Industry Approaches to Food Research
Eileen G. Thompson1
Not Eating Enough, 1995
Pp. 239–250. Washington, D.C.
National Academy Press
Research in the food industry is very different from academic research, both in terms of the total volume of research done and also the pace of the research. Most research in the food industry is designed around business decisions that must be made on some kind of timetable. Thus, the timetable of a decision largely determines the timetable of the corresponding research.
For example, at Quaker Oats each study is designed around a particular business decision. How can a product be optimized? Should a new product be introduced, or should an old product continue? What new food ideas are desired by consumers? A fair amount of packaging research is also done, because packaging is the consumer contact point with a food product.
As a result, industry does not really check hypotheses in the classic way that academic research does. As product after product and concept after concept is tested in industry, some generalities begin to emerge. These
generalities may influence a change in methodology or a change in approach, simply because industry always wants to optimize its research process.
Another feature of food industry research is the constant inflow of validation data. For example, after determining an optimal product and using complicated models to predict sales of, say, $45.6 million, researchers begin tracking the sales data that come in every week. These data are used by researchers to evaluate their initial prediction.
FOOD INDUSTRY FINDINGS
What follow below are generalizations that researchers at Quaker Oats have developed over many years of testing across a sequence of studies.
One of the measures used widely in food research is consumer acceptance testing, often in the form of 9-point acceptability or "liking" scales. However, researchers at Quaker Oats have found that liking is a multivariate, not a univariate, problem.
For example, the Quaker instant oats products have a number of flavors. If Quaker Oats always marketed the best-liked product from among those tested, every product on the shelf would be some combination of apple-cinnamon-raisin because that is the most universally liked flavor for sweet breakfast goods. Therefore at Quaker Oats, one of the things that researchers ask is: How can multiple products be introduced that satisfy the largest population, even though some of the products may not score high in terms of "liking". Multivariate approaches are used to answer the question. When a line of frozen sandwich products was created, factor analyses revealed a segment of the population with a red meat preference. There was also a group of people who wanted products with lots of vegetables and another who wanted barbecue and other spicy flavors. Researchers responded by choosing one or two products from each of the factors so that the broadest product acceptance across the total product line could be achieved. Quaker Oats researchers have also developed an analysis technique that operates almost as a Venn diagram. It finds the combination of products that optimizes overall line penetration.
Another key finding from the food industry is that basic food behavior is difficult to change, and slow to change when it does. For example, when Quaker Oats tests a product, subjects are asked: What would you substitute
this product for? If subjects cannot think of what they would substitute the product for and say, ''Oh well, I am just going to add it to what I do now," there is a real danger. Although there are products that come along that actually change people's eating behavior patterns, it happens rarely. One example of a product that changed food behavior is bagels. Bagels have taken over, and everybody is building bagel factories. Bagels are an ethnic food that have come into wide acceptance because people have found new ways to use them (as sandwiches for example) and have changed their behavior to incorporate them into their eating patterns. In general, however, the idea that a new product will change people's basic patterns of relating to food is not very likely to lead to success.
How Free-Ranging Consumers Eat
Food researchers need to have enormous in-depth insight into how free-ranging consumers eat. When people do not know that they are being observed (via unobtrusive observation techniques), just what do they eat? If information is desired on caloric intake for young male soldiers, then young men in a variety of heavy work exercise environments outside the military should be observed. Anthropologists can be used as observers when there are behaviors that are not quite understood.
If a comparison of caloric intake between garrison soldiers and soldiers in the field is desired, the following would be a suggested approach. Develop a comparable sample of young civilian men who are engaged in various kinds of heavy work activity, but who have no constraints on their eating. For example, find a construction company during the summer that would cooperate and allow anthropologists on the job site to watch what the young people who are building houses eat and how they divide their calories. Then ask the following questions: What proportion of the total calories are coming from beverages? What proportion of the total calories are coming from snacks? How do subjects eat those snacks?
One of the patterns observed in research at Quaker Oats with regard to snacking is what has been called "mindless nibbling." People often eat while engaged in other activities, and the repetitive crunch-taste-crunch pattern of snacking results in a much higher calorie intake than the person may realize. Civilians involved in heavy work activity may take in a fairly high proportion of calories in this manner.
Some beverages can also boost calorie intake, as can supplementary snacking alternatives. For example, in the cereal industry, some products that are being marketed are actually cereal, but they are designed to be hand-held snacks. Perhaps for military personnel there is a product that soldiers could carry with them, so that they could do the mindless nibbling while working at
something else. This might help alleviate the problem of stopping to eat in the field.
