Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 43
APPENDIX B ackground Papers for A Workshop on Methods for tile Collection of Aggregate Data on Food Consumption
OCR for page 44
OCR for page 45
lithe Food System: An Overview HAROLD F. BREIMYER The extensive data that have been amassed and replenished regarding U.S. agriculture and the food system originated in large measure from the needs of operating programs. Their instigation with respect to, among others, food consumption data was more practical or programmatic than intellectual. The emerging agricultural programs of the 1930's revealed a need for national statistics of many kinds that gave impetus to statistical collection programs that have long since been divorced from anything farmlike or rural. I joined the Agricultural Adjustment Administration in May 1936 and found economists there worrying about national income statistics and elas- ticity of demand for oranges and the effect of overseas trade, all with a sobriety never known to classroom academicians. I doubt George Jasny in the Department of Commerce has any idea how much the old operating agency of the Triple-A contributed to the national income data series over which he has exercised fatherly care for so long. Although the National Bureau of Economic Statistics properly gets the first credit for pioneering, I remember vividly how national income estimates Qf the department were spliced with the figures of the young Simon Kuznets in trying to arrive at some background statistical underpinning of demand estimates for farm products. And so it was with data on food production and consumption. To be sure, from its beginning in 1862 the U.S. Department of Agriculture had com- piled certain data on production of farm products. These had gradually been augmented with statistics arising from a variety of sources. Although certain marketing service activities had been earned on for a long time and market news reporting became an early fixture, many of the data can only be de- scribed as of happenstance origin. For example, early in our century, Upton 45
OCR for page 46
46 HAROLD F. BREIMYER Sinclair set in motion popular support for meat inspection, as the public and especially export buyers did not like to know about workers falling into open-manhole lard-rendering vats. Once federal meat inspection activities were well under way, the managers of the agency had a natural desire to collect and report the figures on how much work they were doing. Therein commenced a series of data on production of federally inspected meat, which could readily be converted into carcass weight data on how much federally inspected meat was being consumed. I remember well a meeting in the USDA when the meat inspection people decided they were being "had" as they performed a major statistical func- tion without receiving recognition or compensation. Kenneth Miller, chief economist for Armour and Company, joined in the profuse tribute to the agency for the signal contribution it was making to the vital industry of meat packing. The inspectors continued to report the data, though grudgingly. I cite this instance not only as an illustration of the nonmethodical origin of many data but to indicate how "consumption" was viewed. The data were the carcass weight of beef and lamb and product weight of pork going through processing channels. No correction was made for losses at any later stage. Then and to this day, to a very considerable extent, wholesale-level bias entered into data on disappearance of food products. To complete my philosophical commentary on how statistical series came into being, the sequence very often has been that some program adminis- trator needed new numbers. His statisticians scrambled to find whatever were available and engaged in some resourceful interpolating for those that were not. If the data proved truly useful their compilation was eventually assigned to a specialized agency, where corrections and refinements became attached. A fully clothed respectable statistical series then became a part of our data fund. DATA FOR PRICE ANALYSIS During the 1930's the commodity data for which program managers begged were those lending to statistical price analysis. Henry Schultz and Henry Moore were savants to whom Mordecai Ezekiel and Louis Bean gave al- legiance. Price-determining forces were examined for a large collection of agricultural products. The quantity variable often was production or even total supply, rather than disappearance. The inadequacy of good distribution data was not regretted too much. THE 1938 AGRICULTURAL LAW I turn now to the writing of the Agricultural Adjustment Act of 1938. My chief, O. V. Wells, had a major hand in drafting the language of the law. In
OCR for page 47
