Analysis Of Dataset With 7 Variables And 216 Entries

Describe about the Research Paper For Business and Engineering Fields.

The dataset for analysis contains 7 variables with 216 entries. The value for the 7 variables is as given in the table below.

Table 1

Variable	Sum	Value
Gender	216	Male Female
Nationality	216	United Arab Emirates- Jordan Pakistan South Africa India Lebanon Andorra Greece United Kingdom Nepal Sudan United States Yemen Russia Morocco
Current Age	216	Less than 25 25 – 35 36 – 45 46 – 55 56 – 65 More than 65 years
Educational Level	216	High Diploma Masters Bachelors High School or less Pree Degree
Job Level	216	Lower Management Middle Management Junior Employee Junior Employee Outsource Accountant Top Management Inspector Sales coordinator Officer Office Administrator Grade 8
Field of Work	216	Business Medicine Maintenance Information  Technology Engineering Office Work Aviation Laboratory Logistics Corporate Affairs Tourism and Hospitality Shift Supervisor HSSE Legal Media, Communications or Public Relations
Years Of Experience	216	2-5 6-10 11-15 16-20 21-25 More than 25

The total number of respondents for the questions was 216.

From table 1 above, it is noted that majority of the respondents are male with a percentage of 74.1% while females have a percentage of 25.1. Additionally, the United Arab Emirates have a 59.3% respondent score which is the highest, while Argentina, Yemen, United States of America and Nigeria record among the lowest entries at 0.5% each. From the statistics, majority of the workers are of the age between 25 and 48 making it a total of 48.6%. 

Additionally, most of the workers assessed in the research project are from the engineering and business faculties at 22.2 and 19.0% respectively. Whereas, the level of education with the most entries is the bachelors degree followed by masters. Such may explain the relatively big percentage of the entrants in the business and engineering fields.

Table 4: Descriptive statistics (Mean, Variance...)

Descriptive Statistics
	N	Minimum	Maximum	Sum	Mean		Variance
	Statistic	Statistic	Statistic	Statistic	Statistic	Std. Error	Statistic
HPHRP 4	216	1	5	654	3.03	.078	1.311
HPHRP 5	216	1	5	665	3.08	.073	1.143
HPHRP 3	216	1	5	675	3.12	.072	1.124
HPHRP 2	216	1	5	685	3.17	.074	1.194
HPHRP 1	216	1	5	695	3.22	.069	1.036
POF 3	216	1	5	756	3.50	.061	.791
POF 5	216	1	5	761	3.52	.068	.986
POF 2	216	1	5	762	3.53	.071	1.078
JBT 3	216	1	5	778	3.60	.066	.929
JBT 2	216	1	5	794	3.68	.061	.797
POF 4	216	1	5	808	3.74	.067	.974
POF 1	216	1	5	809	3.75	.062	.833
JBT 1	216	1	5	809	3.75	.059	.758
POF 6	216	1	5	810	3.75	.068	1.007
IWB 2	216	1	5	832	3.85	.062	.843
IWB 1	216	1	5	834	3.86	.063	.864
IWB 5	216	1	5	842	3.90	.068	.985
IWB 6	216	1	5	844	3.91	.059	.745
JBE 1	216	1	5	847	3.92	.066	.929
IWB 4	216	1	5	851	3.94	.058	.727
IWB 3	216	1	5	857	3.97	.065	.915
HPHRP 6	216	1	5	867	4.01	.062	.842
JBE 2	216	1	5	876	4.06	.059	.760
JBE 3	216	1	5	911	4.22	.053	.618

From table 3, the average response from the respondents was that of above average opinion, i.e. they mostly have an average of 3, however, majority of respondents reported better HR in their work places. The variance of the dataset variable range from 0.618 to 1.311

Scale Statistics
Mean	Variance	Std. Deviation	N of Items
88.06	230.917	15.196	24

The Overall Variance is 230.917 and the mean is 88.06.

The Cronchbach’s Alpha statistic for the dataset is 0.943, while the Cronchbach’s Alpha based on Standardized Items is 0.944 for 24 items which according to  Gilem and Gilem(2013), the closer the Cronchbach’s statistic is closer to 1 the greater the internal consistency of the model. Therefore the model used for analysis is consistent.

Table 5: Reliability tests

Reliability Statistics
Cronbach's Alpha	Cronbach's Alpha Based on Standardized Items	N of Items
.943	.944	24

The inter-item Covariance mean is 0.378 while the inter-item correlation is 0.414 for the 24 items analyzed.

