Role of Branding on Hotel Culture and its Effects on Orientation and Coaching Process in the Hotel Industry, Empirical Study of Five Star Hotel in Northern Cyprus.
Dissertation-Chapter 1 Data Analysis
4.1 Introduction
This chapter was directed to examine the suggested hypotheses. This chapter is comprised of five sections. The first section encompassed data screening processes; including missing data treatment, normality and outliers’ examination, and nonresponse bias and common method bias assessment. The second section described the data analysis method used in this study. The third section included the measurement and structural models’ assessment, and hypotheses testing. The last section summarized the chapter.
4.2 Data screening
Data screening refers to a set of procedures that should be executed before estimating parameters on data and testing hypotheses (Hair, Black, Babin, & Anderson, 2010). The main purpose of such activities is to ensure the readiness of data for further analysis (usually regression). An elementary step is to ensure that no coding errors have been made during data entry. Missing data identification and treatment are also necessary steps. Equally important is to examine for outliers and normal distribution of measures. A detailed description for each of these procedures is provided in the next sub-sections.
4.2.1 Data Entry Errors and Missing Data
Many errors might be made during data entry by mistake. This stage was designed to examine for data entry errors. This study examined this issue by inspecting if there are values out of a five-point Likert scale. The descriptive statistics showed that no value was out of the five-point Likert scale. Thus, it can be concluded that there were no errors during data entry. However, missing data is also a vital issue for regression analysis. According to rules of thumb, minor missing data can be replaced by mean or median; meanwhile serious missing data should be eliminated from the analysis (Hair et al., 2010). This study identified 33 questionnaires that suffered from seriously missing data. Thus, they were excluded from the analysis. Consequentially, all responses were completed and valid for further analysis.
-
Normality of Data Distribution.
Data distribution is a precondition for many statistical tests. Substantial departures from normal data distribution may distort the results of the multivariate analysis (Hair et al., 2010). Gefen, Rigdon, and Straub (2011) advocated the use of skewness and kurtosis measures to assess the normality of data distribution. According to the rule of thumb, skewness and kurtosis should be in the range of “+/− 2” as to demonstrate that the data distribution is not violating the normality assumption (Hair et al., 2010). This study examined the skewness and kurtosis of the individual items included in our data set. Table 4.1 showed that none of the items had skewness and kurtosis values outside the range “+/− 2”, indicating that our dataset met the normality assumption.
Table 4. 1: Descriptive Statistics of Skewness and Kurtosis
| Items | Skewness | Kurtosis | ||
| Statistic | Std. Error | Statistic | Std. Error | |
| a1 | -1.089 | .124 | .707 | .247 |
| a2 | -.911 | .124 | .119 | .247 |
| a3 | -.918 | .124 | -.416 | .247 |
| a4 | -.711 | .124 | -.600 | .247 |
| a5 | -.732 | .124 | -.875 | .247 |
| a6 | -.641 | .124 | -.907 | .247 |
| a7 | -.374 | .124 | -1.260 | .247 |
| a8 | -.764 | .124 | -.720 | .247 |
| b1 | -.736 | .124 | .491 | .247 |
| b2 | -1.087 | .124 | .033 | .247 |
| b3 | -1.039 | .124 | .216 | .247 |
| b4 | -1.209 | .124 | .780 | .247 |
