19 June 2015

What is a Bargain?

A while back, I heard the following definition for a bargain:

It is something that you don’t need at a price that you can’t resist.

Although this definition might seem too anecdotic and a bit “unserious”, it is a wonderful illustration of the psychology behind bargains and it explains why sometimes we buy not-so-useful things.

The definition above illustrates that there are two types of utilities involved.

For those of you who don’t know what utility is, it is a fancy word used by economists to describe benefit or pleasure or happiness. For example consuming a bottle of water will bring some benefit / pleasure / happiness to the drinker.

Normative economics takes into account the utility derived from consumption of the good or service. Behavioural economics, however, acknowledges that alongside the benefit of consumption, there is some pleasure (utility) that comes from the purchase itself. This was named by Richard Thaler (1985) as Transaction utility.

In other words, Transaction utility is the pleasure one gets from the deal itself.

Quite interestingly, in the case of a bargain, the consumption utility is close to zero or even negative (i.e. you have to do something with that useless thing and most likely it will bring some head-aches.) while, at the same time, the transaction utility is very high – a price that you can’t resist.

What I believe happens in the moment of purchase is that (at least some) people recognize the low consumption utility, yet can’t resist the great deal. Subsequently, because of confirmation bias, they try to find arguments that support the purchase. More specifically these arguments are directed towards tackling the low consumption utility. In other words, although they realize that the item itself is not particularly useful, they conjure possible uses for the item.  

Thaler, R.H., (1985), "Mental Accounting and Consumer Choice," Marketing Science, 4 (3), 199-214.  

9 June 2015

Why Are You Asking Me Your Question?

A couple of days ago I noticed that some of my connections on Linked In were answering a question posted publicly. I do not want to name names, but since the question concerned two packages for dog food, it is inevitable to show the two packages.

The message on Linked In was:

Hi, Would love some feedback! Our current packaging is the design in the brown Kraft bag on the left and a draft of our new packaging idea is in the creme bag on the right. Which design do you prefer??

Beyond aesthetics and preferences for dog-food bags, there are some very serious issues with this kind of pseudo-research.

First, there is the obvious sampling error. The linked-in connections of the person who asked are not necessarily dog food buyers. I guess the budget for market research was very restrictive.

Second, it is utterly wrong to ask people which one of these two packages is preferred. Which one to pick, is the question that the marketer has to answer. However, the buyer (because the consumer is the dog – hopefully) will never ever have to make this choice.

Choices and preferences are to a large extent dependent on the context of the available options. We also know that choices imply (sometimes unconscious) comparisons. If package B is aimed to replace package A, then the buyers will never have to make the choice between A and B.

Third, the choices buyers make are not made in a void. In other words, the choices dog-food buyers will make are between competing products and one of the products (packages) presented above.

So, a correct way to do things would be to have a between subjects design with two conditions:

Condition A: N competing brands + package A

Condition B: N competing brands + package B

Here N is the number of main competitors on the market. The same competing brands should be used in both conditions.

What should be measured is the choice share for each package in the two conditions.

Fourth, asking people about their preferences is sub-optimal. Preferences are not as stable as we’d like to think. They depend on the other options the choice set has and on many other factors. In this case, it is obvious that asking people online about their future choices in supermarkets is a bit of a stretch.

Fifth, related to asking about preferences and purchase intentions, there is quite a difference between intentions / preferences and actual behavior. So the best way to do this is to measure behavior and not intentions. Though, my friend John Kearon, thinks there’s a better way to do this kind of research.  

The interesting thing is that someone had to make a choice / answer a question. This person then asked other people the question she had to answer. However, the respondents have to answer different questions to make their choices.

Now, the question I have to answer is:

Why I wrote this post on dog-food packaging when I am definitely a cat person.

P.S. The person who asked this question owes me a beer for (free) consulting… A Hoppy Cat beer :)

Later edit (15 Jan 2016): Here's another similar example (also via Linked In), different area, but the same problem:

Good morning all! I would greatly appreciate if you could comment on which logo you prefer most. I can't say much more than it'll be for a design business.   Any comments, likes, shares and feedback will be appreciated.  Thank you!

