Personal, interpersonal, and organizational influences on student satisfaction with clinical education.

Key Words: Clinical education, Education: physical therapist, Professional education, Satisfaction Student.

Professional education faculties should be concerned with student satisfaction as an outcome of the educational process because student satisfaction has been associated with later professional attitudes,[1] career commitment, and retention.[2-4] Particularly important to the health care professions is student satisfaction with clinical education because this component of the curriculum has an impact on choice of areas of practice[5] and selection of work settings.[6]

Literature on satisfaction with clinical education is quite limited. In general, job satisfaction has been amply studied, and theories such as career theory have been used to explain how occupational and psychological variables can influence attitudes toward careers. Models based on career theory[7] or job satisfaction[8-12] typically are multifaceted in order to account for the differences among individuals. In their entirety, models that predict job or career satisfaction, in our view, are not appropriate to use with physical therapist students. Clinical experiences are not the same as jobs, and the role of a student is not the same as that of an employee. Unlike typical employees, students are located at facilities temporarily. Often students complete their experiences at clinical sites that are not those that they would choose for regular employment. Students typically are not paid and, therefore, do not have the same fights and privileges, or work expectations, as employees. Because there are some similarities between employment and clinical experiences, it seems likely that a model for predicting satisfaction with clinical experiences could be derived from combining some of the variables that predict job or career satisfaction.

A widely supported theoretical framework for examining job satisfaction is the job characteristics approach, which was developed to explain relationships between job characteristics and outcomes.[8-10] Tests of this theoretical framework have shown that there is a relationship between job characteristics and attitudinal outcomes such as satisfaction and that changes in job characteristics can influence outcomes.[8] A look at the many survey instruments available to measure job satisfaction shows that there are (1) numerous job characteristics, or facets of a job, that can be related to job satisfaction[13,14] and (2) facets of satisfaction that are common among survey instruments. The literature also supports the notion that global (overall) satisfaction is different from measures of elements of satisfaction.[11] Sums of scores on the elements of satisfaction do not translate into a global measure of satisfaction, so both overall and specific measures are needed to understand job satisfaction comprehensively.[11,15] Due to a lack of instruments available to study global and specific measures of satisfaction during clinical experiences, new measures are necessary.

Students in the health care professions generally appear to be satisfied with their clinical experiences,[5,6,16] but the elements that contribute to this satisfaction are not known. Kahn et al[12] categorized elements of job satisfaction into 3 domains: personal, interpersonal, and organizational. These 3 domains appear to be appropriate for studying satisfaction during clinical experiences. For our study, elements in the personal domain are defined as characteristics that are specific to students as individuals. Facets in the interpersonal domain are those characteristics that reflect the relationship between the student and the clinical instructor. Variables in the organizational domain are based on characteristics of clinical settings that might influence satisfaction.

The purposes of this study were (1) to assess the separate effects that independent variables from the personal, interpersonal, and organizational domains have on overall satisfaction with clinical experiences in physical therapy and (2) to compare the relative contribution each domain makes to the explanation of overall satisfaction. Overall satisfaction was the single dependent variable. Hypotheses were proposed based on job satisfaction theory, career theory, and literature on clinical education.

Hypotheses

Personal Domain

Individuals who are in their first career are more likely than people who have had other careers before physical therapy to be satisfied with their clinical education experiences.[16] People generally are attracted to environments that appear consistent with their general orientation[16] and that provide choices.[17,18] Self-efficacy is associated with choice,[19,20] adjustment to organizations,[21] and performance,[22-25] whereas performance is associated with satisfaction.[23] High levels of life satisfaction also are associated with job satisfaction.[26,27] Based on these assumptions, we hypothesized that high overall satisfaction of physical therapist students with their clinical experiences is associated with early career stage, high self-efficacy, having a choice in selecting the clinical experience, and high life satisfaction. We expected that life satisfaction would be influenced by negative and positive events occurring in the student's experiences at the clinical facility and away from the facility (Tab. 1).

Table 1.
Summary of the independent Variables Initially Hypothesized to
Predict Higher Overall Satisfaction With Clinical Experiences

Domain          Independent Variable
Personal        Students in first career
                Higher self-efficacy
                Having choice of clinical experience site
                Higher life satisfaction (including positive events
                  on-site and off-site)

Interpersonal   Higher scores of CI(a) on professional skills
                Higher scores of CI on teaching skills
                Higher scores of CI on interpersonal skills
                Higher scores of CI on communication
                More frequent meetings with CI
                Higher frequency of receiving supportive
                  feedback

Organizational  Larger and more complex staff
                Greater variety of patients
                Greater variety of skills practiced
                Greater variety of tasks practiced
                Higher numbers of learning opportunities
                More independence in decision making
                Use of institutionalized orientation methods
                  (regularly scheduled, routine, formal sessions)
                Greater consistency between actual and
                  expected clinical practice
                Greater number of CIs

(a) CI=clinical instructor.