The Food System: An Overview 47 the previous 5 years a few persons had got the ear and conscience of Secre- tary of Agriculture Henry Wallace, telling him that the interests of consum- ers ought to be taken Alto account in designing and administering farm programs. Secretary Wallace was basically sympathetic but nearly impotent; however, in the 1938 act, language was inserted declaring a concern for protecting the supply of food for consumers. The act contained in its decla- ration of policy an injunction to assist consumers to obtain an adequate and steady supply of agricultural commodities at fair prices. Language in section 304 declared that nothing should be done to discourage producing supplies of food sufficient to maintain normal domestic human consumption as de- termined by the Secretary from records in the years 1920 to 1929, taking into account current trends in consumption and, significantly, the availabil- ity of substitutes. The years of the 1920's were a wise choice because the poorer nutrition of the 1930's, plagued by depression and drought, was thereby excluded. Thereupon began a scramble for data on which to build credible food disappearance and consumption estimates. The first step was clerical. It was to print tabular cards on which supply and distribution data for the various farm commodities could be compiled. The cards had to have many columns. The physical ability to write and read numbers written in miniature longhand was virtually a requirement of employment in Mr. Wells' unit. It would be a gold star on our national history if we could say that thereafter nutrition became an active ingredient in national farm policy. It did not. One reason, however, is that production outran markets so consis- tently that adequacy of food as such was not a serious issue. Another consideration, however, came on the scene at about the same time. It was the proposal to distribute food to lower-income families. Once again the needs of a program, or potential program, were parent to the statistical progeny. In this case, however, a person of outstanding intellectu- ality was the sponsor. He was Frederick V. Waugh. His instincts were both humanitarian and economic. He turned the emphasis from national aggre- gates on food distribution to stratified data. He sought to apply Engel's law about food buying by income class. Demands grew to compile data on food consumption by regional and ethnic groups but especially by family income. The program involved was the new Food Stamp Plan, which was seen as a way both to improve the diets of low-income families and increase the total demand for farm products. A few crude estimates began to appear as to the nutritional adequacy of householders' diets. An interesting offshoot were several studies, one by George Stigler, showing how cheaply a family could obtain adequate nutri- tion. The menu might be plain and dreary but it would have the minimum necessary nutrients. Even the crudest nutritional analyses implicitly required estimates of
OCR for page 48
48 HAROLD F. BREIMYER waste in distribution of food. As I remember, the waste factor data were extremely poor. Again, getting hold of any kind of estimates required in- genuity more than technical ability. All economists working on those studies were aware of another pitfall, namely, the meaning of "measures of central tendency." We might de- monstrate that families of four with an income of $1,500 (in the valuable dollars of those days) would "consume" a quantity of protein or riboflavin equal to the minimum requirement. Obviously, approximately half would exceed the figure and half would fall short. The skewness of the distribution remained, as I recall, totally unknown. Even if the average exceeded the minimum by 20 percent, many of us suspected underconsumption by a considerable fraction of the families. We had almost no reliable estimates of skewness. Of course, then as now, there were debates on just what figures best represented minimum nutritional requirements. But that is another subject. HOUSEHOLD FOOD CONSUMPTION SURVEYS It would be an injustice to fail to note the variety of statistical studies that had some incidental bearing on food consumption. Almost invariably the compilations had a purpose other than to indicate food consumption. The Census of Business gets expenditure data in its retail trade censuses. The Bureau of Labor Statistics has long assembled data on expenditure patterns for urban worker families as a source of weights for their consumer price indexes. For that matter, the U.S. Department of Agriculture has surveyed farm families for a similar purpose, namely, to update weight formulae for the "prices paid" index. More noteworthy are the decennial nationwide Household Food Consumption Surveys that began, as I remember, in 1955. FOOD DIARIES Probably the most complete data obtainable on families' spending for food are those derived from continuous diaries. I am most familiar with a Michi- gan State University diary study that continued for a number of years and with a Georgia Experiment Station diary enterprise that I believe to have been reinstigated. I regret to admit that I do not know to what extent the data so obtained have been exploited for appraisals of nutritional adequacy. PREOCCUPATION WITH WHOEESAEE DISTRIBUTION DATA In this brief introduction I do not come close to doing justice to the virtual profusion of what I call bits and pieces studies relating to food distribution.
