Question

CASE 14-2 Talbot Razor Products Company

One of the products marketed by Talbot Razor Products Company is an after-shave lotion called Enhance. This brand is sold through drugstores, supermarkets, and department stores. Sales exceed $30 million per year but are barely profitable because of advertising expenses that exceed $9 million. For some time the company and its advertising agency have felt the need to undertake a study to obtain more data on the characteristics of their users as contrasted to those of other leading brands. Both the company and the agency believe that such information would help them find better ways to promote the Enhance brand.

Preliminary discussions between the advertising department and the research department of the advertising agency resulted in the following study objectives.

1. To determine the characteristics of Enhance users versus competitors by such factors as age, income, occupation, marital status, family size, education, social class, and leisure-time activities.

2. To determine the image of the Enhance brand versus competitors on such attributes as masculinity, expensiveness, and user stereotypes (such as young men, factory workers, young executives, and men living in small towns).

3. To discover the meaning to consumers of certain words that were used to describe after-shave lotions.

4. To examine the media habits of users by television programs, magazines, and newspapers.

In discussing the sampling universe, the advertising manager thought the study results should be broken down by heavy versus light users of Enhance. In the manager’s opinion, as few as 15 to 20 percent of the users might account for 60 percent of the total purchases.

It was not clear how many containers a user would have to buy during a specified time period to qualify as a heavy or a light user. The research director and the advertising manager disagreed on a definition of user: The research director thought that anyone who had used the Enhance product within the past year should qualify as a user and therefore be included in the study, while the advertising manager thought that a user should be defined as one who had purchased the product within the past three months. In fact, the advertising manager went on to say, “I am really interested only in those people who say that the Enhance brand is their favorite brand or the brand that they purchase more than any other.”

After much discussion about what constituted or should constitute a user, the research director pointed out that the advertising manager was being unrealistic about the whole sampling problem. A pilot study was conducted to determine how many qualified users could be obtained out of every 100 persons interviewed in Sacramento, California. While the findings were not completely representative, they did provide a crude estimate of the sampling problem and the costs that would result from using any kind of a probability sample. The research director said:

In the Sacramento study we were interested only in finding out how many males 18 years of age or older used after-shave, what brands they had purchased during the past year and the past three months, and what brand they bought most frequently.

All interviewing took place during the evening hours and the weekend. The findings revealed that only about 70 percent of the male respondents were at home when the interviewer made the call. Of those who were home and who agreed to cooperate, only 65 percent were users of after-shave: that is, affirmatively answered the question: “Do you ever use after-shave?” Of those who used after-shave, only 7 percent had purchased the Enhance brand within the past three months, while 15 percent reported having purchased it within the past year. The costs of the Sacramento job figured out to about $6.00 per contact including the not-at-homes, refusals, and completed interviews, all as contacts. The sample size for the Sacramento pilot study was 212 male respondents, and the field costs were $1,272. These costs will be increased substantially if the sample includes smaller towns and farm interviews.

The research director believed that the best sample size they could hope for would be one that provided about 100 interviews with Enhance users plus 100 interviews with users of other brands in each of 10 to 15 metropolitan areas. This would provide a total sample size of 2,000 to 3,000 and would require contacts with between 40,000 and 50,000 respondents. The research director indicated that this size sample would permit breakdown of the results for the United States by heavy versus light users.

The advertising manager, who did not think this would be an adequate national sample, said: I can’t present these results to my management and tell them that they are representative of the whole country, and I doubt if the sample in each of the 10 to 15 metropolitan areas is big enough to enable us to draw reliable conclusions about our customers and noncustomers in that particular area. I don’t see how you can sample each metropolitan area on an equal basis. I would think that the bigger areas such as New York and Chicago should have bigger samples than some of the smaller metropolitan areas.

The research director explained that this way of allocating the sample between areas was not correct since the size of the universe had no effect on the size of the sample. According to the director: If we do it the way you are suggesting, it will mean that in some of the big metropolitan areas we’ll end up with 150 to 200 interviews, while in some of the smaller ones, we’ll have only 50 or 75 interviews.

Under such conditions it would be impossible to break out the findings of each metropolitan area separately.

If we sample each area equally, we can weigh the results obtained from the different metropolitan areas so as to get accurate U.S. totals.

When the discussion turned to costs, the advertising manager complained:

I can’t possibly tell my management that we have to make 40,000 to 50,000 calls in order to get 2,000 to 3,000 interviews. They’re going to tell me that we’re wasting an awful lot of money just to find users. Why can’t we find Enhance users by selecting a sample of drugstores and offering druggists some money for getting names and addresses of those men who buy after-shave? We could probably locate Enhance users for maybe 35 to 50 cents each.

The research director admitted that this would be a much cheaper way, but pointed out that it is not known what kind of sample would result, and therefore it would be impossible to tell anything at all about the reliability of the survey. The advertising manager thought management would provide no more than $30,000 for the study. The research director estimated that the results could be tabulated, analyzed, a report written, and the results presented to management for about $7,000, thus leaving around $23,000 for fieldwork.

Questions for Discussion

1. How should the sampling universe be defined?

2. How large a sample should be collected?

3. How should the sample be distributed geographically?

Answer 1

1) A sampling universe is the data from which our sample is drawn. It includes every survey element that qualifies for inclusion in the research, however a precise definition may vary based on the objective of the research and the target for the research question specifies who or what is relevant to the research. In the above example, the sample universe would be all males over the age of 18 who used after shave lotion.

2) Sample size determination is the most important aspect of statistical research as it impacts accuracy of the results greatly and can hence, declare an entire data collection and analysis research irrelevant and misleading if sample size is not adequate as it's ability to detect an effect may be negligent. Use of Cochran's formula for determining sample size and other formulas for deciding standard deviations and precision should be utilised. The size of the sample should be large enough to include a good overall percentage of the target subjects. The three main factors are size of population, margin of error that can be allowed and level of confidence required. In the above case a sample size which is small can also be of high confidence level and reduce margin of error by including data internally available such as sales figures area-wise and figures of competitors and overall users of aftershave for comparative analysis.

3) The geographical distribution of sample should be based on achieveing best results and most reliable data by focussing on areas which would have more of the target subjects. The requirement is for research on Image of the aftershave and profiling the users who currently buy the product, hence, given constraints of resources mentioned the aim should be to target areas with high population of men over 18 who use aftershave, As there are not too many cultural differences and the social differences can be eliminated by the company as it is well aware of the target consumer, the areas for the survey can be greatly fine tuned to obtain a good sample size which is fairly reliable with minimum costs. This will help to increase largely the size of sample and increase confidence in the analysis. Hence, geographical distribution need not necessarily mean we get a similar sized sample from each location but working within constraints obtain the maximum and most reliable sample by targeting zones within every area which will throw up maximum results.