INTRODUCTION Individuals who examine an event after learning the outcome tend to exaggerate their ability to predict the result prior to having the outcome revealed. This psychological tendency is known as hindsight bias, or the "I-knew-it-all-along" phenomenon (Wood, 1978). It is measured by comparing estimates made by participants after the answers to knowledge problems or the outcome of an event is given, with estimates taken prior to the test or event. Participants` recollections of their original estimates tend to be systematically biased toward the actual solution to a problem (Fischoff, 1975; Wood, 1978). The purpose of this study was to examine the effects of time delay on the amount of hindsight bias displayed by participants. This robust phenomenon has been explored in an applied fashion in a variety of settings including sports events (Roese & Maniar, 1997), tests of difficault knowledge (Pohl & Hell, 1996), perceptions of technological disaster (Brown, Williams, and Less-Haley, 1994), and psychotherapy case histories (Fischoff, 1975). Despite extensive research, no study has focused on the relationship between time and hindsight bias. The reduction and elimination of hindsight bias has been explored in many experimental settings. Arkes, Faust, Guilmete, and Hart (1988) examined the elimination of the bias phenomenon among 194 neuropsychologists, employing the factor of reasoning assessment in their work. In the study, participants were divided in two groups. Those in one group were asked to give one reason why each of three possible diagnoses might be correct while the other participants were not required to list reasons. Arkes et. al found that the frequency of participants succumbing to hindsight bias was lower in the group asked to give explanations for the diagnoses than in the no-reasons group. The results suggested that although the bias phenomenon could not be eliminated, it could be reduced by employing a reasoning measure among participants. In a conceptually diffrent study, Pohl et. al (1996) attempted to reduce the effects of hindsight bias among German college students by informing one group of participants in advance about the experimental design and explaining in detail the bias effect. The experimenters found that informing participants about hindsight bias prior to testing had no effect on reducing the amount of bias displayed. Similar to Pohl et al., Wood (1978) attempted to eliminate the bias phenomenon by telling participants exactly what hindsight bias was and explicitly instructing them to prevent its occurence. The detailed instruction had no sigificant effect as the participants demonstrated the bias tendency despite the persistent attempts at elimination. While past research has used a variety of subjects and settings in an attempt to reduce or eliminate the phenomenon in an experimental contaxt, none have focused on the relationship between time and the reduction or amplification of hindsight bias. Although previous research on the phenomenon has been methodologically sound, the negelction of research on the time variable has left an integral piece of the puzzle missing. Does the amount of time between participants` original answers to general knowledge questions and their recall estimates affect the degree to which recall of their original estimates approach the actual solution? This is the central question the study attempted to answer. Exploring this question and others regarding the effects of time on hindsight bias may yield a broader understanding of the social and psychological factors affecting the tendancy of individuals to exaggerate their ability to have predicted how something turned out. This knowledge can possibly be applied in a positive fashion in order to reduce the faulty thinking of people and perhaps lead them to better decision making. The hindsight bias effect is extremely prevalent in everyday life. The phenomenon has a tendancy to make people overconfident to the point of arrogance, which can often lead to ignorant decision making and disregard for the negative societal consequences of one`s actions. As a result, research on the effect is significant not only to the scientific community but to society as well. The study of hindsight bias is applicable to important political decisions (Leary, 1982), the medical treatment of patients, and academic performance. As shown by Wood (1978) and Pohl et. al (1996), event those who are aware of the hindsight bias phenomenon are not immune to it. Therefore, it is important that this psychological tendency be researched further in an attempt to reduce poor decision making in the future. The primary independant variable in this experiment was time. Specifically, the amount of time elapsed between participants` original answers to general knowledge questions and their attempt at recalling those estimates immediatley after presentation of the correct responses. The three levels of the independant variable included a short time delay (SET) of one day, a medium duration (MET) of one week, and a long delay (LET) of three weeks. The dependent variable was the amount of hindsight bias displayed by participants in the three experimental groups as measured by the "drift" of participants` recall estimates toward or away from the actual solutions. The hypotheses were based on the concept of anchoring and adjustment, researched by Czaczkes and Ganzach (1996) and Chapman and Johnson (1994). Anchoring and adjustment is a persuasive bias in which decision makers are influenced by random or uninformative numbers or starting points (Chapman & Johnson, 1994). It was hypothesized that participants who experienced the greater time delay (LET) would rate their original answers as closer to the correct answers than participants with the shorter time delays (MET and SET), thus showing the greatest amount of hindsight bias of all three groups. Following this prediction, the group consisting of those who were required to wait one week before recalling their original answers (MET) would tend to systematically bias their recollections more than the short-elapsed time group of one day (SET). Based on the work of Czackes et al. (1996) and Chapman et al. (1994), it was hypothesized that increased time delay will lead to increased forgetting of the participants` original answers and thus participants who experienced the greater time delay would be forced to use the actual solution as a cognitive anchor and would subesquently adjust their recall estimates toward this benchmark. Participants exposed to shorter time delays would not bias their answers as much because they would have a more accurate memory of their original answers and would not require the use of the correct answer as an anchor to such a degree.