Interpersonal Domain

There is a correlation between the overall satisfaction students have with their clinical experiences and their ratings of their instructors.[28] In physical therapy education, important characteristics of clinical instructors have been categorized into 4 dimensions: professional skills, teaching skills, interpersonal skills, and communication.[29] Satisfaction in learning environments can be increased when interactions with supervisors are nonstressful,[30,31] supportive,[32] and considerate.[33] As a result of these assumptions, we hypothesized that the overall satisfaction of students with their clinical experiences will increase as the performance of the clinical instructor increases in all dimensions of clinical teaching, as meetings are held more frequently, and as students more frequently report receiving supportive feedback from their clinical instructor (Tab. 1).

Organizational Domain

Students who have completed clinical experiences in stressful situations have reported lower levels of satisfaction. For example, stressful conditions have been identified as those in which care is more critical,[30] staff structure is more complex and interdisciplinary,[6,34] and professional skills are not used.[35] Aspects of work that may be setting-specific and that are found to increase satisfaction include autonomy,[36,37] variety,[6,37] the availability of learning opportunities,[6,31] being supervised by more than one supervisor,[6] and the use of institutionalized methods of orientation.[21] Jones[21] defined institutionalized methods of orientation as those that are (1) given to groups, (2) formal, (3) provided regularly, and (4) routine for all individuals. In contrast, innovative methods are individual, informal, and provided as needed or on request. Based on the combination of these findings, we hypothesized that overall satisfaction will increase at settings where staff size is larger and staff structure is more complex; greater variety of patients, tasks, and skills; higher numbers of learning opportunities; more independence in decision making; the use of institutionalized orientation methods; greater consistency between actual clinical practice and expectations of the student; and greater numbers of clinical instructors (Tab. 1).

Relative Contribution of Domains

Data do not demonstrate that satisfaction is influenced more by variables in one domain than by variables in another domain. For exploratory purposes, we tested the contributions that the groups of personal, interpersonal, and organizational factors made in predicting overall satisfaction. Our expectation was that there would be no difference in how much each domain, as a whole, contributed to the prediction.

Method

The study was conducted in 3 phases, with a different sample used for each phase. The preliminary phase was designed to test the survey instruments. All survey instruments were revised after this phase. Data from the preliminary phase were not included in the analysis. The purpose of the second phase was to test the revised instruments, test the procedures, and collect data. The purpose of the third phase was to collect additional data.

Subjects

The nonprobability sample consisted of students from one physical therapist education program. To vary the sample, students from 2 cohorts were studied at different times. All students had successfully completed the classroom component of the curriculum and 2 previous full-time clinical experiences on schedule, and they were in either the first or the last of 3 final full-time clinical experiences. Seven students from the class of 1993 participated in the preliminary phase of the study. Fifty-three other students from the class of 1993 (15 men, 38 women; mean age=25.3 years) who were in their fifth clinical experience were eligible for the second phase of the study. For the third phase of the study, 62 students from the class of 1994 (16 men, 46 women; mean age=25.2 years) who were in their third clinical experience participated. Students were located at facilities throughout the United States in a wide variety of clinical settings. Participation was voluntary and had no bearing on grades or academic credit. Students made their decisions to participate and signed informed consent documents at the beginning of their clinical experiences. For the second and third phases of the study, only 2 of the 115 eligible students refused to participate. Although 6 students stopped participating at early points in the study, all available data were used for analyses.

Students were enrolled in an accredited professional master's degree program at a private university in the Midwest. The program required 4 academic semesters and 32 weeks of full-time clinical experience. Four-week-long clinical experiences followed each of the second and third semesters. Three consecutive 8-week-long clinical experiences followed the final academic semester. At the time of the study, the class size was 68. When the students matriculated into the program, they had completed their baccalaureate education with a variety of majors at a large number of schools across the country. The most common undergraduate majors of the students in this study were biology and psychology. All students had completed the prerequisite courses and were accepted based on cumulative, core, and math/ science grade point averages (GPAs); Graduate Record Examination (GRE) scores; essays; and recommendations. Although the program does not set minimum standards for GPAs and GRE scores, students in the 2 classes, on admission, had average cumulative GPAs of 3.46 (SD=0.24, range= 2.66-3.96) and average GRE scores of 507 on the verbal scale (SD=84.7, range=230-730), 609 on the quantitative scale (SD=78.7, range=400-800), and 621 on the analytical scale (SD=90.8, range=390-800).

To provide a basis for judging the generalizability of the results, it was necessary to show how the subjects were like students in other programs. To compare the subjects who participated in this study with students in other physical therapist education programs, we used data obtained from a report of the American Physical Therapy Association on a large sample of physical therapist program applicants for the 1993-1994 year.[38] Although there are differences between accepted and applying students, this is the only national data set available. The comparison showed that the students in this study were not different from applicants to physical therapist programs nationwide with regard to age, gender, and academic preparation. Compared with the national data, the combined sample for this study might have underrepresented minority students. Seven minority students were available in the target sample. To protect their anonymity, students were not asked to identify race in the surveys, so actual minority representation in the study is unknown. Because the sample in this study had many similarities to a national sample of applicants to physical therapist education programs and because this program is probably like many other master's level programs, we believe it is likely that the findings can be generalized to other physical therapist students in similarly structured programs.

