1、1 外文翻译 原文 Human Resource Accounting Information: A Comment Concerning Demand Characteristics Material Source: The Accounting Review, Vol. 54, No. 1 (Jan., 1979), pp. 199-204 Author:Doug S nowball A recent experiment reported by Hendricks 1976 sought to assess the impact of human resource accounting
2、(HRA) information on stock investment decisions. Hendricks concluded that “. . . in this experiment stock investment decisions were affected by the addition of human resource accounting information to conventional accounting information“ (p. 300). This paper questions the significance that can be at
3、- tached to Hendrickss finding. Two issues-demand characteristics and the novelty of HRA data-are raised. Because the first issue is relevant to the validity of the results of any laboratory experiment, it is given most attention. THE HENDRICKS STUDY Hendrickss experiment involved a random assignmen
4、t of graduate student subjects to two groups-Group I and Group II. Group I (the experimental group) was exposed to two treatments. For Treatment I, they were presented with conventional accounting statements for two companies, A and B, and were asked to advise an imaginary client on how best to inve
5、st $100,000 between the two companies. For Treatment II, the same task was performed, except that statements incorporating HRA data (in addition to the conventional statements) were provided to the subjects. The control group, Group II, was subjected only to Treatment II. Hendrickss design may best
6、be characterized as a variation of the separate- sample, pre-test-post-test design (see Campbell and Stanley 1966). Figure 1 depicts the design where R indicates random assignment; O refers to the ob- servation of the investment decision; and X refers to the exposure to the experimental event (the p
7、resentation of HRA data). FIGURE 1 2 RO1XO2-Group I R XO3-Group II Typically the O1-O2 difference is ex- amined only to confirm the O1-O3 comparison. However, Hendricks assigned the majority of his subjects to Group I and focused upon the O1-O2 comparison. The O2-O3 difference was examined only to d
8、etermine whether subjects in Group I were preconditioned by first receiving conventional statements. Hendricks did not find a significant difference between the amounts “invested“ in Company B by the Group I subjects and the Group II subjects (the O2-O3 comparison). On the basis of this result, he c
9、oncluded that Group I subjects were not preconditioned by Treatment I and used only those subjects to test his hypothesis. Under Treatment I, Group I subjects invested an average of $22,709 in Company B; under Treatment II the corresponding investment was $48,909. Hendricks attributed the change to
10、the HRA data supplied under Treatment II and interpreted the results as a confirmation of his hypothesis. Commendably, Hendricks referred to demand characteristics as a possible limitation of his study; however, I fail to share his belief that it is unlikely that they influenced the results obtained
11、. The demand characteristics associated with his experiment appear to be so strong as to provide a very appealing alternative explanation for his results. DEMAND CHARACTERISTICS “Demand characteristics“ (hereafter referred to as DCs) refers to the cues that govern a subjects perceptions of what the
12、experimenter expects of him. According to Orne 1969, p. 146, “They include the scuttlebutt about the experiment, its setting, implicit and explicit instructions, the person of the experimenter, subtle cues provided by him, and, of particular importance, the experimental procedure itself.“ If subject
13、s are aware that they are participating in an experiment, they are likely to be highly attuned to those environmental cues that, rightly or wrongly, suggest the experimenters hypothesis. Aronson and Carlsmith 1968, p. 61 have described the human subject as “. . . not responding to the experimental o
14、perations per se, but . responding to his interpretations of what behavior these experimental operations are supposed to elicit from him.“ In other words, aspects of the experimental environment that are irrelevant to the theoretical variable under study may be responsible for 3 behavior changes in
15、the direction predicted, or may obscure the expected effect and lead to systematic outcomes in an unexpected direction. DEMAND CHARACTERISTICS AND THE HENDRICKS STUDY Three features of Hendrickss experiment will be identified as contributing to concern over the impact of DCs. The first two are commo
16、n to many other studies; it is only when considered in conjunction with the more important third feature of Hendrickss study that they become significant. First, concern is raised by the use of student subjects. Student subjects frequently have had prior exposure to laboratory experiments and are es
17、pecially likely to view them as games in which the players task is to determine the true purpose of the experiment. Once “discovered,“ the perceived purpose may have a particularly strong biasing effect upon students responses. Orne 1962 has stated that college students typically share the experimen
