Wolfgang Ziniel
Research and Teaching Assistant, Vienna University, Austria.Measuring the Influence of Third-Party Product
Tests on Product Choice Behaviour
Third-Party Product Reviews (TPPRs) are neutral (as far as the
producers’ interests are concerned) and consumer-orientated product
tests that are carried out by experts. They are published in special-
interest-magazines like PC-World, Runner’s World, Decanter or Wine
Advocate and on the magazines’ web pages respectively. Market
observations provide strong evidence that Third-Party Product Reviews
(TPPRs) significantly influence the success or failure of the products
evaluated (Chen and Xie, 2005). Apart from purely descriptive
contributions, however, there have not been any studies so far that
examine the impact of such test information on purchase behaviour.
This work aims at diminishing this gap in marketing research by
studying the relevance of TPPRs for product choice decisions.
A paired comparison choice experiment applying a conjoint design
is conducted online and analyzed by fitting a Bradley-Terry model in R.
The study took place in the context of quality wines. A 2 (brand,
high/low reputation) x 4 (TPPR, good, bad, editor’s choice, none) x 2
(price level, € 6 und € 10) between subjects orthogonal design (8 cards)
was created and administered among online access panel members
(n=500). Each respondent completed 14 randomly assigned paired
comparison choice tasks. Additionally subject-specific covariates like
perceived credibility, product knowledge or product class involvement
were measured applying dichotomous RASCH models.
The results provide strong evidence that - apart from brand and
price – TPPR strongly shape choice behaviour. Good TPPR and the
editor’s choice attribute showed strong positive worth parameters and
thus exert considerable influence on product choice processes while
high prices caused negative impacts. Interestingly winery reputation
and bad TPPR exerted nearly no influence. This might be a hint that
TPPRs are mainly used for a positive confirmation when intending to
buy a specific wine. It could also be shown that the covariates chosen
significantly moderated the effects.
Firstly this work will help firms to understand the relevance of a
test result for choice processes.
Secondly it demonstrates how to avoid problems of rating scales in
complex and multi-attribute online choice experiments by applying the
Bradley-Terry model and dichotomous Rasch models in psychometric
marketing research.