METHOD PARTICIPANTS
The participants in this study were 66 Carroll College undergraduates (53 women and 13 men, mean age = 18.6 years) enrolled in an Introductory Psychology course. Three seperate classes formed the three seperate groups: Group 1 consisted of 27 participants (21 women and 6 men, mean age = 18.4 years), Group 2 consisted of 20 participants (18 women and 2 men, mean age = 18.7 years), and Group 3 was composed of 19 participants (14 women and 5 men, mean age = 18.7 years). The high percentage of female participants is reflective of the Carroll student population as a whole. All participants recieved course credit. MATERIALS
A general knowledge exam was used in this experiment and consisted of 20 difficult knowledge questions requiring numerical responses (see Appendix A). Similar to Pohl et al. (1996), the questions were selected on the basis that participants had been exposed to the answer in the past but probably did not know the exact solutions to the majority of questions. All questions were presented in the same order during both phases of the experiment. PROCEDURE
In the first phase of the study, participants were groups in a classroom setting. After the experiment was explained and informed consent gained, participants were given instructions on the first phase of the experiment. This phase consisted of completing the 20-question general knowledge exam. Participants were instructed to work individually and as accurately as possible and to complete all the questions. No time constraints were placed on the completion of the exam. The average amount of time of completion was approximately five minutes, with duration ranging from three to eight minutes. Upon finishing, the experimenter instructed the participants that he would be returning on a future date for the second phase of the experiment. All groups received the same instructions and were exposed to the first phase of the experiment on the same day. Phase 2 of the experiment required participants to complete a recall exam that was identical to the general knowledge exam with the exception of different instructions and the inclusion of the correct answer (within parentheses and highlighted by bold type) following each question. Participants were instructed to recall, to the best of their ability, their original answers to each of the 20 general knowledge questions presented in Phase 1. All were instructed to recall answers only to questions answered on the original test. They were provided with as much time as needed to complete the recall exam. The average amount of time taken on the recall exam was four minutes, with duration ranging from three to seven minutes. Participants in Group 1 (SET) were exposed to the second phase of the experiment one day after Phase 1, while Group 2 (MET) completed the second phase one week after Phase 1 and Group 3 (LET) three weeks after Phase 1. Participants in Groups 1 and 2 received a general debreifing session following the completion of their participation in the experiment, while Group 3 was completely debreifed and given an informational sheet by the experimenter covering the purpose of the study at the conclusion of their three-week treatment condition (see Appendix C). At this time Groups 1 and 2 also recieved the complete debreifing form. DESIGN
A quasi-experimental design was used since random assignment in determining the specific treatment groups was not employed. The independant variable (time elpased between participants` original answers and recall estimates) was varied between participants. Group 1 (SET) received a time delay of one day, Group 2 (MET) a time delay of one week, and Group 3 (LET) a time delay of three weeks. ANALYSIS
Using the work of Pohl et al. (1996) as a model, the data of each participant was converted into standardized scores of hindsight bias. This was done by first using the solution and both the original and recalled estimates of all participants, seperately for each question, to transform each value into a z-standardized score. Both original and recalled responses with a z score greater than 3 were removed from the data set. Scores of zero on either the general knowledge or recall exams were included among the data set, but were eventually eliminated because they yielded z scores greater than 3. The remaining values were then calculated again in the same fashion. A difference score *z was computed using the formula *z = Ab.Value(z(E)-z(S)) - Ab.Value(z(R)-z(s)), where E, S, and R denote original estimate, actual solution, and recalled estimate, respectively. The diffrence score * z is a measure of the "drift" of the recalled estimate toward or away from the solution, expressed in units of standard deviation. A positive value shows evidence of hindsight bias, whereas a negative value would occur when the recalled estimate was further away from the correct answer than the original estimate. Diffrence scores were then averaged across participants and time delay conditions. The resulting means were then subjected to a one-way ANOVA, with the error probability set at <.05.
