A Proposal For a Global, Longitudinal Measure of National Consumer Sentiment Toward Marketing Practice

John F. Gaski, University of Notre Dame
Michael J. Etzel, University of Notre Dame
ABSTRACT - There have been a number of published measures of generalized consumer attitudes, perhaps the most familiar being the Index of Consumer Sentiment of the University of Michigan's Survey Research Center. What this paper proposes is an "index of consumer sentiment toward marketing," to be reported at regular intervals to the marketing community. The paper describes the proposed measure and measurement procedure, and also reports some preliminary validation evidence. The authors hope that by subjecting this index to the scrutiny of consumer behavior researchers it can be further refined and improved.
[ to cite ]:
John F. Gaski and Michael J. Etzel (1985) ,"A Proposal For a Global, Longitudinal Measure of National Consumer Sentiment Toward Marketing Practice", in NA - Advances in Consumer Research Volume 12, eds. Elizabeth C. Hirschman and Moris B. Holbrook, Provo, UT : Association for Consumer Research, Pages: 65-70.

Advances in Consumer Research Volume 12, 1985      Pages 65-70

A PROPOSAL FOR A GLOBAL, LONGITUDINAL MEASURE OF NATIONAL CONSUMER SENTIMENT TOWARD MARKETING PRACTICE

John F. Gaski, University of Notre Dame

Michael J. Etzel, University of Notre Dame

ABSTRACT -

There have been a number of published measures of generalized consumer attitudes, perhaps the most familiar being the Index of Consumer Sentiment of the University of Michigan's Survey Research Center. What this paper proposes is an "index of consumer sentiment toward marketing," to be reported at regular intervals to the marketing community. The paper describes the proposed measure and measurement procedure, and also reports some preliminary validation evidence. The authors hope that by subjecting this index to the scrutiny of consumer behavior researchers it can be further refined and improved.

BACKGROUND

The measurement of general consumer attitudes and sentiment has been of interest to academic and commercial researchers for many years. The marketing literature contains a number of efforts to develop global measures of consumer satisfaction with business in general (notably Lundstrom and Lamont 1976, Barksdale and Darden 1972, Hustad and Pessemier 1973), and the "index of consumer sentiment" and "consumer confidence index" of the University of Michigan Survey Research Center and The Conference Board, respectively, receive the attention of even the general public. Michigan's survey of consumer sentiment, initiated in 1946, uses a rotating, monthly, nationwide probability sample of about 800 households, random-digit telephone interviewing, and a five-item instrument (tapping the dimensions of personal finances, economic prospects, and buying plans) to assess general attitude toward buying. The Conference Board, with data collected by National Family Opinion, Inc. from a mail panel of 5000 households, produces a monthly measure of consumer confidence. Opinions relating to business and employment, current and future economic conditions, and personal financial situation are collected. The survey has been in operation since 1967.

In addition, the research firm Yankelovich, Skelly, and White, Inc. tracks trends in consumer social attitudes with its "Monitor" service. Begun in 1970, the data are collected through an annual national sample survey of 2500 consumers, using personal interviews. Attitudinal trends measured include "focus on self," "upward mobility/money," "attitudes toward institutions," and "the family." And Sindlinger & Co. of Norwood, Pa. does virtually the same thing as the Michigan Survey Research Center, except with a four-item instrument, a sample of 260, and daily telephone interviewing.

These consumer polls, of course, are intended as a basis for forecasting future activity and are used in business planning, but they remain, essentially, reflections of current sentiment.

Referring again to the marketing literature, at two-year intervals during the period between 1971 and 1979, Barksdale, Darden, and Perreault (1976; Barksdale and Darden 1972; Barksdale and Perreault 1980) measured overall consumer attitudes toward manufacturers with a five-item Likert scale instrument incorporating these items:

Most manufacturers operate on the philosophy that the "consumer" is always right.

Despite what is frequently said, "let the buyer beware" is the guiding philosophy of most manufacturers.

Competition ensures that consumers pay fair prices.

Manufacturers seldom shirk their responsibility to the consumer.

Most manufacturers are more interested in making profits than in serving consumers.

