“Motivation refers to processes that instigate and sustain goal-directed activities” (Schunk and DiBenedetto, 2020). Employees’ work-related motivation is expected to represent their self-imposed intentions and demands within their own work environment (Locke and Latham, 2004), and to determine the direction, intensity, and duration of an individual’s work (Van Iddekinge et al., 2018). Intrinsic motivation means that employees see work as a reward in itself, and are able to decide for themselves how they want to act and choose what they want to do to achieve their goals (Chen et al., 2018). As intrinsic motivation refers to whether workers get what they want out of their work, whether it enhances their experience, and whether they feel a sense of enjoyment and accomplishment in the process (Kanfer et al., 2013), in this study, we focus specifically on employees’ intrinsic motivation.

Self-efficacy should be construed as a context-specific ability concept (To et al., 2020). Tierney and Farmer (2002) proposed the idea of CSE based on the study of self-efficacy and the theory of creativity, and argued that unlike other self-efficacies that include emotions such as self-esteem and self-confidence in a broad sense, CSE is more oriented toward the ability to evaluate one’s performance of creative activities. When employees are engaged in work that they are interested in and comfortable with, the more confident they are in their abilities, the more they can develop different and effective approaches, and the more active they are in solving the difficulties and dealing with the uncertain risks they encounter during the innovation process, resulting in higher creative performance (Gong et al., 2019). How CSE affects individuals’ ability or confidence they may have in their intentions to carry out a certain conduct has been well researched (e.g., Liu et al., 2016; Newman et al., 2018; Tantawy et al., 2021). In this study, we adopted a slightly different definition of CSE that is more inclined to the self-concept of ability of creative activities in ITI.

Knowledge sharing is a process of transferring knowledge by any means, and generating new knowledge by interacting with others or by discussing the knowledge known to the individual with the knowledge transfer (Wijnhoven, 1998), which can make knowledge available to others in the organization (Bavik et al., 2018). Drawing on the social cognitive theory (SCT) (Bandura, 1986), collective knowledge sharing encourages the development of knowledge as employees engage in creative processes for individual work that is an essential element influencing creative behavior and outcomes (Tantawy et al., 2021). Knowledge sharing must be based on trust between employees and mutual discussion in order to spread knowledge. On the other hand, “knowledge hiding” refers to an individual’s intentional efforts to withhold or conceal knowledge requested by another (Pereira and Mohiya, 2021). Knowledge sharing actions that are ‘goal-directed’ are controlled by the desire to enhance individual ability in creative work (Sun et al., 2022). Four types of knowledge sharing were developed in this study, and the role that those knowledge sharing processes play in ITI employees’ willingness to share or hide knowledge was explored.

According to the motivation-action-outcome (MAO) model (Abiero and Bradfield, 2021), the action captures procedural processes (i.e., action selection) and motivation guides how to perform an action, and is essential to the evaluation of the outcome (Sun et al., 2022). In line with this model, the present study compiled intrinsic motivation, four types of knowledge sharing in ITI, and CSE development in ITI, and proposed a research framework and research hypotheses based on the relationships among the variables as shown in Figure 1.

Motivational processes are those personal or internal processes that result in actions such as choice, effort, or persistence (Schunk and DiBenedetto, 2020). Self-determination theory (Ryan and Deci, 2000) conceptualizes intrinsic motivation as an autonomous form of motivation, which is represented by distinct behavioral regulations (Courtney et al., 2021). When their intrinsic motivation is high, employees are able to enjoy the process of performing their work tasks, and their behavior will be determined less by the external work environment and more by their personal behavior (Llopis and Foss, 2016). People with intrinsic motivation are more willing to actively share their knowledge, and in the process of sharing that knowledge, they will gain more recognition and satisfaction (Kao, 2012; Shen and Chang, 2018). For example, Wang and Hou (2015) drew on the self-determination theory to examine a model of the influence of motivations on employees’ knowledge sharing behaviors, and found that the correlation was positive. However, few studies have attempted to understand how employees’ intrinsic motivation predicts the four types of knowledge sharing behavior; thus, the present study proposed four hypotheses to test the predictive power as follows:

Yoon and Han (2018) showed that knowledge sharing is higher when employees are more innovative. Sharing knowledge can encourage employees to generate novel ideas which can then promote their individual creativity (Jyoti and Dev, 2015). Hong et al. (2004) investigated the relationship between knowledge sharing and organizational climate among R&D employees, which was guided by primary relationship norms to explore the knowledge sharing attitudes, and found that cognitive environmental evaluation could drive creative self-efficacy. For example, knowledge sharing behaviors in online knowledge communities that guides intellectual stimulation for employees in which their novel thinking can be triggered, thus leading to innovative solutions (Tantawy et al., 2021). Based on the above, this study aimed to understand whether ITI employees with the four types of higher knowledge sharing behavior had higher CSE. We therefore proposed the following hypotheses:

Liu and Zhang (2007) showed that employees’ internal motivation is a significant factor in their innovative performance and has a positive effect on both idea generation and execution. As creativity requires a higher degree of IM, it is important to encourage employees to work hard in order to attain breakthroughs (Laforet, 2011). By sharing knowledge, employees with a high level of intrinsic work motivation can be stimulated to be creative in successfully completing their assignments, and new intellectual pathways can also be generated to help them experience higher creativity (Shafi et al., 2020). CSE is related to the ability to produce creative output, and can reflect intrinsic motivation to engage in creative activities (Tantawy et al., 2021). Thus, the following hypothesis that explores the mediated role of the four types of knowledge sharing between intrinsic motivation and CSE in ITI was proposed:

H9: Intrinsic motivation is positively related to CSE mediated by knowledge sharing.

Intelligent transportation jobs include the construction of software and hardware equipment in the traffic control center and various types of roadside equipment. Intelligent transportation implements cloud-based IoT based on the availability of accurate and timely information for the processes of forecasting and planning, resources, logistics service management as well as the many sub-processes in the supply chain (Jiang et al., 2020). The most important goal is to improve the accuracy rate of each type of equipment beyond the requirements of the user unit, and at the same time, to integrate all the various types of equipment installed on the roadside and to share the real-time information with users through the integrated system (Boukerche et al., 2020). Therefore, all tasks require the participation of experienced colleagues, as well as the cooperation of new colleagues and downstream vendors. Each person needs to share knowledge to solve internal and external problems to work out projects, given employees’ uneven experience and individual professional differences.

This study was conducted to investigate the relationship between intelligent transportation employees’ intrinsic motivation and their knowledge sharing CSE by using currently working employees as the empirical subjects. Adopting snowball sampling, a message was sent to the target key personnel in the human resource departments of intelligent transportation companies asking them to deliver the questionnaire link to their colleagues. A message was also embedded in the questionnaire that asked participants to pass on the questionnaire link to friends who worked in technology-related jobs. The survey was conducted from January to February 2021.

Regarding ethical considerations, a message appeared in the introduction section of the questionnaire indicating the purpose of the survey and informing participants that they were free to refuse to reply, and that the collected data would only be used for this study.

A total of 262 questionnaires were collected, and 18 invalid questionnaires were deleted. This gave 232 valid questionnaires for use in the further statistical analysis. The distribution of the sample in this study is analyzed as follows: in terms of gender, 106 (45.7%) respondents were male and 126 (54.3%) were female; regarding age, 29 (12.5%) were 25 years old or below, 48 (20.7%) were 26 ∼ 30 years old, 41 (17.7%) were 31 ∼ 35 years old, 31 (13.4%) were 36 ∼ 40 years old, and 83 (26.7%) were 41 years old or above; as for completed educational level, 73 (40.1%) had graduated from junior college, 105 (45.3%) had graduated from university, and 34 (14.7%) had graduate degrees.

We referred to the relevant literature to develop the questionnaire items on intrinsic motivation, CSE, and knowledge sharing. In this section, we discuss the composition, measurement, and operational definitions of each variable. To ensure content validity, five domain experts were invited to check the accuracy and applicability of the item translation; to ensure face validity, 10 students were invited to respond to the items to check if any statements needed revising. The questionnaire used a 5-point Likert scale, with 1 for strongly disagree, 2 for disagree, 3 for neutral, 4 for agree, and 5 for strongly agree. The reliability and validity of the items and constructs were re-tested according to the confirmatory study (see Tables 2, 3).

