Patient decisions sometimes display biases or irrationalities.  Select two of the following decision biases: hindsight bias, framing bias, loss aversion, appeal of zero risk. For each of the two biases: (a) explain what the bias is, including an example [5 pts.], (b) explain why it’s a bias (that is, why it violates a normative principle) [3 pts.], and (c) describe a descriptive/psychological theory that could explain why this bias occurs.  [2 pts.]
 

Hindsight bias (a) is an error in probability judgment such that after knowing the actual outcome in a situation, the person says that she would have assigned a very high likelihood to that event even if she had not known it was the actual outcome.  In contrast, someone who does not know the actual outcome tends to believe that the likelihood of that outcome occurring is lower.  For example, supposed Dr. H reads a patient’s medical chart, including the definitive diagnosis that the patient is eventually given.  Dr. H. says, in essence, “Oh, I could have predicted that this patient had diagnosis X.  Even if I didn’t know that was the actual diagnosis, I would have given it an 80% probability of being the case.”  In contrast, Dr. F reads the same patient chart without being shown the definitive diagnosis.  When asked what she thinks the likelihood is that the patient has diagnosis X, Dr. F says about 40%. The difference between these two estimates is hindsight bias.  Hindsight is 20-20.
 (b) Hindsight bias is a bias because it represents a failure for the person judging in hindsight to really answer the question asked: if you didn’t know that X was the actual outcome, how likely would have thought it was that X would occur?  If the person really put aside her outcome knowledge, she would give an answer the same as that given by the person who doesn’t know the actual outcome.  (c) The fact that the person judging in hindsight can’t put aside her outcome knowledge provides an indication as to the psychological theory that can explain why this bias occurs. Once someone knows the outcome, it seems obvious and predictable.  All the features of the patient case, in my example, that are consistent with diagnosis X suddenly seem to jump out as the most obvious, salient features of the case.  Features of the case that are consistent with alternative diagnoses, in contrast, are not nearly as salient.

Framing bias (a) is the tendency for people to express systematically different preferences for two choice problems that present equivalent information but with different wordings.  The most famous example is Kahneman and Tversky’s Asian disease problem, which poses a situation where 600 people are expected to die from a new disease outbreak.  The two choices posed below are equivalent to one another, but one is framed in terms of survival and the other in terms of mortality. Note, for example, that saving 200 out of 600 people (A) is the same as 400 of the 600 people dying (C).
Survival Frame Mortality Frame
A. save 200 people for sure
B. 1/3 probability that all 600 people will be saved and 2/3 probability that no one will be saved. C. 400 people die for sure
D. 2/3 probability that all 600 people die and 1/3 probability that no one dies.

 (b) Framing is a bias because it violates the normative principle that decisions should be based on the outcomes of the choice options, not how those outcomes are described.  Since saving 200 people really is the same as 400 people dying in this scenario, those two options should be viewed identically (same for B and D).  If A is better than B, then it can’t be the case that D is better than C, since A=C and B=D. (c) The psychological theory that explains framing entails that people evaluate outcomes as gains and losses relative to a reference point, rather than as objective outcomes.  So, the survival frame entails comparing the outcomes to a reference point of all 600 people already dead – that is, outcomes are framed as gains (bringing people back from brink of death).  The mortality frame entails comparing the outcomes to a reference point of nobody dead yet – that is, the outcomes are framed as losses (pushing people over the brink of death).  Decision makers treat gains and losses differently, being risk averse for gains and risk seeking for losses.  It’s not necessary to mention this, but the reason for this difference is the psychophysical function of how lives (or other outcomes) are valued.  Saving 600 lives is not 3 times as good as saving 200 lives, in psychological terms.  And letting 600 people die is not 1.5 times as bad as letting 400 people die.

Loss aversion (a) is the phenomenon that losses are weighted more heavily than objectively equivalent gains.  For example, gaining $5 is mildly nice, but losing $5 is catastrophically terrible.  An example that illustrates loss aversion is a study by Thaler in which students were asked to place a monetary equivalent on a .001 chance of immediate death.  The question was posed on two ways, one as a gain and the other as a loss.  In the gain version students were asked to imagine that they had already been exposed to a virus that gave them a .001 chance of immediate death. They were asked how much they would pay for the cure.  In the loss version, students were asked how much they would need to be paid to expose themselves (for a research study) to the same virus and the associated .001 chance of immediate death (with no availability of the cure).  Student would pay only about $200 for the cure but demanded sums such as $10,000 to be exposed.
 (b) Why is this a bias? In essence, it is a framing effect.  Both questions ask what monetary amount would offset the difference between 0.000 and 0.001 chance of immediate death.  The only difference between the two questions is that the gain question asks for the monetary equivalent of moving from .001 to 0 chance of death whereas the loss question asks for the monetary equivalent of moving from 0 to .001 chance of death.  That difference is normatively irrelevant because it is based on one’s current reference point.  A .001 chance of death can’t be better simply because that is the risk one currently has, as opposed to the risk one could take on. (c) The psychological account of loss aversion is that, like framing, people tend to evaluate outcomes as gains or losses relative to a reference point.  The two questions in Thaler’s study differ in their reference points.  Whereas framing illustrates that people have different risk preferences for gains vs. losses, however, loss aversion illustrates that the (dis)utility of losses is “scaled up” relative to gains.

