Friday, May 31, 2013

What is Wrong with This Prudential Ad?


Prudential is running a new advertising campaign titled "Let's get ready for a longer retirement"
The text under the picture says:
A typical American city. 400 people. And a fascinating experiment. We asked everyday people to show us the age of the oldest person they've known by placing a sticker on our chart. Living proof that we are living longer. Which means we'll all need more money to live in retirement. and so on.

While I agree that statistically we are living longer than our predecessors and we need to save more for our retirement, I doubt this experiment proves it.

I can imagine how it all came to be:
- Folks, let's emphasize that we are dying at an older age, and need to save more.
- But, Bob, we cannot talk about death in an ad campaign.
- Yeah, let's find something positive in us growing older, sicker, lonelier, absolutely dependent.
- Well, everyone is excited to hear about someone in the stress-less mountains making it to a 100.
- Great idea, let's talk about the oldest person you know!

So, why this chart is not a living proof of us living longer?

Your answers are accepted any time until midnight Eastern Time on Sunday, on our Family Puzzle Marathon.

17 comments:

Jill said...

I think the issue here is that in a town of 400, there's a good chance everyone knows the same 2 or 3 older people.

Concerned Teacher (Happily Retired) said...

Each dot represents only an "observer" of the death age of someone else. What if 10 people knew the same 96 year old, 10 people knew the 97 year old, 10 people - the 98 yr.old, 10 people - the 99 yr.old. Their 40 dots do not represent 40 different people, but only 4 different people. Now, what if 40 people all knew and referred to the same 96 year old? This is NOT a measure of how many different people died at any age.

Ilya said...

In itself it does not indicate we are living longer. At the very least we should compare this data to a similar survey from the past. And even with the comparison, it would not constitute a "proof", since the outcome depends on the "social network" structure. A 40 year old version of this survey may look "better" just by virtue of the older people being known more widely.

Anonymous said...

One reason this experiment does not prove that people in general are living longer is that participants were asked for the age of the oldest person they have KNOWN - which means many people could list the age of the same person of particularly advanced years.

Another problem is that people were asked about anyone they have (ever) known, which means that person may have passed away long ago - which gives no information about how long people are living today and whether there is a trend toward longer life.

The data would be more valid if people were asked for the age of their oldest living relative. (Although there could still be duplication if more than one member of a family participates.)

Yael R.

HavenStats said...

-Large number bias occurred- The sample mean of the data collected will be (probabily) larger than the average life expectancy of a person because the data is just of the oldest person they know. If the distribution is trying to estimate the average life expectancy they need to randomly sample from all the people’s ages of people that have died recently. This distribution illustrates the average age of the oldest person still living. We cannot assume that on average a typical person will live this long because it only shows the oldest person that they know.
- Response bias may have occurred because some people may not know the correct age of the oldest person they know and they also know that their answer will be on TV, so they may have exaggerated their answers.
- All in all, we still think it is a cool ad that highlights some basic statistics in a fun way.

Annie said...

I think there are many reasons this is not proof!
The people chosen to participate may not be a random selection. They may be people pleasers who want to prove the hypothesis, they may work or know someone at Prudential and also want to prove the hypothesis, they may be an unusual group that knew many people over a certain age. People may also know one person who lived past 85, say, but many, many more who died young! It also says nothing about the past so how do we know a chart 50 yrs. ago would be any different?

Anonymous said...

There is no y axis on this chart. The vertical line is an extension of the 65 yo data point on the x axis.

TyYann said...

This is wrong because lots lf people in the same comunity might know the same old person and this would get that old person a higher statistical weight.

TyYann said...

This is wrong because lots lf people in the same comunity might know the same old person and this would get that old person a higher statistical weight.

Lynnet said...

Well, it's just asking how old the *oldest* person you know is. Also, a sample size of 400 people in a country as large and populous as the US is not that large or significant. It's basically just asking how old the oldest people in the country are, not how old the average person lives to be. These ages are outliers, not the average life expectancy for people in the US. This is just based off basic math knowledge and common sense.

anne-marie said...

These are statistics for a typical city so first, I will say that the sample is not representative of the population.
Many factors have to be taken into consideration such a socio-economical, cultural or environmental.
Then, I thought about the generation below 70, the baby boomers. The probability that more people are going to survive their 80's, 90's and so on should be higher (proportional) so the next graph could show us bigger numbers of 80 years old and above and it won't be a proof of us living longer.
Other factors such as climate, demography or geography have to be taken into account.

Anonymous said...

from TracyZ

There are so many things wrong with this ad. The main one in my mind is that the fact that people know or have known someone very old just indicates that some people live a very long time (and most people know someone who have lived into their 90s or over 100 years old). It does not in anyway indicate that the average or median life expectancies are increasing and that people should adjust their financial planning accordingly. It also indicates nothing about how long say 95% of people (2 SD from the median assuming a normal distribution)live. If people make financial arrangements to cover their expected costs to the 95 percentile age, that is more than sufficient for most people, many of whom will live much shorter. The oldest person that people know or have every known is an outlier, falling into a small special uncommon group of people with the right genes, lifestyle, diet, and luck to live a very, very long time.

A few other comments:
1) The ad just asks people to give the age of the oldest person they've known. There is no requirement that the person still be living, or living in the sample town of 400 people, and some respondents might have been thinking of people they didn't know personally, but people they have heard of, for example celebrities or people they read about in the newspaper. In New England (Massachusetts mainly), many towns award the Boston Post Cane to the oldest citizen in the town, and the award ceremony is usually in the news. The Boston Post Cane tradition started in 1909 (http://web.maynard.ma.us/bostonpostcane). Maybe the town for the ad has something similar.

2) The ad says that the dats shown provides "living proof that we are living longer." To use the term "longer", there really should be data for at least two time periods, not just one. How do we know that the oldest people represented by the dots on the chart are older than the comparable oldest people from an earlier time period.

In the past few years, I have media coverage of studies suggesting that the average life expectancy in the US is no longer increasing but falling, a trend that is caused by more sedentary lifestyles (more computer use, less manual labor, in general less physical activity) and less healthy eating (more processed foods) than earlier generations.

Jerome said...

I'm going to answer this in two parts because there really are two parts to this question. The first part is "What's wrong with this picture?"

This picture refers to the Prudential Model of longevity.

1. The sample is very flawed. The sample consists of all the oldest people that anyone knows. All the deaths of the general population that are medical (as in death by childbirth complications), or war related or accidental (as in car accidents), or epidemics have been taken out. Including all deaths would measure a true longevity of a general population, not just the current survivors.

2. The criteria is flawed.
Longevity measures age at death. This sample is measuring those still living.

3. The way the response was collected could be flawed. Given that I overlook the first two objections that I have listed (not easily done), suppose a whole family looks at this chart. One would all say

"Well would you look at that?!! Great Uncle Harry is 93. Let's all vote his age. Anyone know anyone older?"

"No!!" everyone else says.

"So let's honour him," the first speaker says, "Everyone send in a vote."

If they do, they've skewed the results. All that family's votes are for the same person.

4. It trivializes the problems of measuring longevity, which is not necessarily Prudential's fault. They intend this to be light heart-ed. The problem is actually quite complex (see my next response).

Jerome said...

Thank you Maria for a very interesting problem. I feel that your questions are not only thought provoking, but most important entertaining.

I had intended in this part to discuss the problems in measuring longevity itself, but I found the topic just too depressing. Many problems arise from war. This article discusses the many problems confronting those who try to compile premature death.

http://www.cissm.umd.edu/papers/files/deathswarsconflictsjune52006.pdf

The second article I wanted to summarize (which is somewhat more mathematical is this one

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3000019/

I leave those references for those willing to wade through them. They make somber reading indeed.

Jerome's Wife said...

I was surprised first of all that a town is described as having 400 people. I might call a place with 400 people as a village. I am hoping the billboard was not set up in a place that small because it would make the results of the blue stickers on the billboard even more unrepresentative of whether people are living longer. In fact I don't think the premise that it can be determined whether people live longer by simply asking participants to place a sticker on the age yardstick of the billboard for the oldest person they know. It is a flawed premise. There assuredly are people of varying ages that are not represented by the small sample of people reporting people they do know of various ages.

Maria said...

Wow, you are amazing as usual.
Not only the answer I expected but many more insights and interesting information.

Check out all the answers above and especially Jerome's summary and his pointers revealing complexities of measuring longevity as well as TracyZ's savings advice and her suggestion that we may not be living longer. A puzzle point for everyone.

Anonymous said...

TracyZ again:

When I was in college, I had the privilege of interning with S Jay Olshansky, a public health professor who researches the physical limits of how long people can live. In 2001, Dr. Olshansky bet a colleague that by 2050, someone will live to be 130. The colleague thought that someone could live to be 150. More details on the wager and their research is available here: http://www.unisci.com/stories/20011/0112016.htm

Of course, as many people answering Maria's query above posted out, how long the oldest person lives has little to do with how long most people live.

---
Dr. Olshansky is also quoted in a more recent article here, that talks about the declining life expectancy among the least-educated white people in the US. http://www.nytimes.com/2012/09/21/us/life-expectancy-for-less-educated-whites-in-us-is-shrinking.html
A troubling trend.

Post a Comment

Note: Only a member of this blog may post a comment.