its causal, rather than correlational, effects on health behaviors.  One initial example can be found here. We are currently working on improved methods to increase effect size, but this paper nonetheless supports its causal effect on healthy behaviors among college students. 

  1. Peters, E., Shoots-Reinhard, B., *Tompkins, M.K., *Schley, D., *Meilleur, L., *Sinayev, A., Tusler, M., Wagner, L., & Crocker, J. (2017). Improving numeracy through values affirmation enhances decision and STEM outcomes. PLoS ONE, 12(7): e0180674. https://doi.org/10.1371/


Of course there are lots of reasons and nuances behind a low numeracy score... but I suspect low numeracy is the most important factor. 

Regardless of why, I think it's a rather huge and powerful answer to "why should I learn math, even if it's hard?" 

"Of course there are lots of reasons and nuances behind a low numeracy score... but I suspect low numeracy is the most important factor."

That is what it means to control for background characteristics: to figure out whether numeracy is, in fact, the most important factor (in this case, for predicting someone's health). That way you can say, "these two people are alike in every way except for their numeracy score. Is having a higher numeracy score still related to having better health?" Our answer was no: once you compare apples with apples (e.g., both people have a high school degree, are employed, are Hispanic married women with young kids, have health insurance, etc.), a better numeracy score no longer predicts better health. By definition, this means that low numeracy is NOT the most important factor; the socioeconomic variables are.

These were all the socioeconomic variables we controlled for: age, sex, race/ethnicity, educational attainment, employment status, lives with a spouse/partner; has children aged 12 or younger; household size; nativity (born in the U.S. or abroad); mother’s and father’s educational attainment; any vision or hearing problems or diagnosed learning disability; health insurance status; and an English proficiency score. So once you take all those into account, numeracy is no longer related to self-reported health. 

 I'm wondering... if I 'control for' a variable that's really closely related, then might I be insuring a "null" answer?   I mean, if I control for good grades in math courses, then am I not also, basically, "controlling for" numeracy, too?     "Educational attainment" and employment status seem awfully closely related to numeracy, since numeracy is awfully handy in getting an education. 

  That said, I agree that the article isn't the stuff of rigorous research... 

The purpose of controlling for X is to disentangle what you're interested in (numeracy) from all the other things that could be related to the outcome (health). Many studies on literacy or numeracy and health don't do this; therefore, those authors can't conclude that it's really numeracy/literacy that's related to health. It's really about educational attainment (or something else), which is strongly related to people scoring higher on numeracy AND being in better health. So if we don't control for educational attainment (and other variables), then we can't conclude that literacy/numeracy is giving you any extra health benefits. Controlling variables doesn't ensure a null answer; it means that you can be 95% sure that any remaining difference is due to literacy/numeracy rather than all the stuff you controlled for. And it means that you're comparing people who are alike in every way except their literacy/numeracy and health status. 

Slide 7 of this presentation represents this idea visually: http://www.bumc.bu.edu/healthliteracyconference/files/2015/09/Prins-health-literacy-conference-PIAAC-presentation.pdf 

While extremely interesting, I wonder how different views on this issue might affect instruction among adult learners, especially those with low literacy and numeracy skills. 
  • Would students benefit from knowing that their health is affected by their lack of skills?
  • Would instructors be led to change their approaches based on that information?
I can certainly see that grant writers would enhance their chances of getting funded for math remediation or even health "remediation" programs if low numeracy is authoritatively linked to low health.   Otherwise, in terms of using the information for instruction, David's suggestion might point in the strongest direction, "It might be used, for example, in an ESL/ESOL, reading, health, numeracy, or science class." For sure. As the debate here naturally developed, it might also encourage debate among students. Leecy