OCR for page 49
The Food System: An Overview 49 As a selected example, while writing these notes I came across Marketing Research Report No. 1017 of the Agricultural Research Service of USDA. (The agency has since been renamed.) It is titled Marketing Losses of Selected Fruits and Vegetables. I do not know whether Marketing Research Reports truly had reached the number of 1,017 by 1975 (the date of the study), but in any case a scanning of those reports would reveal lots of bits and pieces bearing in some way on quantification of food distribution. The USDA deserves credit, in my opinion, for the great amount of work it has done. But I add quickly, and negatively, that the focus has almost always been on wholesale distribution; and this in turn is explained by preoccupation with assessing price-making influences. Often, the needs of operating programs have been paramount. Marketing orders, for example, are in force for fluid milk and well over 50 fruits, vegetables, and tree nuts. All these are statistically voracious. Manifestly, the wholesale-level orientation implies a similar bias toward the original farm product identification. In distribution data it is easiest to trace through a product such as eggs, particularly if we do not worry too much about stratification of consumption. That is to say, an egg is an egg irrespective of whether it is eaten fried, scrambled, or part of a cake. But if aggregate consumption data are not sufficient and consumption data by classes are sought, we encounter the problem that the various groups of the population do not eat their eggs in the same ratios of fresh versus ingredients of processed foods. The greater the amount of processing, the less meaningful are data ex- pressed in the terms of the farm-identified commodity. This point is so clear, and its implications so complicating, that I need not develop it further. Dr. Van Meir and others on the 2-day program will clarify this problem, I am sure. INSTITUTIONAL FOOD CONSUMPTION I have left until last the topic of institutional food consumption. It has always been left until last. Only in fairly recent times have major projects been undertaken to assemble data on mass or institutional food programs in all their mazelike complexity. In a sense the wholesale bias has continued prominent. I have not had occasion to dig into the institutional food issue in any depth. Persons working on this NRC project on food consumption patterns will doubtless do so. My hunch is that the significance of institutional food services is not minor. Questions can be so simple such as whether noningestion is greater in
OCR for page 50
so HAROLD F. BREIMYER institutional or household food consumption. My guess is that the ingestion ratio is highest in fast food services. My attention has been directed recently to energy utilization in food distribution and preparation; and here the hypothesis is that institutional services can be energy conserving except that the fast food business is energy-wasteful because of its profligate packaging. REID, BURK, STIEBLING, AND OMAR KHAYAM I caption my final remarks with the names Reid, Burk, Stiebling, and Omar Khayam. I am not sure what individuals most deserve gold medals for developing both data and awareness on food consumption. I know that in the U.S. Department of Agriculture all have had to fight for attention with the political figures who dole out dollars for ccc loans, P.L. 480 foreign food aid, and such. My hazy memory tells me that Jean Mayer was preceded long ago by a man named Henry Sherman. With regard to statistical data my mental association brings forth the names of Margaret Reid and Marguerite Burk; and the underrecognized advocate of giving nutrition more promi- nence in the design of farm programs was that grand lady, Hazel Stiebling. She was head of home economics in USDA. But my favorite citation in this connection is from Omar Khayam. He said he was no great guy; he was only trying to reduce the year to better reckoning. Perhaps we ought to award a few plaudits to those who now and henceforth try to reduce food distribution data to better reckoning. I personally applaud all those who will join the National Research Council in doing so. All I can add is that they will find a mixture of a wealth of bits and pieces, and gaping holes; and fitting the pieces together will prove to be an enormous jigsaw puzzle but an in- teresting and worthwhile one.
OCR for page 51