The Problems with the Sample of One

The first few days after moving to the USA, my wife and I stayed at a hotel near our new apartment. One morning, at breakfast, there were two gentlemen at the table next to us. One of them spoke really loud and with profound stamina. Apparently there was some kind of religious convention because Finding Jesus was the main and hottest topic of their conversation.

Just as a note: although I am not religious, I have nothing against religion and practicing it. It simply bothers me when religion is used to promote self-interested bull…….. to way too naïve and vulnerable people.

Coming back to the chat the gentlemen next to us had, the one who was talking (much too) loud was explaining to the other guy how he is prepared to speak / preach to his congregation about (wait for it…) Finding Jesus. He strongly emphasized that he has three or four stories about people who were pretty messed-up and came to his church and in the end Found Jesus and got their lives in order.

The loudly speaking man briefly shared one of his well-crafted stories. As far as I can recall, it is your prototypical alcohol + gambling + debt + lost job + brake up story. However, the character of the story (Jeff if I remember correctly) came to his church and after a while Found Jesus and got his life back on track after a while.

Finding Jesus is not singular. Some people believe that they will be cured of cancer or other terrible illnesses by kissing bones of people long gone – saints. Others believe that they could recover from illness after drinking (herbal) magic potions or other scams from wannabe (unconventional) healers. Others believe that it was because a beautiful lady blew over the dices that they won big in a casino.

For each of these miracles there are stories to back them up. Usually these stories contain one character who was hopeless and resorted to such, let’s say, tricks and by a miracle he or she got what was desired.

All of these are problems of the sample of one.

Whereas the term Sample of one seems to be related mostly to statistics, the issues behind the problems of the sample of one are more related to psychology of judgment than anything else.

First, there is the Availability Heuristic. These so called success stories are overly exposed, most often taken entirely out of context. All the poster-child cases are paraded while the huge majority of not so successful cases are hidden somewhere in a back-closet. Everyone will show off the one person who kissed the knee-cap bone of Saint I don’t know who and who was cured of cancer. However, the other millions of people who also kissed the same piece of dead bone and ultimately died because they were not cured by the holly relic are entirely ignored or even labelled as not true believers.

Second, there is The Narrative Fallacy. We humans are very good at conjuring stories so that what we see (want to see) makes (im)perfect sense. Although all these so called success stories are most likely due to pure chance, we very easily make up a story around them so that they fit with what we (want to) believe.

Third, there is The Coherence Issue. Usually the stories of Finding Jesus or the magical effect of a beautiful lady blowing over dices are exceptionally well crafted. The loud gentleman at the hotel whom I mentioned earlier was very proud that his stories are very good. One thing that makes a story good is its level of coherence. A good story is very coherent. Unfortunately we too often mistake coherence with truth.

Fourth, there is Wishful Thinking. The (sad) truth is many of us want to believe these stories mainly because they fit with what some believe and because they self-serve us. In case we have a difficult illness, we would like to have the extra hope of kissing dead bones and being healed. We would love to be able to do something to improve our chances of success in a casino and overall in our lives.

I know that it is very hard to understand random chance.

I also believe it is even harder to accept that random chance plays such a big role in our lives.

3 June 2015

Buy it! 3% Pure Fat! Classic Framing Example

The Framing Effect is probably the best know implication of loss aversion. A piece of information can be framed either as a gain or as a loss by manipulating the reference point.

In this case, the ham is presented as 97% fat free which implies that the reference point is 100% pure fat. Thus having it 97% fat free seems like a gain.

However, the same information could be framed as 3% pure fat, implying that the reference point is 0% fat. Having the information framed as 3% pure fat would make it feel like a loss.

Naturally, (most) people are more willing to buy something that is 97% Fat Free than they are willing to buy something that is 3% Pure Fat. 

Personal note: Before moving to the USA, I had only heard about this framing, never saw it with my own eyes...

1 June 2015

Dieting by eating Chocolate and Attitudes Towards Gay Marriage: If it Seems Too Good to Be True and You’d Love to Believe it, it Probably isn’t True

On Academic Fraud and Scientific Stunts

I have to admit that I have a weight problem. I’m not obese, but I could surely benefit from losing 10-20 kg (22-44 pounds). Occasionally I try to diet, in the sense of eating more vegetables, less meat, as little as possible fat, avoid sugar (particularly added sugar) etc. And, YES, if you want to diet, you have to give up wonderful things such as bacon, chocolate, bagels and deep-fried food.