They also measured attitudes toward certain marketing phenomena including product quality and advertising. In general, it was found that

In the 1970s, there was substantial stability in consumer attitudes concerning . . . the business sector. . . . (C)onsumers were as skeptical about the operating philosophy of business in 1979 as they were nine (sic) years earlier (Barksdale and Perreault 1980).

At the present time, however, there appears to be no validated index of general consumer sentiment toward marketing practice being regularly or periodically reported to the marketing community. Such an index is to be proposed herein. This type of device would permit the continual monitoring of the basic public perception of, and satisfaction with, the establishment of marketing. In view of the apparent long-standing public hostility coward marketing practice and marketing institutions (Miller 1974, Barksdale and Perreault 1980), it would seem to be desirable to provide such an information service for those in the field for several related reasons: (1) It may sensitize marketers to consumers' perceptions; (2) it would serve to identify the nature of the public relations task (and, possibly, more fundamental casks) facing marketing; (3) it should assist in gauging whatever progress, or lack of i:, is being made; and (4) it may even make a positive contribution toward marketing's public image, i.e., by demonstrating that marketing cares about the public enough to ask the public what it chinks. Most generally, the index would serve as a continuing indicator of "how marketing is doing" in the eyes of the public. The provision of a regular, validated measure of the public impression of marketing should be of interest not only to marketing researchers and practitioners, but to public policy makers and the general public as well.

METHODOLOGY

The Sample

Consumer sentiment coward marketing is to be measured with data collected from the Consumer Mail Panel or the research firm Market Facts, Inc., which is generously making its resources available. This mail panel contains a pool of over 130,000 households and is "nationally balanced," i.e., continually updated to reflect the most recent U. S. Census data in terms of geographic region, annual income, population density (urban/rural location), age, sex, and family size. The sample size is 2000, with a response rate of 75% expected on the basis of historical performance.

Description of the Instrument

Exhibit 1 shows the proposed questionnaire items to be presented to the national sample of consumers to measure attitudes toward the four major elements of marketing practice. The four categories of items are intended to roughly correspond to the four components of the "marketing mix," as they impact on consumers. Promotion is represented by the advertising section and also the personal selling-related items from the final section, which is also intended as a surrogate for distribution/ retailing.

EXHIBIT 1

QUESTIONNAIRE ITEMS

Some scale items were developed by selecting and adapting ones that had appeared in the literature, but most were created ad hoc. This was done in consultation with the Market Facts personnel responsible for data collection. It is anticipated that the scales will undergo "purification" through reliability analysis using Cronbach's alpha. (Pre-test reliability evidence follows.) Perhaps the two items from each scale exhibiting the lowest item-to-total correlations will be deleted, leaving four sets of five-item scales.

Items include both favorable and unfavorable statements, to offset any affirmation/negation response tendency. Responses will be scored from +2 to -2, with most favorable answer (i.e., "agree strongly" with a positive statement, "disagree strongly" with a negative statement) receiving the highest score, indicative of the most favorable attitude.

Scores for the items in each category will be summed, and then importance-weighted according to responses from section #4 in Exhibit 2, scored from one to five for "not at all important" through "extremely important." The measure of overall consumer sentiment toward marketing is then

EQUATION

with xij = scale item response i in category j

wj = importance weight for the marketing mix category j

m = number of items in category

n = number of categories

This represents an attitude score for an individual respondent. The mean score across all respondents can then be calculated and converted to an index number as a measure of general sentiment. [Some readers may question the legitimacy of calculating mean sentiment scores, since division requires ratio data. However, since the principle is well-established that ordinal data may be treated as interval (Kerlinger 1973, pp. 426-7), and since the neutral response ("neither agree nor disagree") may be considered a natural origin, the authors believe it is reasonable to treat responses to the Likert scales as quasi-ratio data. There is precedent for this approach in the literature (Harrell and Bennett 1974).

A similar assumption is made for the importance-weight scale. A score of one, rather than zero, assigned to a response of "not at all important" simply reflects the belief that none of these aspects are likely to be of absolutely no importance to a respondent.]

