Secondary Logo

Journal Logo

Featured Topic: Faculty Recruitment and Productivity: Featured Topic Research Report

The Research Productivity of Faculty in Family Medicine Departments at U.S. Medical Schools: A National Study

Brocato, Joseph J. PhD; Mavis, Brian PhD

Author Information
  • Free

Abstract

Faculty in U.S. family medicine departments have diverse clinical and academic professional roles including patient care as a family physician; teaching medical students, interns and residents as a clinical teacher; and conducting research contributing to the specialty and to society at-large.1 Although much is known about family physicians’ development as clinicians during medical school and residency2–4 as well as their entry into new clinical positions after residency,5,6 little has been written in the literature about family physicians who have decided to pursue an academic research-oriented career track in departments of family medicine.

Developing a better understanding of research activities by faculty in family medicine is vital for a number of reasons. They have a unique opportunity to conduct basic and applied research because of their position as the point of first contact in patient care, as well as their focus on longitudinal care. From a micro perspective, building a core faculty of productive researchers is vital to enhancing the clinical legitimacy of the field of family practice,7–9 as well as the academic legitimacy of family medicine.7,10,11

Geyman12 distinguishes clinical from nonclinical (usually educational specialist) family medicine faculty: nonclinical faculty have traditionally carried the bulk of research production in departments of family medicine. According to Geyman, “the (clinical) family physician must play a central role in identifying and pursuing researchable questions, drawing on other disciplines for help as needed. Research in family practice cannot be delegated to nonclinical researchers, and cannot be meaningful without the combined efforts of the university and the ‘real world’ practice community.”12, p.52 Rodnick furthers this view by identifying physician faculty as the key cohort for developing “a critical mass of (family) medicine faculty.”13, p.306 Rodnick states that these faculty “must overcome their innate inhibitions and force themselves to begin academic writing at the earliest possible moment following their decision to commit to academic careers. What they write about and where it is published is initially of secondary importance.”13, p.306

Given the need for clinical faculty in family medicine to engage in scholarship, one of us (JB) undertook this study to examine faculty’s research productivity in U.S. medical schools. The study, which was part of JB’s doctoral degree course, contains a description of family medicine faculty’s prior research experiences, demographic characteristics, psychological/cognitive characteristics concerning research (i.e., motivation and interest in research), as well as research environmental conditions—factors that might in turn influence research productivity. JB concluded the study with a testing of the model of faculty’s research productivity using these same factors as predictors of traditional forms of research productivity.

Method

Participants

In this study, we focused on full-time faculty with MD or DO degrees (with or without other advanced degrees) who were in family medicine departments at U.S. medical schools. As specified by both Rodnick13 and Henry,14 these family medicine faculty provide the foundations of a clinical research core for the discipline. Further, it is focused on faculty in family medicine departments at U.S. medical schools and did not include additional family medicine faculty at community-based hospitals or clinics. These community-based family medicine faculty normally carry significantly higher clinical and teaching loads, with much smaller research expectations and responsibilities that made it problematic to include them in this study.

Defining research productivity

For purposes of this study, research productivity was defined as the number of peer-reviewed journal articles, national conference presentations, and national grants (both private and government) over the previous two-year period. No distinction was made in type of national grant submitted or funded (research, educational, or other types of grants). A two-year period of self-reporting was selected based on Creswell’s15 suggestion that it is a suitable measure for studies of faculty research productivity. A “peer reviewed” focus was used as a proxy for quality. Research productivity areas were limited to the three highly traditional forms of peer-reviewed scholarship because of the need for family medicine faculty to compile a core body of research comparable to that of other medical specialties. These traditional forms of research productivity are consistent with the three-step process for developing a core body of research in family medicine suggested by DeHaven et al.16 that includes grant funding, presenting, and publishing.

Sampling methods and questionnaire

In 2000, we drew a stratified random sample of 796 faculty who were in family medicine departments at U.S. medical schools in 1999, using a sample size formula developed by Aday.17 This sample included 600 faculty with MD degrees only, 100 with MD degrees plus other advanced degrees, 88 with a DO degree only, and eight who held the DO degree with other advanced degrees. In aggregate, the sample represented 36% of all faculty in family medicine departments nationally. For the groups with MD degrees only and with MD degrees plus other advanced degrees, the Association of American Medical College’s (AAMC’s) Faculty Roster18 drew random samples. Both of these randomly sampled groups cut across all 138 departments of family medicine faculty in the United States and Puerto Rico. The Faculty Roster also compiled a “certainty sample” for the two other groups—i.e., 100% of those holding the DO degree only and those holding the DO degree plus another advanced degree—because of their relatively small numbers.

A 43-item questionnaire was created to assess faculty’s research productivity that contained at least one question for each of the elements of the conceptual model (see Figure 1) developed for this study: prior research socialization, current research interests and activities, current research environments, research productivity, and demographic characteristics. The questionnaire and study design received approval from the institutional review board of Michigan State University. The demographic section of the questionnaire was primarily composed of forced choice categorical measures, but also included continuous measures for age based on year of birth and hours per week spent on research activities.

Figure 1
Figure 1:
The authors’ conceptual model of predictors of the research productivity of faculty in family medicine departments at U.S. medical schools.

Each of the other questions asked respondents for their agreement/disagreement with statements on a five-point Likert scale ranging from “strongly disagree,” “disagree,” “neither agree/disagree,” and “agree,” to “strongly agree” (1 = strongly disagree, 5 = strongly agree), as suggested by Fink19 for scaled survey instruments. A sixth scaled option, “not applicable,” was added based upon the suggestion of the family medicine faculty who pilot tested the questionnaire. However, respondents who indicated “not applicable” for an item were not included when the descriptive data for this study was tabulated to allow for continuous measures. The self-reported research productivity section included a five-point scale ranging from zero productivity units up to five or more productivity units for each productivity outcome measure. The productivity scales emphasized lower levels of research productivity based upon the relatively low anecdotal estimates of productivity identified in the family medicine literature.

Data collection and data analysis

Information from the AAMC’s Faculty Roster18 was used to prepare mailing labels for each of this study’s four sampled groups. The questionnaires were mailed to faculty’s academic departmental offices. Prior to the second mailing of questionnaires, nonrespondents from the Faculty Roster mailing list used in the first mailing were cross-referenced with the American Academy of Family Physicians National Database to determine new preferred mailing addresses (usually either a clinic address or a home address). The second and third waves of questionnaires were mailed to the preferred address to increase the response rate.

Review and pilot testing

The questionnaire’s content and structure were developed from an extensive synthesis of the higher education and medical education literature on faculty research productivity. The questionnaire’s form and content were then reviewed by two national experts on faculty research productivity, Dr. Carole Bland from the University of Minnesota and Dr. John Creswell from the University of Nebraska. Finally, the questionnaire was piloted with a group of five family medicine faculty from the Department of Family Practice at Michigan State University.

Statistical methods

We used means and frequency counts to provide a descriptive summary of the demographic characteristics for questionnaire respondents. The individual questionnaire items used as predictors in the conceptual model of research productivity were rated using a five-point scale: the means and standard deviations were reported for these items. For the measures of research productivity, which were also scaled ratings, three measures of central tendency (mean, median, and mode) and the standard deviation were reported.

To test the conceptual model underlying the research productivity, we used a factor analysis with varimax rotation, reducing the number of variables required for the prediction model. These composite predictors were combined with demographic indicators in a series of multiple regression analyses to test their contribution to each of the research productivity measures.

Results

Response rate

A total of 474 faculty returned completed questionnaires after three mailings. We received responses from nearly every state in the United States as well as Puerto Rico. We determined that 26 faculty from the original study sample were ineligible because they self-identified as having left academic medicine, left departments of family medicine, or had retired. An additional 18 did not have forwarding addresses. Excluding these 44 faculty from the original study sample yielded an adjusted questionnaire response rate of 63% (474/752). A response rate analysis showed no difference in the return and nonreturn rates for each of the study sample groups. Further, we conducted a response bias test via a wave analysis at weeks 2, 4, and 8 of data collection. As part of the wave analysis, responses to two questions from each of the five sections of the questionnaire showed statistically insignificant differences in mean responses between the early, middle, and late respondents.

Respondents’ characteristics

Respondents had a mean age of 45.9 years and reported a moderate amount of experience as a faculty member (mean = 10.8 years). They spent little time on research activities, with a large majority spending little if any time on research (mean = 3.4 hours per week on research). Although there were small pockets of faculty who spent a significant amount of time on research, 38% of respondents reported spending no time on research, 42% spent one to four hours per week, 9% spent four to eight hours per week, and 11% spent more than eight hours per week on research related activities.

Most respondents were men (71%) and overwhelmingly white (89%). In terms of academic rank, 74% of respondents indicated a traditional academic rank (i.e., instructor, assistant professor, associate professor, or professor), while 19% listed a clinical rank. Fifty-five percent identified themselves as an assistant or clinical assistant professor, and 74% were in nontenure track academic positions. Additionally, mirroring the study sample, 88% held the MD degree, while 12% held the DO degree, and 25% reported holding nonclinical advanced degrees in addition to the MD or DO.

Predictors of research productivity

Table 1 summarizes the means and standard deviations for the nondemographic variables of the conceptual framework for this study.

Table 1
Table 1:
Descriptive Analysis of Variables in a Research Productivity Survey of 474 Faculty Members in Family Medicine Departments at U.S. Medical Schools, 2000

Prior research socialization.

In terms of prior research socialization, respondents reported little exposure to research during their academic and clinical training before becoming faculty. Few had substantive experiences engaging in research projects or in developing their own individual research skills. As a result, few had published original research and even fewer had opportunities to disseminate research at national conferences. Respondents also reported little mentorship in research prior to becoming faculty, including have opportunities to participate in research-based fellowships or postdoctoral programs. A few respondents who had participated in graduate training prior to becoming faculty described their research experiences positively.

Psychological and cognitive characteristics.

Many respondents reported interest in and motivation for research. They found research to be personally satisfying, understood the promotion and/or tenure guidelines of their departments, stayed “up-to-date” on literature in their research areas, and further developed their research skills since becoming faculty. Conversely, respondents described themselves as lacking a clear research agenda, having few research projects underway suitable for publication, and feeling a high degree of pressure to participate in research activities.

Research environments.

As shown in Table 1, family medicine faculty described academic family medicine as a discipline not heavily vested in research. The majority of respondents did not have well-developed professional networks of colleagues outside their own academic departments with whom they could discuss research.

Many respondents viewed their academic institutions as possessing a reputation for meaningful research and creating an expectation for faculty to generate scholarly work. Concerning their academic departments, respondents were less clear about the cultural norms and expectations for research. They indicated that although most departments emphasized research, they did not view their departments as possessing a reputation for research, nor did they view their colleagues as being particularly research productive. Further, research support was identified as being disseminated unevenly, especially in regard to protected time for research.

Faculty research productivity

Table 2 shows respondents’ research productivity for the previous two academic years. For all but one type of productivity, respondents reported on average less than one scholarly work per year. Manuscripts submitted for publication were the largest category of scholarship generated, however a majority of respondents produced nothing over the previous two academic years. Overall, 52% did not submit any manuscripts for publication during the prior two years, 58% reported no manuscripts accepted for publication, and 67.3% indicated no proposals/papers accepted for conference presentations. In terms of government grant research funding, 68% reported that they did not submit any grants for federal funding and 75% reported no funded federal proposals. Similarly, 78% indicated no grant proposals submitted for private funding and 84% indicated no funded proposals through private funds.

Table 2
Table 2:
Research Productivity during the Previous Two Academic Years of 474 Faculty Members in Family Medicine Departments at U.S. Medical Schools, 2000

Results of factor analysis of independent variables

The first step in testing the conceptual model for this study was to reduce the number of predictive variables to a more manageable number. Table 3 shows the results of a principal components factor analysis using varimax rotation with Kaiser normalization.

