Measuring Purchase Intent

Purchase Intent information obtained from five-point and four-point scales are not directly comparable; not are transformation from one scale to the other easy to make. Indications are however, that "top-box" response for the two scales tends to be very similar.


Background:

When consumer acceptance of products of concepts is researched, it is usually standard procedure to ask survey respondents their likelihood of purchasing the product. Purchase Intent Information is often obtained on a scale with several intent levels ranging from positive to negative. One particular purchase intent scale that is frequently used is a five-point scale as follows:

Often, however, the researcher opts for a scale that does not give the respondent the choice of a "middle" position — in effect, forcing the respondent to cast a positive or negative vote.

In most cases the choice of a particular scale is based upon the researcher's past experience, familiarity with it, or the fact that previous surveys used it and norms have been established. In some instances, however, it may become necessary to compare data obtained on two different scales or consider a different scale for a particular project.

This discussion focuses on the difference between purchase intent information obtained on two specific scales—the five-point scale as worded above, and a four-point scale worded as above but excluding the middle neutral position.


Research Method:

Market Facts, Inc. investigated the subject of purchase intent rating scales using Consumer Mail Panels and its Data-Gage service. A questionnaire that investigated purchase intent for several products and services was mailed to 4000 female heads of households. Half the sample received questionnaires with the five-point purchase intent scale and the other half received questionnaires with the four-point purchase intent scale. Furthermore, to account for possible order bias within these two groups, the purchase intent scales were rotated—positive to negative for one half and negative to positive for the other half. The number of respondents tabulated exceed 1,500 for each of the two scales.

Purchase intent information was obtained for a variety of products and services, chosen to reflect a broad price range and array to items. Three periods—12 months, 30 days, and 7 days—were used as the time frames.

Purchase intent information was obtained and is reported here for the following:


Results:

Purchase intent data for all nine items are presented in the following bar charts.


Click to enlarge


Closer examination of these data reveals that "topbox" measures of purchase intent (definitely will buy) tend to be the same regardless of whether the four-point or the five-point scale was used. This has important implications for the researcher who is concerned with comparing data obtained on these two scales.


  "Top box (Definitely will buy)
5-pt % 4-pt %
In next 12 months:  
New Automobile 7 7
Microwave oven 2 2
Vacation airline ticket 14 14
In next 30 days:  
Toothbrush 16 19
Flashlight battery 21 23
Electric light bulb 26 28
In next 7 days:  
Movie ticket 13 14
Pudding / pie filling 16 16
Frozen pizza 15 15


Further examination of the data discloses that if it becomes necessary or desirable to redistribute those respondents with neutral purchase intent, in effect converting a five-point scale into a four-point scale, the allocation task is not a trivial one. Market Facts used four standard allocation methods, the results of which are shown in the table below. (Purchase intent data for all three products in each time frame were averaged for this analysis.) The allocations were done by: 1) treating those in the neutral position as "don't know" answers, excluding them and repercentaging, in essence proportionately distributing these respondents to the other categories; 2) distributing them in proportion to the number of respondents in all but the "top-box" position; 3) distributing them in proportion to the number of respondents in the two adjacent purchase intent categories   probably will buy and probably will not buy; and, 4) distributing them equally to the two 'adjacent purchase intent categories.


Click to enlarge


An equal distribution of the neutral respondents to the adjacent two purchase intent categories results in the closest approximation to the results obtained with the four-point scale. However, this is not a particularly good match-positive intent is severely overstated—and suggests that such transformations of data should not be made.

The wording of purchase intent scales in marketing research appears to vary for a variety of reasons. The wording for the scales used in this limited experiment is identical except for the inclusion of a middle answer alternative in the five-point scale. None of the transformations of data attempted in this study worked well. Therefore, it would be inappropriate to attempt to adjust data from scales that are likely to be less congruent than those used here.