summary_eng (1136700), страница 3
Текст из файла (страница 3)
Among 102 sizes included into the recent meta-analysis of the relationshipbetween organizational identification and health (Steffens et al., 2017), 32 effect sizesdid not display any significant correlation, whereas 4 show the negative correlationbetween them: М. Galang and S. Jones used a mixed sample of employees workingfor different organizations and established a negative correlation between theidentification and stress (Galang & Jones, 2014); О. Herrbach used a sample ofengineers to determine a positive correlation between identification and experiencingnegative emotional states (Herrbach, 2006); A. Pisarski and her colleagues studiedthe sample of Australian nurses to conclude that the stronger group identification they12felt, the more frequently their respondents had physical health problems, though thecorrelation level was low (Pisarski, Lawrence, Bohle, & Brook, 2008); R.
Zhang andco-authors analyzed bank employees and managers and found a positive correlationbetween identification and stress (Zhang, Liu, Wang, & Shen, 2011). A series ofempirical research works also yielded the outcomes proving the relationship betweenidentification and the factors that had a negative impact on employee well-being(Escartín, Ullrich, Zapf, Schlüter, & van Dick, 2013; Golden & Wiens-Tuers, 2006;Herrbach, 2006; Mühlhaus & Bouwmeester, 2016; Ng & Feldman, 2008; Pratt &Corley, 2007). Thus, the existing literature provides controversial evidence on thecorrelation between organizational identification and employee well-being.Thepaperabovementionedexplainstwocontroversialpotentialoutcomes:mechanisms1) non-linearthatconditionrelationshipthebetweenorganizational identification and employee well-being; 2) indirect negative effect oforganizational identification on well-being mediated by workaholism that is inreverse to the positive effect of psychological attachment to the employee well-being.Chapter 2 entitled Empirical research into the relationship betweenorganizational identification and employee well-being describes the empirical partof the research.
To prove empirical research hypotheses, it was necessary to measurethe manifestation of workaholism. As there are no Russian-language scales tomeasure workaholism understood as compulsiveness and excessive work concern, aseparate research task consisted in translating the appropriate Dutch Work AddictionScale (DUWAS) into Russian (Schaufeli, Taris, & Bakker, 2006). Paragraph 2.2entitled The analysis of psychometric properties of the Russian-language scale ofthe Dutch Work Addiction Scale (DUWAS) presents the results of testingpsychometric properties of the Russian-language version of the scale. A series ofconfirmatory factor analysis was conducted to prove the construct validity of theRussian-language version of DUWAS and its compatibility with the original version.To prove the replicability and stability of the factor structure, multi-groupconfirmatory factor analysis was conducted.
The two-factor DUWAS model’s fit toempirical data better than that of one-factor. Multi-group confirmatory factor analysis13showed the equivalence of the DUWAS model measuring in male and female groupsand the stability of its factor structure. DUWAS scales are characterized byconvergent and discriminant validity. Therefore, the Russian-language version ofDUWAS was concluded to have acceptable psychometric properties compatible withthose of the original version and the versions in other languages.Paragraph 2.3 describes the empirical research aiming to test the hypothesis onthe non-linear relationship between organizational identification and employee wellbeing and on the indirect negative effect organizational identification has onemployee well-being through workaholism.To test the hypothesis on the non-linear relationship between organizationalidentification and employee well-being the research employed stepwise polynomialregression.
Job satisfaction, emotional exhaustion, work-family conflict, workengagement and workaholism acted as dependent variables, whereas organizationalidentification was the independent one. During the first stage, control variables andthe values of the organizational identification variables were added to the equation aspredictors (linear regression); during the second stage the square of the organizationalidentification variable value was added (polynomial regression). However, thesignificance of the squared term of the regression equation does not necessarily implythe U-shape character of the relationship between the variables. The illusion of thenon-linear relationship may be created due to the distributional asymmetry orinsufficient variance (Le et al., 2011; Morin et al., 2013).
