JOURNAL OF COMMUNITY MEDICINE

INDIAN ASSOCIATION OF PREVENTIVE AND SOCIAL MEDICINE ORISSA CHAPTER

Department of Biostatistics, Bioinformatics, & Epidemiology

Medical University of South Carolina Charleston SC USA

E-mail: gilbertg@musc.edu

Gregory E.Gilbert

Importance of the Research Proposal in Clinical Research

 

This is the first in a series of articles concerning research methodology, statistics, and epidemiology. In this article I will discuss the benefits of expending adequate time on writing a research proposal. A well-defined and well-written research proposal will: improve research, avoid errors, enhance analysis, and reduce time to publication or presentation (as some research proposals may be almost complete at the time of initial consultation with a statistician and thus be ready for publication, oral presentation, or poster presentation). Time spent on analysis will be decreased because more problems are anticipated, “designed out”, avoided, and the analysis is restricted to the narrow path defined by the research hypothesis avoiding, albeit interesting but, needless tangential investigations.

 

As I write this article I find myself wondering, why I am writing it? I know how to write a research proposal. I know others know how to write a research proposal. I have faith “the academy” taught us, as scientists, how to conduct research, yet in practice I rarely observe investigations mirroring scientific pedagogy. This leads me to conclude if investigators were conducting research as they were taught there would be no reason for this article. So, let’s take a look how clinicians can help themselves and a consulting statistician by drafting a research proposal.

 

Keywords: Research, Research Proposal, Research Design

 

It has been my experience scientists, as students, are taught how to “do science”. However, as daily experience “doing” science becomes routine we begin to adhere less to the tenets we were taught. To this end, let me remind the reader of the importance of a literature review preceding any investigation. This literature review need not be exhaustive, but it must be tied directly to the research topic and use sources no greater than five years old – preferable three years old. A more exhaustive review can be done during data collection, data analysis, or as a part of the formal manuscript authoring process. It might be helpful to summarize information from relevant sources in a table similar to Table 1 .

Table   1.  Literature Review Summary Worksheet

                           

                                                                  

                                                                  

                                                                                                                      

                                                                  

                                                                  

                                                                  

                                                                  

                                                                  

                                                                  

           

           

After the literature review is completed, the investigator needs to write a brief introduction. Although the audience for the publication will be the investigator’s peers, the investigator should remember a statistician will be one of the first to read the research proposal and the better the statistician understands the problem at hand, the better the statistician can comment and make recommendations. The introduction should briefly review existing recent literature, introduce the problem at hand, and introduce the research hypothesis. The introduction should clearly set forth the objectives of the study. Further, in addition to the research hypothesis, the investigator should list possible findings supporting and contrary to his or her hypothesis (Good& Hardin 2006 .[ 1]

 

Hypotheses should be narrowly focused, well-defined, testable, and mutually exclusive. Investigators should be in constant search for the “publon” – the least publishable unit. This is not as self-serving as it may appear at first glance. By identifying the “publon” the investigator will narrow the definition of his or her problem and operationalize all facets of the investigation. By doing so the investigator will simplify his or her investigation which will make for easier, less complicated, and clearer writing of the proposal (and manuscript). For example, two colleagues of mine and I investigated the consequences of altering admissions weighting in rural and minority applicants to medical school ([2]Basco Gilbert & Blue 2002; [3]Basco Gilbert & Blue 2001; and [4]Gilbert Blue & Basco 2003). We submitted one paper for publication. One of the reviewers suggested we divide the paper into separate topics dealing with grade point average weighting and selectivity index weighting and in two separate populations – a minority applicant population and a rural applicant population. By dividing the original paper into different topics and focusing on one topic in each paper we were much better equipped to get our point across   and do it so in a clear and concise manner and we produced three much more definitive and better quality publications instead of one.               

 

Once the introduction is written the investigator should draft the methods section. No statistician is going to categorically reject a proposal because an investigator applied misguided statistical methodology before embarking on an investigation. The statistician often has suggestions to improve the research design, sampling frame, and/or statistical methodology proposed. In the methods section the investigator should stress the (biological or other) plausibility of the model proposed, if applicable. The methods section should include the research design including the sampling frame; definitions of explanatory or independent, confounding or control outcome variable(s); statistical methodology to be employed, and a statement of institutional review board approval.

 

After writing the brief methods section the investigator’s focus turns to the results section. The investigator should be familiar enough with the subject matter to “mock-up” a results section with expected findings (leave blanks to insert specific values later). Included will be any tables and/or figures to appear in the manuscript. These “mocked-up” tables and figures are immensely helpful to both the author and the statistician.