RESULTS The resulting *z scores of hindsight bias were computed using the method previously described across participants and then averaged for each time condition. Mean *z scores were -.053,-.061, and .158 for Groups SET, MET, and LET, respectively. Standard deviation for the group exposed to a one day time delay (SET) was .149, while data from the one week group (MET) yielded a value of .214. The standard deviation for the three week time delay group (LET) was .768. These values show a small reverse hindsight bias effect exhibited by Groups SET and MET. Group LET showed a more substantial bias effect as exhibited by the larger positive mean *z value. As shown in Figure 1, the descriptive data exemplify the small and relatively equal tendency for participants in Groups SET and MET to recall their original estimates as actually further away from the correct solution than they were, thus underestimating their original accuracy. Participants in Group LET, who experienced a substantially longer time delay between Phase 1 and Phase 2 of the experiment, overestimated their accuracy by recalling their original answers as closer to the correct answer than they actually were. The group *z means were then subjected to a one-way ANOVA with an alpha level of .05. As shown in Table 1, the differences between the three groups failed to reach statistical significance, F (2,62) = 1.61, p = .21. Results yield the conclusion that time delay did not have a significant effect on the degree to which hindsight bias was displayed by participants. From these findings, it can be concluded that the length of time between participants’ original answers and recall estimates did not significantly affect the amount of hindsight bias exhibited.
DISCUSSION Experimental results did not support the hypothesis that participants who experienced the greater time delay between Phase 1 and Phase 2 of the experiment would display more hindsight bias. There was no significant evidence that time delay amplified or reduced the amount of bias. In interpreting the results, it is notable that Groups SET and MET, who were exposed to time delays of one day and one week, respectively, both showed a reverse hindsight bias effect. With *z means for both groups -.053 and -.061, it is evident that these effects were very small and probably due to chance. Upon further examination of individual data and testing methods, no concrete explanations can be determined as to why a hindsight bias effect was not shown by participants in either condition. A scale attenuation effect may have occurred, but this is unlikely as exemplified by participants in Group LET, who exhibited the expected bias effect (M = .158). The majority of experimental studies on hindsight bias have demonstrated the robustness and persuasiveness of the phenomenon. Pohl et al. (1996) found that even repeated testing with knowledge of one’s personal bias did not result in any reduction of the observed memory distortion. In similar fashion, Wood (1978) failed in an attempt to eliminate the bias phenomenon by explaining in detail hindsight bias to participants prior to testing. According to Pohl et al. (1996), these findings provide strong support for the theories of “cognitive-reconstruction,” proposed by Hawkins and Hastie (1990). Following this theory, there are two types of reconstructive processes that may be responsible for the observed shift in recall. The first type of reconstructive process makes the assumption that participants may have forgotten their original answers, but have an accurate memory of the solution and whether their earlier answers were above or below it (Pohl & Hell 1996). This theory was based on the concept of anchoring and adjustment and served as the basis for the hypothesis of the present experiment. According to Pohl et al. (1996), recall estimates of this nature will typically fall between the original estimate and the solution, thus displaying hindsight bias. The other major theory of cognitive-reconstruction, termed by Pohl et al. (1996) as “solution-based rejudgment,” involves the independent memory decay of both original answers and solutions. According to the experimenters, in this case memory retrieval of either the original estimate or the solution becomes a judgment task for participants. “Rejudgments” of this type will normally be distributed around the solution (Pohl & Hell 1996). If an individual has a more accurate memory of the original answer, this should lead, according to Hawkins and Hastie (1990), to either more perfect recollections or to more cases of insuffiently adjusted reconstructions. Following this concept, a less accurate memory should lead to fewer perfect recollections and increased cases of solution-based rejudgments. Recollections yielded in this fashion do not result in significant findings of hindsight bias. One possible explanation for the unexpected experimental results is that participants in the study used a variation of solution-based rejudgment in their attempt to recall their original answers. This seems likely based on the low *z scores exhibited both towards and away from the solution. Although forgetting the solutions in the present study would have been impossible since they were included on the recall exam, many participants may have overlooked the solution and thus did not use it as a cognitive