It is important for food researchers to be as innovative as possible in the area of unobtrusive observation. Obtaining a large sample size is not what is important. Rather, it is important to vary the nature of the studies. For example, a particular study might provide insight into a problem, even though the sample size (N) is small. Then the next study would verify and quantify the findings in a more controlled form of testing. It is important to spend more time and resources in the hypothesis development process and less on the verification.
At Quaker Oats, researchers have found various ways to hide video cameras, send anthropologists out to do in-home observation, and perform instore observations. A good example of the necessity for unobtrusive observation comes from the Quaker Oats home mix product line, which includes a variety of pancake and cornbread mixes. What researchers have found is that whenever in-home product evaluation is done, where respondents are given a box of products with instructions on the box, they follow the instructions quite accurately. In real life, however, almost no one measures precisely when making pancakes. To get people to behave the way they usually do, the observation must be designed so that testing is not the primary focus of the subject.
Quaker Oats researchers have also benefitted from unobtrusive observation in the form of video study in the pet food aisle. Researchers were trying to learn how much time people spend looking at the various products. How many of the products do they examine? Do they only go to their one brand? What was found in the area of pet treats was totally unexpected. Observers saw children accompanying their mother, and while mother was grabbing the big bag of dog food, a child said, "Wait, wait, wait, get him a treat!" The children actually climbed up the shelves because the pet treats were on top. This led Quaker Oats to reevaluate the role of children in the purchase process. In this case, observation revealed some important information on who influenced the purchase and better shelf location for the product.
Food Research and Physiology
Food researchers also need to consider physiology. Not only is nutrition knowledge needed, but the science of aroma, taste, and craving must also be investigated. For example, in the food industry, microwave products have been less successful than was expected. Products that were meant simply to be
heated in the microwave have not done well; sales have not increased as predicted. What appears to be happening is that microwaved foods produce different odors than when the same foods are baked or toasted. In the microwave, the chemicals produced actually create a burned smell and taste as opposed to the caramelized smell and taste when foods are baked or sauteed. Engineers have actually experimented with designing odors into the packaging so that when the product is cooking in the microwave, it produces more familiar cooking odors. There may be an opportunity here for the military to enhance odors in order to affect appetites.
In the area of cravings and taste preferences, researchers at Quaker Oats who are testing some products have found that exercise seems to affect taste perceptions. Sports beverages are only tested among people who are exercising at a health club or competing in a marathon. Considering the high exercise levels in the field situations, the initial testing of products should be done with people who are exercising. Even for a small test of, say, the variations of pineapple upside down cake, subjects should be recruited who are coming out of some kind of an exercise situation.
Amount of Consumption
Another methodological issue that must be dealt with in product evaluation is how much product subjects actually eat in a test. For example, one test will give respondents multiple products, and they will be asked to take just a couple of bites of each. In another test subjects may be asked to eat all of a product, for example, a full bowl of cereal. Or subjects may be asked to eat an entire meal and then rate a product as part of that meal. Finally, an extended-use test may be conducted where respondents are asked to eat something day after day. Sometimes the same product will get different ratings across the methodologies. Thus if three different versions of a product are being compared, respondents may give a high rating when they just taste a taste, but they may give a different rating when they eat it 5 days in a row.
In a small-quantity testing situation, probably the optimal product is a candy bar, because the sweeter it is, the fattier it is, the richer it is, the better one bite of it tastes. However, if that same candy bar must be eaten day after day, the sweetness could become cloying. Therefore, a less sweet product may actually be optimal. At Quaker Oats, one sweet, fruit-flavored product was considered optimal based on ratings in the small-quantity testing situation. An absolutely top-notch taste rating was obtained. Yet when the product was developed to the point where people were given it in full-size servings, the consumers did not finish it. Thus the taste of the product was optimized in theory, but the product had not been optimized for the amount actually consumed in a normal setting.
Packaging is absolutely critical for the development of food products, because it can encourage or deter consumption. For example, an employer was worried about the weight gain of his employees. A coffee cart passed through the building every day, and it had a small red awning with little bells hanging from it. All of these details were removed, and the cart was painted battle-ship grey, with the result that sales from the cart decreased by 40 percent.