Appeal of zero risk (a) is the phenomenon that certainty (or absence of risk) has special status.  An example comes from the study on pricing pesticides.  If a can of pesticides costs $10.00 and causes a toxic reaction (of specified severity) in 15 out of 10,000 uses, how much extra would you play for a can that causes a toxic reaction in only 10 out of 10,000 uses?  Subject said they would pay about $1 for the lower risk. If the question, in contrast, was how much extra they would pay to reduce the risk from 5 in 10,000 to 0 in 10,000 then subjects would pay more than twice that.  Reducing the risk down to 0 has special appeal such that decision makers are willing to pay a premium for that.
 (b) Giving a special status to certainty violates the normative theory of probability theory, which says that probability is a linear scale.  This means that the difference between 15 vs. 10 in 10,000 risk is the same absolute difference as that between 5 vs. 0 in 10,000 risk.  Since the toxic reaction in question is the same throughout, that means that both reductions give the same benefit and should be valued equivalently.  Another way to argue for the non-normativeness of subjects’ response pattern is to say that zero risk is an unattainable illusion.  Although the risk from this can of pesticides might go to 0, the risk from all sources in life will never go to 0.  What is important is reducing total risk as much as possible for a given level of cost/effort. (c) The psychological account of this phenomenon is that decision makers do not weight probabilities linearly.  That is, although the difference between 15 vs. 10 in 10,000 risk is objectively the same as that between 5 vs. 0 in 10,000 risk, people don’t treat them the same.  Certainty or zero risk gets a special status.  I know this “explanation” sounds like re-stating the phenomenon.  The explanation has to do with the psychophysics of how people treat probabilities.  Just as saving 600 lives is objectively, but not psychologically, 3 times as good as saving 200 lives, so the difference between, say 45% and 40% is objectively, but not psychologically, the same as the difference between 5% and 0%.
 

For each of these two health behaviors—sexual behavior and smoking—briefly describe (i) one health consequence of the behavior [3 pts] and (ii) two psychosocial factors that contribute to the behavior [4 pts.].  Include evidence for at least one of your psychosocial contributing factors for each behavior  (e.g., briefly summarize a study that backs up your point) [3 pts].

Sexual behavior.  (a) The health consequences you could list include pregnancy and sexually-transmitted infections (STIs).  Pregnancy is either a good or a bad outcome, depending on one’s goals, and STIs are always a bad outcome. The likelihood of both consequences decreases with the use of contraception (depending on the type of contraception).
 (b & c) There are a number of psychosocial contributing factors that you could discuss.  Following is a list of the major possibilities.  You only need to present two, and you need only present any level of detailed evidence for one.  You might discuss psychosocial factors that influence sexual activity (e.g., when, what kind) and/or factors that influence use of contraception.
 (i) Age at first intercourse is related to use of contraception at first intercourse, as illustrated in the data presented in the Ogden chapter.  Although it’s questionable whether age is itself a psychosocial variable, this results pattern suggests what type of psychosocial variables might be involved.  Specifically, the data suggest some type of developmental account such that teenagers who have reached a certain developmental state (e.g., capable of planning for sex) are more likely to use contraception.
 (ii) Visceral factors are related to forcefulness of sexual interaction. Loewenstein et al conducted a study to test the idea that current sexual arousal (a visceral drive state) would make it more likely that men would try to coax a reluctant date into having sex.  Heterosexual college men were randomly assigned to one of three conditions. Those in the unaroused group viewed neutral photos.  Those in the previously aroused group viewed sexually arousing photos on the previous day.  Those in the currently aroused group view sexually arousing photos on the study day.  All Ss were then presented with a scenario about a date that progressed toward sexual intimacy at which point the woman asks the man to stop.  Ss were asked to rate the likelihood that they would try to coax the woman into having sex.  Those in the currently aroused group gave higher ratings than those in either of the other two groups.  These results suggest that plans or intentions formed when not aroused do not coincide with actions taken when aroused, and that once one is out of the visceral drive state, it is difficult to remember what it was like in that state or why one acted as one did when in that state.
 (iii) Anticipated regret influences condom use. Regret is the emotion felt when comparing the outcome that resulted from one’s decision to a better outcome that would have resulted had one chosen differently.  Anticipated regret is the anticipation of that emotion even before one makes the initial choice.  Avoidance of later regret is one factor that influences the decision made.  Richard and van der Pligt tested the idea that anticipated regret (and other anticipated emotions) would increase the use of condoms.  College students were presented with a scenario in which they met someone new and decided to have sex.  They were asked either to rate what their feelings would be after using a condom or rate what their feelings would be about using a condom.  Both types of ratings compared condoms to alternative forms of contraception (e.g., feelings about condoms vs. feelings about another contraception).  The “feelings after” condition was designed to trigger consideration of emotions that would be experienced after one’s decision had been made and acted on (i.e., maximize anticipated regret).  Those in the “feelings after” group had more positive ratings of condoms than did those in the “feelings about” group. More importantly, those in the “feelings after” group were more likely to use condoms consistently during the 5 months after the intervention than were those in the ‘feelings about” group.  This latter effect held true only for men (likely because the women started out with such a high level of condom use).  This study suggests that anticipated emotional consequences (rather than just cold, cognitive costs and benefits) are an important determinant of condom use.