EXHIBIT 2

QUESTIONNAIRE ITEMS

Validation and Pre-Test Results

Reliability of each category of the measure will be assessed by the aforementioned coefficient alpha. Discriminant validity can be verified by comparison of the alphas with inter-category correlations, with the former expected to be greater.

Convergent validation for the entire measure may be achieved by identifying correlations with the alternate measures depicted in Exhibit 2 (Sections 2, 3, and "Marketing in General"). Again, responses here are to be scored from -2 to +2 for least favorable to most favorable in each case, and summed across items in the scale. A possible variation of measures 2 and 3 could be produced by multiplying each item score by importance weights (section 4) before summing. (The items in 'Marketing in General" were adapted from the Barksdale-Darden-Perreault instrument.) Some preliminary validation from a Pre-test can now be reported.

The pre-test involved a mailing to 50 panelists, with 41 completed questionnaires returned (21 male, 20 female respondents). The coefficient alphas and item-to-total correlations for each of the four scale categories are presented in Table 1.

TABLE 1

MEASURES OF RELIABILITY FOR COMPONENTS OF CONSUMER SENTIMENT MEASURE

As can be seen, all alpha coefficients are satisfactory, demonstrating internal consistency or reliability. This can be regarded as indirect evidence of validity (Churchill 1976, pp. 251-2), as a necessary but not sufficient condition.

Table 2 shows the correlations among the four categories of the measure. Since each category/scale has an alpha coefficient consider-ably higher than its correlation with any of the other scales, discriminant validity is established.

TABLE 2

CORRELATIONS BETWEEN SCALE CATEGORIES

Finally, some evidence of convergent validity for the composite measure is found in the correlation coefficients between this measure and the alternate global measures of sentiment from Exhibit 2. Correlations are .755, .724, and .724, respectively, between the operative measure and the "Marketing in General" measure, the "Satisfaction" measure (Section 2), and the "Problems" measure (Section 3), with all coefficients significant at the .001 level. (None of the alternate measures were weighted for importance.) Adequate convergence, for pre-test purposes, seems to be present.

In all, the validation evidence provided by the pre-test data is considered to be promising, and a likely strength of the measure relative to others that have been presented in the literature.

OPERATIONAL DETAILS AND PLANS

It is anticipated that this survey of consumer sentiment toward marketing will be conducted annually, at the same time each year. For ease of longitudinal comparison, the mean score will be converted to an index number with the initial measure serving as base period. The information contained in each survey, including the focal index number, will be reported to the marketing community.

At this point, it may appear as though it will be many years before enough evidence has been collected to provide meaningful inter-temporal comparisons. However, since this project also involves replication of the Barksdale-Darden-Perreault (B-D-P) measure, comparison can be made with consumer sentiment in the 1971-79 period immediately. The procedure is as follows:

First, recognize that the 'Marketing in General" scale from Exhibit 2, with the addition of the fourth item from the Price section of Exhibit l, is equivalent to the B-D-P instrument. The only differences are (1) substitution of the word "businesses" for "manufacturers," and (2) a slight, but comparable, restatement of the original B-D-P item #3 described earlier (regarding competition and prices). It is proposed here that these differences are negligible and the items used in this research amount to an adequate replication of B-D-P. (B-D-P item #3 had been excluded from the "Marketing in General" scale because it is the only item that is addressed specifically to one of the "mix" elements, and also for validation purposes, i.e., so as not to have an identical item in two alternate measures.)

Next, since the operative measure of consumer sentiment is being converted to an index number of 100, the same can be done for the B-D-P replication measure, rendering the two interpretable on the same basis. That is, since both measures are indicants of general consumer sentiment, the obtained mean score on the operative measure (-14.9, from pre-test data; range: -280 to +280) can be regarded as equivalent to the mean on the B-D-P replication (-.85; range: -10 to +10), which is done when both are represented by an index number of 100.

