Imagine you’re at a cafe, perusing the lunch menu. Their recommended items are a soup, a traditional chicken sandwich, or a chickpea salad sandwich. You’re too hungry for just soup, so that leaves you to choose between the sandwiches. You always try to go for the vegetarian options, so you opt for the chickpea salad sandwich. But as you’re about to order, you see they have a new falafel wrap that looks incredible. It’s a bit more expensive, but it seems worth it, so you order the falafel wrap. Now you just have to choose a drink...

Without realizing it, you’ve made a series of consecutive choices that ultimately revealed your lunch preference. By deciding between each option presented, an outsider may be able to infer what you wanted for lunch on that day, allowing them to infer how you would usually rank lunch menu items as a whole. For example, upon hearing your decision, a friend might be able to guess what attributes you’re basing your choices around for your lunch dish (for example, How filling will this food be? Does this food fit with my ethical framework?). They might also sense that specific menu item characteristics vary in importance (the food looking appetizing is very important, while the cost is a less critical component), even if how you value these things may change over time.

An everyday situation like this is reminiscent of the more complicated, formal, and sophisticated discrete choice experiment method scientists use to model individuals’ decision-making processes. Like your series of choices between lunch items, discrete choice experiments typically use distinct decisions (e.g., A vs. B vs. C) to estimate the "utility" or value that individuals assign to each alternative based on the attributes of the options. The relative utility values help infer a person's likelihood of choosing one option over another. Despite this setup, DCEs are not primarily focused on creating a strict ranking of the items presented. Instead, they reveal the trade-offs and preferences based on the attributes, leading to utility scores for the different options, which can indirectly inform a ranking.¹

In most standard consumption models seen in economics, consumption is typically modeled as a continuous variable (e.g., the quantity of a good can be any value along a continuum).¹ However, in a discrete choice experiment, the options that are given are discrete, meaning categorical or whole-number terms. For example, an economist may ask, “What is the optimal number of cereal boxes for a family to consume per year?” and use regression analysis or calculus methods to model demand based on average consumption.

Furthermore, discrete choice analysis examines decisions involving distinct alternatives. Rather than asking a continuous variable question like, “How much cereal,” an economist might also ask a discrete choice question like, “What kind of cereal?” to study families’ preferences for different brands. This allows DCEs to study both qualitative and quantitative questions. Although there are an astounding number of cereal options (some might say too many), a DCE could group together different categories or brands of cereal and still provide helpful insight. Most often, a DCE is used when there are only a few distinct quantities or qualities available from which to choose. This might be something like, “How many boxes does your family consume per week?” or, “Which Cheerios variety is your favorite?” With a narrow set of options available, preferences for these types of choices can be more definitively revealed through DCEs. 

Discrete choice experiments are used to examine choices at an individual level and by larger organizations like business firms or government agencies or to estimate consumers' willingness to pay for certain features. Particularly in these larger organizations, it’s essential to have more objective methods of analysis that are easy for extensive scale application. Techniques like logistic regression and probit regression can be used to analyze discrete choice experiments empirically.¹

Choice Set: A collection of alternatives or options presented to participants in a discrete choice experiment from which they must choose. Usually, the set of options is limited to five or fewer so as not to overwhelm the participant. 

Attributes and Levels: Attributes are the specific characteristics of a product or service that are intentionally varied in a DCE, with levels representing the different values or settings of these attributes. For example, in a DCE comparing different medications, the attribute of cost could differ, with different cost levels paired with different potential side effects.

Utility Maximization: This traditional economics principle posits that individuals choose the option in a DCE that provides the highest personal utility or satisfaction. (However, as we know, people’s real-world choices rarely follow a ‘perfectly’ rational calculation.)

Conjoint Analysis: A statistical technique used to determine how people value the different attributes of a product or service. DCE is considered to be a conjoint analysis focusing on choices rather than ratings or rankings.

Latent Class Model: A statistical model used to segment respondents into different groups based on their choices in a DCE, identifying different preference patterns.

Willingness to Pay (WTP): A key outcome of DCEs that estimates how much an individual is willing to pay for specific attributes or improvements in a product or service. Someone’s WTP can be influenced by a number of external and internal factors, and individuals usually differ in their WTP for certain items based on preference and necessity. 

