Burnout in Journalism: Evidence from Regional Television Stations in Ethiopia Выгорание в журналистике: данные на основе региональных телевизионных станций в Эфиопии

The international evidence base shows that the mass communication industry is characterized by elevated levels of employee burnout while there is scant research addressing parallel situations in Africa and more specifically Ethiopia. Accordingly, the objective, of the present study was to determine the prevalence and magnitude of burnout among journalists working for Amhara TV, Oromia Broadcasting Network and Tigray TV using the Maslachburnout inventory and a sample of 123 journalists. The aim was also to determine the contribution of job control, job demands, organizational support and biographical factors to burnout using hierarchical regression methods. A two-step hierarchical regression procedure was employed to determine the best fit to the data. Results indicated that biographical predictors failed to qualify as significant explanatory factors, but the model explained an improved 32 % of the variance in burnout. ANOVA tests showed no significant scores in burnout were attributable to gender, job level, age or work experience differences except for region. Implications are discussed.


Introduction
In the 11th Revision of the International Classification of Diseases (ICD-11), WHO defines burnout as "a syndrome conceptualized as resulting from chronic workplace stress that has not been successfully managed" (WHO, 2020). The condition represents the stress response to affective and cognitive demands that exceed personal resources with presenting symptoms that include becoming increasingly tired, impatient, and irritable (Payne 2001).
Burnout is considered by many to be a problem of the industrialized world, but the developing world especially Africa has also started to show the problem (Maslach, Schaufeli, and Leiter, 2001). The recent social, economic, and cultural development in the continent calls for a similar set of studies to understand these changes and their organization and workforce effects. There is growing recognition that journalism is a stressful occupation, leading to burnout not only due to the daily routines but also owing to exposure to critical and traumatic events (Monteiro & Marques-Pinto 2017). In 2005, journalism was ranked among the 10 most stressful jobs (Center for Disease Control, 2005). It may be argued that the profession may be more stressful in nations that are unstable and democratically immature. In fact, Ethiopia has been one of the most discussed nations in regard to treatment of journalists as well as the use of media for power consolidation by governments (Skjerdal 2010;Skjerdal, 2012). In more recent years, owing to political developments, Ethiopia's regions have assumed a new journalism culture that contests the traditional center-periphery relationships and have deployed media that is more than ever challenging the center. However, little is known about how the journalists are affected by the new media realities in terms of their burnout experiences. Understanding the key issues of burnout and associated factors requires conceptual foundations.
A fundamental conceptual contribution explaining and predicting burnout is the JDC (Job Demands Control Model) (Schaufeli& Bakker 2004). It states that burnout is a function of the level of complexity and amount of mentally demanding work a worker has to perform. The complexity refers to how much control a worker has over the work in terms of mastery of the skills set the job demands as well as level of risk involved professionally in terms error of judgment or pressures of meeting deadlines. In other words, it refers to skills discretion and decision latitude, implying the degree to which a work is at ease in handling job demands because they have the required skills and the freedom or job autonomy in matters of job decision-making (Bakker & Demerouti 2007;Dwyer & Ganster 1991). In simpler terms, job demands can cause burnout in proportion to the level of efficiency they demand, the level of job accountability, the quantitative load as well as the emotional demands in place, pace of work expected, work delivery deadlines, managerial expectations and supervisory relationships as well as ambiguity in job tasks (Skagen, 2016). Another factor related to burnout is organizational support. Workplace social support as perceived by a worker relates to a perception that the organization considers them valuable and sees them as an important organizational member (Aselage & Eisenberger, 2003).Workplace social support is an important construct that explains the value of collegial or supervisory support to employees which can have a meditational role in stress management. The presence of such support gives employees in distress assurances that there are people to fall up on and that they will not be left behind. The support can come in the form of supportive expressions of assurances or message of understanding and fellow feeling or more practical forms of helping in addressing tasks on the behalf of a colleague. It may also be expressed through explanation of work details or demonstration of ambiguous tasks (Giao, Vuong, Huan, Tushar, & Quan, 2020).
Thus associations between low job control, low organizational support, and high job demands with burnout are extensively documented (Magnusson et al, 2008). However to better understand the relationship between job demands, organizational support and burnout among media workers further research is necessary. Research has identified the antecedents and consequences of burnout, including a number of organizational and biographical factors such as gender and age (Varga, Urdaniz & Canti, 1996;Barthauer, Kaucher, Spurk, & Kauffeld, 2020). ). However, there is an acute dearth of studies on burnout in journalism and the limited studies available are dated. An overview of the literature indicates that burnout in broadcast journalism is neglected (with just two studies on the subject (Reinardy, 2012b(Reinardy, , 2013a and North American (with just two studies -one Norwegian and the other Korean). While the shortage of studies is international, there is special relevance to Africa where the state-owned media sector is the dominant employer but which remains unexplored especially in the context of the Ethiopian federation where journalism is highly contested and polarized (Woldearegay, 2021).
In an attempt to fill the void, the present study aims to interrogate journalistic burnout and explanatory constructs in the regional government media sector in three Ethiopian media institutions, focusing on television journalists and comparing prevalence rates among the broadcasters that included AMMA (Amhara Mass Media Agency), TMMA (Tigray Mass Media Agency) and OBN (Oromia Broadcasting Network), which are considered the most politically sensitive in the Ethiopian federation.