 

Next, draft the discussion section of the paper. The investigator should have a reasonable idea of the outcomes of the results. Compare and contrast those expected results with findings in the current literature. In the discussion section introduce possible findings and findings alternative to the research hypothesis listed in the introduction section. In addition to using your expected findings to defend the research hypothesis, use the current literature. In the discussion make sure to include a thorough paragraph on strengths and limitations of the study. Finally, write a brief summary of the expected findings and a one sentence conclusion.

 

There are a number of reasons to go to all this trouble to draft a research proposal rather than naively pose the question, or even meet with a statistician after data collection has been done. The main reason is to prevent lack of focus. With the ease with which complex analyses are done these days it is far too easy to continue to ask “what if” or “how about”. These tangential questions, while interesting and sometimes provocative, prolong analysis and evaluation of the results of the investigation. Additionally, the investigator who invests the time it takes to develop a research proposal is an investigator who is more likely to complete the research. Another important reason for undertaking this process is that the majority of the paper will be written by the time statistical analysis is completed, thus increasing scholarly production. Having a well thought out research idea and plan of attack will be greatly appreciated by collaborative authors as well as the statistician. To assist readers in working through this process I have included a research proposal outline” in Table 11

 

Table 11 Research Proposal Outline

----------------------------------------------------------------------------------------------------------

I.           Literature Review

A.          No more than five sources no more than five, preferably three, years old

II.          Introduction (1 page)

A.          Three or for sentences reviewing literature

B.          State the problem

C.         State the Research Hypothesis

D.         List possible findings and conclusions, including some contrary to the research hypothesis

III.         Methods (1 page)

A.          Research Design

B.          Sampling Frame

C.         Investigational (Explanatory or Independent) Variables

D.         Confounding (or Control) Variables

E.          Outcome Variable(s)

F.          Statement of Institutional Review Board Approval and Adherence to the Tenets of the Declaration of Helsinki

IV.        Results (1 page)

Write with blanks to be filled in later

B.          Include Mocked-Up Tables and/or Figures (as needed)

V.          Discussion (1 page)

A.          Compare and contrast your expected findings with those found in current literature

B.          “Explain away” alternative findings to research hypothesis using expected findings and current literature

C.         Critically discuss Strengths and Limitations of investigation

D.         Summary (additional ½ pages)

E.          Conclusion (additional sentence)

----------------------------------------------------------------------------------------------------------

 

Of course the length of the research proposals will vary. However, as a rule of thumb I would suggest: a one page (typed, double-spaced) introduction (with a detailed hypothesis); a one page methods section (including a discussion of the research design, sampling frame, variable definition, and proposed statistical analysis); a one page results section (plus mocked-up tables and figures); a one page discussion section, a one half page discussion of strengths and limitations; a one half page summary; and a one sentence conclusion. In all, about five pages, about 1000 words. (This article has about 1300 words.) If you do nothing else except fill in the blanks, refine the grammar, add a bibliography and an abstract, the research proposal has become potentially publishable. Certainly, it would make a good candidate for a poster session or an abstract for an oral presentation. Some publications have “Short Communications” and some publications accept 1000 word original research submissions (e.g. Diabetes Care) the completed research proposal could be submitted to one of these publications.

So there you are an operational definition of the contents of a research proposal. It may not guarantee research will be published; but it will be research possessing integrity and completeness worthy of serious consideration.

References

1. Good PI & Hardin JW. (2006). Common   errors  in statistics (and how to avoid them), 2^nd Edition. Hoboken, New Jersey: John Wiley and Sons, Inc.

2. Basco Jr. WT, Gilbert GE, Blue AV. (2002). Determining  the consequences for rural applicants when additional consideration is discontinued in a medical school admission process. Academic  Medicine. 77(10)  Supplement. S20-S22.

3. Basco Jr. WT, Gilbert GE, Blue AV. (2001).   Estimating  the pre-interview rankings of underrepresented minority when ethnicity is not considered in the admission process. Academic Medicine. 76(10). S55-S57.

4. Gilbert GE, Blue AV, and Basco Jr. WT. (2003).  The effect  of undergraduate GPA selectivity adjustment on pre-interview ranking of rural medical school applicants. Journal of Rural Health.  19  (2).   101  -104.

Acknowledgements

The author wishes to acknowledge the referees for their excellent suggests that served to strengthen this editorial; James E. Gilbert, Ph. D. for his insightful comments; William T. Basco, M.D. for the literature summary worksheet; and Matthew Ohland, Ph. D. for the term “publon.