Random Utility Theory: Developed by Daniel McFadden, this is the underlying theory that models individual decision-making in a DCE based on the assumption that both observable (such as physical attributes) and unobservable factors (like personal preferences) influence choices.

The early foundation of discrete choice experimentation can be traced back to American psychologist Louis Thurstone's Law of Comparative Judgment. In his 1927 paper published on the subject, he introduced the concept of random utility, where he theorized that people make choices by comparing the amount of utility (or satisfaction) they get from different options. Still, this utility is influenced by random factors, making it unpredictable. In his paper, Thurstone outlined the random utility model for how people compare and discriminate between stimuli like gym weights or handwriting samples. Essentially, the law uses a series of pairwise comparisons to scale the stimuli. For example, to measure the perceived weight of a series of objects, people can directly compare the weights of the objects in pairs.²

Nearly fifty years later, Nobel-Prize-winning economist Daniel McFadden developed the random utility model (RUM), which formalized how individuals make discrete choices between alternatives. In essence, the model assumes that individuals behave as if they are maximizing their rational preferences. However, their decision-making might be unpredictable due to incomplete information or personal biases. RUM is particularly useful in predicting how people will behave in various situations, even when their actions don’t fully align with classical economic theories of rationality. For example, when deciding between different modes of transportation—such as taking the bus, driving a car, or riding a bike—the model helps to predict which option a person will choose based on the utility (or satisfaction) they expect to derive from each choice. It also accounts for random influences, like mood or external factors, which might cause someone to choose differently than expected (such as choosing to take the bus if they’re too tired to bike). Essentially, the random utility model can capture the strength of people's preferences even if their behavior seems inconsistent, providing a more flexible framework for understanding decision-making in real-world contexts.³

In the 1980s, Australian economist Jordan Louviere expanded on McFadden’s work by introducing choice-based conjoint analysis. This approach adapted traditional conjoint analysis, which was used to measure consumer preferences, into a framework that resembled real-world decision-making more closely. Louviere’s work allowed researchers to simulate actual decision environments, where individuals choose between fully described alternatives rather than rating or ranking them. Ultimately, it’s Louviere who is credited with the development of the choice experiment methodology as we know it today.⁴

By the 1990s, discrete choice experiments had gained significant traction in fields like health economics, and researchers began using this methodology to understand how both patients and healthcare providers make choices about treatments, services, and policies. One prominent development was the use of DCEs to estimate willingness to pay (WTP), enabling economists to quantify the value individuals place on specific healthcare outcomes.⁴

In the 2000s, researchers like the Australian Michiel Bliemer introduced more sophisticated methods to DCE design to ensure efficient data collection, focusing on designing choice sets that would provide the most reliable insights. With the adapted DCE techniques and rapidly improving survey technology, researchers were now able to handle larger and more complex choice scenarios and improved accuracy as questionnaires were designed to reduce respondent fatigue.⁴

The integration of latent class models into DCE analysis during the 2000s and 2010s marked another leap forward. These models allowed researchers to identify segments within a population that share similar preference patterns, making it possible to understand heterogeneity in decision-making. This was especially valuable in fields like marketing, where understanding customer segmentation is crucial for product development.⁴

Most recently, DCEs have been increasingly applied in the digital marketing sphere and integrated with machine learning techniques. Many businesses now use DCEs to predict consumer behavior in online environments, offering personalized recommendations based on discrete choice data. The major advancements in artificial intelligence and computational power have allowed for more sophisticated analysis, enabling DCEs to be used by many businesses’ real-time applications, like dynamic pricing and adaptive choice modeling.

Daniel McFadden

McFadden is a Nobel Prize-winning economist known for his foundational work in discrete choice models. His development of the random utility theory in the 1970s laid the groundwork for modern DCE methodologies and his work on econometrics and consumer choice behavior has influenced how DCEs are structured and applied, especially in areas like transportation, health economics, and policy analysis.

