Introduction
In Organizational behaviour, most of the personality research is based on the big five personality construct. Only a decade ago(Judge, Locke, & Durham, 1997) proposed a higher-order personality construct named the Core Self- evaluations (CSE), which brings up to fundamental appraisals of people as they build their own self worth, competency and capabilities. A core self-evaluation defined how an individual sees her or himself, they also involve how a person perceives and assesses situations(Durán, Extremera, & Rey, 2010).
CSE has become an important research theme of investigation in the organizational sciences. Research indicated that core self-evaluations (CSE) is related to diverse phenomena such as job satisfaction(Dormann, Fay, Zapf, & Frese, 2006)(Judge, Heller, & Klinger, 2008)(Zhang, Wu, Miao, Yan, & Peng, 2014), employee engagement(Shorbaji, Messarra, & Karkoulian, 2011), leadership(Nübold, Muck, & Maier, 2013)(Kim, Liden, Kim, & Lee, 2014), Organizational tenure(McNamara & Pitt-Catsouphes, 2014),workplace creativity (Chiang, Hsu, & Hung, 2014),Work Stress(Morris, Messal, & Meriac, 2013) , Emotional Intelligence (Sun, Wang, & Kong, 2014) and Employee attitude(Thurston, D’Abate, & Eddy, 2012).
(Judge et al., 1997)showed that the core self-evaluations (CSE) wasbased on three criteria (a) evaluation focuses – the degree to which trait involves evolution like self esteem (b) fundamentalist - the concept of self like personality theories and (c) scope - refers to how broad and narrow a trait. On the foundation of Self-esteem, generalized self-efficacy, emotional stability and locus of control were provisional known as CSE traits. Generalized self-efficacy is an estimate of one’s fundamental ability to cope with the wide range of situation(Chen, Gully, & Eden, 2001). Self-esteem is an overall assessment of one’s self worth(Rosenberg, 1965), emotional stability (neuroticism) is the tendency to feel secure(Eysenck, 1990) and locus of control - the individual believes that they manipulate events in their lives(Rotter, 1966).
(Johnson, Rosen, & Levy, 2008) argued that the individual has high core self-evaluations, feel positively about themselves, have power over their lives, and are confident in their capabilities. On the other side, individuals having low core self-evaluations think negatively of themselves, do not feel in control of their own lives, and be short of confidence. Therefore, CSE has the power, to capture the fundamental essence of individual self-evaluation, and it is used globally to measure the personality measurement.
(Judge, Locke, Durham, & Kluger, 1998) argued that each one of these traits was a strong predictor of job satisfaction. They later concluded that these traits have a propensity to be highly correlated, and hence should be evaluated as parts of one latent construct CSE(Judge, Erez, Bono, & Thoresen, 2002).
Core self-evaluations (CSE) has risen as a valid indicator of both affective and target work results. Various studies have indicated the relationship between CSE and job satisfaction(Bono & Judge, 2003). Overall, the empirical research recommended that positive attitudes and behavioural effects are produced by both encouraging environmental conditions consolidated with the "right" worker disposition – such as, high CSE.Besides, people with high scores of CSE reported low levels of stress and conflict, including work, family conflict were more effective in their careers, cope more efficiently with setback, demonstrates more productive responses to input, benefit more from favourable circumstances and opportunities and have higher earnings(Judge, 2009).
The aim of the study was to calculate the reliability and validity of the core self-evaluation scale in the Indian sample through structural equation modelling. Exploratory factor analysis and confirmatory factor analysis were used to identify the factors in CSE. Following that, we presented the method used in the research and then, data were analyzed and the results were discussed to make the final remark on this research.