anchor to bias their recollections of their original estimates as hypothesized. Another possible explanation is that participants, especially those in Groups SET and MET, simply encoded their original responses better in memory than expected. This possibility is explored in the work of Pohl et al. (1996) as a way to reduce the amount of bias displayed in an experimental context. If participants remembered their original answers to the general knowledge questions more successfully, than it is logical that this cognitive strategy would lead not only to a larger number of perfect recollections but also to reconstructions that would exhibit less bias towards (or away from) the solution (Pohl & Hell 1996). The present study is not without flaws. A noteworthy limitation is the use of treatment groups that were not determined by random assignment. Although the use of a quasi-experimental design may have confounded results, there is no evidence that suggests that the groups differed prior to testing. Although participants were not matched on potentially relevant characteristics such as accuracy of short-term and long-term memory or Phase 1 exam score, the number of participants used in each time delay condition was significant enough such that experimental results can be deemed as valid. Other shortcomings involve the materials used to measure hindsight bias and the length of time delay manipulated in the procedure. As shown in Appendix A, the materials consisted of a 20 question difficult knowledge exam requiring numerical responses followed by a recall exam. The test questions were chosen so that participants may have had some previous knowledge of the solution but couldn’t answer more than approximately 10% accurately. In this regard the Phase 1 exam may have been deficient in that certain questions may have been too easy for the majority of participants. For example, 40% of participants answered correctly to the question “How old was Princess Diana when she was killed in a car crash in Paris, France?” and many more answered within three years of the solution (36). The accuracy of responding to the question was unexpected and abnormal in relation to overall exam results, which were significantly less accurate (10 - 15% correct overall). The levels of time delay experienced by participants in the experiment could have been extended so that a two week delay condition was included in the design. Also, a five week duration between Phase 1 and Phase 2 of the experiment may have precipitated increased forgetting among participants and possibly yielded increased hindsight bias. Further research dealing with the effect of time delay on hindsight bias is integral in understanding the phenomenon. Only by studying the conditions in which the bias tendency is amplified or reduced can a comprehensive picture possibly be gained. As a result of the phenomenon’s prevalence in society, it is important that future research focus on the contingencies that precipitate and perpetuate this cognitive error. Failure to learn from our past mistakes undoubtabley will lead us towards an unsuccessful and ignorant future. As articulated by Fischoff (1975) in his foundational work on hindsight bias, if we systematically underestimate the surprises which the past held and holds for us in hindsight, we are subjecting those hypotheses to inordinately weak tests and, presumably, finding little reason to change them. As a result, this perceived knowledge may actually hinder our comprehension of past errors and prevent us from making better decisions in the future. In conclusion, the comprehensive study of hindsight bias is critical to both science and society and deserving of future research attention.
REFERENCES Arkes, H.R., Faust, D., Guilmette, T.J., & Hart, K. (1988). Eliminating the hindsight bias. Journal of Applied Psychology, 73 (2), 305-307. Brown, R.S., Williams, C.W., Less-Haley, P.R. (1994). The effects of hindsight bias and causal attribution on human response to environmental events. Journal of Applied Social Psychology, 24 (8), 661-674. Chapman, G.B., & Johnson, E.J. (1994). The Limits of Anchoring. Journal of Behavioral Decision Making, 7, 223-242. Czaczkes, B., & Ganzach, Y. (1996). The natural selection of prediction heuristics: anchoring and adjustment versus representativeness. Journal of Behavioral Decision Making, 9 (2), 125-139. Fischoff, B. (1975). Hindsight * foresight: the effect of outcome knowledge on judgment under uncertainty. Journal of Experimental Psychology: Human Perception and Performance, 1 (3), 288-299. Hawkins, S.A., & Hastie, R. (1990). Hindsight: Biased judgments of past events after the outcomes are known. Psychological Bulletin, 107, 311-327. Leary, M.R. (1982). Hindsight distortion and the 1980 presidential election. Personality and Social Psychology Bulletin, 8 (2), 257-263. Pohl, F., & Hell, W. (1996). No reduction in hindsight bias after complete information and repeated testing. Organizational Behavior & Human Decision Processes, 67 (1), 49-58. Roese, N.J., & Maniar, S.D. (1997). Perceptions of purple: counterfactual and hindsight judgments at Northwestern Wildcats football games. Personality & Social Psychology Bulletin, 23 (12), 1245-1253. Wood, G. (1978). The I-knew-it-all-along effect. Journal of Experimental Psychology: Human Perception and Performance, 4, 345-343.
