General Research Tips
The Science of Sampling - Mall Intercept Samples

As mentioned in the previous installment in this series (Telephone Sampling), this article was originally written as part of an internal training manual. While a few of the statistics cited within have changed somewhat since the original writing, the content of the article still applies. We'll print this article in several installments:

1. Telephone Sampling
2. Mall Intercept Sampling
3. Mail Samples, Finding Low Incidence Respondents
4. Statistical Issues

Because we've kept this article in its original form, some readers may find it to be a bit technical. As a user of research, it is not important to have a thoroughunderstanding of each of the sampling issues discussed in this article. However, as a user of research, it is very important to realize that these issues exist. Good research, after all, is not simply a matter of writing good questions.

Mall Intercept Samples

A. Background

Mall intercept surveys are widely used and (theoretically) able to reach a large segment of the population. In any given two-week period, about 2/3 of U.S. households shop one or more times at a mall. According to a CASRO membership survey, about 25% of all marketing research and 64% of personal interviews are conducted at malls.

B. Pluses And Minuses

The good things about mall samples are:

• Experimental control.
• Ability to see things.
• Availability of kitchens, etc.
• Cost.

The bad things are:

• Not necessarily even representative of the areas they're in.
• Demographically skewed samples. (Young, female, suburban, middle-income, frequent shoppers).
• Limited to metropolitan areas.

C. Effect Of Mall Samples On Results

For copy, concept, and product tests, data suggest that mall samples understate scores.

• Ossip reported four studies that found lower top box concept scores for mall surveys compared to door to door, even after controlling demographic differences.
• Gannon reported study comparing mall and mail panel for a concept/product test. Mall study got lower concept top box but higher product test attribute ratings.

D. "Ideal" Mall Sampling Plan

According to an article by Seymour Sudman, 1 to achieve a very good sample via the mall intercept method. However, this is what you have to do.

• Randomly select states or regions.
• Randomly select cities within region.
• Randomly select malls within cities.
• Post interviewers at randomly selected mall entrances.
• Interview all days and all times mall open.
• Count traffic so interviews are proportional to traffic based on day of week and time of day.
• Determine frequency of mall shopping and weight sample so that frequent shoppers not over-represented.

E. Practical Plan

Sudman's plan, unfortunately, is so difficult to put in practice that it would rarely, if ever, be used. In particular, the concept of posting interviewers at randomly selected mall entrances would be effective only for on-the-spot interviews. If respondents have to go to an office, the refusal rate caused by having to walk the length of the mall would eliminate whatever benefit there might be to the random sampling plan.

The following is a plan for "good but not pristine" mall samples:

• Select malls so that at least one in each census region, and interviews distributed by region proportionate to universe.
• Establish demographic quotas based on probability sample study.
• Screen out "regulars" - mall employees, recent participants.

F. Quota Screening

We've seen that mall shoppers are a demographically biased group. You can partly overcome this bias by using a quota sample if you know what the characteristics of the sample should be. But what if you don't know? Suppose, for example, you're screening for people who took both French and Spanish in high school and once worked flipping burgers. You haven't the slightest idea what they're like, let alone what the incidence is.

That's where quota screening comes in.

• You set quotas for screenings such that you screen men and women equally, the right number of each age group, etc. Once you've done enough screenings you know what the incidence is, so you can set quotas for the total number of screenings.
• Since you made sure that various demographic groups were screened in their proper proportions, you can be sure that the obtained sample is reasonably representative of the universe.