Students chose their final 3 clinical sites using a lottery process. Facilities were visited by faculty members on a regular basis. The Academic Coordinator of Clinical Education (ACCE) and her assistants called all students at least once during their clinical experiences. All students were visited at least once during the final 3 clinical rotations. In this program, the types of facility and caseloads where students affiliate must vary in type so that students are prepared as generalist practitioners, but certain combinations of types of facility or caseload, or particular sequences, are not required. The clinical sites affiliated with the program are selected by the ACCE and her assistants. The criteria for selection as a clinical site include the following: therapists must have an interest in clinical education, the facility must have an established student program, coordinators of clinical education and clinical instructors must have at least 1 year of experience, facilities need to be able to accommodate students on a regular basis, and opportunities for student learning must be apparent. At the time of the study, the program affiliated with approximately 250 clinical sites located throughout the United States.

Survey Instruments

Four survey instruments were designed for use in this study. The variables measured by each instrument and the times when each instrument was completed by the students are listed in Table 2. Survey questions were selected or adapted from a wide variety of sources. The development of each question and each survey is described in detail elsewhere.[39] For some components of the surveys, reliability and validity were previously tested and established. Seven research experts in survey design, organizational behavior, physical therapy, social science, social work, or psychology reviewed the survey instruments, newly developed questions, and instructions for face validity and clarity. Survey instruments were also previewed for clarity by a physical therapist student, a clinical instructor, and the ACCE. Changes in the survey instruments were made based on all feedback combined. To enhance reliability, we incorporated multiple measures of many variables into the surveys. Other checks on reliability will be described later in this article.

Table 2. Time During the Clinical Experience When the
Surveys Were Administered and the Variables That Were
Measured

Survey 1
Administered: During first 2 days
Variables:
  * self-efficacy
  * level of choice of the facility
  * number of staff in department
  * type of setting
  * student age
  * student gender
  * career (first or second or more)

Weekly Logs
Administered: Last day each week
Variables:
  * no. of meetings between CI(a) and student
  * types of orientation experienced
  * no. of nonroutine learning opportunities
  * rating of variety of skills practiced
  * rating of variety of tasks performed
  * rating of variety of patient diagnoses
  * raring of skill utilization
  * rating of autonomy in making decisions
  * rating of autonomy in implementing care
  * rating of amount of feedback received
  * indication of events at or away from the clinical site with
positive or negative influence on performance

Survey 2
Administered: Midterm
Variables:
  * overall satisfaction with clinical experience(b)
  * life satisfaction
  * satisfaction with facets of experience including autonomy,
variety, learning experiences, orientation, feedback
  * no. of CIs
  * gender of CIs evaluated
  * rating of CI performance on communication, interpersonal
skills, teaching, and professional skills

Survey 3
Administered: During last 2 days
Variables:
  * self-efficacy
  * all items included on survey 2

(a) CI=clinical instructor.

(b) Dependent variable (all others listed are independent variables).

Procedure

For all phases, the primary investigator (JSS) oriented all of the students to the project prior to beginning their clinical experiences. As shown in Table 2, multiple surveys were completed by each student throughout their 8-week-long clinical experiences. Students received their survey instruments in the mail, and they were provided with stamped, addressed envelopes for returning the survey instruments weekly. Each student also received weekly telephone calls from either the researcher or an assistant trained by the researcher to help with this process, reminding the student to complete and return the forms. All data were kept anonymous and confidential, and students were not asked if they were participating in the study. Prior to implementation of the study, all clinical instructors were informed that students were participating in this study, but the instructors had no access to any additional information and students were not to discuss the surveys with them. The ACCE and. her assistants knew about the study, but they also did not discuss the surveys when they made routine telephone calls or visits.

Data Analysis

After determining that changes in the survey instruments were not needed between the second and third phases of the study, data from the 2 groups of students were combined for analysis. Preliminary analyses of all univariate and bivariate data were conducted to determine whether the data met the assumptions for regression. Relationships between the independent variables and the dependent variable (ie, satisfaction) were plotted to determine whether they were linear. Cronbach's alpha analyses on previously established scales measuring self-efficacy and clinical instructor performance were conducted to evaluate internal consistency. Chi-square tests and t tests for independent groups were performed to assess important between-group differences in measurements taken after the first half of the clinical experience as compared with measurements taken after completion of the clinical experience.

Separate regression analyses were used to test what variables in each of the domains predict overall satisfaction. Whether it was the students' third or fifth clinical experience was indicated by one variable in all regression analyses to control for differences between the 2 groups of students. Regression analyses using repeated measures were computed to determine whether differences existed between the first half of the clinical experience and the second half of the clinical experience. Residuals were examined using plots of Student t-test values, Cook's distance estimate, and leverage. When studentized residuals were greater than 3.5 or leverage was above 0.5, cases were omitted from that specific analysis. Residuals exceeding these levels rarely occurred, and when they did, only one or two cases were omitted from the analysis. Because omitted cases were not consistently from the same subject, no subjects were dropped from the study.

