7 December 2012

Are You Sure You Want This?


One of the fundamental assumptions of rationality and normative economics is that people’s preferences are stable. In other words it goes something like this: we can’t judge what someone likes (prefers) but once we know what she likes (wants) we are sure that the preference will not change.

The classical example is: if you prefer apples over oranges and you prefer oranges over peaches then for sure you will NOT prefer peaches over apples.

This is one of the foundation stones of rational choice theory. At the same time it is a very flawed assumption. Let me give you an example:  do you prefer coffee or beer? Apart from people who don’t drink either one or both of them, for the rest of the human species this is a really tricky question. The reason is that it doesn’t say when or in what context this preference is exhibited.

Let’s assume that you go out for a chat and a drink with an economist. The economist asks you what will you have coffee or beer. Now, it is early evening (say… 7:35 PM) and you had a stressful day. You say that you could use a beer. The economist you went for a chat and drink with will assess that your preference is beer over coffee. You have a couple of beers and have a nice chat with the economist (well… as nice as a chat with an economist can be).

Now, imagine that for some reason you meet the same economist a few days later for breakfast at his place. The economist made a nice omelet and some toast. Because he knows your preference already he offers you a beer.

Is this at least a bit weird? Of course it is. Who has beer for breakfast apart for a few alcoholics?  

This was a bit far-fetched example of violations of the stability preference assumption. A valid critique would be that coffee and beer are not in the same category and to a certain extent it is true. At the same time, people change their preferences even within the same category. Next I will explain how this works depending on how we evaluate the options available for products within the same category.

In both theory and real life there are two ways in which we evaluate options that are available. In the separate evaluation mode we evaluate options one by one. For example you want to buy a couch and you have little money so you decide to buy a second-hand couch.

You search on a website with personal selling ads and you call a guy who is moving and sells his old couch. You next go to the guy’s place to see the couch. You get there and start evaluating the couch. Since there is only one couch in this guy’s apartment you evaluate a single couch at a time. The evaluation can be done in various ways. For example you evaluate the couch by thinking how much you would pay for it or you can give the couch a liking score.

Evaluating a product (couch) in this scenario is called in scientific terms “separate evaluation”. If you go to another guy’s house to see another couch, then you will evaluate the second couch on its own (without making a direct comparison between couches).

Taking the same scenario that you want to buy a second hand couch you decide to a second-hand furniture store. There are two couches for sale on that day and they are placed one next to the other. When you are evaluating each couch you inherently start making comparisons between them. This is called in scientific terms “joint evaluation”.

You might think that this is really trivial and to a certain extent you are right. You might think that this is how things are and that’s it. Indeed, this is how things are and sometimes we evaluate (and subsequently choose) things either one by one or jointly. The most interesting thing is that depending on how we evaluate options (separately or jointly) our evaluations and subsequent choices are likely to change.

Hsee, Loewenstein, Blount and Bazerman summarize in an article from 1999 how these changes of preferences work.

A series of studies have adopted the following framework. In order to establish preferences in the separate evaluation mode there were two groups of random people. Each group was presented with one description of a product (or person); the description was brief and mentioned two attributes of the product. Each person in each group was then asked to evaluate the product by either saying how much they liked it or by saying how much they would pay for it. In order to establish preferences in the joint evaluation mode,  a third group of random people was presented with both products and they were asked to say how much do they like or how much would they pay for each product. Next the preferences (likings or willingness to pay) established in the separate evaluation mode were compared to the preferences established in the joint evaluation mode.

The interesting thing was that the preferences in separate evaluation mode were different (opposite) from the ones in the joint evaluation mode. Let me give you an example:

There are two (second hand) music dictionaries:
Music dictionary 1 has: Number of music terms explained: 10.000 | and Condition: “Like new”.
Music dictionary 2: Number of music terms explained:  20.000 000 | and Condition: “Has a torn cover, but otherwise like new”.

I am not going to ask you what you prefer because now, you evaluate the two dictionaries jointly. However, in the separate evaluation mode (one group evaluated the first music dictionary and another group evaluated the second music dictionary) Music Dictionary number 1 was rated higher (preferred).

As you might have preferred yourself, in the joint evaluation mode Music dictionary number 2 was preferred. Thus preferences in separate evaluation mode were reversed when using the joint evaluation mode.

The rationale behind this is quite simple. If we look at the two attributes that describe the music dictionary we see that there is a relatively easy to evaluate attribute, namely the condition of the book. Everyone that has seen a book knows what “Like new” and “Torn cover” mean.

On the other hand, (apart from music specialists) no one knows how many music terms a music dictionary should have. Both 10.000 and 20.000 are large numbers (at least when it comes to words) so the question “how many terms should a music dictionary have?” is a hard one. In other words, the number of terms a dictionary has is a hard to evaluate attribute.

When evaluating each dictionary separately, the easy to evaluate attribute “condition” is prevalent in the evaluation.

Now, when evaluating the two dictionaries together (jointly) things change a bit. The attribute “condition” remains easy to evaluate. But now the two dictionaries can be compared on the hard to evaluate attribute “number of music terms”. This does not mean that suddenly people find it easy to evaluate the number of music terms a dictionary should have, but everyone knows the rule “the more the better”. When the options can be compared against each other on the hard to evaluate attribute, the weight given to the easy to evaluate attribute diminishes.

The implications of these preference reversals are major. On one hand there is the theoretical implication of a severe violation of one assumption of rationality, namely the stability of preferences. On the other hand there are numerous practical implications.

In many instances in life we are evaluating options in separate evaluation mode. For example we date one person at a time (well, most people do so). We see apartments to live in one by one and we have diner with friends one instance at a time.  
Is this good or bad? The truth is that there is no universal answer. What we can do is to know that when evaluating in separate evaluation mode we tend to focus (base our evaluation) on the easy to evaluate attributes.

In other instances in life we are evaluating things in joint evaluation. For example you want to buy a new TV / Stereo music player and you go to a shop like MediaMarkt or Saturn. When you get to the area where the TV / Stereos are presented you are faced with tens of options and by nature you start evaluating them jointly. Now in the case of a new TV, you will see one that has a very good image and a reasonable price, but right next to it there is a larger TV with a Crystal-Clear image and when you look on the shop’s wall and compare one to another, of course you will like more the larger one with crystal-clear image that costs 800 euros more than the other and takes twice the space in your living room.

When you will get the large TV with crystal clear image into your living room you will only notice that it takes a lot of space and you are 800 Euros short. In your living room you are evaluating in separate evaluation mode, thus now you don’t see the difference that you saw in the store. Moreover in a couple of days you will forget about the difference and you will be stuck with a large TV and with a hole in your wallet.  

There is really no universal answer for which type of evaluation mode – separate or joint – is better. Moreover, it is possible to evaluate options in a mixed way. Going back to the second hand couch example, you can see within 3 hours three different couches at three different guys. Each couch will be evaluated separately, but you will make mental comparisons among them. However, this type of evaluation is not the same as evaluating all three couches at the same time in the same location (joint evaluation).     

What I believe to be the key learning here is that preferences can change depending on the mode in which options are evaluated.

Like it?  Spice Up  Your Business


Note: this post is documented from Hsee, Christopher K., George F. Loewenstein, Sally Blount, and Max H. Bazerman (1999). “Preference reversals between joint and separate evaluations of options: A review and theoretical analysis.” Psychological Bulletin, 125, 576-590.  

No comments: