In this nonexperimental study, we used a cross-sectional design and a convenience sampling of fourth-year medical students at 4 MD schools. After obtaining institutional review board approval from each participating institution, we surveyed all fourth-year medical students from 2 higher-ranked private institutions (Boston University School of Medicine in Massachusetts and Northwestern University Feinberg School of Medicine in Chicago, Illinois) and 2 mid-ranked public institutions (University of Nevada School of Medicine in Reno and University of Louisville School of Medicine in Kentucky). These 4 schools were selected to maximize geographic diversity among students and to represent a spectrum of academic competitiveness. After e-mailing each institution’s fourth-year class an initial invitation to participate, we used the Dillman approach and sent 2 reminder e-mails to increase the response rate.
16
The online survey was an original instrument and comprised 12 questions on a 5-point Likert scale with 1 indicating strongly disagree and 5 indicating strongly agree; 1 multiple choice question on factors influencing match selection; 9 demographic questions; and 2 open-ended (qualitative) questions. The face validity of the instrument relied on existing knowledge based on the literature and expert observation and comments from the research team, several of whom authored a recent study about the concerns and attitudes of senior medical students.
17 For example, the team anticipated differences in future residents’ interest in research, in clinical opportunity, and in lifestyles associated with geographic locations.
Descriptive statistics included demographics, Likert scale, and program-specific traits that influence match selection. Data examination of the 4 schools on match selection factors showed similarities between the higher-ranked private institutions and the mid-ranked public institutions; therefore, we categorized the schools into 2 groups. We used the Pearson χ
2 statistic and the independent samples
t test to assess the association of demographics and the program-specific traits that influence match selection with school category. Then we used the χ
2 test of linear trend to assess whether a linear association existed between the program-specific traits that influence match selection and students’ self-reported class rank. Next, we used the Pearson χ
2 to assess the association of responses of students who expressed agreement that the prestige of a residency program would influence its position on their final rank list (+PRESTIGE) with those who expressed disagreement or a neutral response (–PRESTIGE), and of students attempting to enter competitive vs noncompetitive specialties. Competitive specialties were considered those without enough available positions to accommodate all graduating US medical students who would like to match into them, and noncompetitive specialties were considered those with an excess of available positions.
18 Similarly, we grouped the other Likert scale items into agreement or disagreement categories and used the χ
2 test for trend to assess for associations between these responses and the estimated percentage of DO and internationally trained physician faculty provided by students. The data are expressed as frequencies and percentages for all analyses. SPSS version 22 (IBM) was used to analyze the quantitative data, and all
P values were 2-tailed, with statistical significance set at
P<.05.
We analyzed qualitative data using the Pandit variation of the Glaser and Strauss constant comparison, and 3 of the researchers (K.H.M., A.U., and C.K.M.) independently read and coded, collaborated on preliminary coding, then independently recoded each entry using Microsoft Excel 2013 (Microsoft Corporation) before coming to consensus on the common themes.
19 We then calculated the frequency of themes by linguistic position (ie, mentioned first [A list] or second [B list] in the comment). The frequency of themes was then compared with quantitative outcomes to support conclusions.