The number of variables in the analyses was not the same for all domains. Thus, to compare the relative contribution of the domains as evenly weighted entities, a principal-components analysis was required. First, we performed a separate principal-components analysis to determine how many components were present in each domain. After determining that each domain contained only one component, we assigned composite scores to subjects for each domain. A regression model was then tested using satisfaction as the dependent measure and the 3 composite scores, gender, and clinical experience phase as independent variables.

Results

Testing the Assumptions

To determine whether it was appropriate, in our opinion, to combine the data from the second and third phases of study, we compared the groups and found that they did not differ on the ratios of men to women, mean age, or overall satisfaction. The total number of subjects who returned survey instruments was 113. A retrospective power analysis verified that the sample size was adequate for the analyses conducted. The combined sample did meet a power level of 0.8, based on an effect size of 0.35, which we consider large for regression analyses.[40] Of the 1,265 survey instruments and logs sent to the students, 1,214 (96%) were returned.

To verify that the elements of satisfaction chosen for study were related to the construct of overall satisfaction, Spearman rank correlations among these variables were computed. The primary. correlations between the elements and overall satisfaction are shown in Table 3. All of the independent variables were correlated with overall satisfaction, and, although the other correlations are not shown, the independent variables generally were intercorrelated.

Table 3. Spearman Rank Correlations Among Variables Measuring Overall Satisfaction and Satisfaction With Facets of the Clinical Experience (n=106)

                                   Satisfaction Satisfaction
                        Overall        With       With
                      Satisfaction   Autonomy   Feedback
Overall satisfaction    1.00           .44(a)     .40(a)

                       Satisfaction    Satisfaction   Satisfaction
                          With             With          With
                         Learning       Orientation    Utilization
                       Opportunities     Methods        of Skills

Overall satisfaction      .62(a)          .54(a)       .62(a)

                         Satisfaction
                             With
                            Variety
Overall satisfaction         .61(a)

(a) Significant at [infinity] <05 (Spearman 2-tailed test critical value=.197).

Regression analyses verified that the elements were related to satisfaction and showed that the sum of the elements of satisfaction did not explain overall satisfaction. Main effects ([R.sup.2] =.648; F=22.09; df =8,96; P [is less than] .01) and interaction effects ([R.sup.2] =.673; F= 12.21; df = 15,89; P [is less than] .01) were significant. The elements of satisfaction that predicted overall satisfaction included the students' satisfaction with the feedback, the number of learning opportunities available, the orientation provided, and the variety of patients, skills practiced, and tasks performed.

Although no differences based on gender were hypothesized initially, we selectively tested the influence of gender prior to testing the hypotheses due to lack of support in the literature for such differences. The first test showed that women reported higher levels of overall satisfaction than men did. With the maximum possible score being 9, the mean score for the women was 7.6, and the mean score for the men was 6.9 (t =2.69, df=105, P [is less than] .01). The women were younger than the men. Mean age was 24.9 years for the women and 26.2 years for the men (t=2.54, df=112, P [is less than] .01). Men and women did not differ on life satisfaction, self-efficacy, or frequency of reports about negative or positive events reported during the clinical experiences (all t [is less than or equal to] |.88|.) Although choice of clinical settings was achieved through a lottery process, women were assigned to their first choice of clinical setting more often than men. Women obtained their first choice 50.6% of the time, and men obtained it only 41.9% of the time ([chi square] = 70.2, df =4, P [is less than] .001). Women rated their clinical instructors higher than men did. With the maximum possible score being 205, clinical instructors were rated on the average 168.4 by the women and 161.0 by the men (t = 1.96, df=93, P=.05). These unexplained differences were enough to justify statistical control of gender in the final analysis in the study by including gender and the interaction of gender with the independent variables in all regression equations. A variable indicating the match of the students' gender with the clinical instructor was added into the regression analyses testing how well the variables in the interpersonal domain predicted overall satisfaction. Because of the difference in age between men and women, age was included as a control variable in the regression analysis testing the prediction of overall satisfaction by variables in the personal domain.

The analyses were conducted on the average of each variable for the data collected during the 8-week-long clinical experiences. The t tests for paired samples, however, showed that the students' overall satisfaction was higher during the second half of the clinical experience ([bar] X = 7.53) than during the first half of the clinical experience ([bar] X = 7.26) (t = 2.47, df= 106, P [is less than] .02). Regression analyses using repeated measures, therefore, were performed to test other differences between the first and second halves of the clinical experiences. In all except one analysis, there were no other differences between the first and second halves of the clinical experiences. In that analysis, gender and variety interacted differently between the first half of the clinical experience and the second half of the clinical experience. Together, however, they did not contribute to the explanation of overall satisfaction.

Testing the Hypotheses

Personal domain. In the personal domain, overall satisfaction was predicted by the main effects regression model ([R.sup.2] = .415; F = 3.68; df = 15,78; P [is less than] .01) and by the interaction effects model ([R.sup.2] = .586; F = 3.12; df = 29,64; P [is less than] .01). Higher overall satisfaction was noted when students reported fewer negative events in the clinical setting and outside the clinical setting, and when they reported higher life satisfaction. An interaction between gender and reporting of negative events at the clinical setting was found. Specifically, when events were perceived as negatively affecting their performance, the women reported lower overall satisfaction at a higher rate than the men did.