18、ters hope that the study in which they are participating will contribute to science, and, thus, have a very real stake in a successful experimental outcome. Second, Hendricks gave no indication that any explicit reward system operated within his experiment. The DCs issue is not the subjects knowledg
19、e of the experimental purpose per se, but its biasing effect upon responses. As Sigall et al. 1970 have suggested, such biases may not emerge if subjects can be sufficiently motivated by rewards other than those associated with fulfilling the role of a “useful“ subject (i.e., by ensuring the success
20、“ of the experiment). Third, and most importantly, the specific experimental materials and procedures encourage the view that DCs may underlie Hendrickss results. It is difficult to imagine that even the most naive subject in Group I did not accurately detect the purpose of the experiment upon expos
21、ure to Treatment II. Though facing exactly the same task as previously, the subject had additional data available for his judgment. The data represented a substantial addition to the data supplied under Treatment I; the data were not a product of conventional accounting procedures; and, indeed, they
22、 represented an accounting innovation that is the topic of current debate among accountants. The novelty of these additional data is likely to signal that significant changes in decisions are expected by the experimenter. This impression would have been strengthened by a closer examination of the in
23、come statements. Company A had a higher net income for both years than Company 4 B under conventional reporting; the positions were reversed in the HRA statements presented. Similarly, Company A showed a more favorable income trend over the two years reported in the conventional statements but Compa
24、ny B showed superior growth in the net income reported within the HRA statements. Hendricks appropriately pointed out that the subjects merely may have used the net income figures as a basis for their responses (functional fixation). A slightly modified explanation is that the subjects used the net
25、income figures in determining the experimenters expectations and that their responses reflected the perceived expectations. Both the specific materials and the experimental design employed encourage concern over DCs. Recall that Hen- drickss design was characterized earlier as a separate-sample, pre
26、-test-post-test design. Campbell and Stanley 1966 have indicated that this design, unlike the one-group, pre-test-post-test design, controls for the reactive effects of experimental arrangements. However, the same authors warn against excessive reliance on their summaries of research designs. This w
27、arning is particularly germane to Hendrickss study, for the responses of Group II provide no assurance that responses of Group I were uncontaminated by DCs. Even though Group II subjects were exposed only to Treatment II, they nevertheless were exposed to all the experimental materials presented to
28、Group II; the only substantive difference was that no overt decision was required of them after examining only the conventional statements. In short, ample cues were available for DCs to also influence decisions by Group II subjects. DEALING WITH THE PROBLEM OF DEMAND CHARACTERISTICS Whether the res
29、ults of Hendrickss study were biased by DCs is a matter of opinion; that his approach involved little attempt to reduce the possibility of such biases is more certain. Accordingly, Hendrickss failure to report proced ures undertaken to assess the impact of DCs constitutes a serious omission. The rea
30、der is asked to rely on the fact that subjects “. . . responses were kept anonymous and the subjects had no association with or obligation to the experimenter. . .“ (p. 297). Orne 1970 has described a number of procedures to determine what DCs are used by subjects. Of these, both post- and pre-exper
31、imental inquiry procedures could have been employed profitably by Hendricks. The goal of both procedures is to determine the subjects perceptions of the purpose of the experiment. However, 5 post-experimental inquiries, even when conducted by a skilled interviewer, pose two major problems. First, su
32、bjects responses to the post-experimental inquiry itself may be biased by DCs. For this reason, the inquiry should be conducted by interviewers who are ignorant of the true experimental purpose. Second, Orne 1962 has pointed out, “ .the subjects perception of the experimenters hypothesis is based up
33、on his own behavior, and therefore a correlation between these two variables may have little to do with the determinants of behavior“ (p. 781). The pre-experimental inquiry described by Orne 1970 avoids this second problem. In terms of Hendrickss experiment, it would involve describing the experimen
