Risky decisions that affect subjective well-being: validity and distortions

College Student Journal, Dec, 2002 by Ron Hosen, Dina Solovey-Hosen

Risky decisions are important in maintaining subjective well-being if they are based on an unbiased sampling of available information. Biased selection of information tends to produce non-optimal outcomes. Errors fall in the categories of (1) errors of convenience and (2) hedonic rigging. Distortions can also result from the employment of dysfunctional mental accounts. The latter may be more influenced by culture than the errors of convenience and of hedonic rigging.

**********

Risky decisions by individuals involve the construction of mental models from representations of recurring events stored in memory. Whether one has detected a recurring stimulus depends on the classification criteria. The process may be intuitive or it may lend itself to formal description. The result is a cue detector which, when activated, signals the onset of an event and generates a mental prediction. If a valid cue-detector is possible, measurable uncertainty or riskiness exists.

The internal model is based on fragmentary information with the gaps filled by mental processes. A stimulus is also fragmentary, a sample from a universe of potential inputs. The internal model is valid to the extent that it predicts the target event with greater than chance accuracy. It is the result of a sampling process.

Stimulus Sampling

A decision process implies sampling. A sample will always misrepresent the population from which it is drawn. Potential error can be reduced by increasing sample size if the sample selection process ensures that each population member has an equal chance of being selected. If the sampler genuinely wants information about the population, she will employ a sampling process that does not favor any distinctive population subset. The problem of sample size is a budget problem; time, energy and other resources are limited. It is relevant to all decisions under conditions of uncertainty and often is the reason for uncertainty.

Sampling bias is a different source of error. These are biases of judgment impounded in the decision process. They include (1) errors of convenience and (2) errors of hedonic rigging. They are actually cost-benefit calculation errors having different roots. They tend to influence actions by influencing probability estimations and by distorting evaluations of expected outcomes that ultimately affect action choices.

Proactive Risk-Taking Is Functional

Hosen, Stern & Solovey-Hosen (in press) have defined proactive risk-taking in terms of the following criteria: (1) Non-trivial stakes; (2) actor acknowledges the possibility of loss; (3) actor chooses the goal and the means of achieving it; (4) the definition of success and the relevant time frame is specified in advance; (5) actor has some control over the outcome; (6) the goal must be a gain rather than loss avoidance (loss avoidance can be part of the process, however); (7) the risky activity must provide feedback and the possibility of learning and improvement. Proactive risk-taking is clearly very important in achieving sustainable happiness in modern life and both sampling biases and distorted outcome evaluations can undermine Success.

Errors of Convenience

These are the decision heuristics seemingly built into routine cognitive functioning; once adaptive, now merely easy or mindless. These are cognitive shortcuts that arise spontaneously, consistent with a penchant for economical solutions. They include (1) a propensity to attribute causality when two events occur coincidentally in sequence. People seem to prefer to assert causality rather than independence (unrelatedness) when no compelling evidence is present for either; 2) evaluating a stimulus as a member of a class because it resembles the evaluator's stereotype of the class (representativeness heuristic); 3) judging the frequency of an event by the ease with which it can be recalled (availability heuristic); 4) judging a stimulus event against a reference point that differs arbitrarily depending on the sequencing, even if the events are otherwise equivalent; 5) under-correcting a judgment when new or additional information is obtained (Tversky and Kahneman, 1981; Tversky and Griffin, 1991).

Example: A particular motion picture would be judged as "good" if it followed a "very bad" movie but only "fairly good" if it followed a "great" movie.

Example: Someone predicts a rising stock market in 2000 citing rising markets in the last 12 presidential election years, a clear pattern in the predictor's view. In fact, most years of any kind have had rising markets. The political power and the psychological symbolism of the presidency have changed; elections and parties have changed. The volume of trading, speed of trading, and available information have increased by a factor of 200 since 1948; costs have dramatically decreased. The predictor derives meaningless correlations with no explanation of why the (probably spurious) relationships exist. The prediction treats 12 samples of one election each as one sample of 12 elections. A computer programmed to detect correlations is likely to do so by picking up temporary chance deviations from randomness.