Variables in the personal domain that did not contribute to the explanation of overall satisfaction included career stage, age, choice of setting, the occurrence of positive events on-site or off-site, and self-efficacy. The non-outcome self-efficacy measure was a series of 8 items reflecting general feeling of self-efficacy but not specifically related to performance. The result of the Cronbach's alpha test of this scale's internal consistency was .67, which does not meet the .80 level that we consider indicative of good internal consistency.

Interpersonal domain. Prior to testing how much the variables in the interpersonal domain contributed to the explanation of satisfaction, the internal consistency of the scales measuring clinical instructor performance was evaluated. The overall reliability was .87 for the scale on communication skills, .88 for the scale on interpersonal skills, .84 for the scale on professional skills, and .93 for the scale on teaching skills. The variables in the interpersonal domain generally were intercorrelated.

Satisfaction was predicted by variables in the interpersonal domain using the main effects model ([R.sup.2] = .526; F = 10.25; df = 9,83; P [is less than] .01) and the interaction effects model ([R.sup.2] = .574; F = 5.95; df = 17,75; P [is less than] .01). In the main effects model, satisfaction was best explained by the students' reports of the clinical instructor's teaching skills and student gender. In the interaction effects model, the interaction between clinical experience phase and gender best contributed to the explanation of satisfaction. This interaction showed that mean overall satisfaction for women was higher during the fifth clinical experience than during the third clinical experience, whereas mean overall satisfaction for men was higher during the third clinical experience than during the fifth clinical experience. Variables in the interpersonal domain that did not contribute to the prediction of overall satisfaction were the number of meetings between the student and the clinical instructor, the amount of feedback given by the clinical instructor to the student; the interpersonal, communication, and professional skills of the clinical instructor; and the match of gender between the clinical instructor and the student.

Organizational domain. The use of innovative orientation methods by clinical instructors (informal, provided on request, individualized) was reported much more frequently than the use of institutionalized methods (formal, group, regularly scheduled, routine). Ninety-two percent of the students reported that orientation methods were most frequently provided on an individual, informal, or as-needed basis. Due to this near-uniform use of innovative methods and the resultant lack of variation in the data on innovative orientation methods, 3 of 4 innovative methods had to be omitted from the regression equation. Only "orientation provided on request" was included as an independent variable representing innovative orientation methods. Among the variables in the organizational domain, few intercorrelations were found.

The regression analyses showed that, in the organizational model, satisfaction was best predicted by the lack of use of routine orientation; increased variety of patients, tasks, and skills; student gender; and the interaction between clinical experience phase and student gender. Results of the analyses were R2 =.328; F=3.00; df=14,86; P [is less than] .01 for the main effects model and R2 =.418; F=2.04; df=26,74; P [is less than] .01 for the interaction effects model. Variables that did not contribute to the overall explanation of satisfaction included student autonomy, number of instructors, complexity and number of staff, number of learning opportunities, and all of the institutionalized orientation methods.

Relative contribution of domains. After the principal-components analysis showed that each domain contained only one factor, we conducted the regression analyses to test the relative contribution of each domain to the explanation of satisfaction. Only main effects were found to be significant, although inclusion of the interaction effects in the regression equation improved the explanation of overall satisfaction. Both the main effects ([R.sup.2] = .531; F = 17.40; df = 5,77; P [is less than] .01) and interaction effects ([R.sup.2] = .556; F = 10.77; df = 9,73; P [is less than] .01) were significant. Only the component representing the interpersonal domain and student gender contributed to the explanation of overall satisfaction.

Summary of Results

In the tests of the models representing the personal, interpersonal, and organizational domains, overall satisfaction was predicted in each one. Table 4 summarizes the variables that predicted overall satisfaction when the domains were tested separately and the amount of variance that was explained by each combination of variables. With the domains being tested for differences in how they contributed to predicting overall satisfaction as entities rather than groups of variables, the interpersonal domain was the only one to achieve statistical significance. Student gender predicted overall satisfaction in all of the regression analyses.

With only one exception, there were no differences among the variables that predicted overall satisfaction when measurements taken after the first half of the clinical experience were compared with measurements taken at the end of the clinical experience. There were also no differences in the explanation of satisfaction when the students who were in their third full-time clinical experience were compared with a different group of students who were in their fifth full-time clinical experience.

Discussion

Students, in general, were satisfied with their clinician experiences, and this finding is consistent with the literature in the health care professions.[5,6-15] What was perhaps most unexpected was the influence of gender on satisfaction during clinical experiences.

Elements of Satisfaction

To some extent, all of the elements of satisfaction were associated with overall satisfaction, and the direction of their influence was consistent with what has been reported in the literature.[8] Results support the notion that in clinical education, as in employment, overall satisfaction is not equivalent to the sum of elements of satisfaction studied. A significant amount of variance was explained. Overall satisfaction was predicted by the intrinsic elements of satisfaction regardless of when the measurements were made during the clinical education sequence. This finding suggests that, over time, such intrinsic factors contribute equally to the explanation of satisfaction throughout the clinical experience, at least within the short period of 8 weeks.

