Team internal social capital and entrepreneurial learning: a dual-path exploration in new venture teams

Methods

Sample and procedure

Questionnaires were administered to 130 NVTs in Eastern China. The respondents were obtained through the researchers’ social networks using snowball sampling. With the support of management, researchers explained the study’s purpose and procedures. The researchers distributed the questionnaires personally to core team members. Participants were assured that the data would be strictly confidential and that they could withdraw at any time. The University Ethics Committee approved this survey. Three hundred paper-based questionnaires were distributed to the team members and all were recovered. The final data sample included 284 entrepreneurs from 111 NVTs, excluding questionnaires that were incorrectly completed or showed substantial data loss. The questionnaire recovery efficiency was 94.67%.

Participants were from different industries, including wholesale and retail (45.3%), education (14.6%), manufacturing (13.2%), financial and property (7.0%), online services (5.9%), and others (13.9%). The team age ranged from one to eight years. Regarding team size, 51.9% had fewer than eight members, and 48.1% had more than eight. Of the participants, 91.2% were under the age of 46, 54.9% had entrepreneurial experience, and 59.2% were male. Regarding education level, most participants had a bachelor’s degree (39.4%). Table 1 shows the other demographic statistics.

Measures

Team internal social capital

Team internal social capital was assessed using the Entrepreneurial Team Internal Social Capital Questionnaire (Xie and Zhang, 2014). It comprises sixteen items, each rated on a five-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree), with a higher score representing better team internal social capital. For example, ‘The team has a clear development goal that we have recognised.’

Entrepreneurial learning

Entrepreneurial learning was assessed using the Entrepreneurial Learning Scale (Shan et al. 2014). It comprises twelve items, each rated on a five-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree), with a higher score representing better entrepreneurial learning. For example, ‘Failure is important for me to accumulate experience.’

Shared mental models

Shared mental models were assessed using the Entrepreneurial Team Shared Mental Models Scale (Wang and Chen 2010), which has two sub-dimensions: taskwork-shared mental models and teamwork-shared mental models. It comprises 16 items, each rated on a seven-point Likert scale ranging from 1 (not at all) to 7 (very much so), with a higher score representing better taskwork/teamwork-shared mental models. For example, ‘I share a common understanding with my team members about the team task.’

Organisation-based self-esteem

Organisation-based self-esteem was assessed using the Organisation-based Self-esteem Scale (Pierce et al. 1989). It comprises ten items, and each item is rated on a five-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree), with a higher score representing better organisation-based self-esteem. For example, ‘I am an efficient worker in my team.’

Demographic variables

Participants were asked to provide their demographic details, such as age, gender, education level, and entrepreneurial experience, as well as team information, such as industry, team size, and the time of team establishment. Consistent with previous studies, age, gender, education level, team size, and industry were measured as control variables (Liu et al. 2023).

Analytical approach

We analysed the data using SPSS 26.0 and MPLUS 8.0. First, a reliability and validity analysis was conducted, and common method bias and data aggregation were examined. Second, descriptive statistics and a correlation matrix were constructed. Third, multilevel structural equation modelling was performed using MPLUS 8.0. Team internal social capital and shared mental models were aggregated at the team level using SPSS 26.0. We then examined the cross-level effects of team internal social capital on entrepreneurial learning and the mediating roles of shared mental models and organisation-based self-esteem. Given that conventional bootstrapping methods are not applicable in the context of multilevel modelling, we adopted a Monte Carlo approach with 20,000 resamples to estimate the coefficients and 95% confidence intervals for the moderating effects (Preacher et al. 2010; Xie and Feng, 2024).