Table 3
Table 3:
Factor Analysis of Independent Variables, from a Research Productivity Survey of 474 Faculty Members in Family Medicine Departments at U.S. Medical Schools, 2000

The first factor described prior research training before the first faculty position. The second factor combined seven measures of psychological and cognitive predispositions toward research. The third factor defined the current research environment, while the fourth factor related to resources available for faculty research. The fifth factor represented institutional prestige, including the prestige of the medical school, residency program, and graduate school. The sixth factor contained a single variable, perceptions of family medicine as a research discipline. Each of the six components that emerged from the analysis demonstrated a high degree of internal consistency, with Cronbach alphas of .60 for factor 5 (institutional prestige), .64 for factor 4 (resources available for research), .85 for factor 3 (current research environments), .88 for factor 2 (psychological and cognitive predispositions toward research, and.90 for factor 1 (prior research training).

Predicting research productivity

After the factor analysis reduced the independent variables to six composite predictors, we used a series of seven full-model regression analyses to test the contribution of each composite predictor toward predicting each of the productivity outcome measures. The seven productivity models included manuscripts submitted for publication, manuscripts accepted for publication, conference proposals/papers accepted for presentation, national government grants (both submitted and funded), and private grants (both submitted and funded).

Specifically, each prediction model included the new composite variables: prior research socialization, psychological/cognitive characteristics, current research environments, resources for research, prestige regarding research, and perceptions of family medicine as a research discipline. Additionally, the following demographic predictors were included in each model: age, years as a faculty member, hours per week spent on research, gender, ethnicity, advanced degree (MD), advanced degree (PhD/EdD), academic rank (assistant professor), academic rank (associate professor), academic rank (clinical ranks), and tenure stream status. The use of full-model regression analysis required the exclusion of two demographic variables from each model, academic rank (full professor) and advanced degree (DO), because of the statistical need to exclude at least part of dichotomous variables in each prediction model.

Table 4 shows a summary of the regression analyses. For the outcome of manuscripts submitted for publication, the full-model regression accounted for 66% of the variance (R2), suggesting a strong relationship between the independent variables used in the model and this outcome measure. Similarly, in predicting the number of manuscripts accepted for publication, the full-model regression accounted for 62% of the variance (R2), also suggesting a strong relationship between the independent variables used in the model and the outcome measure. An examination of the coefficients for the predictive model for manuscripts accepted for publication showed that psychological and cognitive characteristics, hours per week spent on research activities, and tenure system appointment provided the greatest contributions to the prediction (see Table 5).

Table 4
Table 4:
Summary of Full-Model Multiple Regressions Predicting Research Productivity, from a Survey of 474 Faculty Members in Family Medicine Departments at U.S. Medical Schools, 2000
Table 5
Table 5:
Summary of Full-Model Multiple Regressions: Standardized Coefficients (Beta), from a Research Productivity Survey of 474 Faculty Members in Family Medicine Departments at U.S. Medical Schools, 2000

For the outcome conference papers/proposals accepted, the full-model regression accounted for 50% of the variance (R2). As with the manuscripts accepted for publication predictive model, psychological and cognitive characteristics and hours per week spent on research activities as providing the greatest contributions to the prediction. The respondent’s age was found to be a negative predictor of this outcome.

The government grants submitted and funded models accounted for 51% and 44% of the variance (R2), respectively, which suggests strong relationships. The government grants submitted model identified current research environments, family medicine research as a discipline, years as a faculty member, and hours per week spent on research as the greatest contributors to the prediction. With the government grants funded model, hours per week spent on research activities and years as a faculty member were the most predictive factors.

The private grants submitted and funded models accounted for 32% and 31% of the variance (R2), respectively. The private grants submitted model identified psychological and cognitive factors and current research environments as adding most to this predictive model, while the private grants funded model also showed these two factors as most contributory, along with hours per week spent on research.

Given the different expectations for faculty in tenure system and nontenure system appointments, we examined the research productivity prediction models separately by respondents’ tenure status. The number of tenure-system respondents with complete data was insufficient to test the prediction models. The prediction models were tested with the subset of nontenure system respondents, who made up the majority (74%) of respondents. For six of the seven productivity measures, the magnitude of the adjusted R2 values decreased by eight to ten percentage points for nontenure respondents, ranging from 30% to 57%. The exception was private grants submitted, where R2 increased from 32% to 39%.