We did not expect that the sum of the elements of satisfaction would explain all of the variance, especially because we studied only intrinsic elements of satisfaction. Additional variance remains to be explained. It would be interesting in the future to determine whether satisfaction is influenced by extrinsic rewards such as the grades students receive or the success they achieve with their patients. Other intrinsic factors that may be important to examine are the personal satisfaction that students derive from interactions with their patients and the praise or criticism that students receive from their patients.

Personal Domain

Overall student satisfaction with the clinical experience appears to be predicted by life satisfaction, events involving the student both on-site and off-site, and the interaction of the student's gender with the report of negative events. That life satisfaction predicts satisfaction with the clinical experience is consistent with the job satisfaction literature.[26,27] Why the occurrence of negative events influenced overall satisfaction or whether they are related to life satisfaction was not determined in this study. It was interesting to read what negative or positive events students considered serious enough to influence their performance. The finding that women perceived their performance to be influenced by negative events to a greater extent than men did might be explained in future studies with closer analyses of the types of events being reported. Interestingly, although the specific results are not reported in this article, on-site events were reported by students to influence performance more than off-site events, and negative events were reported more often than positive events. Examples of on-site events included reports such as "I was given more independence this week" and "I changed clinical instructors, and I feel like I'm starting all over again." Examples of off-site events that were reported include "I have special plans for the weekend" and "My housing situation is uncomfortable." Explanations for such findings are beyond the scope of this report. Better classification of events and ratings of their importance to students might yield more information. If negative events can be prevented or students can learn to handle such events more effectively, they may not perceive them to have as great an influence on performance and satisfaction may not be affected to such an extent. Another question that remains unanswered is why students reported lower life satisfaction scores at the end of each 8-week-long clinical experience than at the beginning.

Variables in the personal domain predicted student satisfaction at midterm, at the end of the clinical experience, and across the clinical experience. The amount of variance did not differ when the midterm and final periods were compared, in that more variance is explained at midterm than at the end of the clinical experience. Whether these findings indicate that the variables in the personal domain influence students more during the initial part of the clinical experience than later remains unclear.

Interpersonal Domain

The variables in the interpersonal domain as a group predicted overall student satisfaction. The influence of the individual variables, however, is difficult to interpret due, in large part, to the intercorrelations among the independent variables. The most reliable scale (ie, on teaching skills) was the best predictor of student satisfaction with the clinical instructor's performance.

The total amount of variance explained by the variables in the interpersonal domain differed for the data at the 2 different periods compared with that for the averaged data. The amount of variance explained after the second half of the clinical experience was more than that after the first half of the clinical experience. We were unable to determine whether this finding means that the interpersonal domain had more influence on the students during the second half of their clinical experience than during the first half of their clinical experience.

Organizational Domain

Surprisingly few variables in the organizational domain predicted overall satisfaction. When students reported greater amounts of variety, overall satisfaction increased. When orientation methods were more routine, satisfaction decreased. Perhaps we erred in identifying the 9 orientation methods as individual variables rather than categorizing them into either institutionalized or innovative methods. Entering fewer variables in the regression equation may have raised the amount of variance explained by the variables in the organizational domain that predicted overall satisfaction. Having fewer variables entered into the regression equation would also decrease the risk of Type I or II errors.[40] One possible explanation for the lack of effects on student satisfaction by variables from the organizational domain may be that organizational factors have more influence on job satisfaction, than on satisfaction with clinical experiences, due to the permanency of a work setting.

Overall, the variables in the organizational domain explain less of the variance than do the groups of variables in the personal and interpersonal domains. A direct comparison of the coefficient values, however, cannot be made from model to model due, in part, to the different number of independent variables entered into each model.

Relative Contribution of Domains

There are differences in the way in which the domains contribute to explaining overall satisfaction. As a group, the domains predict the dependent variable. The interpersonal domain appears to have the greatest influence on this prediction, and we believe that this finding may come as no surprise to academic or clinical faculty who believe that it is the clinical instructor who "makes the experience" regardless of the type of facility being used and where. The influence of the interpersonal domain on overall student satisfaction may support the notion of training or credentialing clinical instructors rather than clinical sites.

The relationships of the domains apparently do not vary according to the phase of clinical education or the time during a single clinical experience, at least during the final experiences that succeed academic work. This finding is reassuring, as academic and clinical instructors can begin to plan more satisfying clinical experiences without having to consider whether it is a student's initial clinical experience or his or her last clinical experience. Once the personal, interpersonal, and organizational features are in place, they can be used consistently for a variety of students.