Measurement and Forecasting of Food Consumption by USDA ALDEN C. MANCHESTER and KENNETH R. FARRELL INTRODUCTION In the federal government, responsibility for measuring and forecasting food consumption is in the Economics, Statistics, and Cooperatives Service (ESCS) of the Department of Agriculture. Consumption is an integral part of measurement and forecasting for food and agriculture, including produc- tion, stocks, international trade, and prices. This paper discusses the system in which measures of per capita food consumption are developed, the sources and quality of the data used, and some of the uses of the data. It also considers data on food expenditures and some of its uses. It concludes with a discussion of the known deficiencies of the data base and possible steps to deal with the gaps. THE SYSTEM FOR MEASURING CONSUMPTION The basic concept is of a commodity flow data system (Figure 1). Agricul- tural products are produced on U.S. farms, caught by U.S. fishermen, or imported from abroad. Most move to manufacturing plants for processing and/or preservation. Then they move through the distribution system to retail stores or eating places and to consumers. The USDA system measures all food in the commercial system. Farm home production of foods except vegetables also is measured. Rough esti- mates are made of farm garden vegetables and of nonfarm home production. Food consumption is measured at the national aggregate level for 260 foods (see Appendix to this paper). There is no breakdown between food used at home and that used in restaurants and other away-from-home outlets. The use of supplements (vitamins, stabilizers, etc.) is not measured. 51
OCR for page 52
52 _ a, g 0 C' ~ ._ ' cn V, ~ — ~ ~ s ~ 4 - C C ~ o ~ o Z ~ I c ._ o E 11 j C ~n 4 - a o c ~ Q E ~ —C.D ~ ~ ~ E / . ~ o U. _ E ~ _ - C' / Q o C ~ ~ C s° ~C ~ Q ~C . _ a C) y —m O ~ C I o ~ ~' a, o - ._ . T ~ . ., . ~o . s~ o O ~ ·. C~ c a, ._ c C~ U ·— y Q `,, cn "c a) O ·. o o ._ Cs~ 4_ cn ._ .. C _ ro .- 4 ~ · ·- C53 c a' ._ mm ~ aC, O ~ X ~ C U' ; - C~ Ct - Ct ;^ - ._ o o LU
OCR for page 53
Measurement and Forecasting of Food Consumption by USDA 53 The basic tool is a supply and utilization balance sheet for each commod- ity (Table 11. Supply of each food consists of beginning stocks, production, imports, and in-shipments from territories (mostly Puerto Rico). Utilization consists of exports, shipments to territories, government purchases for military use and export, nonfood use, food use, and ending stocks. The availability of data is such that civilian food use is calculated as a residual after other measured uses are deducted from the total supply. Thus, it is often called disappearance. Estimates of consumption (disappearance) are prepared at two levels for each commodity. The basic measurement is at the primary distribution level, which is dictated for each commodity by the structure of the market- ing system and the availability of data. For some, it is at the farm gate. For most commodities which are processed, it is at the processing or manufac- turing plant. Once the primary level of distribution has been selected, quan- tities of all other components in the balance sheet for that commodity are converted to the primary weight basis using appropriate conversion factors. For example, the primary distribution level for red meat is the slaughter plant, so all quantities are converted to carcass weight. Most users of USDA consumption data are accustomed to the retail weight figures, which translate from primary distribution weight to retail weight by means of conversion factors that allow for subsequent processing and losses in the distribution system. Fresh beef, for instance, loses 26 percent of its weight from carcass to retail cuts. For some uses, a more desirable basis of computation is edible weight. We have calculated per capita consumption on that basis for special articles (Manchester, 19771. That calculation avoids the problems that arise par- ticularly because of the shift from fresh to processed products such as fruits and vegetables. We are developing such additional consumption series as an adjunct to those using primary distribution weight and retail weight. The ideal system for nutritional analysis and many other uses would be one that measured actual ingestion of foods. No data are available to make such estimates for the U.S. population. While it seems highly unlikely that such data will ever become available except from carefully controlled laboratory tests, it may be possible to move closer to actual consumption. For example, the Market Research Corporation of America has conducted menu surveys that record foods actually served and those present. Such data give no assurance that the food was consumed by all present. And, of course, no attempt is made to measure quantity for each individual. THE DATA SOURCES AND QUALITY The supply and utilization data system for food products is entirely depen- dent upon data that are collected for other purposes. No funds have ever
OCR for page 76
76 OLIVER S. CASTLE LA KE ONTA R10 Niagara Orleans Genesee 1 Wyoming , Allegany Taraugus ~ ROCHESTER ; ;' I'm' '''''''1 :§ Monroe Livingsto NEW YORK ~ Ontario 1 1 1 1 1 .~ ~~J \ ~ S;t~llhPn ~ ~ ~- I Chemung ~ PENNSYLVANIA . FIGURE 2 Rochester, N.Y. De Soto ~ | St. Lucie) Lake Martin Glades (Okeechobee . FLORIDA Collier . I. :. ~ ..... . Brov~ Ird i ~—~ ~ Palm Beach .. _ ........................ FT. LAUDERDALE _ -O FIGURE 3 Miami-Ft. Lauderdale, Fla.