Another example is the a problem with one new product concept at Quaker Oats. Researchers were working on a whole line of chicken sandwiches and hamburgers. The product idea was one of high variety, including a barbecue flavor, a mushroom-onion flavor, and a guacamole flavor. The picture of the guacamole product on the photo shown to consumers, which showed "green goo" on the chicken sandwich ruined the entire concept for the testers. Consumers disliked everything about the concept. They disliked all the flavors because the guacamole picture was so unattractive that it turned them against the entire line. Thus the military should address the packaging problem with current technology. Camouflage is important for packaging, but much variety can be designed even with the two colors of beige and brown.
The name of products is also important, and what they are called makes a difference in consumption. For example, one product with the same flavoring could be called "apple cinnamon," "apple crisp," or "baked apple." The new product would be tested under each flavor name to determine which name is more appealing and consumer perceptions of how the product tastes. A restarateur known by the author has found that she can manipulate customer orders for dishes simply by the name she selects for the item.
Complex seasonings—particularly in ethnic foods—have a growing role in American food preferences. In many cases, seasonings are replacing fat and salt as the key taste contributors. For example, salsa now has higher sales in the United States than does ketchup. Quaker researchers are finding that consumers like "hot" food. There are different ethnic-based types of seasonings, for example Cajun, Mexican, and Thai, and there appears to be consumer interest in all of these variations. Researchers at Quaker Oats have found that having personal seasoning control is optimal. The inclusion of hot sauce in military MREs is something that Quaker Oats data would support.
Herbs and spices are also becoming more important influences on American choices. Not only are corporate sales of herbs and spices on the rise, both for use in the home and in food service, but increasingly, products are named by an herb or a spice. Thus, one finds pasta with basil sauce or lemon thyme chicken and such products have high appeal for consumers. This interest
in herbs and spices can help the military execute its plan to rotate meals and achieve up to 24 different varieties. For example, chicken breasts can be prepared in a myriad of ways, including Mexican chicken breast and Italian chicken breast. Both are quite appealing.
When it comes to food, perception is reality. One of the problems in the food industry is that when a product runs into trouble, managers simply announce that in this particular area, consumers are being inconsistent. For example, consumers are labeled "inconsistent" because they have a salad for lunch and then have a chocolate dessert. In fact, what researchers have found, especially among young women, is that they are counting fat grams, and they treat fat grams almost as an allowance. If they can save 5 g of fat on their salad dressing and 3 g of fat by having nonfat yogurt, then by the end of the day, they can spend their allowance on chocolate ice cream. The ice cream provides more pleasure, and they end up with approximately the same amount of fat grams.
Thus, the behavior that is often labeled inconsistent may in fact be very rational and consistent. It is important when results appear to be inconsistent, to call in the video cameras and the anthropologists to try to find out what is really going on.
CELIA ADOLPHI: Eileen, at one time I had heard numbers thrown around that the average product in the grocery store stays on the shelf for about 18 months and is replaced by another product. I don't know what the data show today, but products do in fact turn over. Do you see that having any impact on how we manipulate and manage our ration consumption?
EILEEN THOMPSON: Well, one of the things that is happening in the grocery industry is that there isn't enough room on the shelf, and so we run into problems. Recently a consumer was complaining about how our apple-cinnamon granola bar isn't available any more. It isn't that we don't want to make apple-cinnamon granola bars; it's that we wanted to make a new flavor of granola bar, and in order to get that on the shelf, we had to give something up. So I think it's sort of comparable to the situation: you really only have 12 MREs. If you can only have 12 MREs, then you're going to be in the situation of constantly having to replenish and change that set. If you can have 24 MREs, then the longevity of any one product may stay.
The food industry also develops products very quickly to meet consumer-stated demand and then may not optimize that product. We tend to put the product on the shelf and in essence say, "Well, we will get it out here and then we will fix it. Consumers tend to have very little patience for that. If they try something and don't like it, that's really a problem.
RICHARD JANSEN: A comment about flavor in general is that (I think that Barbara Rolls' talk related to this) if you reduce the caloric density of the diet, the sheer volume will result in weight loss. But if you're talking about people that you do not want to lose weight, like we're talking about here, you've got a real problem with fat because fat in fact holds flavor on the tongue. It's not just whether fat is there or not, but the presence of fat causes the flavor to stay on the tongue longer and so when we are going to try to have this idea that we're going to meet all these health objectives, which are really irrelevant for short-term missions, and try to reduce fat at the same time, you've got to be very worried about flavor. I think that that's one reason why they're throwing away foods that are lower in flavor and lower in fat and eating those things that have better flavor.
