A survey research study of music education faculty: Demographics as related to indicators of job satisfaction and stress

Demographics

A limited number of researchers have examined MEF demographics. Hewitt and Thompson (2006) constructed a MEF demographic profile evidencing that MEF identified more often as male (56.1%), married (78.0%), White (94.0%), with doctorate degrees (71.1%), with an average age of 51.65 and average yearly salary of $53,092. MEF reported spending more time teaching than conducting research. Brewer and Rickels (2012) demonstrated that MEF have different workloads depending on the types of institutions in which they teach. Brewer and Rickels also yielded similar demographic profiles where more MEF were male (55.1%) and the largest portion of MEF time was allocated to teaching (65.0%), adding MEF had an average of 12.25 years of K-12 teaching experience.

Even though such few researchers have examined MEFs’ demographic backgrounds, such information suggests that these faculty members are mostly men, White, and married. Additionally, MEF spend more time teaching, primarily undergraduates, than conducting research. Such demographic information can be important to also understanding job satisfaction and stressors among MEF.

Job satisfaction

Over the past decades, a considerable amount of literature on job satisfaction of general university faculty has emerged. Overall, job satisfaction indicators seem to be associated with work environmental factors and faculty members’ personal lives. These determinants include academic autonomy (Daly & Dee, 2006; Lee & McNaughtan, 2017; Shin & Jung, 2014), support from colleagues and supervisors (Mudrak et al., 2018; Pellegrino et al., 2018), and faculty members’ marital status (Leung et al., 2000; Mamiseishvili & Rosser, 2011; Sabharwal & Corley, 2009). Research on job satisfaction has its roots in autonomy, defined as ‘the freedom of faculty members to research, publish, and teach without restraints imposed by university administrations’ (Eisenberg, 1988, p.1433). Indeed, researchers of several studies have documented that autonomy positively impacts job satisfaction (Daly & Dee, 2006; Lee & McNaughtan, 2017; Shin & Jung, 2014). For example, Daly and Dee (2006) uncovered that high levels of autonomy had a positive and indirect influence on job satisfaction, which facilitated faculty retention. These findings matched those of Shin and Jung (2014) who demonstrated that academic autonomy is a fundamental component linked to job satisfaction. More specifically, Lee and McNaughtan (2017) found that job crafting, defined as autonomy in conducting research and levels of flexibility in music faculty members’ teaching and service, also enhances job satisfaction and vitality, which also affects job satisfaction. They ultimately found that freedom to conduct one’s research is the most critical factor MEF use to gauge their autonomy.

Researchers have also found links between job satisfaction and the support faculty receive from colleagues and supervisors. Mudrak et al. (2018) found that faculty (55.7%) valued support provided by colleagues ‘all the time’ or ‘often’ (p. 336), as well as the job resources provided. Similarly, Pellegrino et al. (2018) reported that MEF regarded mentoring interactions with senior faculty as invaluable, also as positively associated with job satisfaction.

Researchers have also addressed the marital status of faculty members alongside job satisfaction (Daly & Dee, 2006; Leung et al., 2000; Sabharwal & Corley, 2009). Leung et al. (2000) found that married professors reported greater levels of job satisfaction, and single professors demonstrated higher psychological stress (see also Sabharwal & Corley, 2009). However, it remains unclear as to whether marital status affects job satisfaction among MEF. That marriage was defined in traditional terms in these studies might also be modernized to include non-traditional partnerships (e.g. same-gender relationships).

