Health Policy$ense

Nudging Without Shoving - Choice Architecture on Health Insurance Exchanges

2015 LDI HIX Conference Focuses on More Effective Consumer Web Interfaces

This is part of a series of posts covering the Third Annual Health Insurance Exchanges Conference, hosted by LDI’s HIX working group, which took place in April 2015. The full agenda is here and the second post on networks is here.

‘Choice architecture’ remains a hot topic in the world of health insurance exchanges (HIX) and at the LDI HIX conference. How can we structure and present choices that encourage the consumer to select a health plan that fits his or her needs? And at what point does this so-called ‘nudging’ become a paternalistic shove? A number of presenters focused on the theme of letting consumers choose how to choose as a way to preserve autonomy while structuring the choice environment.

Choice architecture is inherently political. Who defines consumer preferences and what’s best for any individual? Does an abundance of information help or hinder choice? The ‘choosing how to choose’ approach is appealing because it gives consumers a sense of freedom, allowing them to see more or less options.

Eric Johnson of Columbia University presented results from an experiment in which participants could use a simplified list of health insurance plans - the default - or switch to the standard layout listing all of the available plans.  Johnson calls the simplified environment the ‘EZPath’ in reference to the toll pass used across the Eastern Seaboard. Do people select the simplified environment, and if they do, are they more confident and satisfied in the plan that they’ve chosen from it?

Johnson and colleagues’ results show that the EZPath helps people make better choices, meaning they select plans that meet their criteria and are less expensive given their needs. People also choose the EZPath more often and return to it more often. However, despite these signals, they aren’t necessarily more satisfied with the shopping experience. Johnson stressed that this needs to be taken into account when designing choice architecture based on consumer feedback.

Johnson also warned of the potential for an “Evil EZPath”. In a NEJM perspective, he and colleagues address seemingly innocuous design decisions, such as the Olympic medal categories for insurance plans on the HIX, and the presentation of monthly, rather than weekly, premiums. They report on a small survey of participants with below-median mathematical ability that found that most preferred gold plans over bronze plans, regardless of which plan was labeled as gold. The survey also showed that participants were significantly less likely to choose the higher-premium, lower-deductible plan when it was presented to them with monthly premiums rather than weekly ones. Johnson and colleagues write:

…we believe that the websites should downplay powerful connotative labels such as bronze, silver, and gold. In addition, they should deemphasize complicated tables of financial information that lay out cognitively overwhelming details about premiums, copayments, deductibles, out-of-pocket maximums, and the like. Instead, they should make it easier for shoppers to estimate total annual costs under a series of plausible scenarios, such as expected utilization based on previous spending history, as well as under best-case and worst-case scenarios.

Another researcher at the HIX conference, Benedict Dellaert of Erasmus University Rotterdam, presented his work with Eric Johnson and LDI’s Tom Baker on the effects of presenting a recommended short-list of options on an online health insurance brokerage. This research builds on studies of consumer decision processes, not necessarily related to health insurance, showing that ranking products strongly promotes better choice outcomes (see here and here).

Dellaert and colleagues studied the largest online health insurance broker in the Netherlands and its use of “partitioning on rank of chosen product”.  In one enrollment period, the Dutch website provided a list of all options raked by price; in the next enrollment period, it provided consumers with a “Top 3” list sorted by price-quality, with the option to see the entire list of choices. Dellaert’s analysis shows that partitioning increases the market share of highly ranked products. Consumers make more in-depth comparisons in a pre-selected set of top alternatives, potentially improving the fit of the plan that they choose.  They don’t necessarily choose the top option on the list, and quite commonly they go for the third one. A marketing expert at the conference remarked that this is in line with how people are known to choose a wine off of a restaurant list.

In a related online experiment, Dellaert redesigned the consumer decision task in a simplified controlled setting. The experiment randomly assigned consumers to a partitioned or non-partitioned list, as well to lists ranked by price alone or a price-quality algorithm.  The results confirmed that partitioning increases the market share of highly ranked products, and that benefits accrue to other highly-ranked products besides the top alternative.  

He also examined the choices people made when choosing to see the full lists.  Surprisingly, a top choice was a plan that did not rank in the top three on a list sorted by price alone or price-quality.  When asked, consumers said they valued characteristics such as a broader provider network, ‘choice of care’ and ‘hospital near you’.  Dellaert concludes that these preferences are important for consumers and any sorting functionality must be able to take this into account. Choice architecture is a fast way to get people to the right choice, he noted, but there is a risk of pointing them to the wrong choice as well.

For more, here is a useful review of evidence on the tools of choice architecture. How to decide what to present to decisionmakers, and how to present it, will remain fertile ground for researchers and policymakers as they seek to improve choice on the health insurance exchanges.