This article is co-authored with Charles J. Courtemanche, Assistant Professor of Economics at the University of North Carolina at Greensboro.

Our forthcoming Journal of Urban Economics paper measures the effect of Walmart Supercenters on obesity, and we find that the diffusion of Walmart Supercenters explains approximately 10.5% of the increase in obesity since the late 1980s. The published version appeared online at about the same time that Wal-Mart Stores and First Lady Michelle Obama have announced a partnership to increase the supply of healthy foods, and we have picked up a bit of attention from the press. We thought this would be a good time to discuss our research and what is illustrates about how the scientific process works.

After working on this paper for a few years, presenting it at conferences and in research seminars at various colleges and universities, and after collecting comments from numerous people, we arrived at the following conclusions. First, the spread of Walmart Supercenters explains about 10.5 percent of the fourteenpercentage-point increase in obesity since the late 1980s. Second, an additional Walmart Supercenter per 100,000 residents would increase the obesity rate by about 2 percentage points. Third, the additional obesity-related health costs of Super Walmart’s expansion are very small relative to the money American households save because of Super Walmart’s lower prices. Our back-of-the-envelope calculations suggest that Walmart Supercenters save the average household $177 per year, with only 5.6% of the savings being offset by higher obesity-related medical expenses.

The paper went through numerous iterations (and different titles) between the first time we presented a very early draft at the University of Southern Mississippi in 2007 and when it was finally published in 2011. In one version, which we presented at seminars and conferences in Fall 2008 and summarized in the pages of Forbes in June 2009, we were unable to identify an association betweenWalmart Supercenters and obesity, but we found slight reductions in obesity attributable to Walmart discount stores and warehouse clubs.

Eventually, the paper made it to the editors of the Journal of Urban Economics. When an article is submitted for publication in an academic journal, it goes through a peer-review process in which other people in the field review it, comment on it, and make a recommendation as to whether the paper should be published in the journal or not.

We were focusing primarily on associations between county retail structure and body weight, which did not allow us to convincingly disentangle correlation and causation. The editor and referee were enthusiastic about the topic, but wanted us to push further in the direction of identifying causal relationships. On the basis of their suggestions, we implemented a regression technique called “instrumental variables” that allowed us, under plausible assumptions, to identify the causal effect of Walmart Supercenters on BMI and obesity.

With this new technique, we were able to show that Walmart Supercenters increase obesity after all. This is what we expected to find when we first started working on this paper in the summer of 2007: people eat more when food prices fall, eating more leads to weight gain, and Walmart Supercenters reduce food prices; therefore, Walmart Supercenters should increase obesity. Since we could identify the effect of Walmart Supercenters very precisely, we decided to leave our discount stores-and-warehouse clubs investigation for when we can get better data, and we re-wrote the paper to focus on the clear Super Walmart effect.

What explains the first round of results? Apparently, our measured effect was confounded by the fact that Walmart enters communities where economic development is projected to occur the most rapidly. We controlled for current income, but we couldn’t really capture expectations about future income and economic activity. Since rising income is associated with lower body weight, obesity rates in thecommunities that Walmart chose to enter were growing more slowly than obesity rates in other areas for reasons that had nothing to do with Walmart. This led us to underestimate the effect of Walmart on obesity.

Research is a conversation, and the story of how this paper went from the idea stage to publication illustrates just how conversational it is. When the conversation works well, scholars work together to discover truth, or at least avoid error.

The process of answering one question generates a lot of others, and this is no exception. There is much more to Walmart’s effect than just its effect on obesity. Even when we restrict our attention to the health effects of the world’s largest retailer, there is a lot that remains unexplored. In future research, we plan to look at how Walmart affects nutrient intake, food security, and a host of other important considerations. Our papers on Walmart aren’t the first words on the subject, and they certainly aren’t the last. We look forward to seeing what we can find out next.

I’m still trying to make good use of Twitter. For more on the Walmart debate from Forbes, see John Tamny’s recent article “Why Wal-Mart is the Embodiment of Economic Stimulus.”