back start next
[start] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [ 24 ] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [68] [69] [70] [71] [72] [73] [74] [75] [76] [77] [78] [79] [80] [81] [82] [83] [84] [85] [86] [87] [88] [89] [90] [91] [92] [93] [94] [95] [96] [97] [98] [99] [100] [101] [102] [103] [104] [105] [106] [107] [108] [109] [110] [111] [112] [113] [114] [115] [116] [117] [118] [119] [120] [121] [122] [123] [124] [125] [126] [127] [128] [129] [130] [131] [132] [133] [134] [135] [136] [137] [138] [139] [140] [141] [142] [143] [144] [145] [146] [147] [148] [149] [150] [151] [152] [153] [154] [155]
24 The Floor Becomes the Ceiling Nobel Laureate Herbert Simon, one of the pioneers in the field, has studied mans capabilities as an information processor for four decades. According to Simon, we react to only a minute portion of the information thrown at us. But Simon states that even the filtering process is not a passive activity that provides a reasonable representation of the world. Rather we actively exclude any data "that is not within the scope of our attention," and can end up with a dangerously inaccurate representation of the world. People, when swamped by information, may select only a small portion of the total, and reach a dramatically different conclusion than what the entire data set would suggest. Says Simon, "The capacity of the human mind for formulating and solving complex problems is very small compared with the size of the problems whose solution is required."" Researchers in many fields began to ponder cognitive limitations. Could they actually prevent the professional from carrying out his responsibilities? Some of the first experiments were conducted on clinical psychologists, who, like psychiatrists, make complex diagnostic decisions in arriving at the proper treatment. In the late 1940s and early 1950s, Paul Meehl, one of the pioneer investigators, made 20 separate surveys of groups of clinical psychologists. He required them to thoroughly examine psychotic and schizophrenic patients, to recommend treatment, then to predict how the patients would respond to the prescribed treatment. He then compared the predictions to the average recovery rates for the standard treatments. Meehl expected the psychologists diagnoses to improve on the standard treatment and to result in higher recovery rates. The averages from standard treatment would be the floor from which the effectiveness of the diagnoses could be gauged. But, in the words of one researcher, "This floor turned out to be the ceiling." The predictions of the groups of clinicians were inferior to the simple averages in 18 out of 20 studies and as good only twice! Further studies showed no correlation between tries of large groups of professional investors have fared worse than the averages for over sixty years. This is the primary reason for the sub-par performance of professionals over time that we witnessed in chapter 1. To outdo the market, then, we must first have a good idea of the forces that victimize even the pros. Once these forces are understood, the investor can build defenses and find routes to skirt the pitfalls.
A Multibillion-Dollar Slip in Investment Theory Scores of studies have made it clear that expert failure extends far beyond the investment scene and that the problem often resides in mans information-processing capabilities. Current work indicates he is a serial or sequential processor of data, who can handle information reliably in a linear manner-that is, he can move from one point to the next in a logical sequence. In building a model ship or a space shuttle, there is a defined sequence of procedures. Each step, no matter how complex the technology, advances from the preceding step to the next step until completion. The type of problem that proved so difficult to professionals is quite different, however; here configural (or interactive) reasoning, rather than linear, was required. In a configural problem, the decision-makers inte retation of a piece of information changes depending on how he evaluates other inputs. Take the case of the security analyst: where two companies have the same trend of earnings, the emphasis placed on growth rates will be weighed quite differently depending on the outlooks for their industries, revenue growth, profit margins, returns on capital, and the host of analytical criteria we looked at previously. His evaluation will also be tempered by changes in the state of the economy, a clinical psychologists training and experience and his or her accuracy. One study found, surprisingly, that psychologists were no better at in-1 80 1 judgments than individuals with no training, and sometimes worse. Do such findings extend beyond the couch? Yes. Radiologists failed to diagnose lung disease from X-rays 30% of the time, although the symptoms were cleariy evident. And, in a classic study of tonsillectomies in the mid-1930s (a fashionable operation at the time), a group of doctors examined 1,000 New York City schoolchildren. The doctors recommended 61% have their tonsils removed. The remaining children were examined by a second group of physicians, who diagnosed that 45% needed tonsillectomies. A third set of physicians, examining the remaining children, recommended the removal of 46% of the rapidly dwindling stock of tonsils. A final examination was carried out on the survivors and, sure enough, 45% needed their tonsils out. At this point, only 65 children remained. Fortunately, the doctors decided to call off further testing before tonsils became an endangered species in the New York City school system.*
the level of interest rates, and the companies competitive environments. Thus, a successful analyst must be adept at configural processing; he must integrate many diverse factors, and if one changes, he must reweigh the whole assessment. Not unlike juggling, each factor is another ball in the air, increasing the difficulty of the process. Are professionals, in or out of the investment field, capable of the intricate analysis their methods demand? A special technique has been designed, using a statistical test called ANOVA (Analysis of Variance), to evaluate the configural capabilities of experts. In one such study, nine radiologists were given a highly configural problem: deciding whether a gastric ulcer was malignant. To make a proper diagnosis, the radiologist must work from seven major cues either present or absent in an X-ray. These can combine to form 57 possible combinations. Experienced gastroenterologists indicated that a completely accurate diagnosis can be made only by configurally examining the combinations formed from the seven original cues.* Although the diagnosis requires a high level of configural processing, the researchers found that in actual practice, it accounted for a small part of all decisions-some 3%. Over 90% came from serially adding the individual symptoms. A similar problem is deciding whether a psychiatric patient is to be allowed to leave the hospital for short periods. The hospital staff has to consider six primary cues that can be present or absent (for example, does the patient have a drinking problem?) and 64 possible interactions. Nurses, social workers, and psychologists showed little evidence of configural thinking, although it was essential for the optimum solution." In another test, 13 clinical psychologists and 16 advanced graduate students attempted to determine whether the symptoms of 861 patients were neurotic or psychotic, a highly configural task. The non-configural results were in line with the first two examples. Curious what he would find in the stock market, Paul Slovic, an internationally respected cognitive psychologist, tested the importance of configural (or interactive) reasoning in the decisions of market professionals. In one study he provided thirteen stockbrokers and five graduate students in finance with eight important financial inputs-trend of eamings per share, profit margins, outlook for near-term profits, etc.- inputs they considered significant in analyzing companies. They had to think configurally to find the optimum solution. As it tumed out, however, configural reasoning, on average, accounted for only 4% of the decisions made-results roughly equivalent to those of the radiologists and psychologists.
[start] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [ 24 ] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [68] [69] [70] [71] [72] [73] [74] [75] [76] [77] [78] [79] [80] [81] [82] [83] [84] [85] [86] [87] [88] [89] [90] [91] [92] [93] [94] [95] [96] [97] [98] [99] [100] [101] [102] [103] [104] [105] [106] [107] [108] [109] [110] [111] [112] [113] [114] [115] [116] [117] [118] [119] [120] [121] [122] [123] [124] [125] [126] [127] [128] [129] [130] [131] [132] [133] [134] [135] [136] [137] [138] [139] [140] [141] [142] [143] [144] [145] [146] [147] [148] [149] [150] [151] [152] [153] [154] [155]
|