Jordan Louviere

One of the pioneers in the development and popularization of discrete choice experiments in marketing and social sciences, Australian economist Louviere introduced choice-based conjoint analysis. His work spans marketing, health economics, and public policy, and his contributions have shaped how researchers measure consumer preferences and forecast decision-making.

Michiel Bliemer

Bliemer is an Australian researcher in the field, particularly in transportation and health economics. His work has focused on the design of efficient DCEs and improving statistical methodologies, authoring numerous publications on experimental design and its application to real-world policy and economic decision-making.

Louis Thurstone

Thurstone was an American psychologist best known for his work in psychometrics and the development of Thurstone's Law of Comparative Judgment in 1927. His theory introduced the concept of random utility, which posits that individuals make decisions by comparing the utility of different options influenced by random factors. This concept became a key theoretical foundation for the random utility model (RUM) used in modern discrete choice experiments.

The development of discrete choice experiments and their role in research has transformed how organizations across sectors make decisions by giving them a clear, data-driven understanding of what people truly value. This not only leads to better-designed products, services, and policies but also ensures that the solutions designed by the leaders in those sectors are aligned with real human preferences, making them more effective and widely accepted.

Marketing and Product Development 

Discrete choice experiments and surveys have enabled companies and marketing teams to tailor products to consumer preferences with unprecedented accuracy. By analyzing trade-offs that different customer segments are willing to make, businesses and product developers can design products and services that meet each consumer’s identified needs. This ability to predict willingness to pay for specific features can enhance both product success and customer satisfaction. With the potential for companies to so closely and accurately track consumer preferences on an individual level, their approach to pricing strategies has shifted, with personalized marketing and dynamic pricing becoming the norm in most industries.^1,4 For digital products, by better understanding how customers make decisions online, companies can also create more intuitive, customer-friendly, and dynamic interfaces, leading to increased engagement and conversion rates.

Public Policy and Environmental Planning

Discrete choice experiments have also had a profound effect on public policy and environmental planning. By assessing how citizens value different policy options like renewable energy investments or public transportation systems, governments can design more effective and efficient policies that reflect the true preferences of their populations. In the environmental sector, DCEs have influenced sustainability initiatives by quantifying the public’s value for non-market goods like clean air or natural habitats, ultimately guiding more informed and impactful policy decisions.⁷

Improving Healthcare

In the healthcare world, discrete choice experiments have revolutionized the way treatments are designed and delivered by uncovering patient preferences for different health outcomes, services, and policies. It can sometimes be easy to conceptualize improvements in healthcare outcomes as simply improvements to medication efficacy, disease mortality, or other quantifiable measures, but healthcare is much more than that; quality treatment should focus on the patient experience, which is often subjective.⁵

Researchers in the healthcare or healthcare policy field often seek ways to quantify and better understand patients' preferences for different treatment options, frequently turning to discrete choice experiments to unpack what aspects of healthcare are most important to patients. Patients in DCEs might be asked about preferences for treatment effectiveness, side effects, or other attributes. Although each patient situation is unique, valuable insight can be gained from centering the opinions of the individuals experiencing the treatment over the outside assumptions of researchers or healthcare providers.⁶

Following COVID-19, we have seen a major increase in media coverage of the shortage of healthcare workers and the often horrendous conditions in which they must work. With a growing need to retain healthcare staff and improve their experiences, DCEs have also been applied to evaluate the employment preferences of healthcare workers like doctors and nurses to inform strategies for recruiting and retaining staff. For example, if there’s flexibility, nurses might have the option to rank their schedule preferences: longer day shifts or shorter night shifts? Multiple days in a row or alternating shift days? By understanding which job or shift attributes are most valued, healthcare systems can design employment offers that attract and keep medical professionals—and keep them doing their best.^5,6

Lastly, from the policymaker perspective discussed above, DCEs can help leaders make informed decisions by providing data on the preferences of all the different stakeholders in healthcare (e.g., patients, providers, hospital administrators). All of these insights can provide a data-driven perspective to the design of healthcare policies, services, and systems as a whole.

Like all experimentation methods, discrete choice surveys and experiments are imperfect. Although they are valuable for understanding differences in preference, there are a number of reasons we must take results with a grain of salt. 