APPENDIX A General Knowledge Exam (Phase 1)Name: _______________________________Age: ______Sex: ______Please respond as accurately as possible to the following questions. Answer spaces are provided on the column on the right. Turn the test over on your desk when finished. Thank you.1. What is the distance between Milwaukee, WI and Chicago, IL? _____ miles2. In what year did South Africa become fully independent of Britishcolonial rule? ______3. How long is a marathon? (answer to the nearest tenth of a mile) _____ miles4. How old was Princess Diana when she was killed in a car crash inParis, France? ______5. How many stories tall is the Sears Tower? ______6. Wisconsin was admitted into the United States as the _____th state. ______7. How many fluid ounces are in a pint? ______8. What is the major league baseball single-season home run record set by Mark McGwire of the St. Louis Cardinals in 1998? _____9. Abraham Lincoln, in office from 1861 to 1865, was the ____th president of the United States. _____10. What is the approximate population of the city of Waukesha, WI? ________11. How many years did the Vietnam War last for? _____12. How many feet in length is a regulation basketball court? _____ feet13. In what year was Dr. Martin Luther King, Jr. assassinated? _____14. The Titanic sank off the coast of the north Atlantic in the year of ____. ______15. What is the distance between Madison, WI and St. Paul, MN? _____ miles16. In what year did World War 1 end? ______17. The Berlin Wall was dismantled in what year? ______18. What is the men’s world track record in the outdoor mile run (1600 meters)? (answer to the nearest second) ______19. In what year did Richard Nixon become the first American presidentto resign from office? ______20. In what year was Carroll College (Waukesha, WI) founded? ______ Recall Exam (Phase 2) Name: _______________________________ Please recall your original test responses to the following questions. List these responses in the answer spaces provided on the right. The correct answers to the questions are encompassed by parentheses and listed in bold type immediately following the respective question. When finished, please turn the sheet over on your desk.1. What is the distance between Milwaukee, WI and Chicago, IL? (93) _____ miles2. In what year did South Africa become fully independent of British colonial rule? (1991) ______3. How long is a marathon? (answer to the nearest tenth of a mile)(26.2) _____ miles4. How old was Princess Diana when she was killed in a car crash inParis, France? (36) ______5. How many stories tall is the Sears Tower? (110) ______6. Wisconsin was admitted into the United States as the _____th state.(30th) ______7. How many fluid ounces are in a pint? (20) ______8. What is the major league baseball single-season home run record setby Mark McGwire of the St. Louis Cardinals in 1998? (70) _____ 9. Abraham Lincoln, in office from 1861 to 1865, was the ____th president of the United States. (16th) _____10. What is the approximate population of the city of Waukesha, WI? (57,000)______11. How many years did the Vietnam War last for? (18) _____12. How many feet in length is a regulation basketball court? (94) _____ feet13. In what year was Dr. Martin Luther King, Jr. assassinated? (1968) _____14. The Titanic sank off the coast of the north Atlantic in the year of ____. (1912)____15. What is the distance between Madison, WI and St. Paul, MN? (258) _____ miles16. In what year did World War 1 end? (1918) ______17. The Berlin Wall was dismantled in what year? (1989) ______18. What is the men’s world track record in the outdoor mile run (1600 meters)? (answer to the nearest second) (3:44) ______19. In what year did Richard Nixon become the first American presidentto resign from office? (1974) ______20. In what year was Carroll College (Waukesha, WI) founded? (1846) ______This concludes the experiment. Thank you very much for your participation.