Survey Method
Respondents were administered a 22-item Maslach Burnout Inventory (MBI), Job Demands and Job Control Scale with 11 items and Perceived Organizational Support that has 7 items. All were likert scale type and required subjects to rate item on a 5-point scale. The administration took an average of 35 minutes.

Method of data Analysis
Sequential /hierarchical multiple regression analysis technique is used to estimate the regression model and the hypothesized relationships in the model. Hierarchical regression is a procedure of exploring the relationships among a dependent variable and several independent variables in steps. Accordingly, independent variables are entered into the regression sequentially to determine their effect on or relationship with the dependent/outcome variable.
Hierarchical regression was used to analyze the effect of an independent variable (e.g. gender of a journalist) on the outcome variable (level of burnout) after controlling for other predictor variables (such as age). At each stage of the sequential variable entry into the regression model, an increment in variance accounted for is noted. The order of entry is decided by the researcher who uses his understanding of relevant theory and research based on the truism that "the data analyst knows more than the computer" (Henderson & Velleman 1981). In essence, what is done is a comparison of sequential models based on their variance improvement with previous models until a decision is made regarding the best fitting model based on the amount of variance explained.

Research questions
• RQ1. What is the level of burnout among journalists working for the different media?
• RQ2. Is burnout related to biographical and demographic factors?
• RQ3. What are the most important predictors of burnout?
• RQ4. What are the relationships between burnout, job demands, job control and organizational support?

Measures
The scales are all adapted measures from published prior studies and tested for suitability using exploratory factor analysis. Burnout is the dependent variable and job demands, job control, and organizational support are the predictors. The three measures are conceptually logically related to the items on the Job Demands and Job Control Scale that predict burnout while Perceived Organizational Support measures a buffering effect on burnout.

Maslach Burnout Inventory (MBI)
Using 22 items, the MBI assesses burnout levels experienced by study subjects. Two example items are "I feel emotionally drained from my work. I feel used up at the end of the workday". An Exploratory Factor Analysis (EFA) conducted to determine the structural reproducibility of the original facets suggested that burnout was conceived generically by the Ethiopian subjects, which is in agreement with some previous conceptualizations reported in the literature (Kristensen et al 2005). In other words, based on the procedure of exploratory factor analysis, burnout was understood by the respondents as a single general concept, not a three-factor construct as reported in some of the literature on burnout (Lee & Ashforth, 1990).

Job Demands and Job Control Scale
The scale has job demands and job control constructs. These contained a total of 11 items asking the job demands placed on a worker as well as the amount of control the worker has over the work. Two examples include "Does your work demand too much effort? And "Do you have a choice in deciding HOW you do your work?"

Perceived Organizational Support
The Survey of Perceived Organizational Support was a valid scale with 8 items.
Respondents were asked to indicate the extent to which they agree with the following statements on a five-point scale. Example items included "My organization cares about my opinions. My organization really cares about my well-being." All measures were piloted and pretested to ensure they are suitable for the context, purpose, and requirements of the study (Gregory, 1996).

Sampling
The television stations studied were selected purposively because they were believed to be at the center of conflict and power struggle, which could render their journalists vulnerable to elevated stress. The sampling of journalists was however based on a random sampling basis and determined using Green's guidelines (Cocco, Serra, Lentini, Deliperi, & Delrio, 2015). Accordingly, 50 + 8(K) where K is the number of predictors is used. Thus, the sample of 123 was deemed more than the minimum size recommended (Gregory, 1996).). The sample of returned questionnaires was 82 % of the 150 copies distributed. The return rate of 82 % is considered more than desirable (Hoppen et al, 2017).