| b5 | -1.087 | .124 | .506 | .247 |
| b6 | -.898 | .124 | .248 | .247 |
| b7 | -.599 | .124 | -1.245 | .247 |
| b8 | -.760 | .124 | .261 | .247 |
| b9 | -.634 | .124 | -.900 | .247 |
| c1 | -.821 | .124 | -.320 | .247 |
| c2 | -1.121 | .124 | .600 | .247 |
| c3 | -.782 | .124 | .012 | .247 |
| c4 | -.887 | .124 | -.405 | .247 |
| c5 | -.897 | .124 | -.497 | .247 |
| c6 | -.800 | .124 | -.661 | .247 |
| c7 | -.782 | .124 | .012 | .247 |
| d1 | -1.014 | .124 | .483 | .247 |
| d2 | -.941 | .124 | .045 | .247 |
| d3 | -.911 | .124 | .128 | .247 |
| d4 | -.807 | .124 | -.298 | .247 |
| d5 | -.724 | .124 | .368 | .247 |
| d6 | -.715 | .124 | .450 | .247 |
| d7 | -1.191 | .124 | 1.265 | .247 |
| d8 | -.770 | .124 | .473 | .247 |
| d9 | -1.034 | .124 | .635 | .247 |
| d10 | -.856 | .124 | -.245 | .247 |
| d11 | -1.026 | .124 | .260 | .247 |
| d12 | -.903 | .124 | -.146 | .247 |
4.2.3 Outliers’ detection
Outliers are defined as a set of observations that far-off from other observations in a random sample from a population (Hair et al., 2010). Outliers distort the results of multivariate analysis and increase estimation errors, and thus outliers should be eliminated. Outliers can be determined by measuring the extent to which a particular observation in a data set is far away from the normal distribution of the sample (Hair et al., 2010). In order to identify outliers in a data set, Hair et al. (2010) advocated the use standardized score which has zero mean and one standard deviation for each observation. Standardized score with a value above 3.5 indicates a potential outlier (Iglewicz & Hoaglin, 1993). In this study, the researcher estimated the standardized values (or Z scores) for each item in the model. Results in table 4.2 showed that none of the observations had a Z value above 3.5. Therefore, the author can conclude that the data set is free from outliers.
Table 4. 2: standardized scores
| Minimum | Maximum | Mean | Std. Deviation | |
| Statistic | Statistic | Statistic | Statistic | |
| Zscore(a1) | -3.03047 | .93501 | .0000000 | 1.00000000 |
| Zscore(a2) | -3.08910 | .96338 | .0000000 | 1.00000000 |
| Zscore(a3) | -2.49196 | .85661 | .0000000 | 1.00000000 |
| Zscore(a4) | -2.20818 | 1.07546 | .0000000 | 1.00000000 |
| Zscore(a5) | -2.16004 | .90200 | .0000000 | 1.00000000 |
| Zscore(a6) | -2.22705 | 1.01610 | .0000000 | 1.00000000 |
| Zscore(a7) | -2.05292 | 1.13776 | .0000000 | 1.00000000 |
| Zscore(a8) | -2.24423 | .94299 | .0000000 | 1.00000000 |
| Zscore(b1) | -3.08217 | 1.22786 | .0000000 | 1.00000000 |
| Zscore(b2) | -2.80092 | .81227 | .0000000 | 1.00000000 |
| Zscore(b3) | -2.60364 | .92873 | .0000000 | 1.00000000 |
| Zscore(b4) | -2.97038 | .87422 | .0000000 | 1.00000000 |
| Zscore(b5) | -2.73041 | .93858 | .0000000 | 1.00000000 |
| Zscore(b6) | -2.82839 | 1.09838 | .0000000 | 1.00000000 |
| Zscore(b7) | -1.78245 | .90424 | .0000000 | 1.00000000 |
| Zscore(b8) | -2.81204 | 1.20738 | .0000000 | 1.00000000 |
| Zscore(b9) | -1.99500 | 1.00040 | .0000000 | 1.00000000 |
| Zscore(c1) | -2.42782 | .97332 | .0000000 | 1.00000000 |
| Zscore(c2) | -2.86919 | .95640 | .0000000 | 1.00000000 |
| Zscore(c3) | -2.66539 | 1.13740 | .0000000 | 1.00000000 |
| Zscore(c4) | -2.31015 | .92908 | .0000000 | 1.00000000 |
| Zscore(c5) | -2.17766 | .88134 | .0000000 | 1.00000000 |
| Zscore(c6) | -2.02296 | .94532 | .0000000 | 1.00000000 |
| Zscore(c7) | -2.66539 | 1.13740 | .0000000 | 1.00000000 |
| Zscore(d1) | -2.87896 | .96960 | .0000000 | 1.00000000 |
| Zscore(d2) | -2.67707 | .98380 | .0000000 | 1.00000000 |
| Zscore(d3) | -2.80990 | 1.00230 | .0000000 | 1.00000000 |