EILEEN THOMPSON: As I said, I'm not a physiologist. I just wanted to raise the fact that I had heard of studies that actually show a distaste for fat in high exercise situations, such as marathons.
IRWIN TAUB: Your comment about the universal appeal of the apple-cinnamon-raisin flavor is very relevant to the design of let's say the 12-menu MRE taste. I take it you're trying to appeal not to just the large group who like the flavor, but also to other niches that have a liking for other flavors. That suggests to me that you produce your products for different groups of people who like different things, and you do not average the responses. This suggests that for an MRE, instead of just having one entree of each type—i.e., 12 in a case—you might have three of the high ranking one, two of the next ranking one, and at least one of all the others. In that way, no one has to really do without what they like and you still have greater consumption of that particular MRE case.
EILEEN THOMPSON: Yes, you need to take both things into account. There is a floor in "liking" scores, below which you can't go. Let's say we are testing 30 products. We wouldn't even look at the bottom 18 scores, but of the top scores, the four we choose might be 1, 2, 5, and 8 rather than 1, 2, 3, and 4. As I said, we've got a computer program that looks at all combinations and indicates which combination of four gives us the largest number of people who like something in the line.
PRISILLA DOLLOFF-CRANE: Both in Barbara's presentation and the comments that were just made, there seems to be a drift toward going back to higher fat levels within the operational rations because of short-term utilization. Based on what we heard earlier, we're spending more of our lives out in the field in nonchoice environments. We are being exposed, especially on the younger troop level, to the knowledge that fat is bad, with the long-term chronic health problems and such. When you put a captured population in nonchoice environments, you are going to have angry people who distrust the system that is feeding them, who do not want to eat a high-fat diet. They know that high fat is bad, but they have no choice. They're spending a lot more than 2 weeks out there, and it's not short term. We are going out there for extended periods, 50–60 percent of our time, and it is totally wrong to go back to high fat. What we need to do is provide them with information that tells them they've got to consume the calories.
HERBERT MEISELMAN: As a follow up to Irwin Taub's point: I think it's important to keep in mind that our soldiers often don't have the chance to choose their meals. Thus, if food is available that they don't want, they may get stuck with it. For that reason, there of course hasn't been fish and liver in the MREs because many people aren't going to accept them. So I think the comparison to the commercial market is very different. Perhaps we should consider for the future that soldiers should be able to choose. I think it would solve many of these problems.
EILEEN THOMPSON: Yes, I would agree with that again. But when I say we don't just pick the top four, the eight from which we would choose would all have high acceptance scores. So we wouldn't be putting the fish and the liver in there, they would be down on the bottom.
PHILIP BRANDLER: When we've gone through some of our analyses in terms of which items to put into the MRE, it's ended up with three beef items were the top three, then the fourth one down was a chicken. We do a beef and a chicken because otherwise you end up with the problem of all beef items. So, we don't deal specifically with acceptance in the absence of any evaluation of whether or not the items fit together. If they're all three red-sauced pasta items, one is probably enough of those. So I think we've taken at least a nod at that without a complex computer program to back up our decision making.
STEPHEN PHINNEY: I hate to interject anecdotal information, but I'm not the first, so I will. I did a fair amount of endurance bicycling and I can tell you from multiple repeats of the same individual, myself, that when I would push out my training envelope too far, those couple days thereafter I have a high aversion to fat. Now, I eat a lot of fat. In fact there have been certain dietary circumstances where I've eaten probably more fat than most human
beings have as a percentage of total calories, and I can do that. It's not that I distaste it or avoid it. But under unusual physical stress conditions, for instance if I trained up to 60 miles, then I go and do a 100-mi ride and the last 20 miles someone comes along and I try to keep up. For the next couple days, I'm probably going to be eating soup and crackers and light carbohydrates. I just have an aversion to fat. So that might reinforce the comments you made. And it may be germane to when you take soldiers out of the garrison. If they exceed their training envelope, they may at least for a time have a similar problem in terms of avoidance of fatty foods.
GILBERT LEVEILLE: I just want to surface the issue, because I think it's very important that it be put in perspective. We may have a situation where the perception is as you stated and we're going to have angry people. But I think for both the civilian and military population, it's awfully important that we take the knowledge that's been accumulated over the last 30 or 40 years on fat and put it into perspective in terms of what components of fat are of a health concern from a chronic disease point of view. Fat per se from a caloric point of view, certainly in the civilian population, is a problem just in terms of caloric excess. But in the military population where you're trying to increase caloric intake, there aren't too many alternatives in order to achieve a calorically dense product. The question is what kind of fat do you use, and you probably need an educational program that gets around the fact that it's not fat per se but the kinds of fat.