Then, future indices of consumer sentiment toward marketing can be related to the historical B-D-P measures if those measures are also converted to index numbers, as they can be from the B-D-P published results. The mean scores of the B-D-P historical measures, and associated index number equivalents, are shown in Figure 1, which portrays the pattern of consumer sentiment observed over time. (The mean B-D-P scores were calculated by scoring responses from -2 to +2 for least favorable to most favorable.) There appears to have been some improvement in consumer sentiment toward marketing in recent years, although it still lies in the negative range. [If the raw data can be obtained from the B-D-P authors, the significance of the observed differences between time periods can be assessed.]

If results similar to those of the pre-test are obtained from the actual survey being conducted, it could be an indication that (1) marketing practice has improved, (2) marketing's public relations efforts are paying off, (3) environmental circumstances have changed in a way that produces less consumer hostility toward marketing, or (4) some evolution of consumer maturity and wisdom has taken place so that consumers are less inclined to blame marketing for their problems.

Interpretation of the previous B-D-P findings is also possible. A low point of measured consumer attitude toward marketing occurred in 1975, a time which marked the end of a decade of public disaffection with national institutions. Aside from experiencing the Vietnam war and Watergate, respondents had recently endured the first oil shortage, a period of rapid inflation, and a severe recession. This was also a period of intense consumerism. The 1977 measure, on the other hand, may have reflected economic recovery and the sizeable reduction in the inflation rate achieved during the Ford administration. Then, the 1979 nadir signaled "Oil Crisis II," another period of rapid inflation, an impending recession, and the national "malaise" reported by President Carter.

FIGURE 1

HISTORICAL PATTERN OF CONSUMER SENTIMENT TOWARD MARKETING

Of course, there may be some concern that the B-D-P replication will produce a spurious result. Reasons for this could include: (1) Sampling differences--although B-D-P also conducted a national mail survey, their response rates were lower (between 42% and 45%), and it may be that their respondents, on average, were more involved in the issue because of more dissatisfaction, and this could be responsible for the lower sentiment ratings reported. However, it does appear that both the B-D-P and Market Facts procedures produce representative national samples. (2) The slight differences between the original and replicated instruments, i.e., changing "manufacturers" to "businesses"--although the measures appear virtually identical, it is possible that "manufacturers" is a more negatively-charged term for consumers. Perhaps the B-D-P measure should be duplicated exactly in future validations.

In spite of the possible limitations, it is hoped that a longitudinal record of consumer sentiment toward marketing will be of some interest and value to marketing scholars and practitioners, as well as other audiences. Thanks to the work of Barksdale, Darden, and Perreault, historical perspective is available from the inception of this Proposed index.

REFERENCES

Barksdale, Hiram C. and William R. Darden (1972), "Consumer Attitudes coward Marketing and Consumerism, Journal of Marketing, 36 (October), 28-35.

Barksdale, Hiram C., William R. Darden, and William D. Perreault, Jr. (1976), "Changes in Consumer Attitudes toward Marketing, Consumerism and Government Regulation: 1971-1975," The Journal of Consumer Affairs, 10 (Winter), 117-39.

Barksdale, Hiram C. and William D. Perreault, Jr. (1980), "Can Consumers Be Satisfied?" MSU Business Topics, 28 (Spring), 19-30.

Churchill, Gilbert A., Jr. (1976), Marketing Research: Methodological Foundations, Hinsdale, IL: The Dryden Press.

Harrell, Gilbert D. and Peter D. Bennett (1974), "An Evaluation of the Expectancy/Value Model of Attitude Measurement for Physician Prescribing Behavior," Journal of Marketing Research, 11 (August), 269-78.

Hustad, Thomas P. and Edgar A. Pessemier (1973), "Will the Real Consumer Activist Please Stand Up: An Examination of Consumers' Opinions About Marketing Practices," Journal of Marketing Research, 10 (August), 319-24.

Kerlinger, Fred N. (1973), Foundations of Behavioral Research, 2nd et., New York: Holt, Rinehart and Winston

Lundstrom, William J. and Lawrence M. Lamont (1976), "The Development of a Scale to Measure Consumer Discontent," Journal of Marketing Research, 13 (November), 373-81.

Miller, John A. (1974), "Who is the Discontented Consumer," in Combined Proceedings, Thomas V. Greer, ed., Chicago: American Marketing Association, 486-90.

----------------------------------------