OCR for page 77
A. C. Nielsen Company Services TABLE 4 Nielsen Major Markets % of U.S. % of U.S. % of U.S. Metro Population Grocery Sales Drug Sales Areas (Jan. 1, 1976) (1975) (1975) New York 7.5 6.9 5.9 Los Angeles 4.7 4.9 5.8 Chicago 3.6 3.4 4.8 . Philadelphia 3.6 3.7 3.2 ''. ~ Baltimore-Wash., D.C. 3.2 3.2 4.3 Boston 3 .2 3.2 2. 8 Detroit 2.9 3. 1 3. 7 cat San Francisco-Oakland 2.3 2.6 3.2 4,, Q ~ Cleveland 2.1 2.3 1.9 & ~ Pittsburgh 2.0 1.8 1.7 Dallas-Ft. Worth 1 .5 1 .6 1 .6 Miami-Ft. Lauderdale 1.4 1.8 1.9 Minneapolis-St. Paul 1.4 1.3 1.2 St. Louis 1.4 1.4 1.3 Atlanta 1.3 1.4 1.3 c ~ Cincinnati 1.3 1.2 1.1 ~ A Houston 1 .3 1 .5 1.2 38 ~ Indianapolis 1.3 1.1 1.4 in. Seattle-Tacoma 1.1 1.3 1.3 O '' Kansas City 1.0 0.9 0.9 ~ ~ ' Memphis 1 .0 0.9 0. 7 it, oQ ~ Milwaukee 1.0 0.9 0.8 _ Portland 1.0 1.2 1.0 Buffalo 0.9 0.9 0.9 Denver 0.9 0.9 1.0 Nashville 0.8 0.9 0.8 Phoenix 0.8 0.9 0.9 Sacramento-Stockton 0.8 0.9 1 .2 Birmingham-Anniston 0.7 0.7 0.6 Charlotte 0.7 0.7 0.7 Grand Rapids-Kalamazoo 0.7 0.8 0.6 Louisville 0.7 0.6 0.6 Albany-Schenectady -Troy 0.6 0.7 0.5 . Oklahoma City 0.6 0.6 0.5 In. Omaha 0.6 0.5 0.6 ~ ~ O San Antonio 0.6 0.6 0.5 it, Q ~ Jacksonville 0.5 0.5 0.7 _ Rochester 0.4 0.5 0.5 TOTAL MAJOR MARKET AREAS 61.4 62.3 63.6 77 NOTE: All population and household data are from Sales and Marketing Management magazine estimates and, unless otherwise stated, refer to the condition as of January 1, 1976. Store count and volume data are Nielsen estimates and refer to 1975.
OCR for page 78
78 OLIVER S. CASTLE Back in 1959, when the research and development of new products were becoming a way of life for large grocery manufacturers, the need for a data source that was both broad in scope, i.e., encompassing virtually all products sold through supermarkets and yet reliable enough to identify and evaluate growth opportunities in the grocery field became apparent. To fill this need, we developed a new system of services called Nielsen Early Intelligence System (NElS). The basic service in this system Data Services is based on a summary of warehouse-to-store shipments of all grocery items from a nationwide sample of 150 supermarkets and reported every 60 days. First, all packaged grocery items are merged into 600 logical product groups but each individual brand within the group is reported separately by brand, size, type, flavor, etc. Some 430 product groups are edible foods, and the remaining 170 are assorted product categories such as detergents, soaps, health and beauty aids, and other general merchandise. We originally thought the primary use, and perhaps sole use of Data Services, would be to identify and quickly appraise product areas of oppor- tunity for manufacturers interested in developing new products and markets outside of their areas of interest. Even though the service was eminently successful in this regard, we soon discovered, as did our clients, that the data had additional uses or applications in a variety of other situations. Table 5 shows a sample report page from Data Services. The product class used in the table is Canned Beef Stew. Note that the unit movement for the total market for October-November 1977 compared to the same period of 1976 reveals an increase of some 2,900 units or over 7 percent versus a dollar gain of about 8 percent. Also provided are data for each brand of beef stew by size, type, flavor, etc. Note that Dinty Moore Beef Stew had a 1977 unit share of 48.7 percent of the total market, up four points from the previous year, and a 53.5 percent share of the dollars spent for beef stews. During this 1977 period, 99 percent of the sample stores purchased at least one size of the brand indicating wide distribution, and average unit movement for the brand was 12.7 cans per week. These same data are provided in identical form for each and every brand in a product group—however many there are. Thus, it is readily apparent that the entire range of food products can be monitored in order to detect any major changes or shifts in consumer preferences or product innovations. Table 6 provides a "shopping list" of the types of information our clients have obtained from Nielsen Early Intelligence Systems. As you can see, the real strength of this service is its ability to answer straightforward questions with direct answers relative to marketing trends in terms of prices, pack- ages, types, etc. Point three may need a moment's explanation. By "controlled brands"