34、tal procedures to subjects (other than those intended to participate in the experiment itself), showing them the specific accounting statements and describing the decision tasks. In esence, these subjects would be “walked through“ the experiment; their participation would be limited only in that a r
35、esponse to the investment decision task would not be required of them. The subjects, uncommitted by behavioral reactions to the experimental tasks, would then be interviewed to determine their perceptions of the experimental purpose. DCs cannot be removed entirely from any experiment; nevertheless,
36、careful pre-experimental inquiry (or post-ex- perimental inquiry at the pilot study stage) might have led Hendricks to revise aspects of his study. Perhaps the ideal method of avoiding biases generated by DCs-by studying the behavior of individuals who are unaware that they are participating in an e
37、xperiment-may have been difficult to accomplish in a study such as his. However, psychologists (e.g., Campbell 1957, 19691 Aronson and Carlsmith 1968) have suggested means by which effects of DCs may be minimized. Some appear inapplicable to most accounting research (e.g., the medical placebo model
38、and the two experiments ploy), and a common approach, deception, raises serious ethical issues. However, Hendricks might have reduced the potential for bias by the use of an improved design and relatively minor changes in procedures (e.g., less-transparent materials, an explicit reward system, and a
39、 focus upon sensitivity to changes in HRA data). NOVELTY Hendricks warned against generalizations of the results of his experiment, yet commented that “. . . the favorable results of this experiment should justify the additional research that would permit more general conclusions“ (p. 301), The 6 in
40、ference is that the results should strengthen beliefs that decisions made in the “real world“ will be found to be affected by the presentation of HRA data within financial statements. However, given the novelty of the HRA data presented to his subjects, the inference seems unwarranted, even if the D
41、Cs issue is set aside. Few individuals presently could claim genuine familiarity with HRA data. Hen- dricks confined his analysis to subjects who indicated an understanding of the meaning of HRA. Yet, it is likely that most had never seen statements containing such data prior to the experiment; inde
42、ed, the subjects had read and heard little about the topic. As a result, their perceptions of the characteristics of the data (e.g., perceptions of the datas objectivity, reliability and, most imporantly, relevance) must have been formed primarily on the basis of the materials presented in the exper
43、iment. How they perceived the HRA data is unclear (probably to themselves as well as to readers of Hendrickss article). The only reasonably safe assumption that can be made is that the data were perceived to be novel. Psychologists frequently have noted the strong and varied reactions generated by n
44、ovel stimuli (for examples, see the review by McGuire 1966). Hence, the responses of Hendrickss subjects may be related more closely to the novelty of the data than to perceptions of the relevance of the data to investment decisions. It may be argued that the subjects responses bear some relationshi
45、p to the reactions of “real world“ investors to the initial introduction of HRA statements. However, “real world“ decisions have economic consequences and reactions to novel data in such an environment are likely to be very different (e.g., more cau- tious) than those in an experimental situation. T
46、he more important issue, though, is the impact of HRA data presented on a continuing basis; the responses of Hendrickss subjects offer no insight into either experimental or “real world“ reactions once such data lose their novelty. The novelty problem, compared to the DCs issue, presents a more form
47、idable obstacle for research on the topic addressed by Hendricks. Careful consideration of DCs can lead to procedures that satisfactorily manage that problem; procedures designed to overcome the novelty problem are likely to represent only a partial solution or may introduce new threats to validity.
48、 Ultimately, the latter concern may lead to a reluctant conclusion that behavioral impacts of drastic innovations in accounting reporting cannot be examined satisfactorily through laboratory experimentation until the changes lose their novelty. 7 译文 人力资源会计信息：一份关于需求特性的评论 资料来源 ： 会计评论， 1979 年 1 月 作者：道格
49、 .斯诺鲍 最近一个被亨德里克斯 1976所报道的实验试图评估人力资源会计信息在股权投资方面的影响。亨德里克斯总结道 “ .在这个实验中股权投资决策是同时受人力资源会计信息与传统会计信息影响的 ”（ p300） 。本论文意在证实此重要性是否与亨德里克斯的发现一致。两个论点 需求特性与人力资源会计信息的创新 被提出。由于第一个观点与每一个劳动力实验结果的有效性相关，它被给予重大关注。 亨德里克斯研究 亨德里克斯的实验包括分配成两个组的毕业生受测者 第一组和第二组。第一组 （ 实验组 ） 致力于两种制度。在第一种制度之下，他们被给予两个公司的传统会计报表， A 公司与 B公司，并被要求为一位虚构的客户对于如何使用 100,000 美元在两公司上投资提出建议。在第二种制度下，进行一项相同的任务，除了向受测者提供的报表是结合了人力资源会计信息的 （ 附加传统报表 ） 。对比组，即第二组，仅受第二种制度的影响。 亨德里克斯的设计可能表现独立样本的多样性时是佳的 （ 见坎贝尔和斯