Wouldn’t it be wonderful if a dieter could hold on to one of these incredibly tasty ingredients?

Of course!

But, what if one such ingredient would actually help to lose weight?

What if EATING FRIED BACON would help you lose weight?

Who wouldn’t want to eat fried bacon and still lose weight?

Without being a nutritionist (I just mentioned I’m overweight), but being able to pass biology in middle-school, I know that eating fat is not going to make anyone slimmer.

Things are similar when it comes to other (guilty) pleasures of food. Eating chocolate, bagels, donuts, ice-cream and sweets etc. is very pleasurable and there is a reason for it.

A few million years ago, when our evolutionary ancestors (pre-homo sapiens species) lived, it was very good for them to eat as much sugar and fat as they could. This is simply because sugar and fat were extremely scarce back then. (Remember, we are talking about a few million years ago).

Whether we like it or not, we have inherited this natural appeal for sugar and fat, but in today’s (western) world these ingredients are (way too) abundant. Therefore it is not in our own benefit to eat as much of them as we can, even if our instincts tell us to.

But what if the facts above are wrong? What if, eating donuts is good for our health? It would be great! We could indulge in sugar and fat (read donuts) without guilt and even with a sense of pride since we are doing something good for our health.

This would be something many of us would love to be true. We want it to be true… and because of this, often we skip the scepticism that is so often needed.

A journalist, John Bohannon, managed to fool lots of people into believing (considering as true) that eating chocolate is actually good for dieting. He devised a real experimental study in which this result was statistically significant.

Unlike many dieting gurus, John Bohannon didn’t do this study in order to sell his books, workshops etc. and make a fortune by selling desirable illusions. He did it to prove a very valuable point: we are not critical enough with the results that are presented, particularly with the methodology of the study.

You can read his full explanation here (open in new tab and read later).

In very, very brief the methodology was flawed and the over-emphasis on statistically significant results is a two-edge blade.

In a different part of the (scientific) world, a new scandal on academic fraud (faking data) erupted. This time it concerned political science and the main reason why this scandal made such a big bang is that the study published (using fake data) became very popular.

One of the reasons for which it became so popular was that the study was published in Science. For the people who don’t know what Science is: for scientists it is similar to the Hollywood walk of fame for actors and the Rock and Roll hall of fame for musicians.

Another reason for which the paper became popular was the fact that Donald Green (a super-star in political science research) was one of the co-authors. Mr. Green claims that he was not involved in the actual running of the study and that the other co-author (a grad student) brought the data.

However, the main reason (in my view) for which the study became so popular is that it told us what we wanted to hear about a rather sensitive topic: gay marriage.

The study claimed that attitudes towards gay marriage can change (from negative to positive) after a 30 minutes visit from a canvasser who disclosed that he or she is homosexual.

You can read a detailed description of the fraud and how it was discovered here

The fact that occasionally some people fake data should not surprise anyone. Academia is not immune to cheating and fraud.

What I find very interesting is that this case has some similarities with the chocolate helps in dieting paper. Obviously both are methodologically utterly flawed. But there is more than just poor methodology.

Telling people that attitudes towards a sensitive topic such as gay marriage can be changed by a 30 minutes interaction with a gay person is very similar to saying that eating chocolate (or bacon) is actually helpful in dieting.

Both are things many of us want to hear and both are a bit too good to be true (once you become critical).

The beauty of behavioural science and other branches of science is that they prove our intuitions wrong, that they show asymmetries (i.e. small input, large output) and that they promise actionable means for practical applications.

There is, however, a difference between counterintuitive and too good to be true.

Whereas counterintuitive things go against our initial beliefs, things that are too good to be true confirm what we want to believe.

Both chocolate helping dieting and changing attitudes towards a sensitive topic after a 30 minutes interaction are in accordance with our desired utopias. Many of us would love to eat chocolate and lose weight; nonetheless this seems utopic. Similarly, many pro-equality supporters would love to have found a way to change conservative views on gay marriage in just 30 minutes; again this seems utopic.   

If it Seems Too Good to Be True and You’d Love to Believe it, it Probably isn’t True