Gender Issues

We did not expect to find gender differences because the job satisfaction literature we reviewed does not support differences between men and women. Nonetheless, gender contributed to the explanation of overall satisfaction in every regression model tested for each domain and for the domains together. An interaction between gender and clinical phase also influenced the explanation of satisfaction when included in the equations testing the interpersonal and organizational domains. Women were more satisfied during the fifth clinical experience relative to the third clinical experience, whereas men were more satisfied during the earlier clinical experience relative to the last clinical experience. We tested one proposed explanation of why women were more satisfied than men: that it might be due to women being more frequently supervised by someone of the same gender. Gender match between the clinical instructor and student, however, was not a variable that contributed to explaining overall satisfaction. In general, differences related to gender cannot be explained by our results. Further study of gender issues is needed. As differences in student satisfaction between men and women are better understood, clinical experiences might be adapted to meet their needs more specifically.

Limitations of the Study

Restricting the study to students from one physical therapist education program limits the generalizability of the data. It also, however, allowed for control of some factors that might have influenced student satisfaction with their clinical experiences, such as academic preparation prior to going into the clinic, events that occurred during the time in which the clinical experiences were completed, and mechanisms by which clinical sites are chosen and supervised by academic institutions. We made an attempt at varying the sample by including students from 2 separate cohorts and completing the study at 2 different times. This study was also limited to one program for practical and cost purposes. As described in the "Method" section, the sample was compared with a national sample of applicants to physical therapist education programs. That differences were not found between our sample and the national sample gives us some assurance that our sample is representative of other physical therapist students. Whether the results can be generalized to other professions is questionable.

The use of common methods such as self-report may lead to measurement errors.[40] To standardize and control measurement error in this study, multiple measurements were taken longitudinally' operational definitions were clear, and hypotheses were directional when appropriate. In planning and implementing the study, we considered the structure of the surveys, the instructions to the subjects, the influence of administering the surveys relatively close together in time, and the internal consistency of selected scales.

Conclusions

As discussed in this article, there are similarities between what explains student satisfaction with clinical education and job satisfaction. Variables from the personal, interpersonal, and organizational domains all contribute to the explanation of satisfaction in clinical education, though interpersonal factors appear to be more influential than variables in the other domains. What contributes; to satisfaction does not appear to be related to the sequencing or timing of clinical education, and gender appears to have a more significant effect in explaining student satisfaction than was predicted by the job satisfaction literature we reviewed.

Future research is needed to test causal relationships among the variables that can be manipulated during clinical education. Larger and more varied samples should be studied to demonstrate the generalizability of the findings. Before changes for improving student satisfaction are implemented, however, the relationship between student satisfaction and performance needs to be understood. Finally, relationships among student satisfaction, performance, career commitment, and job satisfaction also need further exploration to understand the implications for the profession.

Acknowledgments

The primary investigator would like to thank former physical therapist students Jennifer Nicewanner and Audra Robertson, who assisted in implementing the procedures for this study.

References

[1] Arnkoff DB, Glass CR. Clinical cognitive constructs: examination, evaluation, and elaboration. Advances in Cognitive Behavior Research and Therapy. 1982;1:1-33.

[2] Cherniss C. Career commitment in human service professionals: a biographical study. Human Relations. 1991;44:419-437.

[3] Blau GJ. Further exploring the meaning and measurement of career commitment. Journal of Vocational Behavior. 1988;32:284-297.

[4] Welsch HP, La Van H. Inter-relationships between organizational commitment and job characteristics, job satisfaction, professional behavior, and organizational climate. Human Relations. 1981;34: 1079-1089.

[5] Shelley RK, Webb MG. Does clinical clerkship alter students' attitudes to a career choice of psychiatry? Med Educ. 1986;20:330-334.

[6] Showers N. Hospital graduate social work field work programs: a study in New York City. Health Soc Work 1990;15:55-63.

[7] Sonnenfeld J, Kotter JP. The maturation of career theory. Human Relations. 1982;35:19-46.

[8] Glick WH, Jenkins GD, Gupta N. Method versus substance: How strong are underlying relationships between job characteristics and attitudinal outcomes? Academy of Management Journal. 1986;29:441-464.

[9] Hackman JR, Lawler EE. Employee reactions to job characteristics. J Appl Psychol. 1971;55:259-286.

[10] Hackman JR, Oldham GR. Development of the job diagnostic survey. J Appl Psychol. 1975;60:159-170.

[11] Scarpello V, Campbell JP. Job satisfaction: Are all the parts there? Personnel Psychology. 1983;36:577-600.

[12] Kahn R, Wolfe D, Quinn R. Organizational Stress: Studio in Rate Conflict and Ambiguity. New York, NY: John Wiley & Sons; 1964.

[13] Survey Item Bank: Measures of Satisfaction. Vols 1 and 2. Bradford, West Yorkshire, England: MCB University Press Ltd; 1984.

[14] Cook RE, Hepworth SJ, Wall TD, Warr PB. The Experience of Work New York, NY: Academic Press Inc; 1981.

[15] Bolton B. Minnesota Satisfaction Questionnaire. In: Keysor DJ, Sweetland RC, eds. Test Critiques. Kansas City, Mo: Test Corporation of America; 1984.

[16] Schmoll B. Adult development theory in the education of a physical therapist. In: Leadership for Change in Physical Therapy Clinical Education. Alexandria, Va: American Physical Therapy Association; 1986:79-94.

[17] Festinger L. A Theory of Cognitive Dissonance. Evanston, Ill: Row-Peterson; 1957.