| Zscore(d4) | -2.54524 | 1.03661 | .0000000 | 1.00000000 |
| Zscore(d5) | -2.97909 | 1.21330 | .0000000 | 1.00000000 |
| Zscore(d6) | -2.94040 | 1.28591 | .0000000 | 1.00000000 |
| Zscore(d7) | -3.00341 | 1.02892 | .0000000 | 1.00000000 |
| Zscore(d8) | -3.50264 | 1.20403 | .0000000 | 1.00000000 |
| Zscore(d9) | -2.81209 | 1.02663 | .0000000 | 1.00000000 |
| Zscore(d10) | -2.52628 | 1.01922 | .0000000 | 1.00000000 |
| Zscore(d11) | -2.69083 | .99537 | .0000000 | 1.00000000 |
| Zscore(d12) | -2.56438 | .98948 | .0000000 | 1.00000000 |
4.2.4 Non-response bias
Non-response bias is a critical threat in survey investigations and might endanger the empirical results (Armstrong & Overton, 1977). Nonresponse bias refers to the bias that outcomes when respondents differ in significant ways from non-respondents. The convenient way to assess such bias is to compare the characteristics of early respondents with late respondents. Late respondents are usually used as reprehensive for non-respondents (Armstrong & Overton, 1977; Hair et al., 2010). In line with this tradition, this study performed the independent sample t-test to compare the key measures between the early and late respondents. Results in table 4.3 indicated that the observed measures including Brand, Hotel Culture, Orientation, and Coaching of the late respondents were not significantly different from those of the early respondents. Accordingly, no significant non-response bias existed in the data set.
Table 4. 3: the independent sample t-test for the key measures
| Variables | t-test for Equality of Means | |||||
| t | Df | Sig. (2-tailed) | Mean Difference | Std. Error Difference | ||
| Brand | -.154 | 58 | .878 | -.02500 | .16205 | |
| Hotel Culture | -1.718 | 58 | .091 | -.22593 | .13150 | |
| Orientation | -1.838 | 58 | .071 | -.26667 | .14506 | |
| Coaching | -.123 | 58 | .903 | -.01667 | .13599 | |
4.2.5 Common method bias
Since this study collected perceptual data which is collected from a single source, the data might be susceptible to systematic biases due to the instrument used (Fuller, Simmering, Atinc, Atinc, & Babin, 2016). This issue refers to common method bias (CMB). The conventional approach to examine this issue is Harman’s one-factor test. The primary reasoning in this test is that if responses are systematically biased by a common scaling, then one factor would comprise most of the observed variance in all observed variables (Jarvis, MacKenzie, & Podsakoff, 2003; Podsakoff & Organ, 1986). This study performed a Principal Component Analysis to inspect if a one-factor accounts for most of the observed variance. The result indicated that no single factor accounted for the majority of the variance. This finding provides empirical evidence that CMB is not an issue for this study.
4.2.6 Multicollinearity assessment
Multicollinearity is a critical threat in a multiple regression and would mistakenly inflate path coefficients and provide exaggerated t-values. Multicollinearity refers to a situation in which one independent variable in a model can be linearly predicted from the other variable(s) with a considerable degree of accuracy. This study examined for multicollinearity by estimating the well-known criterion namely the variance inflation factors (VIF) (Hair et al., 2010; Henseler & Chin, 2010). Table 4.4 should the result of VIFs’ estimation. The results indicated that the highest variation inflation factor in the research model was less than 1.62, indicating that the regression estimates will not likely suffer from multicollinearity bias.