Results

Reliability and validity analysis

We conducted a factor analysis and found that some items had lower factor loadings, which could affect the scale’s overall reliability (ranging from 0.40 to 0.83). Given that Cronbach’s α was just above the minimum acceptable threshold of 0.70 (e.g., 0.74/0.78), we decided to remove the items with low factor loadings to enhance the internal consistency of the measurement. Table 2 presents the results before and after these modifications; they show an improvement in Cronbach’s α to 0.85/0.89. The reliability of each scale was tested again, and the results showed that Cronbach’s α and composite reliability (CR) were all above 0.8, indicating good reliability (Geldhof et al. 2014). Moreover, the validity of each scale was tested, and the results showed that the average variance extracted (AVE) of each scale was greater than 0.5, indicating good convergent validity (Fornell and Larcker, 1981). Finally, by comparing the fit indices of different models, we found that the five-factor model (Model 1) had better-fit indices than other models, indicating that the variables have good discriminant validity. Tables 2 and 3 show the specific results.

Common method bias

As all measures were evaluated using the same source, common method bias may have influenced the overall study results. Thus, Harman’s single-factor test was used to assess common method bias. The explanatory variance of the first factor before rotation was 31.1%, which was less than the critical criterion of 50%. Therefore, the effect of common method bias on the current study was limited (Podsakoff et al. 2003).

Data aggregation

In this study, team internal social capital, taskwork-shared mental models, and teamwork-shared mental models represent team-level variables. Therefore, this study first converted the individual data to the team level. Intergroup variability (ICC) and Rwg were used to perform aggregation tests on the three variables (Bliese, 2000; James et al. 1984). Statistical analysis indicated that the Rwg, ICC (1), and ICC (2) values of the team’s internal social capital were 0.95, 0.21, and 0.78, respectively. Additionally, the Rwg, ICC (1), and ICC (2) of the taskwork-shared mental models were 0.87, 0.63, and 0.67, respectively, and the Rwg, ICC (1), and ICC (2) of the teamwork-shared mental models were 0.94, 0.68, and 0.67, respectively. All variables met the criteria of Rwg exceeding 0.7, ICC (1) exceeding 0.05, and ICC (2) being higher than ICC (1).

Descriptive statistics and correlation analysis

Table 4 shows the descriptive statistics and correlation matrices of all the variables. Entrepreneurial learning was positively correlated with organisation-based self-esteem (r = 0.34, p < 0.01). Team internal social capital was positively correlated with taskwork-shared mental models (r = 0.43, p < 0.01) and teamwork-shared mental models (r = 0.44, p < 0.01). These results provide preliminary support for our research hypotheses.

Hypothesis testing

We used multilevel structural equation modelling to test the hypothesised model. First, a null model was established with entrepreneurial learning as the dependent variable. It was found that χ² = 0.17, p < 0.001, and ICC (1) = 0.30 > 0.12, which allowed for the following multilevel analysis.

Figure 2 shows that team internal social capital is positively related to entrepreneurial learning (β = 0.31, p < 0.001); the results support H1. This study tested the mediation effect and 95% bias-corrected Confidence Interval (CI) using the Monte Carlo approach based on 20,000 samples (Preacher and Selig, 2012). From the results presented in Table 5, the mediating effects of taskwork-shared mental models on the relationship between team internal social capital and entrepreneurial learning were significant (indirect effect = 0.20, SE = 0.10, 95%CI = [0.16, 0.07]). Moreover, teamwork-shared mental models mediated the relationship between team internal social capital and entrepreneurial learning, with significant indirect effects (indirect effect = 0.16, SE = 0.08, 95%CI = [0.08, 0.26]). Thus, H2a and H2b are supported.

Structure equation models. Note: *p < 0.05, **p < 0.01, ***p < 0.001.

Additionally, the mediating effect of organisational self-esteem on the relationship between team internal social capital and entrepreneurial learning was significant (indirect effect = 0.35, SE = 0.14, 95%CI = [0.15, 0.51]). These results support H3.

These findings are consistent with the results of Mckeown (2010), which identified the positive relationship between team internal social capital and entrepreneurial learning. However, our results emphasise the significant mediating roles of team-shared mental models and organisation-based self-esteem. This suggests that the mechanisms through which team internal social capital influences entrepreneurial learning are more complex, highlighting the need for further investigation into these mediating factors across different contexts. Building on this foundation, we now explore the qualitative findings of Study 2 to further enrich our understanding of these relationships.