Data analysis
Data analysis software SPSS version 24 was used to manage the burnout study's database. Analysis was performed on all usable questionnaires following a quality audit. Descriptive statistics, including number (n), percentages (%), mean, and standard deviation (SD), were calculated for the demographic variables. A statistical test of analysis of variance (ANOVA) was used to compare group differences on the measurements of interest regarding the three TV stations studied. Pearson's correlations were computed between burnout and allied dimensions of support and control (Emerson, 2015). Multiple sequential regressions were used to examine the relationship of demographic and biographic variables, burnout, stress, and perceived organizational support. Statistical parameters that included an F-test of model adequacy, R, R square, adjusted r square, unstandardized coefficients (B), standardized coefficients (Beta), and p-values for each sequential model in the regression are reported as they relate to prediction of burnout as an outcome variable and inter-variable association between the constructs of interest, including the relationship between burnout and organizational support (Hoyt, Leierer & Millington, 2006). Together with these tests, a psychometric evaluation of scale reliability was implemented. The burnout scale had an impressive reliability score of .81. Organizational support scale had a reliability alpha of .80 and job control and job demands scales had alphas of.77 and .80, respectively.
Burnout scores were 55 % for TMMA, 59 % for OBN and 61 % for AMMA. The mean was 58 %. Based on scoring guidelines (16)  To have an initial preview of the extent of correlation and its location in the inter-factorial space, Pearson's correlation was computed for each predictor ahead of the hierarchical regression. Table 1 shows the correlation matrix. A moderate correlation was observed between burnout generically and organizational support (r=.224, p<0.05), burnout and job demands (r=.585, p<.001). Strong correlations were observed between organizational support and job demands (.338**), organizational support and job control (r=.397, p<.001), and organizational support and burnout (r=.224, p<.05*). Job demands were correlated with burnout (r=.485,p<.001).
Inspection of the correlation matrix further shows job control and burnout were correlated (r=.262, p<.05**) as were job control and organizational support r=.397, p<.001), job control and job demands (r=.693, p<.001).  As shown in Table 2, a two-step hierarchical regression procedure was employed to determine the best fit to the data with R2 as the statistic.
Individually and together three of the predictors in Model One: job demands, job control and organizational support have differing significance in explaining burnout. Taking job demands as one example, it does come into view to add unique explanatory power when the effects of the other independent variables are held constant (Beta = .622, t = 4.951, p = <.01) while the constructs of support and job control failed to qualify. In Model two biographical predictors failed to qualify as significant explanatory factors. Thus job experience (Beta=.149 t= 1.232, p= .221), age p=.200);gender (Beta=.120,t=1.321,p=.190), and job role (Beta=.134,t=1.408,p=.163). However the models do explain an important part of the variance in burnout, constituting a total of 32 %. This figure based on guidelines is considered significant (Chin 1998;Hair et al 2012).
The results of the regression showed that in Model One, job control, organizational support, and job demands explained 26 % of the variance (R Square=.264, F (4,123) =8.388, p, 001 and with the addition of job control, organizational support, and job demands, Job role of the Respondent, Gender of the Respondent, Age of the Respondent, Region, Job experience of the Respondent in Model Two explained variance rose to 32 % (R Square=.323). F (9,123)=4.553, p<.01.
As explained above, in both models individual factors did not produce appreciable significance levels. Thus, the variables of conceptual and biographical interest were individually unable to predict burnout.
Similarly, ANOVA was used to compare groups on attributes including gender, age, and job title. There was no significant difference across the groups.
However, there was a significant effect of region on burnout at the p<.05 level for the three regions [F(2, 120) = 4.274, p = .016_]. The result of the regression presented in Table 3 showed that in Model One job control, organizational support, and job demands explained 27 % of the variance (R Square=.269, F (4,123)=8.388, p,001 and with the addition of job control, organizational support, job demands, Job role of the Respondent, Gender of the Respondent, Age of the Respondent, Region, Job experience of the Respondent in Model Two explained variance rose to 32 % (R Square=.323). F (9,123)=4.553, p<.01. As an important procedure, the F-test in Table 4 presents a report of model significance as an important preliminary test of assumptions.