4.3 Data analysis
This section provided a detailed description of data analysis-related subjects. These included data analysis procedures, assessment of the measurement model, assessment of the structural model, and testing the suggested hypotheses. The last section summarised the data analysis section. As mentioned earlier, the method of analysis in this study is PLS-SEM. It allows for multi-layers of equations to be examined simultaneously alongside measurement. This study followed the succeeding procedures to analysis the data using PLS-SEM as suggested by Hair et al. (2011a). First, the author assessed the psychometric properties of the measurement model as recommended by the rule of thumb (Hair et al., 2011a). Both reliability and validity of the measurements were examined at this phase. Second, the author assessed the quality of the structural model and tested the suggested hypotheses. The hypotheses were assessed by estimating the direct effect and indirect effect between the covariates.
4.3.1 Measurement model assessment
The measurement model was examined by analysing its internal consistency, convergent and discriminant validity. As for internal consistency, the composite reliability of each construct in the model was estimated and a value of 0.70 and above is recommended (Hair et al., 2011a). Table 4.4 showed that the values of composite reliability ranged between 0.78 for Branding and 0.84 for Coaching, demonstrating an adequate level of internal consistency. Average variance extracted (AVE) and item loadings in order to examine the convergent validity (Hair et al., 2011a). Theoretically, convergent validity is attained when each construct accounts for at least half of the variance in its items. In addition, items loading should be 0.6 and above indicating that a great amount of the variance in each item is accounted for by its latent variable (Hair et al., 2011a). we dropped all the items that had a loading value of less than 0.6. Table 4.4 showed that the item loadings were well above the threshold of 0.6. In addition, AVE values ranged between 0.504 for Orientation and 0.57 for culture, indicating a satisfactory convergent validity.
Table 4. 4 Constructs reliability and convergent validity
| Construct | Composite Reliability | AVE | Item | Loading |
| Brand | 0.7805 | 0.543 | a4 | 0.7843 |
| a5 | 0.6988 | |||
| a6 | 0.7248 | |||
| Culture | 0.8023 | 0.5759 | b2 | 0.7702 |
| b5 | 0.8069 | |||
| b6 | 0.6953 | |||
| Orientation | 0.8018 | 0.5046 | c1 | 0.7054 |
| c4 | 0.6164 | |||
| c5 | 0.7815 | |||
| c6 | 0.728 | |||
| Coaching | 0.8432 | 0.5184 | d1 | 0.6805 |
| d3 | 0.7461 | |||
| d9 | 0.7332 | |||
| d10 | 0.7316 | |||
| d12 | 0.7069 |
As for discriminant validity, the square root of the AVE was estimated (Fornell & Larcker, 1981) and items cross-loading (Sarstedt, Ringle, Smith, Reams, & Hair, 2014). A construct exhibits discriminant validity if the square root of the AVE is higher than the correlation with other latent constructs (Fornell & Larcker, 1981); which is the case of all the constructs in the measurement model (see table 4.5). In addition, items loading should be higher on their postulated construct than any other constructs. Items cross-loading analysis, in table 4.6, showed that each item had a higher loading value on their postulated construct than any other constructs in the model. The results above show a great deal of discrimination validity.
Table 4. 5: Discriminant validity-AVE Squared correlations
| Construct | Orientation | Brand | Culture | Coaching |
| Orientation | 0.5046 | |||
| Brand | 0.2886 | 0.543 | ||
| Culture | 0.1731 | 0.2716 | 0.5759 | |
| Coaching | 0.3285 | 0.2593 | 0.2531 | 0.5184 |
Squared correlations, AVE in the diagonal.
