Science 30 August 2013:
Vol. 341 no. 6149 pp. 976-980
DOI:10.1126/science.1238041
Research Article

Poverty Impedes Cognitive Function

Anandi Mani, Sendhil Mullainathan, Eldar Shafir, Jiaying Zhao | 5 Comments

Being poor is a state of mind. [Also see Perspective by Vohs]

Add a new comment

These postings do not necessarily represent the views/opinions of Science.

What an interesting perspective. I have recently joined the ranks of the disabled who can no longer work in the community & have had to curtail other activities due to chronic pain & illness. This article helped me to understand why I feel as if my decisions are now less than optimal, leading me to rely on my husband & daughter to help me carry out optimal solutions in health care, finances, work & education, & general life activities. I am very glad that I can at least continue to self educate at home in my professed advocation in the "voodoo" science of psychology. These types of studies and information only help the science & art of psychology. I look forward to doing more research and self-education in this area as my practice has always been in behavior restructuring. This type of research helps me to better understand why clients "do what they do" & help me formulate better behavioral modifications to help them self-improve. Thank-you for giving us all new perspectives & avenues to explore & exploit!

Submitted on Sun, 11/24/2013 - 23:36

The claims in this study depend on specific patterns of significant and non-significant results across a variety of measures and comparisons. The observed empirical perfection in producing these patterns is extremely unlikely given the inherent variability that should be present due to random sampling. The probability of selecting a random sample that produces the necessary results can be estimated by supposing that the reported sample statistics reflect the population values. The probabilities of the reported patterns for the four experimental studies are: 0.32, 0.76, 0.45, and 0.37. The probability of all four such studies producing the necessary pattern is the product of these values, 0.04.

The findings from the field studies also produced successful patterns, so the probability for all six experiments must be lower than 0.04. The description of the first field study does not provide enough detail to estimate the probability of success, but it must be less than 1.0. For the second field study, the probability of a significant result is estimated to be 0.53. Thus, the probability of successful outcomes for all six experiments is no larger than 0.02.

The reported findings seem too good to be true. Given the size of the reported effects and the sample sizes, it is not believable that experiments with random samples would all match the desired significance patterns. It could be that additional experiments that did not support the theoretical claims were not reported, that the reported experiments were run improperly, or that the theoretical claims were over-fit to both the signal and noise in the data. Regardless of the details, the empirical findings in should not be interpreted as proper scientific evidence for the proposed relationship between poverty and cognitive function.

Details of the power analysis can be downloaded from http://www1.psych.purdue.edu/~gfrancis/Publications/SciencePoverty/

Submitted on Sat, 10/05/2013 - 12:53

This is an excellent article and a very well-designed study. The writing is fairly accessible, and given the importance of these findings I wish that the authors and Science would ensure public access to it. I found it through a Facebook share of an article in the Washington Post by a fellow student and would love to repost it, but the news article doesn't do justice to the scientific report.

In the interest of making the public aware of new knowledge regarding the consequences of poverty, I ask that the article be made publicly accessible.

Submitted on Sun, 09/01/2013 - 08:05

Is the statistical significance of their claims a bit overstated here since they don't account for the matched nature of the data? In Table 1 they have sample sizes in the 900s but in reality they only have 462 independent units. Their 'true' sample therefore should be less than what they report (each person is counted twice and doesn't provide independent information). Some of their p-values are pretty low so this may not change their results in a big way, but I thought it was worth mentioning.

Submitted on Sat, 08/31/2013 - 11:33

This title is quite saddening to read. In the comparatively accepted, established and straightforward field of physiology, we must control for most (if not all) variables before claiming anything about the dataset, and can rarely derive causation from the results. With complex human behavior and internal processes going on simultaneously, affected by multitudes of variables of which no one has any control, conclusions about already underprivileged people are being drawn on extremely weak bases. Here, poverty (i.e poor people since it takes a human to be poor) is insinuated to be causing cognitive impediment even in the title, which is just astonishing. Hopefully then, in America, not many poor people will be reading this elite journal to take offense. Here in Scandinavia though, where we treat people more on the basis of human value, some people are, and myself being raised in a poor neighborhood is a (n=1) living disproof of the thesis being proposed.

Luckily I was raised in Sweden, where everybody have an equal opportunity to educate themselves, regardless of background and this should be the lesson learned; we are all the same, so let's start viewing each other as such.

Submitted on Fri, 08/30/2013 - 08:15