OCR for page 79
79 3 m an l Be: . o Cal 5 - ._ l ad Cal ._ V) Cot ~ .= ~ cat V) as .c~ A: ~4 ._ - - Ct Cal is ;^ ~ A: Cal > ~ O 3 ~ o ~ CQ Pa Cat o - o ~ ~ Al o U' o ._ ;: ~ em z o 1 EN O . ~ m _ Sit ~ ~ 00 ~ ~ . . . . . . . \0 ~ ~ ~ ~ V) ~ ~ _ ~ ~ ~ ~ O . . . . . . . 00 US ~ _ 00 ~ _ _ O _ 00 ~ O ~ ~ ~ O ~ V) ~ ~ Cat \0 O · O O O 00 ~ ~ 00 ~ _ _ 00 _ _ · C~ O 0\ ~ ~ ~ \D ~ ~ _ ~ C~ · x · ~ ~ ~ ~ - · c~ oo ^ ~ ^ ^ ^ ^ (~) ~ o ~ ~ - oo ~ - ~ - - oo ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ o ~ O U~ ~ ~ u~ ~ ~ ~ d. u~ ~ r~ oo ~ ~ O · ^o ^ ~ o x ~ ~ oo ~ - ~ - - ~ ~ ~ ~ o ~ ~ ~ ~ ~ ~ o o o ~ ~ ch · ~ · ~ oo - oo ~ . ^ ~ ^ ^ ^ ^ - o ~ ~ ~ ~ ~ - ~ - - t - vo ~ ~ oo ~ ~ c~ ~ ~ ~ . . . . . o o o - o o u~ c-) 'e cl . - v) o . - c) c~ m _ 0 ~ ~ ~ 0 0 0 u~ ~ o — ~ ~ — 0 0 ~o _ 3 3 3 3 v~ v~ c~ v~ 3 3 mmmm 0 ~ 0 0 0 0 0 ~ 0 ~ ~ m" ~ ~ m" ~mm — = — s ~ ~ ;^ ~ — ;~. ;~ O g ·~ — O ~ ~ ~ E~ ~ ~ ~ ~ E~ ~ ~
OCR for page 80
80 O LI VE R S . C AS T LE TABLE 6 NElS Data Services for Identifying Significant Consumer Buying Trends 1. Price (Lower/Higher) 2 . P. a c k a g e ( C o n v e n i e n c e / D u r a b i 1 i t y / N o v e 1 t y ) 3. Brand (National, Regional vs. Controlled) 4. Type (Moist/Dry Chunky/Regular) 5. Size (Larger/Smaller) 6. Formula (With Borax/Ammonia With Raisins/Dates) 7. Nutrition (Natural Ingredients/All Beef) 8. Convenience (Instant Breakfasts/Spray Cleaners) 9. Flavor/Color/Design (BBQ Flavor/Floral Print Design) 10. Diet (Low-Calone/Cholesterol) 11. Ecology (Regular vs. Recyclable) we mean private label, chain's own, or brands such as TOPCO, which, for example, may be marketed by two different chains one in Cleveland and one in Kansas City but are exclusive to these chains in their respective areas. These controlled brands are reported in the same detail as other brands size, type, etc. but not identified by chain—only as "controlled brands. " Other trends that are also identifiable through our Data Services report may relate to nutrition, convenience, diet, or be ecologically based on container type. Although clients often purchase Data Services alone, many utilize it in conjunction with Nielsen Product Pickup Service, since by means of the pickup service it is possible to obtain samples of new products just intro- duced into the market. As the name indicates, Product Pickup Service is just that; it's a facility for obtaining samples of any grocery product introduced into the market—at any place or at any time. By means of 350 field agents, covering all principal cities and the areas between, we can supply one sample unit, or many hundreds of cases, of a specific brand to our clients. We are prepared to handle all perishable products frozen, fresh, or other fragile items and ship them anywhere, in special containers whenever necessary. We can pick them up on a regular schedule weekly, monthly, quarter- ly or on a one-time-only basis. We can also secure samples in 23 addi- tional countries through our overseas companies. Over the years, we have obtained retail samples for an ever-widening variety of reasons or purposes. Table 7 lists a few of the more common reasons. They vary from wanting to check a competitive product for any number of reasons to qualitative assurance reasons relative to taste, flavor, color, freshness, nutrition maintenance, and so forth.