Table 4. 6: Discriminant validity- Items Cross loading
| Item | Brand | Culture | Orientation | Coaching |
| a4 | 0.7843 | 0.3445 | 0.4344 | 0.4087 |
| a5 | 0.6988 | 0.3949 | 0.4373 | 0.3744 |
| a6 | 0.7248 | 0.4149 | 0.3038 | 0.3374 |
| b2 | 0.339 | 0.7702 | 0.3592 | 0.3996 |
| b5 | 0.4909 | 0.8069 | 0.3346 | 0.3837 |
| b6 | 0.3441 | 0.6953 | 0.2468 | 0.3641 |
| c1 | 0.3998 | 0.3449 | 0.7054 | 0.3208 |
| c4 | 0.3697 | 0.2057 | 0.6164 | 0.4722 |
| c5 | 0.3484 | 0.3368 | 0.7815 | 0.3768 |
| c6 | 0.4059 | 0.2796 | 0.728 | 0.4753 |
| d1 | 0.2505 | 0.2851 | 0.4683 | 0.6805 |
| d3 | 0.3841 | 0.3697 | 0.3947 | 0.7461 |
| d9 | 0.3332 | 0.39 | 0.3916 | 0.7332 |
| d10 | 0.3545 | 0.312 | 0.3816 | 0.7316 |
| d12 | 0.4613 | 0.4202 | 0.4378 | 0.7069 |
4.3.2 Structural model assessment
Next, the R-square value and path coefficients along with their t-statics and standard errors were estimated to assess the structural model and test the proposed hypotheses. The R-square value shows the predictive ability of the model and the ability of exogenous constructs to explain the endogenous constructs (Hair et al., 2014a). The results of the structural model estimate are presented in table 4.7 and depicted in figure 1. The results show that the model explains 27%, 31%, and 33% of the variance in Culture, Orientation, and Coaching, respectively. These figures show an adequate predictive power in the suggested model (Hair et al., 2014a). The results show that there is a positive and significant relationship between branding and hotel culture (Path coefficient= 0.521, p< 0.000) as stated by H1. Thus, hypothesis H1 was accepted. The results also show that Branding had a positive and significant effect on both employees coaching (Path coefficient= 0.339, p< 0.000) and employees’ orientation (Path coefficient= 0.440, p< 0.000). Therefore, hypotheses H2 and H3 were empirically supported. Similarly, hotel culture had a positive and significant effect on both employees coaching (Path coefficient= 0.326, p< 0.000) and employees’ orientation (Path coefficient= 0.187, p< 0.01) as stated in the hypotheses H5 and H6. Accordingly, hypotheses H5 and H6 were empirically accepted.
Table 6: Path coefficients bootstrapping estimates (Direct effects)
| Effect | Path coefficient | Standard bootstrap results | |||
| Standard error | t-value | p-value (2-sided) | p-value (1-sided) | ||
| Brand -> Orientation | 0.4398 | 0.0577 | 7.6208 | 0.000 | 0.000 |
| Brand -> Culture | 0.5211 | 0.0497 | 10.4934 | 0.000 | 0.000 |
| Brand -> Coaching | 0.3392 | 0.0525 | 6.4647 | 0.000 | 0.000 |
| Culture -> Orientation | 0.1869 | 0.0695 | 2.689 | 0.0073 | 0.0036 |
| Culture -> Coaching | 0.3263 | 0.0664 | 4.9113 | 0.000 | 0.000 |
Figure 1: the estimate of the structural model
With regard to hypotheses H7 and H8 that proposed hotel culture as a mediator between branding and each of coaching and orientation, the author estimated the significance of the indirect effects between branding and each of coaching and orientation, as recommend by Nitzl et al., (2016) and Preacher and Hayes (2008). The author also estimated the Bootstrap Confidence intervals as a further test to examine the significance of the indirect effects (Nitzl et al., 2016; Preacher & Hayes, 2008). Table 4.7 showed the PLS estimate of the indirect effects. The results indicate that the indirect effect between branding and employees coaching was positive and significant (Path coefficient= 0.170, p< 0.000). Likewise, the indirect effect between branding and employees coaching was positive and significant (Path coefficient= 0.097, p< 0.01). In addition, none of the Bootstrap Confidence intervals of both indirect effects includes zero. These figures provide a strong empirical support for H7 and H8.
