英语第五单元段落匹配_在线真题试卷与模拟练习_英语第五单元段落匹配_考试宝

更新时间：试题数量：购买人数：提供作者：

有效期：个月

章节介绍：共有个章节

收藏

我的练习

我的错题
(0道)

我的收藏
(0道)

我的斩题
(0道)

我的笔记
(0道)

专项练习

顺序练习 0 / 0

随机练习 自定义设置练习量

题型乱序 按导入顺序练习

模拟考试 仿真模拟

题型练习 按题型分类练习

易错题 精选高频易错题

学习资料 考试学习相关信息

搜索

题库预览

The public awareness of scientific fraud has increased remarkably since the late 1980s when a controversy made front-page news, a controversy about a paper investigated for fraud which had as coauthor a Nobel laureate. During the 1990s scientific fraud was disclosed on numerous occasions (Lock, 1996). In fact, it was recently suggested that fraud was now is “endemic in many scientific disciplines and in most countries” (Fulford, 1998) and may undermine the credibility of scientific research. However, the clandestine character and consequential lack of reliable information make it difficult to study scientific fraud.The characteristics and frequency of scientific fraud, therefore, are generally unknown, and its impact on medical research is unclear. Biostatisticians routinely work closely with physicians and scientists in many branches of medical research and have unique insight into data. In addition, they have the methodological competence to detect fraud and could be expected to have a special professional interest in the validity of results. Biostatisticians therefore could provide unique and reliable information on the characteristics of fraud in medical research. The objective of this study was to assess the characteristics of fraud in medical research by surveying members of the International Society for Clinical Biostatistics (ISCB).

The ISCB is an international society for biostatisticians, which, in April 1998, had 442 members in 31 countries or regions on six continents. Most of the members (69%) were from the European Union (Table 1). The ISCB has a subcommittee on fraud, which was set up following the expression of concerns about fraud and inappropriate responses to it. In April 1998 a questionnaire designed for anonymous response was mailed to all members, with an accompanying letter inviting participation in the survey. An electronic version was also opened on the ISCB home page on the Internet. No prepaid return envelopes were used and no formal reminders were distributed. The deadline for response was the end of July 1998. Mailed responses were sent to the survey office in Lund, Sweden.

The questionnaire was designed cooperatively by members of the ISCB Subcommittee on Fraud and contains 31 items divided into two sections: one on age, sex, position, nationality, etc., and the other on experience with and opinions on scientific fraud. Prior to the actual survey the questionnaire was pilot-tested on a small number of ISCB nonmember biostatisticians in the United States and in the European Union and was revised in accordance with the experiences from this pilot study.

In the questionnaire, fraud was defined as a deliberate attempt to mislead others in the design, conduct, analysis, or reporting of a study. Furthermore, it was stressed that careless or incompetent research was not necessarily condidered fraud; an “intention to cheat” was required. In addition, the responders were specifically asked not to characterize poor methodology and borderline practices as fraud when replying to the questionnaire. The sign test was used to evaluate the difference in subjectively estimated prevalence of fraudulent projects between epidemiological papers and papers from clinical trials. Spearmans rank correlation coefficient was used to investigate a potential relation between response rates and prevalence of statisticians having been engaged in fraudulent projects.

One hundred sixty-three members and three nonmembers completed questionnaires. We report here information from the 163 members who responded. The response rate for members was 37% (see Table 1). Several responders expressed their concern about anonymity (nationality, age, and sex could identify them). Eleven members also declined to disclose their nationality. Most responders mailed or faxed their questionnaires to the survey office; only 23 (14%) used the Web.

Table 2 shows the distribution of the responders by age, sex, and type of employers. Nineteen members had governmental employment, 72 university, 52 private, and, while 35 members reported multiple employment, 19 members had multiple employer types. Counting responses rather than responders, of the 29 responses stating governmental employer, six stated employment by a regulatory agency and 23 by “other governmental”. Ninety-one responses stated employment by a university. Private employment was stated by 77 responses. Of these, 41 responses stated employment by a pharmaceutical/biotechnical/device company, 18 by a contract research organization, 8 by “other private company,” and 10 responded that they were self-employed. Comparison of responders and nonresponders was not possible because ISCB does not record these demographic data in its membership records.

