Might consumers interpret GMO labels as a warning label?

Opponents of mandatory labeling of GMO foods often argue that requiring mandatory labels could mislead consumers - making them think there is a safety risk when the best science suggests the opposite.  This is no minor issue, as citizens in Oregon and Colorado will vote on mandatory labeling initiatives this November (previous voter initiatives in California and Washington narrowly failed; legislation in Vermont has already passed).  

Here, for example, is an unlikely critic of mandatory GMO labeling, Cass Sunstein (Obama's former "regulatory czar") writing for Bloomberg.com:

“... GM labels may well mislead and alarm consumers, especially (though not only) if the government requires them. Any such requirement would inevitably lead many consumers to suspect that public officials, including scientists, believe that something is wrong with GM foods — and perhaps that they pose a health risk.”

I have made related arguments in the past, and have even published some prior academic work giving some empirical evidence backing the concern.  However, the evidence is far from conclusive.

Marco Constanigro at Colorado State University and I decided to investigate the issue more directly in a couple studies we conducted last year, which are now published in the journal Food Policy.  

Our research strategy sought to determine whether consumers who were exposed to foods that had GMO labels subsequently indicated higher levels of concern than people who hadn't been shown such labels.  

In the first study, we used apples as the context.  Respondents were randomly assigned to one of three groups.  One group (the control) made choices between apples that did not mention GMOs at all - that had a decoy attribute: ripening with ethylene.   Another group made choices with mandatory ("contains") GMO labels, and another group with voluntary ("does not contain") labels.  The following shows examples of choices we presented to people in the control and treatment groups.

After making several choices between apples like this with different labels, then we asked each set of consumers a bunch of questions about how safe they thought it was to eat GMOs, how concerned the were about GMOs relative to other issues, etc.

Here's the first key result: There was no consistent statistically significant difference in the average level of concern for GMOS expressed by people shown different labels.  That is, the mere presence of the GMO label did not lead to a greater level of concern about GMOs.

However, we can also study the actual apple choices that people made, and use those choices to infer aversion to GMOs.  And here, another set of interesting results emerges:  Consumers' willingness-to-pay to avoid GMOs is more than twice as high in the presence of mandatory "contains" GMO labels as compared to voluntary "does not contain" GMO labels.  Also, willingness-to-pay to avoid ethylene ripening (a common, and heretofore uncontroversial, industry practice) is as high as that to avoid GMOs.

In the second study, respondents were divided into one of two groups.  The first control group was shown an unaltered box of cheerios and was simply asked to click on the areas of the box they found most and then least appealing.  A second treatment group did the same but for a box of cheerios that had, in small print on the bottom left-hand-side of the package the label "partially produced with genetic engineering."   After looking at these packages, we then asked each set of respondents a series of questions about how safe they thought it was to eat GMOs, how concerned the were about GMOs relative to other issues, etc.  The idea is that if GMO labels signal safety then those people who say the mandatory label should subsequently indicate a higher level of concern than those who did not see such a label.

Here are "heat maps" associated with the initial the results where we simply asked people to click on the areas they found most/least desirable.  The top pictures show the clicks for most desirable and the bottom pictures the clicks for the least desirable (clearly people in the GMO treatment noticed the GMO label and found it unappealing):

Here's the key result: There was no statistically significant difference in the level of concern for GMOS expressed among people shown the box with the GMO label vs. the group shown the box without the GMO label.  

Thus, neither study supported our hypothesis that the mere presence of GMO labels would lead people to believe GMOs are more or less safe.  

Here's how we concluded the paper:

“We interpret the evidence as suggesting (at least in the context of our studies) that any signaling effects, should they exist, are likely small and below the ability to consistently detect given our sample sizes of approximately 200 participants per treatment. Nevertheless, we do not believe the results completely rule out the possibility of a signaling effect.

A true labeling mandate imposed by law may well send a different signal about the nature of scientific and public concern than labels shown by researchers on a survey. It is likely impossible for a researcher to impersonate governmental authorities (and the media and culture surrounding a “real world” label implementation) required to fully reproduce the potential signaling effect of a labeling requirement. Our approach – exposing consumers to GM labels via a choice experiment or modified packaging – only simulates exposure to GM labels in a market-like setting, and it must be acknowledged that “real world” effects are possibly more pronounced.

