Tipping the Scale: The Role of Discriminability in Conjoint Analysis

Anocha Aribarg, Katherine A. Burson, Richard P. Larrick
Article Snapshot
Current average rating    
Key Takeaways
  • ​Marketing researchers give hypothetical choices  to consumers to identify which product attributes, such as price or gas consumption of a car, are most important.
  • The attributes's scale can vary, such as expressing a gas consumption as gallons per 100 miles or per 1,000 miles.
  • Generally, the more "expanded" the scale is (e.g. gallons per 1,000 miles), the more consumers prefer the products that perform well on the expanded attribute.
  • Because of diminishing sensitivity, very extreme expansion of a scale actually reduces the impact of an attribute in preference.
  • Shifts in choices due to scale expansion impact conjoint researchers' inferences about screening and attribute relative importance.

​Article Snapshots: Executive Summaries from the Journal of Marketing Research

[Full Article] [Google Scholar]

Simply expanding an attribute’s scale can shift choice toward alternatives that perform well on that attribute and thus impact metrics derived from conjoint analysis such as attribute importance and screening.


Practitioners cannot simply assume that expanded attributes might lead to greater relative importance weights and increased screening: At some inflection point, that trend flattens and then reverses.
 
Holding attribute rance and number of levels constant in a conjoint design, we show that expanding theattribute scale produces systematic shifts in the relative importance of attributes.

Research

Scale expansion influences consumers' choices. Choice-based conjoint is a widely used method that infers importance of product attributes by observing choices. Previous research had not examined the impact of scale expansion onmetrics derive from conjoint analysis. We hypothesize that scale expansion would initially increase consumers' ability to discriminate between alternatives, making them more likely to choose alternatives that are superior on expanded attributes, then decrease it due to diminishing sensitivity, impacting derived relative importance and screening propensity.

Methods
In three choice-based (one incentive compatible) conjoint studies with over 1,700 participants recruited from both a nationwide online panel and a university, we demonstrate the impact of attribute representation in their choices among cars and snack bars.

Findings
We demonstrate the impact of attribute representation in choices. Specifically, we observe an inverted-u pattern for inferred relative importance derived from partworths, measures of perceived differences in attribute values, and screening propensity in response to scale expansion. This is important for academics or practitioners employing the conjoint method to appreciate because both consumer choice and their own inferences will depend on the specific scale expansion utilized.

Implications

Representation of important product attributes are changing (e.g., fuel usage on the Monroney sticker, recent mandate to change serving size information). Because choice depends on scale expansion, consumers should recognize this bias while policy makers can consider implementing the most socially beneficial levels. Practitioners should appreciate that their scale’s expansion will determine their conjoint results. By illustrating the curvilinear relationship between scale expansion and multiple measures, we offer policy makers and practitioners some insight into the limits of scale expansion.

Questions for the Classroom

  • What is the difference between this effect on conjoint analysis, and the well-known scale range and number of levels effects?
  • What is the "right" level of scale expansion (either for an accurate measure of attribute importance or for a choice architecture tool)?
  • Can the screening results provide additional support for the authors' claim that perceptions of differences between alternatives are impacted by scale expansion?

Article Citation: Anocha Aribarg, Katherine A. Burson, and Richard P. Larrick (2017) Tipping the Scale: The Role of Discriminability in Conjoint Analysis. Journal of Marketing Research: April 2017, Vol. 54, No. 2, pp. 279-292.

doi: http://dx.doi.org/10.1509/jmr.14.0659
[
Full Article]
[Google Scholar]


Author Bio:

 
Anocha Aribarg, Katherine A. Burson, Richard P. Larrick
Anocha Aribarg is Associate Professor of Marketing and Michael R. and Mary Kay Hallman Faculty Fellow, Ross School of Business, University of Michigan (e-mail: anocha@umich.edu). Katherine A. Burson is Associate Professor of Marketing, Ross School of Business, University of Michigan (e-mail: kburson@umich.edu). Richard P. Larrick is Professor of Management, Fuqua School of Business, Duke University (e-mail: larrick@duke.edu).
Add A Comment :
 

Become a Member
Access our innovative members-only resources and tools to further your marketing practice.