TPC Journal-Vol 11-Issue-1

34 The Professional Counselor | Volume 11, Issue 1 Procedure To answer our research question, we used data from a larger study of novice professional counselor burnout, which included both quantitative and qualitative data. After receiving IRB approval, we obtained lists of names and email addresses of counselors engaged in supervision for licensure from the licensing boards in seven states: Florida, Nebraska, New Mexico, Oregon, Utah, Washington, and Wisconsin. We aimed to recruit a nationally representative sample by purposefully choosing at least one state from each of the ACA regions. In addition, states were selected based upon our ability to obtain a list of counselors who were engaged in supervision for licensure from the respective licensure boards. We were able to survey at least one state from each ACA region except the North Atlantic Region. After removing invalid email addresses, we invited 6,874 potential participants by email to complete an online survey in Qualtrics. This survey was completed by 560 counselors, yielding a response rate of 8.15%. This response rate is consistent with other studies that employed a similar design (Gonzalez et al., 2020). All participants were asked, Do you believe you are currently experiencing symptoms of burnout?, to which participants responded (a) yes or (b) no. Participants who responded yes were then prompted with the direction, Describe your symptoms of burnout, using an open-ended text box, which did not have a character limit. A total of 246 participants (43.9%) responded yes and qualitatively described their symptoms of burnout. On average, participants provided 30.31 words ( SD = 36.30). We answered our research question for the current study using only the qualitative data, which aligns with the American Psychological Association’s Journal Article Reporting Standards for Qualitative Research (JARS-Qual; Levitt et al., 2018). Data Analysis To answer our research question, we analyzed participants’ open-ended responses using content analysis, which allows for systematic and contextualized review of text data (Krippendorff, 2013). As recommended by Krippendorff (2013), we followed the steps of conducting content analysis: unitizing, sampling, recording, and reducing. We first separated the responses of the 246 participants into discrete units. For example, “feeling exhausted and back pain” was coded as two units: (a) feeling exhausted and (b) back pain. This process resulted in a total of 1,205 discrete units. We reduced our data into categories using an inductive approach, which allowed for new categories to emerge from the data without an a priori theory (Krippendorff, 2013). Although there are multiple conceptualizations of burnout (Maslach & Jackson, 1981; S. M. Lee et al., 2007) that could have informed our analysis (i.e., deductive approach; Krippendorff, 2013), we chose an inductive approach to capture the conceptualization of burnout for novice professional counselors—generating categories based on participants’ explanations of their own symptoms of burnout (Kondracki et al., 2002). To that end, we developed a codebook by randomly selecting roughly 10% of the discrete units to code as a pretest. Our first and third authors, Ryan M. Cook and Janelle L. Jones, independently reviewed the discrete units, met to discuss and develop categories and corresponding definitions, and coded the pretest data together to enhance reliability. This process yielded a codebook that consisted of 12 categories. Cook and Jones then used the codebook (categories and definitions) to independently code the remaining 90% of the data across three rounds (i.e., 30% increments). After each round, Cook and Jones met to discuss discrepancies and to reach consensus on the final codes. The overall agreement between Cook and Jones was 97% and the interrater reliability was acceptable (Krippendorff α = .80; Krippendorff, 2013), which was calculated using ReCal2 (Freelon, 2013). At the end of the coding process, Cook and Jones reviewed their notes for each code and further organized them into subcategories based on commonalities. The second author, Heather J. Fye, served as the auditor (see Researcher Trustworthiness section) and reviewed the entire coding process.