There are at least two other reasons to believe that some forms of signaling are alive and well. First, study 1 reveals that mandatory “contains” labels generated significantly higher implied willingness-to-pay to avoid GE food than voluntary “does not contain” labels. The differences in responses to mandatory vs. voluntary labels may result from the asymmetric negativity effect, which may in turn result from differences in what these two labels signal about the relative desirability of the unlabeled product. The differences in the “contains” vs. “does not contain” may also send different signals and change beliefs about the likelihood that the unlabeled product is GE or non-GE. Second, in study 1 we found aversion to our “decoy” attribute – ethylene ripening – in the control that is on par with aversion to GE food. During fruit storage, atmospheric ethylene is often controlled to slow or accelerate the ripening process (see Sinha et al., 2012), but we are not aware of any significant controversy over its use. Ethylene is a natural plant hormone, and many consumers use the same mechanism when they put a banana in a fruit bowl to induce ripening. Should produce ripened with ethylene also be required to be labeled? Did the mere presence of the attribute on our survey signal consumers that it is an attribute that should be avoided?”

Who are the vegetarians?

One of the challenges researchers face in trying to learn about the characteristics of vegetarians is that there are so few of them.  I've seen estimates that put the percentage of vegetarians in the US population as high as 13%, but most estimates are closer to 5%.  That means that if one does a survey that has 1,000 respondents (which is a pretty typical sample size for pollsters), you'll only have about 50 vegetarians in the sample - hardly a large enough sample size to say anything meaningful.

We've been running the Food Demand Survey (FooDS) for 19 months now, and each monthly survey has over 1,000 respondents.  I took the first years' data (from July 2013 to July 2015), which consists of responses from over 12,000 individuals.  This sample is potentially large enough to begin to make some more comprehensive statements about how vegetarians might differ from meat eaters in the US.

Applying weights to the sample that force the sample to match the population in terms of age, gender, region of residence, etc., we find that 4.2% of respondents say "yes" to the following question: "Are you a vegetarian or a vegan?", which means that 95.8% say "no".  

There is some sampling variability from month-to-month, but overall, the trend in the percentage of respondents declaring vegetarian/vegan status has remained relatively constant, and if anything, has trended slightly downward over time.

So, how do self-declared vegetarians/vegans differ from meat eaters?  The following table shows differences/similarities in socio-economic and demographic characteristics.

Some of the biggest differences appear for age, race, overweight status, and politics.  Vegetarians tend to be younger, less white, skinnier, and more liberal than meat eaters.  Two unexpected results are that vegetarians indicate a much higher rate of food stamp participation (which is a bit surprising since the share of households with >$100K in income is higher for vegetarians than meat eaters) and a much, much higher rate of food-borne illness.  

In our survey, we also measure respondents' "food values" (for detail on the approach, see this academic paper we published).  This approach requires people to make trade-offs (they cannot say all issues are most important).  Respondents are shown a set of 12 issues and are asked to place 4 (and only 4) of them in a box indicating they are the most important issues when buying food, and to also place 4 (and only 4) issues in a box indicating they are the least important issues when buying food.  We measure relative importance by subtracting the share of times an item appears in the least important box from the share of times it appears in the most important box.  Thus, relative importance is on a scale of +1 to -1, and average scores across all 12 items must to sum to zero.  

Meat eaters tend to rate taste and price as relatively more important food values than vegetarians.  Vegetarians tend to rate animal welfare and the environment as more important food values than meat eaters.  Even still, vegetarians rate nutrition, taste, price, and safety as more important food values than animal welfare or the environment.  

The survey also shows people a list of 16 issues and respondents indicate how concerned they are about each issue (1=very unconcerned to 5=very concerned).  As the table below shows, vegetarians are more concerned about all issues than are meat eaters, even an issue like GMOs which is (at present) primarily a plant issue.  The difference in level of concern between vegetarians and meat eaters is particularly large for gestation crates, battery cages, and farm animal welfare.  

Given some previous discussion on the economics of Meatless Monday, I ran some statistical models to determine whether vegetarians tend to spend more or less on food than meat eaters.  