Table 3 gives a summary of the main results. A clear majority (126 responders; 76%) considered career and power to be the main motive for fraud, not financial reward. One hundred and four members (65%) considered the impact of fraud on the progress of medical science to be a major problem (27 members) or somewhat important (77 members). One hundred and three responders (63%) reported that the organization in which they worked did not have a system for handling suspected fraud, or that they did not know if this was the case, or did not reply to the question. Eighty-one responders (51%) knew of at least one fraudulent project in their proximity during the last 10 years. Forty-three responses indicated that the fraud had been related to fabrication or falsification of data and 31 to suppression or selective deletion of data, whereas 16 and 32 indicated deceptive design or analysis and deceptive reporting of results, respectively. Forty-nine responders (31%) reported having been engaged in a project in which fraud took place or was about to take place. In 14 cases the responder intervened to attempted to prevent the fraud from occurring; in 11 cases the responder reported it to a relevant person or organization; in 10 cases the responder was not aware of the fraud as it happened or did nothing to prevent it. Fourteen responders reported other actions or did not reply to the question. Information was not obtained on whether an attempt to prevent fraud was successful. Twenty responders (13%) had been requested to support fraud themselves. In three cases, this concerned fabrication or falsification of data; in 11 cases suppression or selective deletion of data; and in 7 and 12 cases deceptive design or analysis and deceptive reporting of results, respectively. Eight of the 20 responders reported that the attempted fraud eventually was changed into acceptable practice. Nineteen responders (12%) had participated as an expert in a formal investigation on suspected fraud.

The subjectively estimated prevalence of fraud in published reports was somewhat greater for epidemiological studies than for clinical trials (interpolated median values 0.80% and 0.69% respectively; p = 0.047). Response rates appeared to be lower in countries or regions with higher frequency of statisticians having participated in fraudulent projects (rs = -0.41, p < 0.001).

This survey of biostatisticians found that knowledge of fraudulent projects is surprisingly common. Although only 163 of the 442 members (37%) responded to the survey, 81 (51%) reported that they knew of at least one fraudulent project in their proximity in the past 10 years. Even if these 81 responses included all episodes of fraud known to the 442 ISCB members, the prevalence estimate of ISCB members knowing of fraudulent projects in their vicinity in the past 10 years would be 18%. Although the overall low response rate to this survey limits its generalizability, the high proportion of responders knowing about fraudulent projects provided the primary motivation for this report.

The prevalence of persons knowing about fraudulent projects among biostatisticians in the current survey is greater than that among the members of the American Association for the Advancement of Science (AAAS) who were surveyed in 1994. In a random sample of 469 researchers and administrators in universities, industry, and government, 27% reported that they knew of at least one fraudulent project over the last 10 years (Hamilton, 1992). The participation rate in the AAAS survey was similar (31%) to ours. If all episodes of fraud known to AAAS members had been reported, the prevalence estimate would have been 8%. The proportion of participants knowing about fraudulent projects is also substantially greater than that in a 1992 Norwegian survey of 152 medical principal investigators whose study protocols had been assessed by ethics committees. In that study also 27% of the responders knew about one or more cases of scientific fraud (Jacobsen and Hals, 1995). However, that study focused on ethical issues in general rather than on fraud; the response rate was also substantially higher (70%).

There are several explanations for the high proportion of responders who knew of the occurrence of fraud, other than the low (37%) response rate. Biostatisticians are often privy to confidential data and are knowledgeable enough to understand their implications. Therefore they may be in a unique position to observe scientific transactions in the prepublication stages. Second, because of their technical knowledge, biostatisticians may be harsh judges of scientific collaborators’ optimistic interpretation of data. In our instructions we attempted to distinguish between fraud and carelessness, and we requested that poor methodology and borderline practices not be considered fraud. However, our instructions might have been ambiguous, and we cannot be sure that the responders were not overzealous in what they considered to be fraud. Third, we cannot be certain that the reported episodes of fraud were at all important, or that the same episode was not reported by multiple respondents. However, the latter does not seem likely because of the diverse countries of origin of the responders.

Discussions of fraud have in the past often focused on clinical trials. Reasons for this may be that results of clinical trials have more direct medical consequences than epidemiological studies and that regulatory authorities control the conduct in trials. However, our survey seems to suggest that fraud might be more prevalent in epidemiological studies than in clinical trials. A possible explanation for this finding may be the more controlled nature of the clinical trial, which makes it less amenable to fraud (Buyse, George, Evans et al., 1999).

The impact of fraudulent research has been discussed previously (Garfield and Welljams-Dorof, 1990). It appears that the scientific community quickly recognizes and avoids citations from work known, or suspected, to be fraudulent. This further stresses the importance of control systems that reveal fraudulent publications as soon as possible. The low proportion of biostatisticians who reported working for organizations with systems for handling suspected fraud reiterates the need for more such systems. The low number of biostatisticians who reported having participated as experts in formal investigations on fraud may suggest that few investigations of fraud occur.

If we were to do this survey again, we would use prepaid return envelopes. We had hoped that allowing replies through the World Wide Web could save the cost of postage, but that did not prove to be the case. We would also ask more specific questions on the nature of any fraudulent episodes to which the responders had been exposed.

We believe that it is important to change the general attitudes toward fraud in medical research. Improved education, emphasizing ethical standards in research, and more and better systems for handling suspected fraud could be expected to play roles in prevention. However, more research in this field is necessary to determine which educational, organizational, or administrative measures actually prevent fraud. In addition, low participation rates appear to be a major problem in many surveys. Survey techniques and procedures therefore might have to be specially developed to improve validity and reliability in this field.

1