Table 4. 7: Path coefficients bootstrapping estimates (Indirect effects)
| Effect | Path coefficient | Standard bootstrap results | Bootstrap Confidence intervals | ||||||
| Standard error | t-value | p-value (2-sided) | p-value (1-sided) | 0.50% | 2.50% | 97.50% | 99.50% | ||
| Brand -> Orientation | 0.0974 | 0.0371 | 2.6242 | 0.0088 | 0.0044 | 0.0118 | 0.0294 | 0.1698 | 0.1985 |
| Brand -> Coaching | 0.1701 | 0.0405 | 4.1958 | 0.000 | 0.000 | 0.0793 | 0.0985 | 0.2529 | 0.2874 |
Overall, we conducted several analyses to examine the proposed model. The measurement model expressed a satisfactory level of reliability, and convergent and discriminant validity. The structural analysis provided empirical support for the suggested hypotheses. Table 4.8 summarized the result of hypotheses testing.
Table 4. 8: summary of hypotheses testing
| No. | Hypotheses | Result |
| H1 | There is statistical significance influence of branding on hotel culture. | Accepted |
| H2 | There is positive statistical significance influence of branding on hotel’s employees’ orientation. | Accepted |
| H3 | There is positive statistical significance influence of branding on hotel’s employees’ coaching. | Accepted |
| H4 | There is positive statistical significance influence of hotel culture on hotel’s employees’ orientation | Accepted |
| H5 | There is positive statistical significance influence of hotel culture on hotel’s employees coaching. | Accepted |
| H6 | Hotel culture mediates the relationship between branding and orientation. | Accepted |
| H7 | Hotel culture mediates the relationship between branding and coaching. | Accepted |
4.4 Summary of the chapter
This chapter conducted several analyses to examine the suggested hypotheses. Several data decreeing activities confirmed that the data were suitable for regression analysis. The psychometric properties of the measurement model were examined, and the results confirmed the reliability and validity properties of the model. The results of hypotheses testing provided support for all the hypotheses. Specifically, both branding and hotel culture were important predictors for orientation and coaching. And hotel culture was an important mediator that transformed the effect of branding on orientation and coaching. The next chapter discussed the results and their implications in detail.
References
Alsaad, A. K., Yousif, K. J., & AlJedaiah, M. N. (2018a). Collaboration: the key to gain value from IT in supply chain. EuroMed Journal of Business, 13(2), EMJB-12-2017-0051. http://doi.org/10.1108/EMJB-12-2017-0051
Alsaad, A., Mohamad, R., & Ismail, N. A. (2017). The moderating role of trust in business to business electronic commerce (B2B EC) adoption. Computers in Human Behavior, 68, 157–169. http://doi.org/10.1016/j.chb.2016.11.040
Alsaad, A., Mohamad, R., & Ismail, N. A. (2018b). The contingent role of dependency in predicting the intention to adopt B2B e-commerce. Information Technology for Development, 0(0), 1–29. http://doi.org/10.1080/02681102.2018.1476830
Armstrong, J. S., & Overton, T. . (1977). Estimating nonresponse bias in mail surveys. Journal of Marketing Research, 14(August), 396–402.
Dillman, D. A., Smyth, J. D., & Christian, L. M. (2014). Internet, phone, mail, and mixed-mode surveys: the tailored design method. John Wiley & Sons.
do Valle, P. O., & Assaker, G. (2016). Using partial least squares structural equation modeling in tourism research: A review of past research and recommendations for future applications. Journal of Travel Research, 55(6), 695–708.