Without controlling for any differences in income, age, etc. that were found in the initial table above, I do not find any statistically significant differences in spending patterns.  Meat eaters report spending about $94/week on food eaten at home and vegetarians report spending about $3 less (a difference that isn't statistically significant); meat eaters report spending about $46/week on food eaten away from home (e.g., at restaurants) and vegetarians spend about $9.80 more (a difference that isn't statistically significant).  Even after I control for differences in income, age, etc., I do not find any significant differences in food expenditures between vegetarians and meat eaters.  The biggest determinants of food spending is income (high income individuals (>$100K in income) spend $35/week more away from home than low income (<$40K in income)) and household size (larger households spend more).  Younger people spend about the same as the older on food a home, but spend more eating out than do the old.  

A short lesson on experimental auctions

One of the most robust findings from the research on what consumers are willing to pay for non-market goods (for example, foods made with new technologies that are not yet on the market) is that people tell researchers they are willing to pay more than they actually will when money is actually on the line.  One review showed, for example, that people tend to overstate how much they are willing to pay in hypothetical settings by a factor of about three.  That means if someone tells you on a survey that they're willing to pay $15, then they'd probably only actually pay about $5.

One way to deal with this problem of hypothetical bias is to construct experimental markets where real money and real products are exchanged.  The key is to use market institutions that give consumers an incentive to truthfully reveal their values' for the good up for sale.  I wrote a whole book with Jason Shogren on the subject of using experimental auctions for this purpose a few years back.

I recently filmed a short primer on the consumer research method for an on-line course being created by my colleague Bailey Norwood.  He graciously put it up online for anyone's viewing pleasure.  

What drives ingredient-based food fears?

That was the question asked in this article just published in the journal Food Quality and Preference.  The authors, Brian Wansink, Aner Tal, and Adam Brumberg surveyed over 1,000 mothers to study which food ingredients they found fearful, and they consider how such fears can be alleviated.  

The abstract:

“This study investigates food fears that are ingredient-based, focusing on the case of high-fructose corn syrup. The results of a national phone survey of 1008 U.S. mothers offer five preliminary sets of observations: first, consumers with a fear of a specific ingredient – such as high-fructose corn syrup – may exaggerate and overweigh perceived risks. Second, such consumers may often receive more information from the internet than from television. Third, they may be partly influenced by their reference group. Fourth, ingredients associated with less healthy foods mainly hurt evaluation of foods perceived as relatively healthy. Fifth, food fears may be offset when an ingredient’s history, background, and general usage are effectively communicated. These findings suggest new insights for understanding how public health, industry, and consumer groups can more effectively target and address ingredient fears.”

From the conclusions:

“When health risks exist, food fears are merited. In other cases, ingredient fears and avoidance may be wrongly based on the stigmatization of an ingredient or on misinformation. These results offer new preliminary insights about who is most prone to ingredient avoidance, where they receive their information, what types of ingredients are most susceptible to being feared, and how fears might be mitigated.

There appear to be at least two non-mutually exclusive motivating factors behind ingredient avoidance. First, some individuals may overweigh the perceived risks of the avoided ingredient. Second, some individuals who avoid ingredients may have a greater need for social approval among their reference group than those with a more moderate view (though such effects were small in our sample). This is a key contribution to the literature on risk because it underscores a novel potential motivation – akin to the Prius Effect – behind ingredient avoidance. ”

Hypothetical bias - Catcher in the Rye edition

Economists tend to be skeptical of the answers people give on surveys - of people's stated preferences.  It's what we do that drives economic outcomes and well-being.  The trouble is that what we say on surveys often diverges from what we do when shopping.  The standard research finding is that, on average, people say on surveys that they are willing to pay about 2.5 times what they will actually pay when real money is on the line - a phenomenon referred to as hypothetical bias.  I've spent a lot of my career trying to devise survey techniques that get closer to the truth and I think we've made progress, but caution is warranted.

A lot of the research on hypothetical bias started in the 1980s and 90s when survey techniques started being more widely used to ask people to state their willingness-to-pay for environmental amenities.  Thus, you can might imagine my surprise when, recently re-reading Catcher in the Rye (published in 1951) the last chapter contained this clear, concise insight on hypothetical bias. 

A lot of people, especially this one psychoanalyst guy they have here, keeps asking me if I'm going to apply myself when I go back to school next September. It's such a stupid question, in my opinion. I mean how do you know what you're going to do till you do it? The answer is, you don't. I think I am, but how do I know? I swear it's a stupid question.