Measuring Purchase Incidence Rates

The manner in which a purchase incidence question is worded has a significant effect on the results obtained.


Background:

When consumer research is conducted on a particular product (or product category) it is often necessary or desirable to estimate the incidence of purchase of that product. Establishing an accurate incidence rate of purchase can have important consequences on the cost of the research study: further interviewing can be restricted to only purchasers of the product or quotas can be established to insure their representation in the sample and, hence, maximize interviewer efficiency.

Obviously, an accurate estimate of the incidence of purchase may, in fact, be a major objective of the research. Also, if the purpose of the research is to determine how users of a product differ from nonusers, an accurate method of delineating these two groups is necessary.

This discussion focuses on the incidence rate estimates developed for two specific questioning procedures.


Research Method:

Market Facts, Inc. investigated the issue of incidence rate definition with Consumer Mail Panels using its Data Gage service. Questionnaires were mailed to female heads of households and purchase data were obtained for the following products:

Two different methods of obtaining purchase information were investigated:

  1. A "direct" or one-step question in which respondents were asked to indicate which products they had bought in a given time period, and
  2. An "indirect" or two-step question in which respondents were asked to indicate which products they had ever bought, then also indicate when their most recent purchase of that item was made.

Furthermore, two purchase time periods were investigated: three months and six months.

Four demographically balanced panels of 1,000 households were developed and each received a different version of the questionnaire.

Panel 1. One-step, direct question for a three month time period.
"Below are listed several products. Please "X" each product you or anyone in your household bought in the PAST THREE MONTHS."

Panel 2. Same as Panel 1 but for six month time period.

Panel 3. Two-step, indirect question.
"Below are listed several products. Please "X" each product you or anyone in your household ever bought. For each product ever bought, "X" the box that best describes when the product was purchased most recently:

Panel 4. Same as Panel 3 but with time period options in reverse order.

Approximately, 3,200 questionnaires were returned and processed (80% useable return rate).


Findings:

The data reveal that the different procedures produce different purchase incidence rates. For the three month time period the reported purchase incidence rates were significantly higher using the one-step, direct method of questioning, than they were using the two-step, indirect question.

Comparison of these two questioning variations for a six month time period shows the same tendency as the one-step direct method. That is, higher purchase incidence rates were detected in the one-step direct method than they were in the two-step indirect method. The differences detected in the longer time period, however, were not as pronounced as those detected in the shorter time period.

Comparing three to six month differences, purchase incidence increases by between only five and eight percentage points with the one-step direct method. The increases in these incidence rates are

suspiciously low considering that the time period has doubled and the type of products themselves. The two-step indirect method, reveals differences ranging from nine to 18 percentage points. These appear to be more realistic increases.


  "One-step direct question claimed purchase in:
3 months % 6 months % Difference %
White glue 46 54 8
Aspirin 68 73 5
Replacement auto tires 32 39 7
Record albums 41 48 7
  "Two-step indirect question claimed purchase in:
3 months % 6 months % Difference %
White glue 32 50 18
Aspirin 57 72 15
Replacement auto tires 24 39 15
Record albums 32 41 9


A closer look at the two questioning methods suggests that the one-step direct method minimizes respondent involvemant. That is, the respondent is not asked to think about any item for very long. The two step indirect method forces the respondent to get involved. The "ever purchased" question sets the stage. Then, the four time periods allow the respondent to choose one that fits best. This approach probably encourages the respondent to consider the response more thoughtfully.

In summary, the data suggest that when a one-step direct question is used to obtain purchase incidence there is a strong possibility of overstatement. Whether this overstatement is due to respondents' confusion about the time period, or a function of the products being investigated has not been ascertained.

When comparing data for three months and six months, the increase in incidence rates was more realistic for the two-step indirect method. Further, the two-step method encourages the respondent's involvement thereby providing a more accurate response.

In conclusion, a comparison of the two methods suggests that the two-step indirect method of questioning produces a more accurate rate of purchase incidence.