OCR for page 81
A. C. Nielsen Company Services TABLE 7 Some Reasons for Using Nielsen Product Pickup Services 1. Competitive Products- New end established 2. Average Age of Shelf Product 3. Taste and Flavor 4. Color 5. General Appearance Label, Contents, etc. 6. Nutrition Maintenance 81 One of the more interesting programs we are now conducting is for the Food and Drug Administration, which utilizes the resources of not only Data Services and Product Pickup, but also our Food Index Service. The key here was Nielsen's unique ability to provide accurate market data coupled with a highly trained, nationwide field force capable of handling in-store assign- ments. Here's how these services are being used to solve a problem facing the FDA. They wanted to assess the extent to which nutritional labeling had been accomplished, on a category-by-category basis, and then to measure the growth of the nutritionally labeled items versus competing items not so labeled. The FDA contracted for the Data Services Directory, and matching com- puter tape, covering the progress of 430 edible grocery product classes in which they were concerned with nutritional labeling. These in turn were regrouped into very broad categories such as products containing flour, canned vegetables, dairy products, and so forth. Using the flour category as an example, they grouped all items in which flour was the primary ingredient thus including pancake mixes, cake mixes, all-purpose flour, cookie mixes, etc. These broad categories were classified relative to con- tribution to total food sales, then each product class was ranked relative to its importance within the broad category. And finally, each individual brand was ranked relative to its importance within its product class. Once these listings were completed, the FDA, by means of a sampling procedure, developed a brand listing of retail samples from each category to be secured from supermarkets. This listing was passed on to Nielsen's Product Pickup Service's central headquarters in Northbrook for fulfillment. A number of leading brands on the list with virtually 100 percent dis- tribution could be found in almost any supermarket and presented no prob- lem. On the other hand, a good many others were regional or were fairly obscure brands with limited distribution. To secure retail samples of these brands, we returned to the data bank and found the cities and areas where these brands were in distribution. From then on it was a simple matter to transmit pickup orders to the field agents
OCR for page 82
82 OLIVER S. CASTLE in the respective areas and cities to secure the brand samples needed. Here is another example of the advantages of store-by-store data. In this case it was simply to retrieve shelf samples quickly and at low cost. But most often it is to determine what is happening, in what marketplace, and often why. From these samples, the FDA iS currently sorting out: ~ . The brands requiring nutritional labels and, in fact, so labeled, from 2. The brands requiring nutritional labels but not bearing them. 3. The brands not requiring nutritional labels but bearing them anyway. 4. Those not requiring nutritional labels and not having them. Once all of the brands are placed in the proper groups, the FDA will go back to the Data Services Directory to establish the current dollar volume of each classification. To determine future growth, they will continue to monitor each group for the next year to assess gains and losses for each of the four. In addition to the above, the FDA iS also using the product samples to check whether the contents match the contents' label in terms of water, solids, etc. And, if the label carries a picture of the contents, does the picture truly represent the contents? Another use of the product samples by the FDA iS to run a nutrient assay of each of the products bearing nutritional labels to see if the nutritional values claimed by the label are in fact true. A fourth use of the samples, and related to the one above, is an in-depth analysis of all of the products labeled as "dietetic. " If the label claimed the product contained only four calories and no sugar, the contents were analyzed to verify the label claim. As we understand it, the USDA iS going to perform similar analyses on any products containing meat sandwich spreads, chili, canned meat, etc. since the FDA has no jurisdiction over meat products. I'd like to make one final comment on this overall study. As you probably know, many food products are delivered directly to the store and never move through a wholesaler's or chain warehouse. For example, close to 90 percent of all soft drinks are delivered directly to the store; many dairy items milk, butter, cheese, ice creams—are handled in the same way. Most crackers, cookies, some frozen foods, and other specialty items are also store-door delivered. In order to provide data on all of these items, we utilized Nielsen Food Index Services since they cover all items, regardless of source or type of distribution. Since what is happening in the marketplace is such an important factor in the marketing of any product, we added another service, Store Observation Service, to assess the elements that so often influence the customer's deci- sion to buy or not buy.
