The Conjoint Cookbook
A practical, interdisciplinary guide to designing, fielding, and analyzing choice experiments with R
blah blah blah intro stuff here
Conjoint experiments or forced choice experiments (where respondents are shown randomly shuffled combinations of features and are asked to choose their most preferred option, and then they repeat that over and over again) have long been popular in marketing, and since 2014 they’ve become popular in polisci, public policy, and social science more broadly. But even though the actual approach is used the same way across disciplines, the estimands that disciplines are interested in are completely different.
In marketing, they care about consumer preferences and market shares, so they use conjoint data to build simulations of hypothetical product profiles, ultimately measuring market preferences.
In polisci, they care about casual effects—e.g. how much does the favorability of a political candidate change if they are a lawyer vs. not a lawyer. In this world, analysts don’t look at aggregate market preferences, but look at the exact causal levers and effects associated with different experimental conditions.
The two worlds know nothing about each other.
Main point of book—bridging the two worlds of polisci-style causal effects (AMCEs and marginal means and AFCPs) with marketing/econ-style preference and utility descriptions + providing hands on code examples of the methods, from basic OLS from polisci to hierarchical Bayesian multinomial logit in marketing
Running examples
Product packaging
Study 5 from Sokolova, Krishna, and Döring (2023) (and data from ResearchBox)
| Features/Attributes | Levels |
|---|---|
| Price | $2, $3, $4 |
| Packaging | Plastic + paper, Plastic + sticker |
| Flavor | Nuts, Chocolate |
If these two granola bars were your only options, which would you choose?
| Price | $3 | $4 |
| Packaging |  |
 |
| Flavor | Chocolate chips | Nuts |
| Choice |
Political candidates
From Hainmueller, Hopkins, and Yamamoto (2014)
| Features/Attributes | Levels |
|---|---|
| Military service | Served, Did not serve |
| Religion | None, Jewish, Catholic, Mainline protestant, Evangelical protestant, Mormon |
| College | No BA, Baptist college, Community college, State university, Small college, Ivy League university |
| Profession | Business owner, Lawyer, Doctor, High school teacher, Farmer, Car dealer |
| Gender | Male, Female |
| Income | $32,000; $54,000; $65,000; $92,000; $210,000; $5,100,000 |
| Race/Ethnicity | White, Native American, Black, Hispanic, Caucasian, Asian American |
| Age | 36, 45, 52, 60, 68, 75 |
Minivans
From chapter 13 in Chapman and Feit (2019)
| Features/Attributes | Levels |
|---|---|
| Passengers | 6, 7, 8 |
| Cargo area | 2 feet, 3 feet |
| Engine | Gas, electric, hybrid |
| Price | $30,000; $35,000; $40,000 |