Stressors

Job-related stress is also an important issue, especially since stress can trigger psychological and physical illnesses (Haseeb & Sattar, 2018) and can be related to job turnover (Hamann & Sherbon, 1988). Scholars have examined job-related stress concerning faculty salary (Hamann & Sherbon, 1988; Sabharwal & Corley, 2009), hours of class preparation and sleep (Bernhard, 2007), parental responsibilities (Carr, 1998; Misra et al., 2012; Pellegrino et al., 2018), and uncertainties surrounding tenure (Bradley et al., 2017; Pellegrino et al., 2018). Researchers have also commonly reported that salary plays a significant role in faculty stress levels (Hamann & Sherbon, 1988; Sabharwal & Corley, 2009). Hamann and Sherbon (1988) found that as much as faculty regarded their salaries as insufficient, this resulted in higher levels of stress among men versus women faculty. Sabharwal and Corley (2009) corroborated this finding that the average annual salary of men faculty was higher than women faculty, suggesting that gender inequities in salary in fact matter.

Fewer researchers have focused on links between hours spent on class preparation and stress. Bernhard (2007) found that stress is associated with ‘emotional exhaustion, depersonalization, and low personal accomplishment’ (p. 118), adding that there exists a strong, positive relationship between depersonalization, emotional exhaustion, and the average hours of class preparation per week. Regarding relationships between sleep and stress, Bernhard (2007) demonstrated that stress (i.e. emotional exhaustion and depersonalization) and sleep are negatively correlated. Put differently, as hours of sleep decrease, emotional exhaustion, and depersonalization increase. Given that the often-recommended hours of sleep per day for adults is 7-to-9 hr (Hirshkowitz et al., 2015), the average daily hours of sleep reported (6.78) may exacerbate stress.

Researchers have also explored parental responsibilities of faculty members. Hamann and Sherbon (1988) found that female faculty with children had fewer peer-reviewed publications. Misra et al. (2012) found that women spent more time caring for their children regardless of academic rank. Pellegrino et al. (2018) added that female MEF with children experienced more challenges in balancing their professional lives. It appears, then, that parental responsibilities are related to stress, also by gender.

Finally, Misra et al. (2012) suggested that job demands are different across faculty, underscoring that the uncertainties involved with tenure-track requirements punctuate anxiety and stress. Bradley et al. (2017) found that tenure-track MEF felt unsure if their scholarly outputs were sufficient, complaining about a lack of clarity in tenure requirements. Pellegrino et al. (2018) also found that most MEF respondents (70.0%) felt very stressed during their tenure processes, and this was especially true for MEF in larger institutions.

Participants

We employed multiple sampling methods to survey our study participants (Nardi, 2018). More specifically, after obtaining university Institutional Review Board (IRB) approval in November 2020, the National Association for Music Education (NAfME) distributed our survey instrument via Google Forms on behalf of us as researchers. The NAfME invited 820 members whose membership profile indicated that they taught at the university level. However, due to a small number of responses (n = 37), we employed a purposeful sampling technique (Nardi, 2018) adding MEF who worked at the 100 US colleges and universities to our sample (Universities.com, n.d.). We obtained an additional list of 279 MEF and contacted them via an email invitation containing a link to the survey. Our sampling framework yielded 1,099 potential respondents and an overall response rate of 12.4% (N = 136). We informed participants of the aims and ethical implications of the study before they volunteered to participate.

Survey instrument

We collected data via an online survey instrument as per best survey research practices (Edmondson et al., 2012). More expressly, we developed a 67-item instrument that corresponded to constructs and concepts derived via researchers’ previous studies and findings (Bernhard, 2007; Bradley et al., 2017; Carr, 1998; Daly & Dee, 2006; Hamann & Sherbon, 1988; Lee & McNaughtan, 2017; Leung et al., 2000; Mamiseishvili & Rosser, 2011; Mudrak et al., 2018; Pellegrino et al., 2018). We framed the 67-item instrument around our three primary sections of interest: (a) demographics, (b) indicators of job satisfaction, and (c) indicators of stress.