Thus the correlationsbetween the predictor and the dependent variable were also calculated separatelybefore and after the turning point (the point where the sign of the relation changes,i.e. the lowest point of the U-curve or the highest point of the inverted U-curve)reflecting the non-linear relationship between them.
If the signs of these correlationsdiffer, it testifies to the non-linear character of the relationship.To test the hypothesis on the non-linear negative effect the organizationalidentification has on employee well-being through workaholism, the path analysiswas used. As it is supposed that the relationship between the independent variable(identification) and mediator (workaholism) is non-linear, the mediation effect may14depend not only on the ‘independent variable-mediator’ or ‘mediator-dependentvariable’ relationships, but on the value of the independent variable itself. To takethis into account, the instantaneous indirect effect (Θx) was used reflecting thechanges in job satisfaction / work engagement / emotional exhaustion / work-familyconflict extent derived from the impact of the organizational identification throughworkaholism at the specific level of organizational identification.
If the mediatorvariable is a linear function of the independent variable, Θx is a constant at differentvalues of the independent variable, i.e. the indirect effect achieved through themediator variable is equal regardless of the value of the independent variable. Threevalues reflecting the low level (–1 SD), the medium level (average value) and thehigh level (+1 SD) were taken as specific values of organizational identification forwhich the value of Θx were calculated.Polynomial regressions calculated for each well-being indicator demonstratethat with three of them the coefficient of the squared term is statistically differentfrom zero (job satisfaction: B = 0.059, [95% CI = 0.019 – 0.100], p = .004; emotionalexhaustion: B = -0.071, [95% CI = -0.118 – -0.024], p = .003; work engagement:B = 0.063, [95% CI = 0.019 – 0.100], p = .004).
Comparing the linear and polynomialmodels for each of the three well-being indicators shows that in all the three cases thepolynomial model describes empirical data better. However, the difference in theexplained variance between these two models is minimal, with ΔR2 varying from.003 to .004, i.е. adding the squared term to the model increases the explainedvariance by less than 1%. Comparing the correlations of organizational identificationwith each of these three indicators of the employee well-being before and after theturning point reveals that in all the cases correlations retain their signs and remainalmost similar in strength.
This implies that the growth in the organizationalidentification level, the level of job satisfaction and work engagement increases aswell, whereas the level of emotional exhaustion goes down both before and after theturning point. The visual comparison of the lines reflecting the linear and the nonlinear models of the relationship between organizational identification and well-beingindicators (see Figure 1) does not allow to conclude that there is a significant15difference in the level of fitting empirical data in linear and square equationsdescribing the relations between organizational identification and each of the fourwell-being indicators.
Thus, there is no reason to believe that organizationalidentification is in non-linear correlation with job satisfaction, emotional exhaustion,work engagement, and work-family conflict. Hypotheses 1а, 1b, 1c and 1d were notconfirmed.Figure 1. Linear and non-linear relationship between organizational identification and jobsatisfaction (A), emotional exhaustion (B), work-family conflict (C), work engagement (D). Thevariable values were standardized. The grey color is used to indicate the 95% confidence interval.16Similar analysis strategy was used to test hypothesis 2а on the non-linearrelationship between organizational identification and workaholism.
The polynomialregression shows that the coefficient for a squared term of the equation is statisticallydifferent from zero (B = 0.036, [95% CI = 0.018 – 0.053], p < .001). The Wald testdemonstrated that adding it to the model leads to the statistically significant increasein the volume of the explained variance of workaholism by 1%. Concurrently, thecorrelations between organizational identification and workaholism before and afterthe turning point are different: before the turning point organizational identificationand workaholism do not correlate, while after this point there is a positive correlation.This means that when low-level organizational identification increases, this growth isnot accompanied by the growth in workaholism. However, if the level oforganizational identification is high, its further increase goes hand in hand with therise in workaholism.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