Smoking (a).  The health consequences of smoking you might list include death from cardiovascular disease (RR = about 2), lung cancer (RR = about 9), other forms of cancer (head & neck, kidney, bladder), or COPD (chronic obstructive pulmonary disease, which includes chronic bronchitis, asthma, and emphysema).  You could also say that smoking increases all-cause mortality at every age (although of course, everyone eventually dies of something).
 (b & c) Again, there are a number of psychosocial contributing factors that you could discuss.  Following is a list of the major possibilities.  You only need to present two, and you need only present any level of detailed evidence for one.  You might discuss psychosocial factors that influence who starts to smoke, and when, who continues to smoke and why, and who quits smoking successfully.
 (i) Sex and ethnicity influence age of initiation smoking.  Robinson & Klesges (1997) demonstrated that among middle-school children, boys are more likely to smoke than girls and European-American children are more likely to smoke than African-American students.  Although one might argue about whether sex and ethnicity are psychosocial factors, these demographic variables were also associated with having family members and friends who smoked, perceived easy access to cigarettes, perceived instrumental value of smoking (e.g., looking cool), rebelliousness, and lower social support.  Each of these differences was in the direction such that boys and/or European Americans were at higher risk for later smoking.
 (ii) Inaccurate beliefs about smoking are associated with initiation of smoking.  Leventhal et al., (1987) examined beliefs among grade school – high school students.  A number of inaccurate beliefs were associated both with current smoking status and with likely future smoking status (e.g., differentiated between kids with and without friends or family who smoked).  Smokers were more likely than non-smokers to think they personally were less susceptible than average to the effect of smoking.  Smokers were more likely than non-smokers to overestimate the percentage of adults and peers who smoked.  Smokers were more likely than non-smokers to underestimate the percentage of adults and peers who disapproved of smoking. Smokers were less likely than non-smokers to understand the severity of addiction.  All of these misconceptions are psychosocial factors that potentially contribute to the initiation of smoking. (The study is correlational, so doesn’t show a causal link).
 (iii) Reasons to keep smoking.  Leventhal and Avis (1976) documented 3 categories of smokers.  Pleasure/taste smokers smoke for the pleasure of the smoke and can be dissuaded from smoking if the cigarette tastes bad.  Habitual smokers smoke as an automatic behavior pattern (much like biting nails) and will reduce smoking if forced to note each time they light up.  Addicted smokers smoke to get the nicotine and will smoke sufficient amounts to get the level of nicotine they are used to.  For example, if given low-nicotine cigarettes they will take more puffs from each cigarette and smoke more cigarettes.  Each of these 3 is a psychosocial factor that contributes to the maintenance of smoking in (at least some) current smokers.
 (iv). Factors that influence quitting.  We didn’t discuss much detailed evidence here, but psychosocial factors you might list that influence quitting include intervention from physician, intensive behavioral support, multi-modal therapy, self-efficacy, social support, or personal knowledge of someone with a smoking-related illness.  This topic would not be a good choice for your “detailed evidence” factor, but would be fine for the “no detailed evidence” factor.