Validity of Hypothetical Choices VS. Real-World Behavior

One of the most debated issues in DCEs is the question of external validity. Critics often ask whether the hypothetical choices participants make in any experimental setting accurately reflect their real-world behavior—and DCEs are no exception. If you’ve ever been faced with a high-stress situation where you needed to quickly make a decision, you may know that the way we often feel and the choices we make in real life can be different from what we would’ve predicted in a hypothetical version of the situation. Thus, results from discrete choice experiments or questionnaires can be biased, as respondents may not act the same way in a controlled environment as they would when facing real, immediate financial or personal consequences.

For instance, in a healthcare setting, choices made in a survey may differ significantly from those made under the pressure of actual costs or time constraints. Compared to simply taking a survey, might things feel different if you're physically holding a hospital bill for tens of thousands of dollars? Or in front of a loved one dying in a hospital bed? Proponents of DCEs counter this by trying to design experiments to mimic real-world conditions as closely as possible and using techniques like incentive compatibility (where an incentive is provided for everyone to truthfully reveal their true preferences) to reduce bias. But, of course, the gap between hypothetical and real behavior can never be completely eliminated, rendering external validity a contentious issue.⁵

Complexity and Choice Overload

Another controversy revolves around the cognitive demands discrete choice experiments place on participants. DCEs often present multiple attributes with several levels in each choice set, which can overwhelm respondents and lead to choice overload, where too many choices make it harder for participants to select a choice. Some researchers and practitioners argue that when individuals are presented with too much at once—or they experience information overload—their choices may become less reliable or consistent, affecting the validity of the data. 

While some researchers believe that carefully designing and simplifying questionnaires can help respondents manage even the most complex choices, simplifying questionnaires can sacrifice the richness of the collected data. People are often subject to the status-quo bias, which describes our resistance to change, even if our preferences may otherwise differ. Thus, researchers often struggle to balance the experimental complexity and data quality when considering the cognitive load placed on participants. 

Ethical Concerns and Use of DCE Data in Policy

Finally, the use of DCEs in policy-making has sparked ethical debates, particularly around willingness to pay (WTP). Using WTP as a metric for essential services like healthcare or environmental protection can be incredibly misleading and result in inequitable outcomes.⁵

For example, individuals with lower incomes may be less “willing” to pay less for important services, but further investigation would reveal they are simply less able to pay. Without the ability to take this into account, policy decisions may inadvertently prioritize the preferences of wealthier individuals, giving wealthier groups more influence in policy change. Critics also encourage DCE researchers to question which factors should be valued, whose values (trial groups vs. all–trial population), and when participants should be questioned (still receiving the intervention or afterward).⁸

Prostate Cancer Treatment

As discussed above, in the healthcare sector, discrete choice experiments are increasingly being used to understand patient preferences when it comes to choosing between different treatment options. In one recent study of prostate cancer patients from the USA, Canada, and the UK, a discrete choice survey was used to understand patients’ treatment preferences. The survey revealed that patients are most concerned with treatment efficacy—but there are a number of other factors taken into consideration, like the route of drug administration and frequency of monitoring visits. Some of these factors were just as important to patients as potential major side effects like rashes, nausea, and fatigue.⁹

Understanding these preferences is important in order for providers to make informed suggestions and treatment plans for their patients. Many patients, particularly if they’re in the midst of something like battling prostate cancer, may not fully understand all the possible treatment options available to them or their tradeoffs, and DCEs provide a more streamlined way for providers to design treatments that truly fit the needs of the individual.⁹

Public Acceptance of Energy Technologies

For new energy technologies to be successful, public acceptance of their implementation is key. Unfortunately, many studies seeking to understand public perception of different green technologies don’t account for the effects of labeling and time or don’t account for the heterogeneity of the general public. This leads to a biased and incomplete understanding of public acceptance.⁷

Researchers who set out to study public preferences for energy technologies conceptualized three forms of public acceptance: socio-political acceptance, market acceptance, and community acceptance. Using discrete choice experiments, the team found that public preferences for energy technologies are temporally stable, even when people are faced with urgent or recent news of technology failures. The researchers also found that the way technology is labeled has an impact on respondents’ feelings toward renewable and natural gas technologies. (In this case, when labels remain unobserved, nuclear energy and biomass take prominence, and certain socio-demographic groups differ greatly in their sensitivity to labeling and in the temporal stability of their preferences).⁷

From this study and others, it’s clear that discrete choice experimentation allows for more nuanced understandings of public opinion and perception. In green technology,  the effects of labels, time, and heterogeneity can be hugely influential to rates of public support, and DCEs can help us understand those differences. 