When only nontenure respondents were considered, there were some changes in the dependent variables with significant contributions to the prediction models. For submitted manuscripts, the only change in the prediction model was the addition of ethnicity (beta = .16) as a positive predictor. For accepted manuscripts, both ethnicity (beta = .14) and academic rank of assistant professor (beta = −.32) were added to the original prediction model derived from the full sample. There was no change in the predictors comprising the model for accepted conference proposals. For submitted government grants, the two composite variables (current research environment and family medicine as a research discipline) dropped out of the model with clinical rank added (beta = −.33) as a negative predictor. The prediction of funded government grants changed with the addition of academic rank of assistant professor (beta = −.48) and academic rank clinical (beta = −.47) and the removal of years as a faculty member. For nontenure system faculty, the prediction of submitted private grants saw the addition of academic rank assistant professor (beta = −.56), academic rank associate professor (beta = −.37), academic rank clinical (beta = −.50) and ethnicity (beta = .15) to the model. Finally, for funded private grants, hours of week spent on research dropped out of the model.

Discussion

The demographic and productivity snapshots we provide in this study support the few national studies of family medicine faculty research conducted by Culpepper and Franks,20 as well as Hueston.21 Additionally, our data lend a considerable amount of credence to the anecdotal literature in family medicine that describes the lack of a critical mass of researchers and research being conducted in departments of family medicine on a national basis.7–10 The reasons that few family medicine faculty spend time on research may be largely two-fold. As mentioned extensively in the family medicine literature, faculty have been increasingly called to support their academic activities through greater clinical responsibilities. Because of these heightened clinical demands partnered with significant teaching expectations, there is likely to be erosion of time available for family medicine faculty to conduct research.

In addition to the demographic characteristics of family medicine faculty, we also provide some unique insights on the respondents’ psychological and cognitive characteristics, particularly faculty’s perceived tension between their clinical and academic roles. Respondents were aware of the need to produce research and to develop their research skills to become productive scholars. Yet, we provide strong evidence that they face a great deal of countervailing pressure related to their other roles, diminishing the likelihood they will engage in research. This tension is manifested in the extremely high level of role stress reported by a large majority of our respondents. In short, the tension between academic and clinical roles is also exacerbated by the mixed messages from medical schools and departments of family medicine regarding the relative importance of both of these faculty roles.

Our findings related to family medicine faculty’s prior research socialization offer little evidence of cumulative research productivity: few reported significant engagement in research activities before their first faculty position. While there is an organizational climate that supports research, there is little evidence of an infrastructure necessary for research productivity. This is reflected in the extremely limited opportunities for research socialization in medical school, internship, and residency for future family medicine faculty, related to research mentorship, participation in research projects, and experiences producing and disseminating research. This lack of research socialization is distinctive and largely different from other nonprimary care specialties that have their own clinical practices rooted in a more well-defined knowledge base centered in disciplinary scholarship.9,22 In most departments, a critical mass of faculty who conduct research appears to be lacking. Perhaps the greatest departmental impediment to research is the lack of protected time for research—a benefit consistently mentioned in the higher education literature as strongly influencing faculty research productivity.23 As a result, few departments can be considered highly research productive, nor do they tend to convey high expectations for research. Because family medicine faculty have missed this valuable socialization as part of their training, accelerated research socialization for new faculty becomes vital to assuming academic responsibilities, including producing research.

Family medicine faculty are products of their disciplinary, institutional, and departmental environments for research. From the results of this study, it is clear that family medicine faculty do not consider the discipline of family medicine to be strongly rooted in a research culture. Part of this perception might be the relative youthfulness of the specialty. But a larger part might be what others have identified as family practice’s “countercultural roots”: the discipline seeks to distance family practice from other medical specialties through the creation of a unique populist clinical identity, as well as a nonconformist, “anti-intellectual” academic identity.7,13,24

It is not surprising that few of our respondents reported any research produced over a two-year period given their uneven prior research socialization, mixed motivation to conduct research, as well as the conflicting organizational messages they received regarding research. Undoubtedly, the lack of faculty research productivity is troublesome for a specialty and discipline that desires growth and definition through a critical mass of faculty research. Our results provide no evidence that the research productivity of faculty in family medicine departments is increasing, even given the consensus in the family medicine literature on the need for such scholarship.25–27