Methodology
The objective of this research was twofold, firstly to the reliability and validity of core self evaluation in the IT sector employees in India, and secondly to test a model fit test of core self evaluation in Indian context. The population of the research defined as those who were working in information technology(IT) company and completed the minimum one year in the company. The sample of the survey was conducted on 425 employees by using structured survey in NCR (National Capital region) However, analysis was done using only 410 samples due to 15 responses had to be diminished because of responses error in the field. The sample was collected three information technology (IT) companies in the National capital region (NCR). Respondents for this work have been selected using quota and convenience sampling technique. Selection of the company based on revenue of the organization like a small company, medium company and large company. Every company was allotted the certain quota like a large company for 150 respondents, medium company for 140 respondents and 135 respondents for small company. Now convenience sampling was used for collecting the employee responses in different organizations. The Data collection was done in two ways sending the mailed to the employees and visited to the organization.
Instrument
The core self-evaluations scale (CSES) was applied to evaluate CSE in each participant. 5 point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree) was used to measure 12 statements in CSES. The latent construct CSE was having four specific traits of self-esteem, generalized self-efficacy, emotional stability, and locus of control. The sample statements such as “Overall, I am satisfied with myself” and “There are times when things look pretty bleak and hopeless to me” were used in the research. Item coding of the statements and CSES scale were given in appendix 1.
Result analysis
Statistical Package for social science (SPSS) and Analysis of Moment Structure (AMOS) package for structural equation modelling were employed to execute the data analysis. In this research, exploratory factor analysis, confirmatory factor analysis, reliability, and validity analysis were performed. KMO test is to verify the adequacy of the data and Bartlett’s test of sphericity for evaluating the hypothesis that is not identity matrix. KMO value was 0.842, which was higher than cut-off value 0.5. The Bartlett’s test was significance (chi-square=2993.604, df=66, p=0. 001) (Table1).
Table 1
KMO and Bartlett's TestKaiser-Meyer-Olkin Measure of Sampling Adequacy. / .842
Bartlett's Test of Sphericity / Approx. Chi-Square / 2993.604
df / 66
Sig. / .000
The Communalities explained the sum of the squared factor loadings of a variable in all the factors. The highest communality value was 0.854 (I am confident I get the success I deserve) and the lowest communality value was 0.757 (I do not feel in control of my success in my career) which was higher than standardized value 0.5 (Table 2).
Table 2
CommunalitiesInitial / Extraction
I am confident I get the success I deserve. / 1.000 / .854
Sometimes I feel depressed. / 1.000 / .829
When I try, I generally succeed. / 1.000 / .818
Sometimes when I fail, I feel worthless. / 1.000 / .803
I complete tasks successfully. / 1.000 / .784
Sometimes I do not feel in control of my work. / 1.000 / .815
Overall I am satisfied with myself. / 1.000 / .804
I am filled with doubts about my competence. / 1.000 / .817
I determine what will happen in my life. / 1.000 / .789
I do not feel in control of my success in my career. / 1.000 / .757
I am capable of coping with most of my problems. / 1.000 / .794
There are times when things look pretty bleak and hopeless to me. / 1.000 / .814
Extraction Method: Principal Component Analysis.
Exploratory factor analysis is a multivariate technique broadly utilized as a part of social, behaviors, and management sciences. Exploratory factor analysis is regularly applied to investigate the dimensionality of an estimation instrument by multivariate data structures.EFA was conducted utilizing the principal component analysis method and varimax rotation. 12 statements were extracted in four factors which was indicated by eigenvalues 1 or greater than 1 and total variance explained by 80.64 percent (Table 3). In this research first, second, third and fourth factors wereexplained as 44.84 percent, 14.01 percent, 12.33 percent and 9.43 percent variance respectively (Table 3). Screen plot shown in figure 1.
