Predictably Misbehaving

Lecture 6: Projection Bias

Joshua Foster at UW-Oshkosh

Predicting the Future

People have a tough time imagining a future different from today.

Why are people wrong this way?

Our imaginations are imperfect models of reality.

If we imagine what something/someone could be like:

• We typically see a sketchy image of it in our heads.
• Our brain is activated in similar ways as when we are actually looking at it/them.

Imagining the future is a similar process. Example:

• "How would you feel if you found 100 bucks?"
• I bet you visually imagined what would happen
• And that image shaped your response.

Our ability to do this is limited, and affected by our current state.

• When disgusted, it's hard to imagine affection.
• When satiated, it's hard to imagine hunger.

What is Projection Bias?

Preferences Can Fluctuate Over Time

There are many ways preferences change over time. Examples:

1. Physical: hunger, addiction, aging, etc.
2. Experience: adventure, etc.
3. Environment: friends, weather, etc.

When making decisions today:

• It is important to understand changes in future preferences.
• Making summer plans in the winter.
• Choosing a major and career.
• Having children.

Traditional economic perspective:

• People are very good at predicting their future preferences.

Mis-prediction of Preferences

Studies in psychology suggest people exhibit a systematic bias.

Projection bias isn't just mis-prediction.

• It is a mis-prediction with a systematic direction.
• People don't believe their preferences are transient.
• People aren't able to predict all meaningful factors.

We will examine some evidence and discuss its consequences.

Evidence of Projection Bias

Grocery Shopping

Hunger

We consider hunger, because:

• It's perhaps the clearest evidence of projection bias.
• People have lots of experience with being hungry.

Meaning: mis-predictions aren't due to a lack of experience.

Studies show those shopping on an empty stomach buy more.

Consider the following simple experiment:

• Randomly give a muffin to individuals entering a supermarket.
• Monitor how much food they buy.

Gilbert et al. (2002) in Org. Behavior and Human Decision Processes

Exit surveys showed:

• Those who did not eat muffins:
• 51% purchased unplanned food items.
• Those who did eat muffins:
• 34% purchased unplanned food items.

But this study isn't perfect. It could be:

• Shoppers may be buying food for later that day.
• So those who had a muffin buy less.

Read and van Leeuwen (1998) in Org. Beh. and Human Dec. Proc.

• Office workers were asked: "Would you like a healthy or unhealthy snack in one week?"
• Sometimes asked when hungry (late afternoon)
• Sometimes asked when satiated (after lunch)
• They knew snacks would be given either:
• When they were hungry (late afternoon), or
• When satiated (immediately after lunch)

Necessary argument:

• Those hungry will be tempted by unhealthy food.
• Removing hunger allows focus on 'smart' decisions.

Thirst

Van Boven and Loewenstein (2003) in Personality & Soc. Psy. Bul.

Experiment procedures:

• Ask visitors at a campus gym to fill out a survey.
• Two treatments:
• Not thirsty (before exercising)
• Thirsty (after exercising)

Hypothetical question about hikers who packed light:

• "Which do you think the hikers regretted not packing more: water or food?"
• If not thirsty: 52% regret not packing water.
• If thirsty: 92% regret not packing water.

Those who are thirsty are more likely to think hikers should also be.

Drugs

Giordano et al. (2002) in Psychopharmacology

• Study on addiction using actual drug addicts.
• Addicts to receive a dose of heroin substitute (BUP)
• Some could receive it now.
• Some had to wait 5 days.

Asked for their monetary valuation for an extra dose:

• 10, 20, 30, etc.
• Some were asked right before a dose of BUP.
• Some were asked right after a dose of BUP.

Addicts valued the extra dose more if they were deprived.

They under-appreciate their future cravings when satiated.

Low- and High-brow Movies

Loewenstein (1999):

• Choose a movie today: 56% choose low-brow.
• Choose a movie for next week: 30% low-brow.

Low- and High-brow Movies

Goldstein and Goldstein (2006) on Netflix customers:

• Watch and return low-brow movies right away.
• Let high-brow movies sit around much longer.

Professors and Tenure

Gilbert, et al. (1998) in J. of Personality and Soc. Psy.

Immune Neglect

The tendency to underestimate ones own powers of adaptation to unfavorable events.

Study procedures:

• Current assistant professors at U of Texas asked:
• How would you respond to not getting tenure?
• Predicted significant unhappiness.
• (This is an event five years in the future.)

Responses from former U of Texas professors:

• Well-being returned to current state.
• Did not take as long as predicted.

Accurate in predicting short-run impact of decision.

• But exaggerated the long-term impact.

HIV Test Results

Sieff, et al. (1999) in American Journal of Psychology

• Surveyed those entering a clinic for a HIV test.
• Predict happiness in 5 weeks after getting:
• A test indicating HIV-positive
• A test indicating HIV-negative
• Five weeks after result, filled out the same survey.

People overestimated both:

• How good they would feel after a favorable result.
• How bad they would feel after an unfavorable result.

Are We Getting it Wrong?

Naive Discounting Vs. Projection Bias

Naive hyperbolic discounting is a mis-prediction:

• You over-estimate your self-control in the future.

Projection bias is a state-dependent mis-prediction.

• You're more likely to predict future temptation to over-eat when hungry.
• You're more likely to predict future smoking when you haven't had a cigarette for a while.

Sometimes, they're hard to distinguish in the real world.

• Though, theoretically it is straight-forward to see the difference.