[18] Brehm JW, Cohen AR. Explorations in Cognitive Dissonance. New York, NY: John Wiley & Sons; 1962.

[19] Brief AP, Aldag RJ. The "self" in work organizations: a conceptual review. Academy of Management Review. 1981;6(1):75-88.

[20] Wood R, Bandura A. Impact of conceptions of ability on self-regulator mechanisms and complex decision-making. J Pers Soc Psychol. 1989;56:407-415.

[21] Jones G. Socialization tactics, self-efficacy, and newcomers' adjustments to organizations. Academy of Management Journal. 1986;29: 262-279.

[22] Bandura A, Schunk DH. Cultivating competence, self-efficacy, and intrinsic interest through proximal self-motivation. J Pers Soc Psychol. 1981;41:586-598.

[23] Locke EA, Latham GP. Work motivation: light at the end of the tunnel. Psychological Science. 1990;1:240-246.

[24] Martin JJ, Gill D. The relationships among competitive orientation, sport-confidence, self-efficacy, anxiety, and performance. Journal of Sport and Exercise Psychology. 1991;13:149-159.

[25] Ellen PS, Beardon WO, Sharma S. Resistance to technological innovations: an examination of the role of self-efficacy and performance satisfaction. Journal of the Academy of Marketing Science. 1991;19:297-307.

[26] Tait M, Padgett MY, Baldwin TT. Job and life satisfaction: a reevaluation of the strength of the relationship and gender effects as a function of the date of study. J Appl Psychol. 1989;74:502-507.

[27] Rain JS, Lane IM, Steiner DD. A current look at the job satisfaction/life satisfaction relationship: review and future considerations. Human Relations. 1991,44:287-307.

[28] Sinnicrope RF, Cournoyer DE. Validity of student ratings in field instruction. Journal of Social Work Education. 1990;26:266-272.

[29] Emery MJ. Effectiveness of the clinical instructor: students' perspective. Phys Ther. 1984;64:1079-1083

[30] Dodds AE, Lawrence JA, Wearing AJ. What makes nursing satisfying: 2 comparison of college students' and registered nurses' views. J Adv Nurs. 1991;16:741-753.

[31] Gelmon SB, Williams JI. Perspectives on the physiotherapy profession: a survey of Canadian Physiotherapy Association members. Physiotherapy, Canada. 1983;35:21-29.

[32] McCloskey JC, McCain BE. Satisfaction, commitment, and professionalism of newly employed nurses. Journal of Nursing Scholarship. 1987;19-20-24.

[33] Schriesheim CA, Murphy CJ. Relationships between leader behavior and subordinate satisfaction and performance: a test of some situational moderators. J Appl Psychol. 1976;61:634-641.

[34] Fortune AE, Feathers C, Rook S, et al. Student satisfaction with field placement. Journal of Social Work Education. 1985;3:92-104.

[35] Brutvan EL. Intra-role conflict: a result of naive attempts toward professionalization. J Allied Health. February 1985;14:3-11.

[36] Hanson CM, Jenkins S. Ryan R. Factors related to job satisfaction and autonomy as correlates of potential job retention for rural nurses. Journal of Rural Health. 1990;6:302-316.

[37] Price JL, Mueller CW. A causal model of turnover for nurses. Academy of Management Journal. 1981;24:543-565.

[38] 1993 Applicant Report: A Description of Applicants to Physical Therapy Entry-Level Education Programs for the 1993-1994 Academic Year. Alexandria, Va. American Physical Therapy Association, Division of Research, Analysis, and Development; 1993.

[39] Stith JS. Personal, Interpersonal, and Organizational Factors Influencing Student Satisfaction With Professional Clinical Education Experiences. St Louis, Mo: Washington University; 1994. Dissertation.

[40] Cohen J, Cohen P. Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences. Hillsdale, NJ: Lawrence Erlbaum Associates Inc; 1983.

JS Stith, PhD, PT, is Instructor and Associate Director for Professional Education, Program in Physical Therapy, Washington University School of Medicine, 4444 Forest Park Ave, Box 8502, St Louis, MO 63108 (USA) ([e-mail]). This study was completed in partial fulfillment of the requirements for Dr Stith's doctoral degree at Washington University. Address all correspondence to Dr Stith.

WH Butterfield, PhD, is Professor Emeritus, George Warren Brown School of Social Work, Washington University, St Louis, Mo.

Mj Strube, PhD, is Professor, Department of Psychology, Washington University.

SS Deusinger, PhD, PT, is Assistant Professor and Director, Program in Physical Therapy, Washington University School of Medicine.

DF Gillespie, PhD, is Professor, George Warren Brown School of Social Work, Washington University.

The study protocol was approved by the Human Studies Committee at Washington University School of Medicine.

This work was partially supported by a doctoral research award from the Foundation for Physical Therapy Inc.

The project was presented as a poster at the joint Congress of the Canadian Physiotherapy Association and the American Physical Therapy Association, June 4-8, 1994, Toronto, Ontario, Canada.

This article was submitted September 3, 1996, and was accepted November 18, 1997.