In addition to the descriptive findings, our predictive findings provide several interesting conclusions. Of the research productivity measures we used, the regression models were best at predicting manuscript submissions and publications. Psychological and cognitive factors along with time dedicated to research played a key role in predicting publication-related outcomes. To a lesser extent, there was also a substantial amount of variance accounted for in the prediction of conference presentations and government grants. The composite factors were less important in predicting these outcomes, where indicators of experience and effort carried the weight in the prediction models. The model was weakest in predicting private grants: both effort and the composite factors played a role in the model. This suggests that that there are unique combinations of factors associated with the various measures of research productivity; no single model best predicts all research-related outcomes.

Additionally, despite the large number of predictive variables involved for multiple productivity outcomes, a relatively small set were statistically significant. Of these, many of the same predictors were important for different productivity outcomes—possibly signifying potential relationships between various research activities, such as between manuscript acceptance and conference presentations, and manuscript acceptance and grant funding. However, across a majority of the productivity measures, the psychological and cognitive factors and time spent on research were the strongest predictors of research productivity. It is meaningful to note that demographic variables such as gender and ethnicity, advanced degree, faculty rank, and to lesser degree age were not strong predictors of research productivity. Unfortunately, the lack of research productivity seems to be a shared phenomenon within the discipline.

When we tested the prediction models using the subsample of nontenure system respondents, the amount of explained variance decreased in most cases. Generally the composite variables derived from the conceptual model were stable in their contributions to the prediction models of research productivity. Among the demographic variables, those associated with academic rank were more likely to become significant predictors of research productivity: lower ranks were associated with lower productivity. For three models, ethnicity also made a significant contribution to the productivity outcomes, favoring majority respondents. It is a limitation of this study that the number of tenure-system respondents was insufficient to test these productivity models.

While our study significantly advances understanding of the research activity of clinical faculty in departments of family medicine, our study had a number of additional limitations. Our study was limited by the sample selected. Replicating the study with nonclinical faculty from family medicine departments, and faculty from community-based residencies and from academic disciplines other than family medicine would provide a richer context for understanding the uniqueness of scholarship within family medicine and its place within the variety of disciplines in academic medicine. Additionally, examining research productivity more broadly to include other forms of nontraditional scholarship would be an appropriate extension to test the goodness of the models’ fit.

In summary, when predicting research productivity of faculty in departments of family medicine, that faculty actually spend greater time overall on research activities is more important than demographics. Additional, faculty need to possess a psychological and cognitive predisposition to research including further enhancing their research skills, establishing a definable research agenda, fostering research networks, having multiple research projects underway, maintaining an in-depth knowledge of a research area of specialty, as well as clearly understanding research expectations for promotion and tenure.

Our results suggest a number of implications for deans of medical schools, department chairs, and individual faculty within family medicine, including that considerable time needs to be invested in recruiting and selecting faculty with an innate interest in primary care scholarship who are further willing to expend time conducting research. Further, emphasis needs to be placed on providing substantial faculty development in research skills, mentorship, and networking, among other areas, to lay the foundation for faculty research activities. This concurrent attention to both targeted and purposeful faculty recruitment practices, as well as continuous faculty development will begin to move the specialty and academic discipline closer to their scholarly aspirations.

The authors would like to acknowledge the significant contributions of Carole Bland, PhD, from the University of Minnesota, John Creswell, PhD, from the University of Nebraska, and James Fairweather, PhD, from Michigan State University toward providing substantive feedback in the development of this study. The authors would also like to thank the American Academy of Family Physicians Foundation and the Society of Teachers of Family Medicine for their financial support, as well as the Association of Departments of Family Medicine for their endorsement of this study, which made this research possible.