1
Table 3
Total Variance ExplainedComponent / Initial Eigenvalues / Extraction Sums of Squared Loadings / Rotation Sums of Squared Loadings
Total / % of Variance / Cumulative % / Total / % of Variance / Cumulative % / Total / % of Variance / Cumulative %
1 / 5.382 / 44.849 / 44.849 / 5.382 / 44.849 / 44.849 / 2.481 / 20.672 / 20.672
2 / 1.682 / 14.019 / 58.868 / 1.682 / 14.019 / 58.868 / 2.469 / 20.573 / 41.244
3 / 1.480 / 12.336 / 71.204 / 1.480 / 12.336 / 71.204 / 2.373 / 19.777 / 61.021
4 / 1.133 / 9.438 / 80.642 / 1.133 / 9.438 / 80.642 / 2.355 / 19.621 / 80.642
5 / .381 / 3.175 / 83.818
6 / .352 / 2.933 / 86.750
7 / .341 / 2.840 / 89.590
8 / .323 / 2.688 / 92.279
9 / .282 / 2.348 / 94.627
10 / .252 / 2.101 / 96.728
11 / .215 / 1.794 / 98.522
12 / .177 / 1.478 / 100.000
Extraction Method: Principal Component Analysis.
1
Factor loading is defined as the correlation between factor and variables. In this research, four factors were identified as first (self-esteem), second (neuroticism), third (generalized self-efficacy), and fourth (locus of control) (Table 4). In self-esteem, factor highest loading (I am confident I get the success I deserve) was 0.88 and lowest loading (When I try, I generally succeed) was 0.85. Neuroticism factor having the highest loading (Sometimes I feel depressed) was 0.88 and lowest loading (Sometimes when I fail, I feel worthless) was 0.86. 0.84 was the highest loading (I am filled with doubts about my competence) and 0.81 was the lowest loading (I complete tasks successfully) in generalized self-efficacy. The last factor locus of control was having the highest loading (I determine what will happen in my life) was 0.85 and the lowest (I do not feel in control of my success in my career) was 0.81. All the variables were having loading equal to 0.7 and greater than 0.7, which indicated that factor extracted adequate variance of that variable (Table 4).
Figure 1
1
Table 4
Rotated Component MatrixaComponent
1 / 2 / 3 / 4
I am confident I get the success I deserve. / .882
Overall I am satisfied with myself. / .860
When I try, I generally succeed. / .853
Sometimes I feel depressed. / .884
There are times when things look pretty bleak and hopeless to me. / .867
Sometimes when I fail, I feel worthless. / .860
I am filled with doubts about my competence. / .844
I am capable of coping with most of my problems. / .826
I complete tasks successfully. / .814
I determine what will happen in my life. / .852
Sometimes I do not feel in control of my work. / .837
I do not feel in control of my success in my career. / .810
Extraction Method: Principal Component Analysis.
Rotation Method: Varimax with Kaiser Normalization.a
a. Rotation converged in 5 iterations.
Factor 1- Self-esteem, Factor 2- Neuroticism, Factor 3- Generalized self-efficacy, Factor 4- Locus of control
1
Reliability
Reliability is identified with the consistency of the measurement that is the degree to which the instrument produces the same results on rehashed trials(Babbie & Mouton, 2001). Reliability estimates the correlation between test score and theoretical true value of the variable. Cronbach’s Alpha is used to assess the reliability(Cronbach, 1951). The Cronbach’s Alpha value should be greater than the standardized value 0.7(Litwin, 1995). The cronbach’s Alpha for the core self-evaluations (CSE) was 0.88, 0.89 for self-esteem, 0.88 for Neuroticism, 0.87 for generalized self efficacy and 0.86 for locus of control (table 5). Reliability criteria were met for core self-evaluation and factors.
