A series of articles examine complexity in individual and group decision-making, particularly the problems posed by nonseparable preferences. A person has nonseparable preferences when her preferred outcome from one set of alternatives depends on the outcome of or options available in another set of alternatives. For example, a person who prefers that taxes increase only if spending on education increases has nonseparable preferences for taxes and education spending.
Why do nonseparable preferences matter? Much of the research in the social sciences assumes that people have separable preferences, or that their preferences across issues are independent. The format of most public opinion surveys forces people to separate their responses to issues that may be linked in their minds. For example, surveys ask separate questions about education spending and tax increases without allowing people to express how their opinion on one of the issues depends on the outcome of the other issue. As a result, public opinion surveys often produce an inaccurate measure of the public’s preferences. I designed new survey formats to detect nonseparable preferences and found high rates of nonseparable preferences on a wide range of issues: immigration, health care, gun laws, abortion, marriage equality, pollution regulations, drilling for oil in the Arctic National Wildlife Refuge, English as the official US language, police surveillance, trade policy, and many more. Nonseparable preferences can explain many of the phenomena that social scientists use to show that people have inconsistent, unstable, or nonsensical opinions, even when people have stable and well-formed preferences.
My research on nonseparable preferences has received three national awards and has been supported by two grants from the National Science Foundation. I am currently working on a book manuscript in which I show that most political issues are multi-dimensional, people often have nonseparable preferences across issues, and public opinion surveys limit the kinds of preferences people are able to express. As a result of ignoring nonseparable preferences, social scientists have inaccurate measures of public opinion on many issues and have underestimated the stability and coherence of people’s opinions. The manuscript also shows that people know more about politics than we give them credit for.
Some of the findings from this research include:
People who have nonseparable preferences are more likely to display inconsistent survey responses than people with separable preferences. (2001. A Theory of Nonseparable Preferences in Survey Responses. American Journal of Political Science. 45(2):239-258. Note: Due to a printer's error that occurred after I approved the page proofs, several mathematical symbols are missing from the article and can be found in errata.) Nonseparable preferences produce a form of measurement error in public opinion surveys and can create question-order effects. (2001. Nonseparable Preferences, Measurement Error, and Unstable Survey Responses. Political Analysis 9(2):1-21.)
The problems posed by nonseparable preferences extend to group decision-making. When multiple issues appear in a referendum and some voters have nonseparable preferences, the outcome of the vote may be the least preferred outcome to a majority of voters. The best way to overcome problems posed by nonseparable preferences is to allow deliberation, sequential votes on each issue, or to leave the votes to legislatures. (Lacy and Niou. 2000. A Problem with Referendums. Journal of Theoretical Politics 12(1, January): 5-31.)
Nonseparable preferences also explain electoral dynamics, such as party polarization (Lacy and Niou. 1998. Elections in Double-Member Districts with Nonseparable Voter Preferences. Journal of Theoretical Politics 10(1, January): 89-110.) and divided government (Lacy, Niou, Paolino, and Rein. 2019. Measuring Preferences for Divided Government: Some Americans Want Divided Government and Vote to Create It. Political Behavior.)
If a candidate in a two-candidate election is currently in a disadvantaged or losing position, she can introduce a new issue to win the election, but only if some voters have nonseparable preferences for the new issue and original issue. (Lacy and Niou. 2013. Nonseparable Preferences and Issue Packaging in Elections. In Schofield, Norman, Gonzalo Caballero, and Daniel Kselman, eds., Advances in Political Economy: Institutions, Modeling, and Empirical Analysis. Berlin: Springer-Verlag.)