APPENDIX B Informed Consent Form (Group 1) This experiment measures the hindsight bias effect. It will be conducted during a small portion of normal class hours on Thursday, October 22 and Friday, October 23. Actual testing will take approximately five to ten minutes each of the two days. As a participant in this experiment, you will be asked to complete, to the best of your ability, a general knowledge exam consisting of 20 fill-in-the-blank questions on the first day of testing. On the second day of testing you will be presented with a different test that will require approximately the same amount of time to complete. The experiment requires that you work individually and as accurately as possible. The experiment is being conducted by John DenBoer as a course requirement for Experimental Psychology. Names are required only for data analysis and all individual results will be kept strictly confidential. As a participant, you are free to withdraw from the experiment at any time without penalty. Also, as a student in Psychology 101 the results of the experiment will not in any way affect your grade in this class nor your academic standing at Carroll College. You will receive extra credit for your participation as determined by Dr. Kasimatis. Please provide your signature and the date below if you agree to the terms of the experiment as they are presented to you and agree to offer your participation. __________________________________ ________ Signature of Participant Informed Consent Form (Group 2) This experiment measures the hindsight bias effect. It will be conducted during a small portion of normal class hours on Thursday, October 22 and Thursday, November 12. Actual testing will take approximately five to ten minutes each of the two days. As a participant in this experiment, you will be asked to complete, to the best of your ability, a general knowledge exam consisting of 20 fill-in-the-blank questions on the first day of testing. On the second day of testing you will be presented with a different test that will require approximately the same amount of time to complete. The experiment requires that you work individually and as accurately as possible. The experiment is being conducted by John DenBoer as a course requirement for Experimental Psychology. Names are required only for data analysis and all individual results will be kept strictly confidential. As a participant, you are free to withdraw from the experiment at any time without penalty. Also, as a student in Psychology 101 the results of the experiment will not in any way affect your grade in this class nor your academic standing at Carroll College. You will receive extra credit for your participation as determined by Dr. Kasimatis. Please provide your signature and the date below if you agree to the terms of the experiment as they are presented to you and agree to offer your participation. __________________________________ ________ Signature of Participant Informed Consent Form (Group 3) This experiment measures the hindsight bias effect. It will be conducted during a small portion of normal class hours on Thursday, October 22 and Thursday, November 12. Actual testing will take approximately five to ten minutes each of the two days. As a participant in this experiment, you will be asked to complete, to the best of your ability, a general knowledge exam consisting of 20 fill-in-the-blank questions on the first day of testing. On the second day of testing you will be presented with a different test that will require approximately the same amount of time to complete. The experiment requires that you work individually and as accurately as possible. The experiment is being conducted by John DenBoer as a course requirement for Experimental Psychology. Names are required only for data analysis and all individual results will be kept strictly confidential. As a participant, you are free to withdraw from the experiment at any time without penalty. Also, as a student in Psychology 101 the results of the experiment will not in any way affect your grade in this class nor your academic standing at Carroll College. You will receive extra credit for your participation as determined by Dr. Kasimatis. Please provide your signature and the date below if you agree to the terms of the experiment as they are presented to you and agree to offer your participation. __________________________________ ________ Signature of Participant
APPENDIX C Debriefing Form You have just participated in research on hindsight bias conducted by John DenBoer, an Experimental Psychology student at Carroll College. This research was designed to test the effects of time delay on the degree to which hindsight bias is displayed by participants. Hindsight bias is the tendency for people to exaggerate, after learning an outcome, their ability to have foreseen how something turned out. Research has repeatedly shown that all types of people, from nueropsychologists to football spectators, succumb to this tendency in a variety of settings. In this experiment hindsight bias was measured among three different groups (class sections) of students enrolled in Psychology 101 at Carroll College. All participants were asked to complete the same general knowledge exam on October 22. Group 1 (the 10 a.m. class) was exposed to a time delay of one day and on October 23 they were asked to recall their original answers to the 20 questions on the general knowledge exam. Group 2 (the 11 a.m. class) was exposed to a time delay of one week and on October 29 were asked to recall their original answers to the general knowledge test taken one week earlier. Group 3 (the 12 p.m. class) experienced a time delay of three weeks between taking the first exam and being asked to recall their first exam answers on November 12. As you can see, the three treatment groups experienced exactly the same procedure except for the time delay between participants’ taking the general knowledge exam and when they were asked to recall their original estimates (one day, one week, or three weeks). The experimenter hypothesizes that time delay is directly related to the amount of hindsight bias. Therefore, Group 3 will show more hindsight bias than Group 2 and Group 1 will show the least amount of hindsight bias. Hindsight bias is measured by the numerical amount participants bias their recall estimates toward the actual solution. For example, a participant answers 90 to the question “How many stories tall is the Sears Tower?” When presented with the correct answer and asked to recall their original answer, he or she erroneously recalls that they answered 100 on the first exam. Since the actual answer is 110 stories, the participant exhibited the hindsight bias effect by biasing their recall answer of 100 towards the actual solution of 110. If you have any questions or comments about the experiment, please contact John DenBoer via e-mail at jdenboer@carroll1.cc.edu. All results of the experiment are confidential. Thank you for your participation. |