The purpose of the first-order validation factor analysis was to remove non-identifiable items. First, the original items of each construct were subjected to first-order validation analysis to examine the relationship between each item in each construct. Second, the fitness was checked to see if it was within the criteria, which according to Hair et al. (2019) included a χ²/df ratio of less than 5, a GFI greater than 0.80, an AGFI greater than 0.80, and an RMSEA less than 0.10, as shown in Table 1.

In this study, Cronbach’s α and composite reliability (CR) were used to examine the reliability analysis of the constructs and the overall questionnaire, and SPSS 23 statistical analysis software was used to perform the analysis. The higher the value, the higher the reliability consistency of the scale. To enable a more accurate measurement, this study also relied on the Cronbach’s α and CR value of the structural equation model to analyze the data, where good values are usually above 0.70 (Cook and Beckman, 2006). The Cronbach’s α values of the constructs ranged from 0.85 to 0.94, and the CR values of the constructs ranged from 0.85 to 0.93, as shown in Table 2.

According to Fornell and Larcker’s (1981) criteria for assessing convergent validity, a higher convergent validity represents a high FL value, which means that the tested variables can be converted to a potential variable, and generally an acceptable FL value should be at least 0.50. The FL values of the constructs ranged from 0.77 to 0.87, as shown in Table 2.

Cook and Beckman (2006) stated that the average variance extracted (AVE) value needs to be more than 0.50 to represent the effect of averaging on a construct. This value is the sum of the squares of the standardized factor loadings of the constructs, and if this criterion is reached, it means that each question can be explained by each construct. The AVE values of the constructs ranged from 0.59 to 0.77, as shown in Table 2.

The average variance extracted for each construct must be greater than the squared correlation coefficient between the two compared constructs in order to represent the discriminant validity of each construct (Fornell and Larcker, 1981). As shown in Table 3, the square root of the AVE for each construct ranged from 0.77 to 0.86, all of which were greater than the correlation coefficients between constructs.

The study model was finally analyzed for overall fitness using the AMOS 20.0 statistical software (Hair et al., 2019). RMSEA = 0.05, GFI = 0.88, AGFI = 0.85, NFI = 0.91, and NNFI = 0.96, where greater than 0.90 meets the criteria. CFI = 0.97, IFI = 0.97, and RFI = 0.90, where greater than 0.80 meets the criteria. PNFI = 0.81 and PGFI = 0.72, where greater than 0.50 meets the criteria.

Intrinsic motivation and automatic response have a positive relation (β = 0.50^***, p < 0.001); intrinsic motivation and rational reflection have a positive relation (β = 0.46^***, p < 0.001); intrinsic motivation and ridiculed reflection have a positive relation (β = 0.32^***, p < 0.001); intrinsic motivation has a positive relation with deprived reflection (β = 0.39^***, p < 0.001); automatic response has no relation with CSE (β = 0.13, p > 0.05); rational reflection has a positive relation with CSE (β = 0.45^***, p < 0.001); ridiculed reflection has a negative relation with CSE (β = −0.17*, p < 0.05); and deprived reflection has a negative relation with CSE (β = −0.20^**, p < 0.01), as shown in Figure 2.

The explanatory power of intrinsic motivation for automatic response is 25%; the explanatory power of intrinsic motivation for rational reflection is 21%; the explanatory power of intrinsic motivation for ridiculed reflection is 11%; and the explanatory power of intrinsic motivation for deprived reflection is 15%. Automatic response, rational reflection, ridiculed reflection, and deprived reflection all have explanatory power of 25% for CSE, as shown in Figure 2.

In terms of indirect effects, intrinsic motivation had an indirect positive relation with CSE (β = 0.14*, p < 0.05), as shown in Table 4.

Each industry has its own characteristics, and each company also has a different culture. As people are usually influenced by the herd effect, when it is discovered that most people in a group perform a particular behavior, there will be an invisible pressure that drives others to follow the behavior of the majority. Therefore, it is recommended that subsequent researchers can conduct research on different industries and conduct further discussions.

As technology advances and helps virtual communities and online platforms flourish, this has led to increasingly diversified ways of knowledge sharing, and a growing number of people paying attention to the effectiveness of different knowledge sharing systems. Therefore, in the future, we can also explore the impact of different ways of knowledge sharing on employees’ CSE or work performance.