The first section included 12 demographic items (e.g. gender and age). We collected institution types in which participants taught following the Carnegie Classifications of Institutions of Higher Education (Center for Postsecondary Research, 2020), or whether participants taught in the following institutions: (a) Research 1 (R1) universities with very high research activity, (b) Research 2 (R2) universities with high research activity, (c) Doctoral/Professional Universities (D/PU) with less than $5 million in research expenditures, (d) Master’s 1 (M1), (e) Master’s 2 (M2), (f) Master’s 3 (M3), and (g) Baccalaureate Colleges (BC). All M1 through M3 grant master’s degrees but vary in terms of the number of degrees offered (i.e. large, medium, and small). BCs confer baccalaureate or higher degrees which represent at least 50% of all degrees, and fewer than 50 master’s degrees or 20 doctoral degrees are awarded each year.

The second and third survey instrument sections included items using 5-point Likert-type scales ranging from 5 (strongly agree) to 1 (strongly disagree) and open-ended questions to solicit richer information (Miller, n.d.). Likert-type scale item responses were aggregated, and composite scores were calculated based on average scores ranging from 5 to 1 for all the subconstructs under the second and third sections.

The second section included one dependent variable—job satisfaction—and three independent variables—autonomy, collegial support, and marital status. Job satisfaction levels were assessed using items 26 to 28. Higher scores denoted greater job satisfaction. Autonomy, collegial support, and marital status were measured with items 29 to 40. Higher scores were positively associated with each independent variable.

The third section included one dependent variable—stress—and five independent variables—salary, teaching, sleep, parental responsibilities, and certainties surrounding the tenure process. Job-related stress levels were assessed with items 43 to 45. Higher scores indicated lower job-related stress levels. Each stressor was measured with items 46 to 59, with higher scores positively associated with each stressor. See the full survey instrument in the supplemental material. See also Table 1 that illustrates item numbers and one example question per construct.

OPEN IN VIEWER

Table 1.

Items per study variable.

Item numbers	Study variable	Example question
26–28	Job satisfaction	I feel satisfied with my job.
29–31	Autonomy	I can choose which classes I want to teach.
32–37	Collegial support	I find interactions with colleagues helpful.
38–40	Marital status	My spouse/partner pay(s) attention to my concerns about my job.
43–45	Job-related stress	I do not feel stressed about my job.
46–48	Salary	I receive a satisfactory salary.
49–53	Teaching	I feel energized when preparing classes.
54–56	Sleep	I feel refreshed after sleeping.
57–59	Parental responsibilities	My child/children does/do not distract me from my research.
60–65	Certainties surrounding the tenure process	I received clear documentation regarding requirements for achieving tenure from my institution.

Instrument reliability

Cronbach’s alpha analyses confirmed the satisfactory reliability of our survey instrument. Overall alpha estimates on both our job satisfaction and stress constructs were acceptable (α = .82 and α = .73; Vaske et al., 2017). We also piloted this instrument and had three MEF who were not participants in this study review and critique the content of this instrument before distributing to respondents. We did all of this to determine whether this survey instrument measured the concepts that we developed it to measure (Edmondson et al., 2012). After the pilot, we officially collected data in January of 2021.

Data analyses

To answer research question one, we used frequency distributions for all nominal data (Thompson, 2009). For our continuous data, we calculated means and standard deviations (SDs). We explored central measures of tendency for each Likert-scale item, including mean responses and SDs. We also combined item responses per construct to generate composite scores.

To answer research questions two and four, we conducted simultaneous multiple regression analyses. We confirmed the assumptions of linearity, normality, and homoscedasticity, and each variance inflation factor (VIF) score was well below 10, indicating that multicollinearity assumptions were also met. We used the first regression model to determine if autonomy, collegial support, and marital status significantly predicted job satisfaction. We used the second model to test if salary, teaching, sleep, parental responsibilities, and (un)certainties surrounding the tenure process significantly predicted stress levels. To offer a measure of practical significance concerning the magnitude of the effect (Field, 2018), we calculated Cohen’s (1988) f ² as an index of effect size, also categorizing effects as small (f² = .02), medium (f² = .15), and large (f² = .35).

