Passion

Sampling and Data Collection

The data were collected from women entrepreneurs located in RRR Australia. The rationale for focusing on this group was discussed in the Introduction. An online survey was conducted to measure the entrepreneurial leadership and venture growth intention of Australian RRR women entrepreneurs. Purposive sampling was employed with respondents recruited in two tranches through non-probability snowball sampling. First, an email invitation to participate in the survey was sent to 750 women entrepreneurs subscribed to the only Australia-wide women entrepreneurship development program, The WiRE Program. This was followed up with an email to several women associations operating in RRR Australia, requesting these associations to invite women entrepreneurs in their association to participate in the survey. Potential respondents were invited to click on a survey link. One-hundred-and-nine survey responses were received after multiple reminders. Ten respondents were excluded due to missing data, which left 99 usable surveys. We suspect we could have achieved a higher response rate if Australia had not been struggling with the aftermaths of COVID and extreme flooding in rural areas at the time of distributing the survey, which significantly negatively impacted especially small businesses.

Data Collection Tools

The survey instrument was pretested with 10 participants in the study area to check the validity and appropriateness of wording, formatting, and sequencing of questions. The questions were refined based on the pilot outcomes. Three scales were utilized in this study to measure the constructs: Entrepreneurial Leader Identity (ELI), Entrepreneurial Passion (EP) and Venture Growth Intention (GI). The construct ELI was measured by four items adapted from: "Developing and nurturing a venture/business is an important part of who I am (EL1)", "I think of myself as an entrepreneur (EL2)", "I think of myself as a leader (EL3)" and "When I describe myself, I would include the word leader (EL4)".

To measure EP, we drew on the four items of scale from Cardon, Gregoire, Stevens and Patel. However, one item was dropped after assessing the measurement model due to low factor loading. The three items utilized were: Nurturing a new venture/business/initiative through its emerging success is enjoyable (PA1)", "It is exciting to identify unmet market gap (PA2)" and "Inventing new solutions to problems is an important element of who I am (PA3)"; The excluded item was "Assembling the right people to work with me or my business is exciting (PA4)".

A single item, "My intention is to grow my venture as large as possible", measured GI. The scale was adapted from Edelman, Brush, Manolova and Greene. Measurement items of all the scales were measured by 5-point Likert scales ranging from "strongly disagree" to "strongly agree".

Data Analysis Method

Structural equation modelling (SEM) is a multivariate method for testing and evaluating multivariate causal relationships. SEM examines direct and indirect effects on hypothesized causal relations. In general, there are two approaches to SEM: covariance-based and component-based SEM. A large sample is needed to perform covariance-based SEM, whereas component-based SEM can be performed on a small sample. As our study hypothesized multivariate causal relations and has a small sample size, with the help of Software Smart PLS 3, component-based partial least squares structural equation modelling (PLS-SEM) was applied to test the relationship among the study constructs ELI, EP, and GI. The method was chosen because our study has a small sample size, and PLS-SEM works better with a small sample size compared to CB-SEM. Besides, PLS-SEM has more flexible requirements concerning sample distribution and measurement scales.

PLS-SEM requires two steps to be completed. The measurement model should be evaluated in the first step to ensure its validity. In the second step, the hypothesized relationship should be tested, and the measurement model should be evaluated to ensure its validity. Following the established guidelines for PLS-SEM, both the measurement model and the structural equation model were validated in our study. The measurement model was evaluated by assessing reliability and validity prior to evaluating the structural model. Cronbach's alpha and composite reliability (CR) were used to evaluate the reliability of the constructs. Discriminant validity was assessed by the Fornell-Larcker criterion and the Heterotriat-Monotrait ratio of correlation. Convergent validity was accessed by the Average Variance Extracted (AVE). The structural model was assessed based on explained variance (R2), predictive relevance (Q2), the significance of paths, and bootstrapping generated 5000 samples to compute T-values to test the model, as Hair, Ringle and Sarstedt suggested.

Assessing the Common Method Bias

Common method bias is the bias produced in estimates due to the common method used to assess both independent and dependent variables. When surveys collect data on both independent and dependent variables simultaneously, the estimated effect of one variable on another may be skewed by common method variance; that is, systematic variance shared among the variables introduced into the measures by the measurement method rather than the theoretical constructs being measured. In this study, common method bias was assessed by Harman's single factor test.