Rethinking Voter Preferences: A New Approach to Understanding Election Day Choices

Understand the power of discrete choice experiments in action through TDL director Turney McKee’s use of discrete choice surveys to reshape how we understand voter choices. In this DCE, voters were presented with a series of pairs of hypothetical candidates, with each candidate randomly assigned a set of attributes. By analyzing the trade-offs that voters are willing to make, researchers can build a fairly nuanced picture of the importance of the issues themselves, along with robust estimates of preference for individual policy stances.

Using Behavioral Science to Improve Team Dynamics

Researchers at The Decision Lab explored how behavioral science can enhance decision-making processes within teams. Through a combination of a discrete choice survey and an implicit association test, they uncovered some of the biases affecting project prioritization in team settings. This article summarizes their approach, findings, and recommendations for leveraging behavioral science to improve group dynamics.

1. Haghani, M., Bliemer, M. C. J., & Hensher, D. A. (2021). The landscape of econometric discrete choice modelling research. Journal of Choice Modelling, 40, 100303. https://doi.org/10.1016/j.jocm.2021.100303
2. Louis L. Thurstone. "A Law of Comparative Judgment." Psychology Review, 34 (1927): 273-286. 
3.  McFadden, Daniel (1974). "Conditional Logit Analysis of Qualitative Choice Behavior". In Zarembka, Paul (ed.). Frontiers in Econometrics. Academic Press. pp. 105–142. ISBN 978-0-12-776150-3.
4. Gell, Tim (2023). Choice-based conjoint analysis. Drive Research. Retrieved September 25, 2024, from https://www.driveresearch.com/market-research-company-blog/choice-based-conjoint-analysis/#:~:text=Choice%2Dbased%20conjoint%20analysis%20is%20an%20advanced%20survey%20technique%20in,a%20Choice%2DBased%20Conjoint%20Study
5. de Bekker-Grob, E. W., Swait, J. D., Kassahun, H. T., Bliemer, M. C. J., Jonker, M. F., Veldwijk, J., Cong, K., Rose, J. M., & Donkers, B. (2019). Are healthcare choices predictable? The impact of discrete choice experiment designs and models. Value in Health, 22(9), 1050-1062. https://doi.org/10.1016/j.jval.2019.04.1924
6. Wang, Y., Wang, Z., Wang, Z., Li, X., Pang, X., & Wang, S. (2021). Application of Discrete Choice Experiment in Health Care: A Bibliometric Analysis. Frontiers in public health, 9, 673698. https://doi-org.gate3.library.lse.ac.uk/10.3389/fpubh.2021.673698 
7. Van Rijnsoever, F. J., van Mossel, A., & Broecks, K. P. F. (2015). Public acceptance of energy technologies: The effects of labeling, time, and heterogeneity in a discrete choice experiment. Renewable and Sustainable Energy Reviews, 45, 817-829. https://doi.org/10.1016/j.rser.2015.02.040
8. Tinelli, M., Ryan, M. & Bond, C. What, who and when? Incorporating a discrete choice experiment into an economic evaluation. Health Econ Rev 6, 31 (2016). https://doi.org/10.1186/s13561-016-0108-4 
9. de Freitas, H. M., Ito, T., Hadi, M., Al-Jassar, G., Henry-Szatkowski, M., Nafees, B., & Lloyd, A. J. (2019). Patient Preferences for Metastatic Hormone-Sensitive Prostate Cancer Treatments: A Discrete Choice Experiment Among Men in Three European Countries. Advances in therapy, 36(2), 318–332. https://doi-org.gate3.library.lse.ac.uk/10.1007/s12325-018-0861-3