To answer research questions 3 and f5, we conducted two sets of one-way ANOVA analyses on participants’ levels of job satisfaction and stress to determine if scores varied by institution type. We based interpretations on statistical significance at or below p = .05. We confirmed assumptions of independence, linearity, normality, homoscedasticity, and noncollinearity (Field, 2018), after which we calculated Eta-square (η²) as an index of effect size to examine the proportion of variation in the dependent variable accounted for by groups as the independent variable, also categorizing effects as per Cohen’s (1988) η² categories of small (η² = .01), medium (η² = .06), and large (η² = .14).

Finally, for our qualitative analyses, we employed NVivo (QSR International Pty Ltd, 2020) to yield comprehensive pictures of participants’ perceptions and thoughts (Wong, 2008). We performed a word frequency analysis on the comments to use the output for uncovering recurring themes. Such auto-coding examinations are less susceptible to research bias, which enabled us to identify the main themes and indicator words as codes (Fielding et al., 2013). As such, we also coded and sorted a set of categories to extract main ideas by quantitatively presenting open-ended questions as follows: ‘X% of responses mentioned concepts Y’ (Fielding et al., 2013, p. 3266).

Participant demographics

Factors Affecting Job Satisfaction

Simultaneous multiple regression results revealed that the overall model significantly predicted job satisfaction, F(3, 106) = 10.353, p < .001, adjusted R² = .205, Cohen’s f² effect size estimate of 0.258. Predictor variables—autonomy, collegial support, and marital status— explained 20.5% of the total variance in job satisfaction. Overall, two predictors—autonomy (β = .257, p = .005) and collegial support (β = .290, p = .002)—were significant predictors of participants’ reported levels of job satisfaction with a medium effect size (Cohen, 1988). Marital status did not significantly predict job satisfaction levels. Overall, both autonomy and collegial support contributed the greatest to the regression model, but collegial support predicted more of the variance than autonomy (see Table S6 in the Supplemental Material).

Comparisons of Job Satisfaction by Institution Types

ANOVA results indicated that there was no significant and a small effect of institution type on mean job satisfaction, F(6,129) = 0.908, p = .492, η² = .041. The mean score of M3 group (M = 4.07) was slightly higher than those of other groups, but no significant differences revealed that all seven groups were essentially equal in job satisfaction levels. Accordingly, no post hoc pairwise comparisons with a Bonferroni correction were conducted (see Table S7 in the Supplemental Material).

Factors affecting job-related stress

The regression model was statistically significant (F[5, 56] = 5.790, p < .001, adjusted R² = .282, Cohen’s f² effect size estimate of .393). Stressors—salary, teaching, sleep, parental responsibilities, and certainties surrounding the tenure process—explained 28.2% of the variance in job-related stress with high effect sizes observed (Cohen, 1988).

More specifically, results indicated that of the five variables entered as predictors, three variables—sleep, parental responsibilities, and certainties surrounding the tenure process—were significant predictors of participants’ reported levels of job satisfaction, with standardized coefficient β = .281, p = .024, standardized coefficient β = −.245, p = .046, and standardized coefficient β = .320, p = .015, respectively. This suggests that greater levels of sleep, parental responsibilities, and certainty surrounding the tenure process resulted in lowering job-related stress level. In addition, standardized beta coefficients further identified certainties surrounding the tenure process to be the most influential predictor (see Table S8 in the Supplemental Material).

Comparisons of Stressors by Institution Types

ANOVA results indicated that institution types had a statistically significant difference in terms of the mean stress score with a medium effect size observed, F(6,129) = 2.356, p = .034, η2 = .099. This suggests that at least one of the group means was statistically significant from the other group means. The effect size was medium, suggesting that the magnitude of difference between group means was medium (Cohen, 1988).

The post hoc tests of pairwise comparisons using Bonferroni correction revealed that the mean score of R1 group was statistically higher than that of M1 group (R1 = 3.82, M1 = 3.42; p = .010) (see Table S9 for details in the Supplemental Material). Given that the only statistically significant different result for mean stress arose between R1 and M1 institutions, this would suggest that R1 faculty were more likely to report low levels of stress than M1 faculty.

Open-ended questions