Episode 39: Jonathan Sicsic and Preferences for Nursing Homes and Home Care

And I guess the idea with including levels that don't seem that someone could have a preference for is because you want people to trade off these aspects, right? So I can imagine no one would actually want that.

Like, I don't know what would be a bad level attribute here, but...

Yeah, yeah, it's a good point.

So when we develop discrete structure experiments, we need to think about a lot of aspects at the same time.

So we need to think that to find the dimensions that are really the most important to people, but also dimensions that people actually trade against.

So there is the idea that there is no dominant attributes or levels.

So this is why this is difficult to do a stated preference on long-term care where we add the location of care as an attribute.

Previous studies have used, for instance, whether the care was at home or in a nursing home as an attribute and found that home care was a dominant attribute.

So this is why we break down the decision process into two and even three stages.

And then regarding the attributes, we didn't find really that there was one single attribute that was most dominant.

We imagined that cost was really important.

But its cost was around 40% of the total choice variance explained.

So it means that 60% of the choice variation comes from the three other attributes.

So that's fair.

Regarding the levels, we found that actually we build a score.

So this is a controversial choice of giving a score to the quality dimensions.

because it's difficult to place a number on a quality, for instance.

But we felt that for a policymaking perspective, we have more and more countries that develop scoring cards to score the quality of nursing homes or institutions.

This is, for instance, the case in the US.

There are also debates in France.

The high health authority tried to develop a scoring algorithm where nursing homes are scored A, B, C, or D.

So A is the best.

And actually, what is quite interesting is that they released this scoring algorithm one year ago.

And we look at the documentation and they actually use the same dimensions as we did in our paper, while we didn't have any interaction with this high health authority.

So we found similar dimensions.