References

1 Bland CJ, Simpson DE, Hekelman FP, Stritter FT. Introducing the dedicated issue on faculty development. Fam Med. 1997;29:231–2.
2 Randall CS, Bergus GR, Schlechte JA, McGuinness G., Mueller CW. Factors associated with primary care residents’ satisfaction with their training. Fam Med. 1997;29:730–5.
3 Hosokawa M, Zweig S. Future directions in family medicine: results of a Delphi study. Fam Med. 1990;22:429–33.
4 Merenstein JH, Schulte JJ. A residency curriculum for the future. The STFM Task Force on Residency Curriculum for the Future. Fam Med. 1990;22:467–73.
5 DeWitt DE, Robins LS, Curtis JR, Burke W. Primary care residency graduates’ reported training needs. Acad Med. 2001;76:285.
6 Turner BJ, Laine C. Differences between generalists and specialists: knowledge, realism, or primum non nocere. J Gen Intern Med. 2001;16:422–4.
7 Doherty WJ, Christianson CE, Sussman MB. Introduction. In: Doherty WJ, Christianson CE, Sussman MB, eds. Family Medicine: The Maturing of a Discipline. New York: Haworth Press; 1987:xi–xvi.
8 Stephens GG. Developmental Assessment of Family Practice: An Insider’s View. In: Doherty WJ, Christianson CE, Sussman MB, eds. Family Medicine: The Maturing of a Discipline. New York: Haworth Press;1987:1–15.
9 Pellegrino ED. Family Practice Facing the Twenty-First Century: Reflections of an Outsider. In: Doherty WJ, Christianson CE, Sussman MB, eds. Family Medicine: The Maturing of a Discipline. New York: Haworth Press;1987:23–47.
10 Colwill JM. Family Medicine in the Medical School. In: Doherty WJ, Christianson CE, Sussman MB, eds. Family Medicine: The Maturing of a Discipline. New York: Haworth Press;1987:71–85.
11 Culpepper L, Becker LA. Family Medicine Research: Two Decades of Developing Its Base. In: Doherty WJ, Christianson CE, Sussman MB (eds). Family Medicine: The Maturing of a Discipline. New York: Haworth Press, 1987:129–57.
12 Geyman JP. On the developing research base in family practice. J Fam Pract. 1978;7(1):51–2.
13 Rodnick JE. Research in family medicine: counterculture continued. Fam Med. 1987;19:251–307.
14 Henry R. Developing research skills for medical school faculty. Fam Med. 1997;29:258–61.
15 Creswell JW. Faculty Research Performance. Vol 4. Washington, DC: Association for the Study of Higher Education, 1985.
16 DeHaven MJ, Wilson GR, Murphree DD. Developing a research program in a community-based department of family medicine: one department’s experience. Fam Med. 1994;26:303–8.
17 Aday LA. Designing and Conducting Health Surveys: A Comprehensive Guide. 2nd ed. San Francisco: Jossey-Bass Publishers; 1996.
18 Faculty Roster. Washington, DC: The Association of American Medical Colleges, 1999.
19 Fink A. The Survey Kit. Thousand Oaks, CA: Sage Publications, 1995.
20 Culpepper L, Franks P. Family medicine research. Status at the end of the first decade. JAMA. 1983;249:63–8.
21 Hueston WJ. Factors associated with research efforts of academic family physicians. J Fam Pract. 1993;37(1):44–8.
22 Culpepper L, Becker LA. Family medicine research: two decades of developing its base. In: Doherty WJ, Christianson CE, Sussman MB (eds). Family Medicine: The Maturing of a Discipline. New York: The Haworth Press, 1987:129–64.
23 Creswell JW. Conceptual explanations of research performance. In: Finklestein MJ (ed). The ASHE Reader on Faculty and Faculty Issues. 2nd ed. Needham Heights, MA: Ginn Press, 1990:240–8.
24 Stephens GG. Family medicine as counterculture. Fam Med Teach. 1979. 1979;16(5):14–8.
25 Weiss B. Publications by family medicine faculty in the biomedical literature: 1989–1999. Fam Med. 2002;34:10–6.
26 Curtis P, Dickinson P, Steiner J, Lanphear B, Vu K. Building capacity for research in family medicine: is the blueprint faulty? Fam Med. 2003;35:138–40.
27 Kruse J, Bradely J, Wesley R, Markwell S. Research support infrastructure and productivity in U.S. family practice residency programs. Acad Med. 2003;78:54–60.
© 2005 Association of American Medical Colleges