prior to program implementation generally precludes such unhappy conclusions. Some readers may be disappointed that our method does not refer to statistics and psychometrics. The omission is intentional and does not represent oversight, lack of respect, or rejection. We recognize that statistics and psychometrics can elaborate and refine any applications of this method. However, our method does not require measurement and evaluation expertise.
Thus, program planners and evaluators who do not have training in conventional measurement and evaluation theory may employ the method with full confidence. 0 LITERATURE CITED
1 Cronbach, L. J. et al. Toward reform ofprogram evaluation. San Francisco: Jossey-Bass, 1980, 438 pp. 2 Wholey, J. S. Evaluation: Promise and performance. Washington, D.C.: Urban Institute, 1979, 234 pp.
SOUP A recent multimedia promotional campaign for a brand name soup claims that "scientists link eating patterns to health of American people." A news release further explains that". . . one eating pattern consistently contained a higher percentage of healthy people than the others. This eating pattern, known as Eating Pattern One, was characterized by the consumption of more soup, more dairy products, less sugary foods and beverages - as well as a variety of other foods." In order to determine the validity of this provocative claim, one might examine the data base, the research design, and the definitions of eating pattern and of health. The research basis for the advertising claims is reported in a paper by H. S. Schwerin et al. (American Journal ojClinical Nutrition 34:568-80, 1981). The authors reanalyzed 24-hour food intake recall data and some of the clinical and biochemical assessment data collected originally in the 1968-70 Ten State Nutrition Survey and in the 1971-74 Health and Nutrition Examination Survey I (HANES I). A factor-analysis technique applied to correlations among intakes of 15 food groups revealed 7 statistically different clusters, or "eating patterns." The emergent eating patterns identify some unique inclusions and exclusions in group food intake, but do not offer a clear picture of total diets. For example, low meat consumption and high poultry consumption characterize group V; high fish and fat consumption describe group VII; and high dairy and soup and low sugary food conVOLUME
l3
NUMBER
3
1981
AND
3 Schutz, R. E. The conduct of educational R&D in an era of limits. Los Alamitos, Calif.: SWRL Educational Research and Development, 1981,25 pp. 4 Hanson, R. A., O. E. Behr, B. T. Meguro, and J. D. Bailey. Development and verification of instructionally sensitive achievement tests. Los Alamitos, Calif.: SWRL Educational Research and Development, 1980, 24 pp. 5 Hanson, R. A., R. E. Schutz, and J. D. Bailey. What makes achievement tests tick. Los Alamitos, Calif.: SWRL Educational Research and Development, 1980, 41 pp.
HEALTH
sumption characterize eating pattern I. Obscured by the factor-analysis approach is the fact that the mean grams of solid food (i.e, exclusive of dairy products, soups, and non-sugary beverages) consumed by persons in eating pattern I was less than twothirds of the mean weight of food consumed by the population as a whole. This observation relates to the skewed age distribution among the eating patterns. The HANES I data reveal that 44OJo of the persons in eating pattern I are 9 years old or younger whereas only 19% of the rest ofthe population (i.e., those in patterns II-VII) are 9 years old or younger. Thus, eating pattern I is influenced highly by children's food behavior. The investigators took as a measure of "health" the absence of clinical or biochemical signs and symptoms of poor nutritional status. Despite the availability of data on blood lipids and blood pressure and the association of these factors with diet and disease, these measurements did not enter into the estimation of "health" of the sample. Obesity was considered a clinical symptom for infants, but not for other children and adults. With these definitions in mind, the percentage of persons in eating pattern I with no clinical symptoms was significantly greater than was the percentage of symptom-free persons in any other eating pattern. Eating pattern I also was among the groups with the highest percentage of people without biochemical symptoms. However, when broken down by age groups, the
"health" advantages of eating pattern I are far less impressive. With respect to clinical symptoms, eating pattern I displays no significant advantage in any age category over eating patterns IV, V, VI, or VII and no advantage over any other eating pattern for persons 9 years or younger, between 18 and 24, or 45 and over. With respect to biochemical symptoms, eating pattern I reflects no advantage among persons 9 years old and under or persons between 45 and 64. Overall, the failure to standardize data for age distribution is a serious flaw in the data analysis. Considering the HANES I data, 92, 74, and 55OJo, respectively, of the age groups from 1 to 2, 3 to 5, and 6 to 8 years old displayed no clinical symptoms whereas only 29% of the population 10 years old and over displayed no clinical symptoms. Since the age distribution of the eating pattern I sample was highly skewed towards the younger age groups, age alone accounts for most ofthe apparent superiority of eating pattern I. The use of the term health to describe an assessment that ignores blood lipids, hypertension, and childhood and adult obesity is also questionable. Thus, the research by Schwerin et al. provides little evidence with which to evaluate the "healthiness" of soup in the diet. Nonetheless, the authors present an interesting way to view food intake data. Eating patterns potentially offer a meaningful and useful model for expressing food behavior and for targeting nutrition education. S.M. O.
JOURNAL OF NUTRITION EDUCATION
89