Discussion
The study aimed to determine the prevalence and magnitude of burnout as well as the most important explanatory factors from among a number of variables hypothesised to be relevant. Descriptive statistics indicated that the journalists studied have mild degrees of burnout. The present findings are in substantial agreement with studies that addressed burnout prevalence among journalists (Reinardy 2009;Reinardy, 2011). However, given the context of Ethiopia and its implied effect on media personnel and their operations, journalists may be underreporting their perceptions, a phenomenon that is encountered in several surveys.
An important drawback in self-report scales is the validity threats from social desirability bias. Social desirability refers to the tendency of study subjects to respond in a more socially recommended manner when asked about their realities to avoid disapproval. Thus respondents would want to look more satisfied, happier etc. than they actually are (Sjöström, & Holst, 2002). It is suspected that between 10-75 % of variance may be explained by socially desirable responding (Vesely, & Klöckner, 2020). Seen in this light and the realities of Ethiopia, it may be suspected that socially desirable responding may have lowered the burnout ratings. The thesis may be valid in the context of collectivist societies such as Ethiopia's which emphasize positive self-presentation irrespective of one's circumstances (Eysenck 1990). Indeed, admission of elevated burnout may be contrary to selfenhancement and impression management as important socially coping skills (Parmač Kovačić, Galić,, & Jerneić, 2014).
However, the findings also go beyond studies reported for media workers in stable democracies where organizational life may be less chaotic and more predictable. In fragile national ecologies, burnout may issue from newsroom routines that deviate from norms, controversial media material production and dissemination, ethical self-doubts or censure over news and other journalistic content, editorial judgment fears, social-professional overload involving subjects, as well as a climate of fear and instability as journalism is in essence political work that may further draw hostility from certain groups or actors.
On the other hand, issues of ethnic nationalism may be present in newsrooms, which may mean that journalists may wish to appear above the game, and as a result report artificially lower levels of burnout at a particular point in time. However, it is difficult to quantify the contribution of this potential factor, which is unlikely to offer the most persuasive explanation as nationalism is unlikely to feature prominently in the day-to-day work life realities of journalists for too long. Nonetheless, because surveys are snapshots they may at times be influenced by seasonal political developments that affect journalists as citizens and/or professionals, biasing survey results in the positive or negative direction (Kam & Meyer, 2015). In the more typical news media context, a job-relevant journalistic theory applicable to the African context is now guard dog journalism (Alemnew & Gebru, 2018). In this kind of journalism, the pressure on journalists can be unrelenting because journalists are used instrumentally in the fight among political elites (Donohue, Tichenor and Olien 1995) and carry the brunt of public scrutiny that is based on models of liberal media (Dworznik, 2006;Furedi, 2006), which may contribute to burnout. The lack of professional independence and diminished sense of control may further exacerbate burnout levels (Lauk, 2009). On the other hand, news organizations in conflictridden African contexts may be too embroiled in turmoil to provide support to staff. As the present study shows, whatever support was provided (which could also be contaminated by social desirability or other concerns) was inconsequential. Thus, while it was previously hypothesised that increased levels of perceived support would lead to lower burnout (Greenglass Burke , & Konarski 1997), the present study showed there was no significant effect of organizational support on burnout.
While this study did test job control and job demands, there are no previous studies to compare results with in media settings, except perhaps workload and autonomy, which are differently operationalised (MacDonald et al 2016). There were modest inter-organizational burnout differences which could be explained by contextual dynamics. However, the known limitations of cross-sectional research apply to the present study and the findings need to be interpreted cautiously. Bigger and more diverse data are needed to have a fuller and more reliable diagnosis of the relevant factors of burnout in the journalistic workplace. This includes the detection of the role of demographic factors including age and gender in predicting or explaining burnout (Brewer & Shapard, 2004). In this study, these factors did not systematically relate to burnout. The lack of association agrees with the findings of Sepehrmanesh et al (2010) who found no relationship between demographic factors and burnout scores as well as those of Reinardy's (2011) who reported that younger journalists had greater burnout experiences because they lack experience to handle demands of the job and relational skills necessary to handle job pressures. With increased age and attendant experience more stability ensues and together with it more control of burnout (Bakker, Demerouti & Schaufeli, 2002).The present study replicates the findings of Reinardy (2011a) that used an ANOVA design to compare burnout along journalistic professional titles but found no differences among reporters, editors, and managers.

Conclusions and Implications
The purpose of the study was to determine the prevalence and magnitude of burnout among journalists working in three television stations. Overall, the evidence in this study supports the conclusions of previous studies on the phenomena of burnout in the media industry from a differing media ownership structure. It shows that burnout is rampant across the regional-state owned media sector irrespective of journalists' biographical characteristics. The findings may not be surprising in view of the job characteristics of media work that can often involve hazards especially in transitional societies that are experiencing communal tension. However, it may also suggest that there could be far-reaching professional consequences of burnout such as inattention and limits to the exercise of caution as desiderata in high-speed high-risk jobs such as journalism. Journalists' emotional wellbeing is important in conflict societies; in particular where journalism and media need to be constructive actors as preexisting tensions of diverse etiologies can easily turn in to perilous eruptions that defy any restorative attempts. Avenues for further research are suggested that include a more in-depth analysis of the complex relations among job satisfaction, burnout and intention to quit, using mixed methods, that may help to shed more light on the media workforce.

Ethical Declaration
The study was conducted in conformity with the Declaration of Helsinki-Ethical Principles for Medical Research Involving Human Subjects, and local legislation. In accordance with normative protocols, 1. Informed consent was obtained from the respondents, 2. Their dignity and wellbeing were protected at all times, and 3. The study's data was anonymised and confidentially treated throughout the study.

Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.