Purpose and study framework
The purpose of the study was to examine the effect of self-employed physicians’ perceptions of economic exchange (distributive justice) and noneconomic exchange (perceived organizational support) with the hospital they practice at on their customer-oriented boundary-spanning behaviours (COBSBs). More precisely, we study the external representation (ER), internal influence (II), and service delivery (SD). In addition, we study the moderating effects of physicians’ professional and organizational identification on the relationships between perceived organizational support, distributive justice, and COBSBs (ER, II, and SD). The conceptual framework is based on social exchange theory. Figure
1 provides an overview of the study framework.
Research question 1: Is economic exchange (distributive justice) positively related to physicians’ COBSBs (external representation, internal influence, and service delivery)?
Research question 2: Is noneconomic exchange (perceived organizational support) positively related to physicians’ COBSBs (external representation, internal influence, and service delivery)?
Research question 3: Does organizational identification positively moderate the positive influence distributive justice and perceived organizational support on physicians’ COBSBs (external representation, internal influence, and service delivery)?
Research question 4: Does professional identification positively moderate the positive influence of distributive justice and perceived organizational support on physicians’ COBSBs (external representation, internal influence, and service delivery)?
Study design
Data were collected from a survey of self-employed physicians practicing at a convenience sample of six hospitals in Flanders (Belgium). The physicians were invited (and two times reminded) by their CMO to participate in the online survey. The invitation included a letter explaining that this study of Ghent University was supported by the Flemish association of head physicians, the medical board, and executive team of the hospital in which they practice. A concise explanation of the study aim was also included. Out of the 761 physicians from six hospitals in Flanders who were invited to participate, 180 physicians completed the online survey (initial response rate = 27%). After checking the results for missing values, the final sample consisted of 130 physicians. Participation in the study was voluntary and anonymous. The medical ethics committee of the University Hospital of Ghent approved the study.
Setting
Belgian physician-specialists practice prevailingly as self-employed professionals. From a financial point of view, physicians have therefore a distinct revenue stream. The hospital is reimbursed for the operating expenses (non-medical activity) by a hospital budget. This budget covers the hotel costs, cost of nursing, etc. The physician is entitled a medical fee for the medical activities, mainly reimbursed by fee for service. However, notwithstanding physicians operate as self-employed practitioners, they need the organizational support that enables them to practice medicine. To cover these costs, a negotiation takes place to determine the share of fees that should be transferred to the hospital (a contract governing the financial relationship).
Measures
The survey was collated from previously published instruments, which have demonstrated sound psychometric properties in past research. All question items were translated to Dutch and then back translated in order to ensure that the meaning had been retained—for which three independent translators were used. We used a five-point Likert-type scale (1 = strongly disagree; 2 = disagree; 3 = neither agree nor disagree; 4 = agree; and 5 = strongly agree). Following previous research, the items were aggregated to create a scale score.
Physicians’ perceptions of distributive justice were measured by the four-item scale developed and validated by
[35]. The items explicitly addressed the financial, economic relationship. Two sample items are as follows: “Does your financial agreement with the hospital reflects the effort you have put into your work?” and “Is your contract appropriate for the work you have completed?” The internal consistency of the instrument was sufficiently high (Cronbach’s α = 0.93).
Physicians’ perceived organizational support was assessed using the eight-item scale of Eisenberger et al.
[11] (1986). Cronbach’s α for this was satisfactory (0.94). Two sample items are as follows: “Help is available from my organization when I have a problem” and “My organization is willing to help me, if I need a special favour”.
We measured physicians’ customer-oriented behaviours using a six-item shorted version of the scale of Bettencourt and Brown
[8]. We used a short version in order to limit the length of our questionnaire. For each type of COBSB, we used two items. Sample items are as follows: “I encourage friends and family to come to this hospital for its products and services” (external representation); “I encourage other coworkers to contribute ideas and suggestions to improve services” (internal influence); and “I take time to understand patients’ needs on an individual basis” (service delivery). The internal reliability for the three scales was acceptable (respectively, α = 0.81, α = 0.94, and α = 0.80). A confirmatory factor analysis (principal component analysis with oblique rotation) extracted three factors with an eigenvalue greater than 1, which corresponded with the three forms of customer-oriented behaviour. A total of 88.3% of variance was explained by the three factors.
The extent to which self-employed physicians identified with the hospital was measured using the five-item scale of Mael and Ahforth
[29]. A sample item is as follows: “When someone criticizes the hospital, it feels like a personal insult”. The internal consistency was acceptable (Cronbach’s α = 0.83). Following Hekman et al.
[15], the extent to which physicians identified with the medical profession was measured with the same basic items and rating scale used to measure OI, but with all references to the hospital changed to “medicine” or synonyms. A sample item is as follows: “When I talk about physicians, I usually say ‘we’ instead of ‘they’”. Cronbach’s α was acceptable (α = 0.80).
Control variables
A demographic questionnaire was included to obtain descriptive information. Gender, age, tenure, and profession (surgery or internal medicine) were included to rule out potential alternative explanations for our findings. Previous research has suggested that these variables are important to social exchange
[15].
Analysis
The Statistical Package for the Social Sciences (SPSS), version 22 for Windows, was used to conduct descriptive and statistical analyses. Correlation analyses were performed to assess possible multicollinearity between the control variables. The age and tenure variables correlated highly (r = 0.843), and so age was not used as a control variable. In addition, since profession (surgery or internal medicine) did not correlate with the dependent, independent, and moderating variables, this control was not included. Descriptive statistics were used to describe the sample and study variables. Confirmatory factor analyses confirmed the distinctiveness of the measures used in this study. Pearson correlation analysis was used to test whether the variables were related. To test our research questions and analyses, the data underwent hierarchical regression analysis. To avoid multicollinearity, the independent variables were centred
[36]. The first step of the analysis involved entering the control variables, gender, and organizational tenure into the model. In the second step, the centred independent variables were added, and the centred moderating variables were then entered. Having multiplied the centred independent variables by the centred moderators, these two-way interaction terms were entered, while controlling for their main effects and the control variables (gender and organizational tenure). Following Bal et al.
[37], we argue that interaction effects may be more difficult to detect (especially in field studies), and so an alpha level of 0.10 was used to estimate interaction effects
[38]. To understand the form of these interactions, we plotted the regression lines at 1 standard deviation below and 1 standard deviation above the median.