6. Bias in Thinking and Decision Making
Key concepts:
Heuristic: is a mental shortcut that focus on one aspect of a complex problem and ignoring others.
Heuristics can result in patterns of thinking and decision making which are consistent, but inaccurate.
Link to dual process model of system 1 thinking with heuristic.
These patterns of thought are described as cognitive biases.
Cognitive bias: are systematic errors in thinking that affects the decisions and judgments that people make.
Anchoring bias: Tendency to rely too heavily on an initial piece of information offered (the "anchor") when making decisions. (Emphasize the presence of anchor in the study)
Illusory correlation – cognitive bias
Study
Tvsesky
· Aim: To investigate the influence of anchoring bias on decision-making.
· Design:
o sort participant evenly to two conditions.
o First condition will show participant an ascending multiplication from number 1 to 8.
o Second condition will show participant a decreasing multiplication from number 8 to 1.
o They have only 5 seconds to estimate the product of the problem.
· Result & Conclusion:
o The median for those offered by the “ascending group” was 512, “descending group” was 2250. The actual product was 40320.
o One’s process of decision making is influenced by anchoring bias.
· Link-back:
o In the first group, the few numbers that appears to the participant are small, like 1, 2, and 3. On the other hand, the few numbers that appears to the participants in second group are large, such as 6, 7, 8. The first group get a smaller estimated product than the second group.
o This shows that people tend to rely on the first piece of information, therefor showing anchoring bias influence decision making.
· Evaluation:
o The study is simplistic and easy to replicate. This is a lab experiment, which means the procedure will be standardized and highly in detailed. The researcher can make a video or slideshow to present all the details and instructions. Makes replicable and makes the study reliable.
o The study uses independent sample design which fails to control participant variability. People assigning to different groups may have different experience in math education. Some group did faster and makes result inaccurate. To eliminate this confounding variable, researcher could use matched pair design.
Strack
· Aim: to see how the anchoring effect could influence guesses of Mahatma Gandhi age when he died.
· Design:
o 60 male and female German University students were divided into two groups and were asked two questions.
o Condition 1: if Gandhi was older or younger than 9 years old when he died. Condition2: if he was older or younger than 140 years old.
o They were then asked how old they think he was when he died.
· Result & Conclusion:
o High anchor, average 67. Low anchor, average 50.
o Anchoring bias did influence the guess of Gandhi’s age when he died.
· Link-back:
o This study demonstrates the anchoring effect – numerical information given to participants before making a judgment can influence their later judgements.
o To be specific, the first question provided for participants contains a numerical information that infer the approximate age of Gandhi.
o By comparing the two conditions, we can see that the information did affects one’s decision making.
· Evaluation:
o The researcher well controlled all the variables in the experiment to infer a clear causation between the anchor in question and their estimation of Gandhi’s age. The experiment is carried out in a lab condition where environmental factors are eliminated. The question is all the same, except for the anchor presented in the first question is different. This makes the effect of anchor in the first question fully presented.
o The study uses independent measure design, which means participant variability exists. Some participants may know the actual death age of Gandhi before the experiment is conducted. This will dramatically screw the result of this experiment. To improve, researcher should ask if participant know the actual death age of Gandhi after the experiment. Researcher could also use random allocation to reduce participant variability.