RICHARD ATKINSON: This is the same for fat as not pouring out the baby with the bath water. In trying to get the field rations down to 30 percent fat, we have seen exactly the same thing that Gil Leveille's saying. You can't beat fat for caloric density. And yes you would like to go to 30 percent fat, or go lower in the garrison situation, but Steve Phinney, maybe you could comment on what happens to fat when you're in a high-exercise situation. I think exercise probably covers a multitude of sins. If they're exercising more, they're probably burning off that fat. If you eat a high-fat diet, it tends to put fat in the visceral depots, which is bad. However, when you exercise, that tends to mobilize the visceral depots. Steve, do you have any comments on fat and exercise?
STEPHEN PHINNEY: I would just say that if you're within your training envelope, and this has to do I think with lumberjacks and other people in high-stress situations, they're trained to that environment. They have reached a metastable state in that they don't exceed their envelope and therefore they can better utilize a mixed calorie diet. It sounds like a truism, but a well-trained soldier who is put to an assigned task will probably not have that much disturbance of caloric intake. However, if you take people from a relatively sedentary environment and put them into a high-activity environment, you
have two factors. One is you've changed their environment, and the other is you've added an additional energy expenditure that traditionally people in most human studies, do not compensate for. Thus, you have exercise, which may have its own effects, and then an additional caloric deficit induced by the exercise per se. Keep them within their training envelope. You will probably do very well on what I would call a moderate-fat diet and that's the 30 to 40 percent range.
DAVID SCHNAKENBERG: We've looked at what happens to the serum cholesterol of soldiers out in the field with exercise, but it's kind of a confusing picture. We don't understand our customers. If you're in a situation where you could have a package ration system, the MRE, and we've had individuals lose 4 to 5 pounds in 10 to 12 days. Repeatedly we will see average serum cholesterol dropping say from an average of about 200 mg down to 180 or 185 in those 10 days. I suspect you could go to Pritikin's fat factory and they can't do any better…and we are cheaper.
If you look at the field ration tests in 1985, where we had many soldiers out in the field for 6 weeks eating different rations, our highest fat ration was our A Ration. It was about 40 percent of calories—not inordinately high. In this study the soldiers' body weight stayed relatively stable for the entire period. There were quite a few eggs offered in that menu and in that instance, serum cholesterol rose slightly on the average for the group over a 6-wk period. For those who were fed the packaged T Ration and one MRE, which had an overall low fat content of about 30 percent of calories, serum cholesterol actually went down. Karl Friedl will talk about another situation of severe caloric deprivation over extended periods of time in a range of studies where cholesterol levels had increased at the end of the studies.
KARL FRIEDL: They went from about 150 to 220.
EILEEN THOMPSON: So it's again a rather complex issue.
DAVID SCHNAKENBERG: Can I just clarify one point here? The point I was trying to make is that the situation that people are in and conflicts of their daily lives, like their exercise, not only affects the physiology, but it also may affect taste perceptions. If you're testing which ration does better among people who just sit around, you may not have the same kind of taste preferences that someone out there in that active environment will have.
JOHN VANDERVEEN: In that context, with the hot-weather testing that was done in a climatic chamber, people were working in the heat and then were offered beverages at various temperatures. The effect of a slight cooling of the beverage down to 70¹F had a dramatic impact in terms of volume of fluids actually consumed.
WILLIAM BEISEL: I think the analysis that was made this morning in all of these studies is that the issue to consider maybe a low level of carbohydrates in the rations. We've got a long way to go in this meeting, and we're going to discuss a lot of things, but I think that we certainly are not low in terms of fat—that is, the proportion of fat and protein. However, the area of questions to me is carbohydrates. We need to come back to that and think a little bit more about that.
ROBERT NESHEIM: Ed Horton isn't here to defend himself, but at our last meeting he gave a whole presentation on metabolic events during exercise and he pointed out very clearly that the duration of exercise had a great deal to do with whether you were burning fat or carbohydrates. During the initial parts of exercise, it's primarily carbohydrates, and as exercise is protracted—as in marathoners or long-term bicyclers—after 30 minutes or so, you get up to using mostly fat as the source of energy. So we have to take exercise duration into consideration too.