Table 5
Cronbach's Alpha / N of Items / N of respondent / EvaluationSelf-esteem / .892 / 3 / 410 / Good
Neuroticism / .885 / 3 / 410 / Good
Generalized self-efficacy / .872 / 3 / 410 / Good
Locus of control / .862 / 3 / 410 / Good
Core self- evaluations / .887 / 12 / 410 / Good
Validity
Validity is characterized as the extent to which that specific scale measures what it is intended to measure as a typical phenomena. So, the instrument should be accurate in what it intends to measure. Various types of measure are used for validity like content validity, construct validity and criterion validity(Carmines & Zeller, 1979). Nevertheless, in social science and management research, we used convergent and discriminant validity. Convergent validity of the construct was estimated by composite reliability (CR) and Average variance extracted (AVE)(Aibinu, Ling, & Ofori, 2011). The composite reliability measure is utilized to determine how well a construct is measured by its assigned indicator. The composite reliability, value should be greater than 0.7(Chin, 1998)(Joe F Hair, Ringle, & Sarstedt, 2011). The average variance extracted (AVE) is used to assess internal consistency of the construct by measuring the amount of variance that a latent variable capture from its measurement items and to the measure of variation because of estimation error(Fornell & Larcker, 1981). Adding in Average variance extracted (AVE) value should be greater than 0.5(Joseph F Hair, Black, Babin, Anderson, & Tatham, 2006). In this research Composite reliability for generalized self-efficacy, self esteem, Neuroticism, locus of control was having value of 0.87, 0.89, 0.88, 0.86 respectively which was higher than standardized value of 0.7 (Table 6). The Average variance extracted (AVE) values were raging from 0.67 to 0.73 which was higher than cut-off value 0.5. Discriminant validity shows the degree to which a known contrast is different from other constructs(Hulland, 1999). It is measured by comparing Average variance extracted (AVE) of the construct itself and otherconstructs. To test the discriminant validity, diagonals value should be higher than the particular rows and column corresponding values (Hulland, 1999) (Table 7). In this study, all the criteria were met for convergent and discriminant validity.
Table 6- Convergent validity analysis
Construct / Variable / Item code / Factor Loading / Composite Reliability / Average varaiance Extracted (AVE)Core self evaluations / Self esteem / CSE1 / 0.892 / 0.893 / 0.736
CSE7 / 0.825
CSE3 / 0.856
Neuroticism / CSE2 / 0.860 / 0.886 / 0.721
CSE12 / 0.851
CSE4 / 0.837
Generalized self-efficacy / CSE8 / 0.850 / 0.873 / 0.696
CSE11 / 0.824
CSE5 / 0.828
Locus of control / CSE9 / 0.790 / 0.863 / 0.677
CSE6 / 0.881
CSE10 / 0.795
CR: Composite reliability, AVE: Average variance extracted
Table 7 -discriminant validity analysis
GENER / SELFE / NEURO / LOCUSGENER / 0.834
SELFE / 0.506 / 0.858
NEURO / 0.447 / 0.376 / 0.849
LOCUS / 0.599 / 0.459 / 0.392 / 0.823
GENER: generalized self-efficacy, SELFE: self esteem, NEURO: Neuroticism, LOCUS: Locus of control
Confirmatory factor analysis
Exploratory factor analysis (EFA) is usually tied in to investigate the dimensionality of an estimation instrument by multivariate data analysis while confirmatory factor analysis (CFA) is a statistical method applied to confirm the component construction for a particular lot of observable variables(Joseph F Hair et al., 2006).EFA is usually applied in the circumstances in which the actual factorial structure or even the actual dimension of the scale for a given population is not known, commonly in the situation associated developing new scale. On the other hand, confirmatory factor analysis is used within conditions exactly where you have some knowledge in the dimensionality of the variables under study, sometimes it's based on empirical finding or theory.
In this research, four components (self-esteem, Neuroticism, generalized self-efficacy, locus of control) were identified through exploratory factor analysis (EFA) andconfirmatory factor analysis (CFA) was used to verify these factors in the factor analysis with the help of Analysis of Moment Structure (AMOS) (Figure 2). Structural equation modelling (SEM) was done with the aid of statistical package AMOS(Arbuckle, 2007). The hypothesised model was assessed with the assistance of CFA. In this model, total numbers of variables were 28, out of which 16 were exogenous and 12 were endogenous variables.The standardised regression weights for all the variable were significant at p<0.001 (Table 6). Measurement was performed to insure adequacy of the parameter estimates of the model. Three criteria were used, first feasibility of parameter estimates, second the appropriateness of the standard error and third statistical significance of the parameter estimates. In this study, all the parameters (regression weights, covariance) were feasible and standard errors were minimum (Table 8) (table 9). P values were significant at 0.001 and critical ratios were greater than 1.96, which indicated the goodness fit of the model.
