2.1 Study Samples
From April to May, 2020, general practitioners from primary medical institutions in Jiangsu Province were investigated. Located in southeast coast of China, Jiangsu province has the second highest GDP among all the provinces in mainland China. Jiangsu province has 13 cities [19, 20]. In recent years, Jiangsu province has attached great importance to the establishment of the general practitioner system and issued various documents on the incentive mechanism of general practitioners for many times.
In the design of the study, logistic regression was intended to use to analyze the incentive effect, so the sample size was determined to be about 50 times of the independent variable, thus 900 general practitioners working in primary medical institutions in Jiangsu province were expected to be investigated [21]. From April to May 2020, 10 primary medical institutions were randomly selected from each of the 13 cities divided into districts in Jiangsu Province to conduct a survey. The respondents were all general practitioners who had worked in the hospital for at least 1 year.
2.2 Variable Definition
The self-made questionnaire was used in this survey(questionnaire is shown in appendix). The measurement of incentive effect including job satisfaction, willingness to work hard and turnover intention. The indicator of incentive effect pays attention to general practitioners' own motivation feeling, avoiding the use of performance alone, which is easily affected by multiple factors. Job satisfaction was measured by Cammann job satisfaction Scale (5 questions, Cronbach’s a = 0.740). Willingness to work hard was measured by TRAAM scale (4 questions, Cronbach’s a = 0. 905), and turnover intention was measured by Mobley turnover intention scale (5 questions, Cronbach’s a = 0.822) [22–24]. The job satisfaction scale, willingness to work hard and turnover intention scale are all Richter five scales. The overall incentive effect was measured from three dimensions: job satisfaction, willingness to work hard and turnover intention. The total Cronbach’s a of the incentive effect scale was 0.910.
The study assumed that factors that may influence the incentive effect include: age, gender, marital status, educational background, years of working as a general practitioner, average monthly income, weekly working hours, daily sleeping duration, and physical exercise frequency. Age is a continuous variable. We classified Gender as being either "male" or "female", marital status was classified as "unmarried", "married" or "widowed/divorced, etc.". Educational background was classified as "junior college", "Undergraduate", "Master" or "Doctor". The years of working as a general practitioner was classified as "5 years and below", "6–10 years", "11–20 years" or "21 years and above". Average monthly income was classified as "3,000 Chinese yuan and below", " 3,001–4,500 Chinese yuan ", " 4,501-6,000 Chinese yuan " or " 6,001 Chinese yuan and above". Weekly working hours was classified as "40 hours and below ", "40.1–45 hours", "45.1–50 hours" or "50.1 hours and above". Daily sleeping duration was classified as "6 hours or less", "6.1-7 hours", "7.1-8 hours" or "8.1 hours or more". Physical exercise frequency was classified as "once a month or no", "several times a month", "several times a week" or "every day".
2.3 Data Collection
We adopted the method of chance sampling, and the questionnaire was filled by electronic link. 7 general practitioners (130*7 = 910 general practitioners) were investigated in each hospital, and the number of rejected general practitioners was counted. A total of 837 valid questionnaires were recovered, with an effective recovery rate of 91.98%.
2.4 Statistical Analysis
After collecting the questionnaire data and recalculating the reverse scoring questions, the reliability of each dimension of incentive effect and incentive effect scale was calculated. Then, the scores of each dimension are calculated to calculate the average scores of job satisfaction (5 questions), willingness to work hard (4 questions), turnover intention (5 questions) and overall incentive effect (14 questions in total). The outcome variable with each mean score ≥ 3 points was defined as 1, and the outcome variable with < 3 points was defined as 0.
Independent variable, categorical variable was assigned, such as gender (female = 0, m = 1), marital status (single = 1, married = 2 widowed, divorced = 3). Rank variable such as educational background, years of working as a general practitioner, average monthly income, weekly working hours, daily sleeping duration, and physical exercise frequency coded in the order level.
After completion of coding, forward: LR method was used for binary logistic regression analysis. OR value and the 95% confidence interval between OR value was calculated.