Without scientific integrity, there can be no evidence base.

In a recent article published in Nature, Martinson and colleagues (1) reported on a 2002 survey of scientific misconduct committed by mid-career and early-career scientists. The study drew national attention because it was the first to provide empirical evidence of the prevalence of scientific misconduct, using a large random sample taken from databases maintained by the National Institutes of Health's Office of Extramural Research.

The encouraging news is that the prevalence of the most serious misconduct reported by scientists was low. For 6 of the top 10 misbehaviors--including falsification, plagiarism, failure to present contradictory data, and overlooking the flaws in another scientist's data--reported frequencies were under 2%. A disturbing finding, however, was that a third of responding scientists admitted in this anonymous survey that they had engaged in 1 or more of 10 behaviors that the authors of the study determined were potentially sanctionable deviations from accepted levels of scientific integrity.

As with most survey studies, this investigation had several limitations that warrant caution in the interpretation of the findings. Low response rates and subsequent potential non-response bias were an obvious problem, with a response rate of 52% among the mid-career scientists and a rate of only 43% among the early-career group. Even more troubling--to me, at least--was the ambiguous wording of some of the survey questions. For example, 15.5% of the scientists admitted to "changing the design, methodology, or results of a study in response to pressure from a funding source." As one who often has had to change my study design and methods in response to the objections of peer reviewers, I would be hard-pressed to judge such behavior as "sanctionable misconduct." In my experience as an author, an unwillingness to alter study design and methods in the face of reviewer recommendations-no matter how right you might be--is a surefire strategy for not getting your research funded! By contrast, changing the results of a study in response to external pressure would, I believe, meet most scientists' definition of misconduct. Unfortunately, the wording of the survey question in the study by Martinson et al makes it impossible to determine the meaning of the responses.

Study limitations aside, however, I find myself in agreement with the authors' major conclusion: Far too little attention is being paid to factors in the broader US research environment that may compromise scientific integrity. The pressure on scientists to secure funding, publish their findings, and secure their promotions is enormous. Martinson et al embrace the National Academy of Science recommendation that attention needs to shift from the identification and sanctioning of "bad apples" within the scientific community toward the creation of conditions that uphold the "responsible conduct of research." (2) I support this recommendation as well as the call for future research to identify the major elements of the scientific environment that either encourage or retard the responsible conduct of research. This is an under-examined area of physical therapy academic life.

While we await the results of future research, I believe that journal editorial boards and reviewers have an important role to play in helping our profession support and reward the responsible conduct of research. John Marburger, III, PhD, Director of the US Office of Science and Technology Policy, has advocated the use of 3 important measures of scientific integrity that I believe editorial boards can and should use to judge the integrity of work that is submitted for publication. Marburger refers to these measures as "truth, diligence, and humility! (3)

Truth. Despite the myriad methods used in science, all scientists have empiricism in common. Marburger explains that empiricism means "either we ourselves, or others on our behalf, actually look at reality and not just speculate about it. If we do not do it, then we have to rely on others to observe and report the actual behavior of phenomena in a manner as independent as possible of any strategy or explanatory framework being tested." Passion about a research hypothesis is fine as long as the application of empirical observations is independent of that passion.

Putting Marburger's definition to use in manuscript review, I would argue that the quality of a study's empiricism is the first area where scientific integrity enters into the process--what Marburger calls the truth dimension of scientific integrity. From an editorial perspective, the truth dimension encompasses the degree to which an author sufficiently describes the precise means of observation so as to allow the reader to replicate those means. This is probably the dimension of scientific integrity that is most commonly applied in the manuscript review process. This standard is relevant regardless of whether the manuscript being evaluated is based on quantitative or qualitative methodology.

Diligence. For Marburger, diligence represents that aspect of scientific integrity that refers to the degree to which the methods described in a manuscript capture all of the relevant variables needed to test a research question, as well as the degree to which all of the relevant previous research (pro and con) is tracked down and reported. An author's neglect of relevant research or misinterpretation of past research would constitute deviation from the diligence dimension of scientific integrity. The breadth and depth of manuscript reviewers' familiarity with a research area are critical to the appropriate application of the principle of diligence in the peer-review process.

Humility. The third and perhaps most difficult dimension of scientific integrity to evaluate within an editorial context is the degree to which an author has modified observational strategies in response to empirical findings and not vice versa. For Marburger, the worst violation of this third principle is ignoring data that do not fit preconceptions. More subtle (and far more difficult to detect) is the abandonment or alteration of the hypothesis in the face of evidence.

To meet the standard of humility, scientists have to (1) admit in their manuscripts when they have guessed wrong, (2) avoid post hoc alterations of their research hypothesis, and (3) avoid ignoring the data when discussing the implications of their research.

When the goal of a study is to test relationships among phenomena, scientific humility requires the scientist to articulate a carefully formulated hypothesis in advance of collecting data. Scientific humility in the face of evidence is probably the most difficult dimension of scientific integrity to achieve for physical therapist scientists and, in my view, is the dimension that most differentiates the scientist from the advocate. If we are truly going to develop an evidence base in our profession, scientists must be willing to evaluate and disseminate findings that run counter to the conventional clinical wisdom of the day.

Protecting the scientific integrity of research will require going well beyond identifying falsification, fabrication, and plagiarism to come to grips with the advocated standards of scientific truth, diligence, and humility. (4) This is true for all health care disciplines--including physical therapy.

References

(1) Martinson B, Anderson M, de Vries R. Scientists behaving badly. Nature. 2005;435:737-738.

(2) Panel on Scientific Responsibility and the Conduct of Research. Washington, DC: National Academy of Sciences; 1992.

(3) Marburger J. Scientific integrity in government. Paper presented at: American Physical Society "April Meeting" on Scientific Integrity in Government; April 17, 2005; Tampa, Fla. Abstract K6.00002. Available at: http://www.thecre. com/pdf/20050711_quality.pdf. Accessed September 12, 2005.

(4) Special Report: Taking on the cheats. Nature. 2005;435:258-259.

Alan M Jette, PT, PhD, FAPTA 

Acting Editor in Chief

[e-mail]