Figure 2- Confirmatory factor analysis
SELFE: self esteem, NEURO: Neuroticism, GENER: generalized self-efficacy,, , LOCUS: Locus of control
The model was hypothesized and goodness of fit criteria used with the support of SEM, AMOS. For assessment, the model numbers of goodness of fit criteria were used in the research. The chi square value indicates the model fit and suggest that the ratio ( /DF) in the range 1 to 3 acceptable fit for the model(Hu & Bentler, 1999). Goodness of fit index (GFI) is a measure of the relative sum of variance and covariance(Byrne, 2013). The GFI value ranges from 0 to 1 and a cut-off value of GFI 0.9 which indicates a good fit of the data(Miles & Shevlin, 2007). Adjusted goodness of fit index (AGFI) differs from GFI just in the actuality that it changes in the quantity of degrees of chance in the predetermined model. AGFI value accepts greater than or equal to 0.9 and it represents a well fitting model. Root mean square residual (RMR) value accepts less than 0.05 for good fit of the data(Byrne, 2013). Root mean square error of approximation (RMSEA) is the most evaluative criteria for the model and RMSEA indicates the error of approximation in the population(Browne & Cudeck, 1993). RMSEA express the complexity of model and value 0.01 to 0.08 indicates the goodness fit of the data(MacCallum, Browne, & Sugawara, 1996). Normed fit index (NFI) also known as the Bentler-Bonett normed fit index and itscut-off value greater than 0.9 which represents the best fit of the model(Bentler & Bonett, 1980). Tucker-Lewis coefficient (TLI) is also known as non-normed fit index (NNFI). The NNFI adjusts the NFI for the number of degrees of the freedom in the hypothesized model(Bentler & Bonett, 1980). The Indicated value of NNFI is 0.9 for good fit of the data. Comparative fit index (CFI) indicates that model fit data good as in the logic that hypothesized model sufficiently describe the sample data(Bentler, 1990). The CFI standard value is 0.9 for good fit of the data(Hu & Bentler, 1999). Incremental fit index (IFI) cut-off value is0.9, which indicates a good, fit of the data.
Table 8- Parameter estimates (Regression weights)
Estimate / S.E. / C.R. / P / LabelCSE1 / <--- / SELFE / 1.000
CSE7 / <--- / SELFE / .922 / .045 / 20.697 / ***
CSE3 / <--- / SELFE / 1.031 / .047 / 21.718 / ***
CSE2 / <--- / NEURO / 1.000
CSE12 / <--- / NEURO / 1.062 / .053 / 20.100 / ***
CSE4 / <--- / NEURO / 1.017 / .051 / 19.768 / ***
CSE8 / <--- / GENER / 1.000
CSE11 / <--- / GENER / .942 / .050 / 18.835 / ***
CSE5 / <--- / GENER / .897 / .047 / 18.936 / ***
CSE9 / <--- / LOCUS / 1.000
CSE6 / <--- / LOCUS / 1.133 / .063 / 17.978 / ***
CSE10 / <--- / LOCUS / .986 / .059 / 16.633 / ***
The hypothesized model found a better fit in CFA. The ratio (CMIM=95. 683/DF=48) was 1.993 which indicated goodness fit. In this model GFI=0. 956, AGFI=0. 943, NFI=0. 968, IFI=0. 984, TLI=0. 978, CFI=0. 984 values were greater than 0.9, which showed goodness fit of the model. RMSEA and RMR value were 0.04, 0.03 respectively, which was less than 0.08 that indicated better fit of the model. So Structural equation modelling (SEM) confirmed four factors in core self-evaluationscale, which supported the prior research.