Often it is hard to tell if deviations from good scientific practices are due to dishonesty or error. Both dishonesty and error can produce similar consequences, but dishonesty has the intent to deceive. Therefore, harsh judgments and consequences are held for dishonesty, as it is a serious violation of scientific ethics, whereas honest errors are often forgiven. Definitions of dishonesty are not rigid but organizations including the National Academy of Sciences, the National Academy of Engineering, and the Institute of Medicine have definitions that involve deception, misrepresentation, fabrication, falsification, or plagiarism [11]. In addition, other forms of misconduct according to these organizations include harassment, misuse of funds, violation of regulations, and vandalism [11].

Dishonesty in the production and analysis of data may take the form of fabrication or falsification. Fabrication is the unfounded creation of data, making up data in the absence of experimental results [1, 7]. A notorious example of fabrication was by William Summerlin who claimed that he successfully transplanted skin from one unrelated mouse to another, after using a solution that suppressed the immune response and therefore rejection. However, in 1974, it was found that Summerlin had used a black marker on white mice to fabricate the transplantation of skin from a black mouse to a white mouse [7].

Falsification involves the manipulation or misrepresentation of experimental data and is more difficult to define and prove than fabrication [11]. Falsification often occurs during the analysis and statistical handling of data. The most common forms of falsification involve the misrepresentation of results in terms of trimming, cooking, and fudging of the data [12]. Trimming or “tightening up” of data involves the failure to report results that do not support the scientist’s hypothesis. Fudging of data occurs when scientists try to make results appear better. This type of falsification includes the omission of ambiguous, outlying, or unexplainable data points. “Cooking” the data refers to designing experiments in order to obtain results that the scientist already suspects will have positive results, thereby avoiding tests that are likely to have negative results. As such, it can be difficult to distinguish the boundary between misrepresentation of data, poor methodology, or making sense of ambiguous results [13]. Since rules for analyzing data are not defined, judgment must be exercised on how to trim data in order to convey clear and objective results [1].

Plagiarism is considered a form of dishonesty; whether intentional or unintentional, it is regarded as intellectual theft [7]. Plagiarism occurs in a variety of situations in which one takes undeserved credit for another person’s ideas, techniques, methods, or written work. It can be due to improper citation, attribution, self-plagiarism, or paraphrasing. It is not uncommon in science to base research on previous methods or techniques for which credit must be given. Other situations in which plagiarism may occur is during the peer review process in which scientific reviewers may be tempted to steal unpublished ideas. Although unintentional plagiarism may be viewed as an error whether due to ignorance or carelessness, steps should be taken to correct it and avoid it in the future. Many documented cases are investigated each year, and many more undocumented; it is expected that much of plagiarism is unintentional [11].

It is important to distinguish between types of errors in research. Practical errors include human error and experimental error. Examples of human error include using an instrument inappropriately, errors in performing calculations, and errors while recording data. Human error is most commonly due to inattention, disorderliness, or hastiness in research. Experimental error is inherent in using an instrument and should be taken into account when reporting the data and results. Theoretical errors are due to errors related to the interpretations and analysis of data. It also includes self-deception or bias in analysis. These errors can be at the start of a project, during its construction and design, or while interpreting the data. Self-deception refers to the occasion when a scientist is convinced that the results of the experiment have significance or validity even when the data do not support it. For many scientists their entire life’s work is devoted to a specific field. Between the pressure to produce results and the loss of perspective and self-criticism, scientists may be deceived to see what they want to see when evaluating their data.

Biases are systemic flaws in research and can lead to errors. The definition of what is considered a bias is controversial. Resnik suggests that one person’s bias may be another scientist’s valid assumption or methodology [7]. Biases often originate from political, social, and economic aspects of science. Longino asserts that biases can be minimized by freedom and openness in research and by being open to criticism [16]. Conflicts of interest can be considered a special form of bias and result in the loss of objectivity in scientific research. Conflicts of interest originate from personal or financial interests that conflict with professional obligations. In the academic setting potential conflicts of interest include power, tenure, funding, and publications, whereas in business or medicine, it may be self-referrals, gifts, or recruitment bonuses [17]. In themselves, conflicts of interest are not all unavoidable or unethical. Instead, conflicts of interest present unique situations in which a scientist must have the insight to be extra diligent when making decisions.

The increase in academic researchers desire to collaborate with business and industry has brought concerns over conflict of interest more to the forefront. As a result, the Association of American Medical Colleges (AAMC) proposed guidelines for dealing with conflicts of interest. The AAMC supports full disclosure, aggressive monitoring, and misconduct management [18]. Full disclosure requires revealing all information that could restrict investigators’ activities. It should include details of both individual and family financial and professional interests that are related to the research at hand. Next, these disclosures should be reviewed by supervisors, chairpersons, or institutional review boards [18]. The reviewers should first confirm if a conflict of interest exists and to what degree it might affect the research. If the conflict of interest is prohibitive, the reviewers have a responsibility to remove the researcher from the relationship with either the industry or the research being pursued. In academic research it is often the institutional review committee that is responsible for developing a conflict of interest policy, determining when there is a conflict of interest and what actions need to be taken in order to maintain the integrity of the research environment [1].