The effect of paired comparisons on triple choice sets : a thesis presented in partial fulfilment of the requirements for the degree of Master of Business Studies in Marketing at Massey University, Palmerston North, New Zealand
As consumers become aware of different brands they might purchase, it is likely they will consider those brands by making a series of paired comparisons, before finally settling on one option they prefer most. Choice theory suggests that preferences are formed early, so by influencing a consumer to prefer one option in favour of other options at the start of a choice episode, this can have a systematic effect on subsequent, and in particular final choice. Simonson, Nowlis, and Lemon (1993) assert that consumers who make paired comparisons of alternatives that vary in price and quality before selecting from a triple set of the same options are more likely to choose the cheapest option, than those who evaluate just the triple set comprised of the same options. Four experiments tested this claim but the predicted effect failed to occur. Moreover, results from one experiment had the reverse effect, the preference share of the cheapest option decreased, while the share of the more expensive options increased. This was a statistically significant result. This contra finding is in agreement with the large body of published evidence that suggests consumers, when it is possible for them to do so, prefer higher quality to lower quality options. The effect of background factors on choice was of concern, so the effect of gender, household income, and age on choice was tested. Results from these tests were inconsistent, and showed that only young males from high-income households were significantly effected by the stepwise treatment. There was concern that heterogeneity in the sampled group of respondents might have confounded the measurement of treatment effects. To help reduce the influence of background factors, all results were weighted. However, Simonson et al. did not account for heterogeneity, so it is possible their treatments have interacted with some background factor associated with the context of choice, individual difference between respondents, or the product attributes. For this reason, the claim by Simonson et al. is open to criticism. Alternative explanations for Simonson et al. (1993) findings are advanced. New research is required into the effects of paired comparisons on choice.