DERECHO Y LA JUSTICIA*
2- La mirada antropológica al campo del derecho
The first hypothesis this e x p e r i m e n t w a s designed to explore, that creeping determinism would be more in evidence when outcome knowledge was given after the event description rather than before, received only weak support. Where outcome knowledge was given before the event description no evidence of creeping determinism was found. However, two scenarios out of the four produced results consistent with the creeping determinism hypothesis in the condition where outcome knowledge was given at the end of the event description. The pooled results in Tables 6.1 and 6.2 tell a similar story. Overall, less than 50 per cent of subjects or assessments given in hindsight were consistent with creeping determinism. Again only one scenario ("The British-Gurkha Struggle") provided clear evidence for both the outcome reported to have
174.
occurred and the outcomes reported not to have occurred.
In summary, then, differences between foresight and hindsight (after) were much less consistent than those obtained by Fischhoff (1975a). Limited support was obtained for the suggestion from Experi ment 1 that creeping determinism is mediated by the position of outcome knowledge. Reporting the outcome of an event produced a statistically significant increase in its perceived likelihood for only two of the four cases and then only when outcome knowledge was presented at the end of the scenario. At the moment the available data would seem to suggest that presenting outcome knowledge at the end of a scenario is a necessary, but not sufficient, condition for hindsight-foresight differences to occur.
The second hypothesis, that subjects would exhibit greater con fidence in likelihood assessments made in hindsight than foresight, received no support. This might have been expected in view of the failure to detect consistent effects in hindsight. In the one case
k_ .
where creeping determinism was found ("The British-Gurkji Struggle", hindsight after) confidence levels were opposite to that predicted,
but not significantly so. ;
The third hypothesis, concerning the effect of different information orders, manipulated (as indicated in Experiment 1) by ordering statements scaled by Thibaut and Walker, on hindsight judgements also received little support. The hypothesis, it will be recalled, was that hind sight judgements would exhibit stronger creeping determinism when information pointing at the reported outcome was presented last rather
175.
than first. Inspection of figures 6.1 and 6.2 reveals the medians
associated with each scenario to be in the predicted direction. For
the outcome which occurred (Number 3) assessments are higher, for both hindsight conditions, in Jury Case B. Further inspection of Figure 6.2 reveals the opposite to be the case for the outcome reported not to have occurred (Number 2) . As none of these differences were significant, however, the results will not be discussed in any more detail.
However, the interesting question remains as to why quite strong evidence for recency effects in hindsight judgements was found in Experiment 1 but not here. This failure to replicate the results may stem from two considerations. First, it may be due to the procedural differences between the two experiments. In Experiment 1 an item-by item sequential presentation together with "surprise" instructions
in hindsight was used. Experiment 2 presented subjects with a continuous prose passage and instructed subjects to make judgements in the light of all information presented. Both the sequential presentation and "surprise" instructions may account for the failure to replicate order effects here. For example, the sequential presentation procedure may cause people to attend to the information more especially that towards the end. In recalling it in order to formulate likelihood assessments the information which appeared last may be more available from memory and hence exert greater influence over assessments.
The second consideration stems from the way in which information order was manipulated in the two scenarios. In Experiment 1 it was reported as being achieved in a non-rigorous or objective way - the
176.
bulk of "lawful" or "unlawful" statements either appeared first or last, and were ordered in such a way as to make the scenario easy to comprehend.^- Such a procedure falls short of a rigorous order manipulation in a number of ways. First, rather than have the bulk of lawful or unlawful statements first or last it would be more appropriate a manipulation of order to have all of each sort appear in such a way. Second, the scale values and categorisation given to these statements by Thibaut and Walker's (1975) judges may differ on both counts to those given by British judges. Third, not all the statements taken from Thibaut and Walker were scaled by them, speci-r fically, those briefly outlining the incident. Finally, statements were scaled according to degrees of lawfulness or unlawfulness but not in relation to a specific charge.
In view of these shortcomings a further experiment was designed to investigate the effect of different information orders on hindsight
judgements. The ordering of information was achieved using a more
rigorous and objective procedure in line with the points discussed above.
An oversight and shortcoming with this, ar.d the previous experiment, was that subjects were not informed about a specific charge
177.
CHAPTER 7 .
EXPERIMENT 3: THE EFFECT OF INFORMATION ORDER UPON HINDSIGHT