Testing the Factor Structure of The Compassionate Coaching by Leader Index (CCLI)

Abstract: The Compassionate Coaching by Leader Index (CCLI; Shuck & Ghosh, 2021) was designed to operationalize the experiencing of coaching. The Index was developed through a series of semi-structured interviews conducted face-to-face with executive leaders. Interview lasted 75–90 minutes. To fully capture the essence of compassionate coaching, participants had to meet two criteria: (a) they had to be identified as compassionate as well as (b) recently lead through a highly stressful situation. In total, the team interviewed 16 participants whose background included professional and major league sports teams, executives at Fortune-100 businesses and Universities, and Non-Profit leaders. Ultimately, twelve themes emerged.
1. Self-other Boundary for Compassion
2. Dignity in Compassion
3. Compassion as Proactive
4. Compassion as Reactive
5. Compassion as Identity Work
6. Personal Experience as a source for capacity for Compassion
7. Compassion as Emotional Intelligence
8. Compassion as walking the line between empathy and accountability
9. Compassion as practicing and seeking out authenticity
10. Compassion being enabled by a relational context
11. Building Personal Bandwidth for Compassion
12. Building Organizational Bandwidth for Compassion
Due to a dearth of existing items, Shuck and Ghosh (2021) developed a set of behaviorally based questions aligned to the twelve themes. Although initial steps for scale development were followed (Nimon, Zigarmi, Houson, Witt, & Diehl, 2011; Rich, Lepine, & Crawford, 2010; Shuck, Adelson, & Reio, 2017; Worthington & Whittaker, 2006), the factor structure of the CCLI has yet to be tested. The purpose of the present study is to design a survey, collect data, and conduct a sequential-exploratory-confirmatory procedure to examine, refine, and confirm the Index factor structure (cf. Durvasula, Netemeyer, Andrews, & Lysonski, 2006).
The survey will be created in Qualtricsxm using the blocks function and administered through Prolific©. A function available in Qualtricsxm to balance survey respondents equally between U.S. based males and females, mirroring U.S.-demographics, will also be employed. The survey will begin with a prescreening question within the Prolific© database of prescreening questions to ensure respondents have supervisory experience. Only those with supervisory experience will be permitted to move forward. Completion of the 75-item survey (including 1 item for Prolific, 1 for BOT checking, 1 for implied consent, 3 for instructional manipulation checks, 56 for the CCLI, 3 for the CFA marker variable, and 10 demographics) will take no more than 14 minutes (Dillman, 2007). For CCLI items, counterbalancing will be employed with randomization at the item index and construct levels to reduce common method bias (Reio, 2010). Demographic related questions will be placed at the conclusion of the survey (Chambers & Nimon, 2018). Once respondents complete the survey, they will be redirected back to Prolific.
After data cleaning, data will be split into exploratory and confirmatory subsamples. Exploratory factor analyses (EFA) will provide an accessible technique to explore factor structures, reduce the item set, and identify items that best fit a simple structure (Worthingon & Whittaker, 2006). For the EFA, principal axis factoring and promax rotation will be used. When considering how many factors to extract, we will consider conceptual consideration of the items (see Hair et al., 1998), screen plots, eigenvalues, and Wood, Tataryn and Gorsuch's (1996) guidelines for overextractions. Confirmatory factor analyses (CFA) will be used to confirm the factor structure derived from EFA. We will employ the maximum likelihood estimation technique for testing CFA models as it focuses on creating parameters that reproduce the covariance matrix in the population as opposed to the sample (cf. Thompson, 2004). Finally, we will assess common method variance and bias using the CFA marker variable technique as outlined in Williams et al. (2010).