OCR for page 83
A. C. Nielsen Company Services 83 Availability or distribution is perhaps the most critical of all from everyone's point of view manufacturer, retailer, and customer. Is the pro- duct there, on the shelf, where it can be purchased? But also important are the selling prices, your product and your com- petitors'; the number of shelf facings (i.e., Is it easily visible?; shelf location What shelf, what section, and where is the section in the store; and, Is the product being promoted or featured, and if so, how? These are the questions most often asked, but many others of equal importance can be handled by this service. To answer questions such as these, Nielsen employs a nationwide field staff specifically trained for this work. The staff, 80 in total, are located in 50 standard metropolitan statistical areas, in all areas of the country. The administration of the service is centralized in Northbrook, as are all Nielsen services. In your invitation offering us the opportunity to present "our wares" before you this afternoon, you also requested that we address ourselves to certain questions, the first being, What gaps exist in the data base? Nielsen Food Index, Nielsen Major Markets, and Nielsen Early Intelli- gence Service samples are all drawn from a universe with the following definition: all grocery stores, as defined by the U.S. Census Bureau, plus supermarket departments of general merchandise stores, i.e., the full-line grocery sections in mass merchandisers. This definition leaves out fringe- type stores handling small amounts of packaged groceries. This measurement gap, expressed in terms of consumer grocery products, could be estimated for a typical product category to be 5 percent. Nielsen Early Intelligence System's Data Service's samples do not include small grocery stores. Since these small stores account for approximately 15 per- cent of total dollar volume, this could raise the gap for Nets to close to 20 percent. How complete are the data? I believe this can be answered in the follow- ing summary of Nielsen services: · Nielsen Food Index Services including Major Market Service- provide consumer sales plus retail inventories, purchases, distribution, out-of-stock, the prices paid by consumers, in-store promotional data, and retailer advertising. · Data Services provide product movement to a sample of supermarkets. From this source, data are produced covering brand movement in both dollars and units per week per store, shares, distribution, and retail prices, and all brand details are broken down by size, flavor, type, and so forth. · Product Pickup Service is a complete service and limited only by the client's desires. Store Observation Service, as the name implies, is based on observable data in the selling area of the store.
OCR for page 84
84 OLIVER S. CASTLE How reliable are the data? Samples for these services are drawn from a sampling frame defined in Table 1. Practically maximum geographical dis- persion is achieved since, as an example, the 1,300-store national sample is spread through 606 counties. Similar dispersion is achieved in smaller areas sampled. Quinquennial Census material, updated with material from the Census Bureau Current Retail Trade Programs, is used both to determine the sam- pling frame and to estimate universe data from sample results. Sampling list material on a current basis is obtained from: · progressive grocer lists, · chain lists, and · gathered by survey. Sample units are selected with disproportionate sampling rates with larger stores having the greater chance of selection over small stores. This differ- ence in selection rates is taken into consideration when sample data are projected to universe estimates. At the same time that stores are audited to obtain data on individual grocery products, total retail dollar movement is obtained for each store for the audit period. This dollar movement is combined with the individual grocery product data to form a ratio. This ratio, having less variance from store to store than the absolute product data to be measured, gives the estimating process higher efficiency. In summary, the sample design and estimation process has the following characteristics: a stratified sample design using practically maximum geographic dis- pers~on; 2. disproportionate sampling with greater chance of selection for large stores; and 3. ratio estimation to achieve greater precision. Finally, I would like to comment on what I believe are the advantages that distinguish Nielsen marketing research services from any other. First, in most instances, the data produced flow from the marketplace and result from consumer actions. As such, when we report consumer sales, it means the products in question have been purchased by consumers, not merely moved out of a warehouse. It's the same with prices these are the prices paid by consumers. The distribution measurements are made in the store, as are the retail inventories and the data relating to sales influencing factors.
OCR for page 85
A. C. Nielsen Company Services 85 Second, nearly all the services are based on scientifically drawn samples. We believe that samples provide the greatest value in terms of dollars ex- pended, accuracy achieved, and breadth of data. And last, because of our access to the stores making up the samples, we can achieve a flexibility that's impossible to provide otherwise. I believe the study we are currently doing for the FDA iS a good example of that.
OCR for page 86
Representative terms from entire chapter: