Monday, 14 January 2013

"The end of history illusion" illusion

An intriguing study was published in Science recently with the eye-catching title "The End of History Illusion". It spawned plentiful uncritical media coverage, including the You Won't Stay The Same article in the NYT, and already the effect has its own Wikipedia entry. But does it really exist or has hype and stylish presentation generated an illusory illusion?

Jordi Quoidbach, Dan Gilbert and Timothy Wilson claimed to have shown that people of all ages underestimate how much their personalities, preferences and values will change in the future. Ideally the psychologists would have asked people to predict the amount they'd change, and then followed them up later to compare the actual change with the prediction. They didn't do that.

Instead, their favoured approach was to survey over 7,000 people aged 18 to 68 and to ask some of them - "the reporters" - to recall their personality ten years ago, using a personality questionnaire, and to ask others - "the predictors" to estimate their expected personality ten years hence. All participants also answered questions about their current personality. The key contrast was the amount of difference between recalled vs. current personality (documented by reporters) and the amount of difference between predicted and current personality (documented by the predictors). Across every decade of life, the former was much larger. People thought their personality had changed more than they thought it would change.

But there are some serious problems here. First off, there's no actual data on real change. How do we know this is a delusion of prediction? Perhaps the reporters are wrong about how much they've changed. Quoidbach's team realised this so they looked at data from a separate longitudinal study that really had followed people over time (once in 95-96 and again in 04-06), allowing examination of actual personality change.

Unfortunately, this MacArthur Foundation Survey of MidLife Development in the United States (MIDUS) used a completely different measure of personality. "Direct comparison of the data was not possible," Quoidbach et al confessed. Nonetheless, after comparing the amount of personality change in the MIDUS survey with amount of change estimated by reporters in their own study, the researchers said the change was "almost identical", and "substantially larger" than the change predicted by the predictors in their study.

But there's another serious issue with the new research, which was highlighted in a recent blog post: that is, the predictors might well have believed that their personality will change, but they didn't know which direction it will change in. Take extraversion. Perhaps they will become more shy, perhaps more gregarious? Without knowing the direction of change, the most accurate prediction is to report no change in extraversion from their current score.

Quoidbach's group hinted at this problem by acknowledging that participants may not "feel confident predicting specific changes". To overcome this, they asked over a thousand people to answer a non-specific question, estimating how much they felt they'd changed as a person, or how much they would change. It doesn't entirely deal with the direction of change issue, but again, "reporters" estimated more change than "predictors" predicted.

Yet another survey with thousands more participants asked them to recall or predict changes in their values over ten years - things like hedonism and security. Again, people reported more change in their values than they predicted. Recalled change was more modest in older participants, but again the difference in recall and prediction occurred at every decade. However, this survey has the same problems as before - the issues of memory distortion and predicting bi-directional change - and in this case they went unaddressed.

Gathering yet more data, the researchers surveyed thousands of people about their preferences in the past compared with now, and asked others about their likely preferences in the future, compared with now - things like favoured holidays, taste in music and food. The idea of this study was to eliminate memory bias. Quoidbach et al reasoned that people have an accurate sense of their past preferences, although they didn't reference any data to support this claim. People recalling the past again appeared to have changed more than was anticipated by those looking ahead.

A final, far smaller study attempted to address the practical implications of our failure to anticipate how much we will change in the future. One hundred and seventy adults were split in two groups. One stated their current favourite band and said how much they'd pay to see them in ten years. The other group reported their favourite band ten years ago and said how much they'd pay to see them today. There was a big difference - those looking ahead said they'd pay 61 per cent more to see their current favourite band, as compared with the price the retrospective group said they'd pay to see their former favourite band today. "Participants substantially overpaid for a further opportunity to indulge a current preference," the researchers said.

But was this a fair comparison? The retrospective group know things about their former favourite band that the future group couldn't possibly know about their current favourite band. For example, perhaps members of the retrospective group didn't like their chosen band's follow-up albums, perhaps they already saw them in concert many times over the last ten years. Maybe the future group were optimistic about the future creative output of their current favourite band. In short, there are so many other factors at play here, besides participants' beliefs about the stability of their own preferences.

Convinced by their own demonstrations of the End of History Illusion, Quoidbach et al speculated about the possible causes. One explanation, they suggested, is that the effect is a manifestation of our gilded view of ourselves: "most people believe that their personalities are attractive ....," the researchers wrote, "having reached that exalted state, they may be reluctant to entertain the possibility of change." But this explanation seems to ignore the self-doubt and pessimism that blights many people's lives. Quoidbach's other proposal is that the End of History Illusion is a manifestation of the fluency heuristic - because it's tricky to imagine change in the future, we infer that change is unlikely.

These speculations are premature. It would be easier to believe in the End of History of Illusion if there was data on actual change, rather than a reliance on participants' memories of themselves in the past. Even if the effect is real, it's also not clear if this is a general bias about the future that extends beyond our beliefs about ourselves. What predictions would we make about the future change of other people? Or about human culture in general? Here's one thing that surely won't change - slick psychology papers with eye-catching titles getting lots of attention.


Quoidbach J, Gilbert DT, and Wilson TD (2013). The end of history illusion. Science (New York, N.Y.), 339 (6115), 96-8 PMID: 23288539 

[thanks Thom Baguley for help with understanding some of the methodological issues]

Post written by Christian Jarrett (@psych_writer) for the BPS Research Digest.


Nora Miller said...

Enjoyed the post, Christian! I agree with your concerns about both the perceived outcomes and the possible reasons for the differences. Quick personal story: in my 30s I moved to a different part of the country and started my first "real job". I bought what I considered professional clothing, like suits, and found myself choosing quite muted colors, like gray, brown, and navy. These colors also took over my choices for weekend clothes. Fast forward 30 years: I moved again to Arizona, in SW USA. With the Spanish influence here, you see a lot more color around, in tiles, paint, pottery, signage, and clothing, and that has influenced me to expand my personal color choices. I felt daring in doing so, like breaking out of my shell. At some point, I looked in my closet and thought how *crazy* my former self would find the color palette. So imagine my shock when I found a picture of myself from that era in a red sweater, and other in a red plaid shirt! The point is, I clearly misremembered my own preferences from 30 years ago. If you don't accurately remember what you liked, how can you evaluate how much you have "changed" or how much you might continue to change in the future? I don't see how this study could ever justify conclusions without actual evidence of change or its lack.

Christian Jarrett said...

hi Nora, thanks for kind feedback and a great example of the tricks of autobiographical memory.

Anonymous said...

I read part of the Science article and came away thinking the less of Science. I have several issues with the experiment. First, the sample which was taken from a web site of a popular TV Show. I'm pretty sure that's faulty sampling. Better, I guess, than getting ones students as the subjects, but still . . .
Also why the fancy way of breaking up the sample into 10 year groups according to age and comparing the look-back and look-forward group? The people in the two groups are different yet their answers compared as if they were from the same group. Why not simply ask each person in the sample both sides of the question and compare? What is happening at the particular time in the world might and probably would influence the different age groups differently and not have anything to do with personality changes.

Anonymous said...

Thanks for a well thought out analysis. I was disappointed with Science for publishing this and am gald to see I am not alone. A Science Prof

bianca said...

I'm so happy to see this review of the "End of History Illusion" article. Even though I don't disagree with the concept, I feel like the conclusions were overgeneralized and the study a little questionable. I don't know, but I strongly suspect the subjects were found on the American Idol website. It has the large audience numbers, a broad age-demographic, and of the top-rated TV series it is the most-likely to be heavily female-skewed (the subjects were 80% female according to the article) among the online community. It also fits with the music-related questions. I think it's fair to say that that show's audience doesn't have the best judgment when it comes to the long term.

Post a Comment

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