Daughter: “Daddy, can me and pretend friend Lulu have a snack”
Me: “Sure, do you want two choices or three choices??”
Daughter: “We want three choices”
Me: “Ok, your choices are carrots and hummus, strawberries with whipped cream, or apples”
Daughter: “Strawberries with whipped cream on the yellow plate for me. Lulu wants the purple plate…”
This is a daily interaction. My daughter is five, and she wants to be in control with significant personal autonomy. We want her to at least believe that to be the case, so choice architetecture is a skill that my wife and I have spent a lot of time to develop. If we don’t give her a choice, and just hand her strawberries with whipped cream, she’ll refuse the strawberries. If we give her fifteen choices, one of two things happen. If she is not exhausted, she asks for a more limited choice set. If she is exhausted, she meltdowns as the cognitive load of deciding what she wants for a snack is too much for her tired brain.
The key to our choice architecture is for mom and dad to curate the option space down to a reasonably narrow set of choices where none of the choices are bad. Last night, I had a mild preference for her to choose carrots and hummus as I would have enjoyed carrots and hummus leftovers from Lulu’s plate, but I was functionally indifferent to the three choices she had. She was able to make a good choice, feel like she was in control and be a BIG GIRL.
Adults tend not to go into incomprehensible crying fits when they encounter too many choices which overwhelm their cognitive processing ability. Instead, they choose quickly and they tend to choose poorly. Medicare Part D is an illustrative case via Incidental Economist:
How could reduction in choice possibly be a good thing? Turns out, there’s a substantial literature on just this question. According to a 2012 Kaiser Family Foundation survey, 40% of seniors responded that the degree of multiplicity of plans was confusing and made it difficult to select the best plan. In a recent paper published in Health Affairs, Chao Zhou and Yuting Zhang summarize similar evidence for Medicare Part D.
Jason Abaluck and Jonathan Gruber [ungated working paper version here] observed that Part D enrollees had difficulty making their initial plan choices when Part D started in 2006. They found that beneficiaries paid more attention to plan premiums than to their own total out-of-pocket health expenses. Florian Heiss and colleagues, using 2007 and 2008 Medicare Part D data to study plan choices, found that fewer than 10 percent of consumers enrolled in the least costly plans in 2007 and 2008 and that beneficiaries could save on average about $300 per year if they switched plans. [Links added.]
Jason Abaluck and Jonathan Gruber found that the suboptimality of choices by beneficiaries was persistent over time.
I was on Healthsherpa last night. Choice overload abounds.
For my age, income and family status, there were fifty two choices on the Exchange that I could buy. If I entered the zip code of my older sister’s favorite show from the early 90s, there were forty eight choices. If I used my parent’s zip code, twenty three choices appeared. If I used my first apartment’s address, sixty three choices appeared. If I made the prospective buyer 29 years old, another eight to twelve choices of catastrophic policies were added for all zip codes. If I dropped my income to qualify for cost-sharing assistance, another six to eight choices per address popped up.
If I had to buy insurance, I would be in choice overload and buy based on simple heuristics that are not even satisficing on anything other than stopping the process quickly much less optimizing for the best health plan that meets my anticipated healthcare needs given my financial constraints.
One of two things should happen for the 2015 open enrollment period on the federal exchange. The first is a structured decision support tool could be added. This tool would allow for the full array of choices to be made available to any shopper. However, it would ask several filtering questions. These could be a simple as: “do you want low premiums but high out of pocket expenses or higher premiums and lower out or expenses” and “do you already have doctors that you want to keep in network, or do you really not care who your doctor is”. A few more simple decision support questions would quickly narrow down the shopping scope to three or four choices. That is a managable cognitive load where satisficing choices have already been brought into the potential optimizing option space.
The other option is for the federal exchange to transition from a clearinghouse model where any minimally qualified plan is displayed and available for sale to an actively managed market model where the Feds either set strict definitions as to what can be sold, or choose a subset of plans to sell from the qualifying plans. I don’t think this will happen due to the need to have insurance company buy-in as the losing companies will have a legitimate reason to scream.