Elrehail, H., Emeagwali, O. L., Alsaad, A., & Alzghoul, A. (2017). The Impact of Transformational and Authentic Leadership on Innovation in Higher Education: The Contingent Role of Knowledge Sharing. Telematics and Informatics. http://doi.org/10.1016/j.tele.2017.09.018
Fornell, V., & Larcker, C. (1981). Evaluating structural equation models with observable variables and measurement error. Journal of Marketing, 18(1), 39–50.
Fuller, C. M., Simmering, M. J., Atinc, G., Atinc, Y., & Babin, B. J. (2016). Common methods of variance detection in business research. Journal of Business Research, 69(8), 3192–3198. http://doi.org/10.1016/j.jbusres.2015.12.008
Gefen, D., Rigdon, E. E., & Straub, D. W. (2011). An Update and Extension to SEM Guidelines for Administrative and Social Science Research. MIS Quarterly, 35(2).
Gujarati, D. N. (2009). Basic econometrics. Tata McGraw-Hill Education.
Hair, J. F., Black, W. C., Babin, B. J., & Anderson, R. E. (2010). Multivariate Data Analysis: A global perspective. Analysis.
Hair, J. F., Hult, J. G. T. M., Ringle, C. M., & Sarstedt, M. (2014a). A Primer on Partial Least Squares Structural Equation Modeling (PLS-SEM). SAGE Publications.
Hair, J. F. J., Sarstedt, M., Hopkins, L., & Kuppelwieser, V. G. (2014b). Partial least squares structural equation modeling (PLS-SEM): An emerging tool in business research. European Business Review, 26(2), 106–121. http://doi.org/10.1108/EBR-10-2013-0128
Hair, J. F., Ringle, C. M., & Sarstedt, M. (2011a). PLS-SEM: Indeed, a Silver Bullet. Journal of Marketing Theory and Practice, 19(2), 139–151. http://doi.org/10.2753/MTP1069-6679190202
Hair, J. F., Sarstedt, M., Ringle, C. M., & Mena, J. a. (2011b). An assessment of the use of partial least squares structural equation modeling in marketing research. Journal of the Academy of Marketing Science, 40(3), 414–433. http://doi.org/10.1007/s11747-011-0261-6
Henseler, J., & Chin, W. W. (2010). A Comparison of Approaches for the Analysis of Interaction Effects Between Latent Variables Using Partial Least Squares Path Modeling. Structural Equation Modeling, 17, 82–109. http://doi.org/10.1080/10705510903439003
Iglewicz, B., & Hoaglin, D. C. (1993). How to detect and handle outliers (Vol. 16). Asq Press.
Jarvis, C. B., MacKenzie, S. B., & Podsakoff, P. M. (2003). A critical review of construct indicators and measurement model misspecification in marketing and consumer research. , 30(2), Journal of Consumer Research, 30(2), 199–218.
Nitzl, C., Roldan, J. L., & Cepeda, G. (2016). Mediation analysis in partial least squares path modelling, Helping researchers discuss more sophisticated models. Industrial Management and Data Systems, 116(9), 1849–1864. http://doi.org/10.1108/IMDS-07-2015-0302
Podsakoff, P. M., & Organ, D. W. (1986). Self-reports in organizational research: Problems and prospects. Journal of Management, 12(4), 531–544.
Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40(3), 879–891. http://doi.org/10.3758/BRM.40.3.879
Ringle, C. M. (2012). A Critical Look at the Use of PLS-SEM in MIS Quarterly. MIS Quarterly, 36(1).
Sarstedt, M., Ringle, C. M., Smith, D., Reams, R., & Hair, J. F. (2014). Partial least squares structural equation modeling (PLS-SEM): A useful tool for family business researchers. Journal of Family Business Strategy, 5(1), 105–115. http://doi.org/10.1016/j.jfbs.2014.01.002
Sekaran, U., & Bougie, R. (2010). Research Methods For Business (Fifth Edit). John Wiley & Sons, Inc. New York, NY, USA.
Zikmund, W., Babin, B., Carr, J., & Griffin, M. (2012). Business research methods. Cengage Learning.