Table 6: Inter-Item test statistics

Summary Item Statistics
	Mean	Minimum	Maximum	Range	Maximum / Minimum	Variance	N of Items
Inter-Item Covariance	.378	.078	.834	.756	10.739	.019	24
Inter-Item Correlations	.414	.086	.934	.848	10.827	.020	24

Construct validity of the questionnaire is conducted by use of the Pearson correlation coefficient test. From table 7, the Pearson correlation coefficient of the variables is 0.000 which is less than 0.05 at 95% level of significance, indicating that all the questionnaire variable are valid for analysis. 

In order to explore the underlying theoretical structure of the dataset, an exploratory factor analysis is conducted to examine the relationship between the variable chosen and the respondent.

The sample is from a normal distribution

The sample size is of size n=200 or more

From the component table, it is clear that only four components are extracted using the principle component analysis (Williams and Brown, 2010). 

Component Matrix
	Component
	1	2	3	4
HPHRP 1	.555	.539	.209	.049
HPHRP 2	.561	.570	.311	.021
HPHRP 3	.593	.489	.281	.033
HPHRP 4	.552	.525	.274	.089
HPHRP 5	.663	.257	.191	.143
HPHRP 6	.675	-.050	.136	-.197
POF 1	.709	.261	-.214	.138
POF 2	.755	.279	-.257	.080
POF 3	.695	.269	-.267	.167
POF 4	.730	.186	-.359	.103
POF 5	.471	.064	-.396	.177
POF 6	.753	.284	-.309	.137
IWB 1	.632	-.391	-.062	.302
IWB 2	.679	-.366	-.007	.061
IWB 3	.730	-.409	.257	.161
IWB 4	.661	-.553	.165	.230
IWB 5	.748	-.353	.310	.146
IWB 6	.647	-.532	.234	.295
JBE 1	.705	.065	-.161	-.302
JBE 2	.600	-.301	-.235	-.354
JBE 3	.639	-.358	-.367	-.085
JBT 1	.712	-.183	.094	-.446
JBT 2	.709	-.044	.095	-.475
JBT 3	.705	-.005	.150	-.388
Extraction Method: Principal Component Analysis.
a. 4 components extracted.

Therefore the number of components extracted are 4

From the table below the Kaiser Meyer sampling statistic for adequacy of the questionnaire is 0.894 which is less than 0.9 and greater than 0.5, hence indicating the questionnaire is adequate for analysis since it is above the recommended value of adequacy (Pett et al., 2010). Whereas the Bartlett’s test of sphericity has a sigma value of 0.000, which is less p-value of 0.05, therefore the significance level is small enough to reject the null hypothesis (Williams et al., 2010). Therefore, the correlation matrix in table 5 is not an identity matrix. Hence the sampling sufficiency of the variables is accepted.

Descriptive Statistics

Table 7: KMO output results

KMO and Bartlett's Test
Kaiser-Meyer-Olkin Measure of Sampling Adequacy.		.894
Bartlett's Test of Sphericity	Approx. Chi-Square	4045.398
	df	276
	Sig.	.000

Table 8: Total variance explained

From the scree plot given below, four factors were extracted, i.e. the point of inflection of the scree plot produced is 4 at Eigen value >1 (Ghodsi and Zhu, 2006).

The extraction method of the rotated component matrix is principle component matrix with 6 iterations and a Varimax with Kaiser Normalization rotation. Factors with a loading of more than 0.5 are assumed to contribute more to the given variable; hence from the rotated component matrix the factors are bold to indicate their contribution to a factor.

Rotated Component Matrix
	Component
	1	2	3	4
HPHRP 1	.026	.754	.238	.136
HPHRP 2	.030	.828	.159	.155
HPHRP 3	.095	.772	.185	.176
HPHRP 4	.072	.781	.193	.102
HPHRP 5	.304	.607	.276	.159
HPHRP 6	.347	.327	.172	.508
POF 1	.193	.421	.620	.193
POF 2	.166	.426	.665	.263
POF 3	.175	.396	.664	.161
POF 4	.191	.298	.720	.249
POF 5	.155	.080	.614	.080
POF 6	.170	.408	.722	.213
IWB 1	.701	.023	.375	.125
IWB 2	.622	.063	.294	.348
IWB 3	.805	.205	.130	.291
IWB 4	.864	.031	.158	.225
IWB 5	.788	.278	.104	.302
IWB 6	.893	.082	.118	.157
JBE 1	.146	.256	.416	.598
JBE 2	.285	-.096	.335	.655
JBE 3	.416	-.155	.521	.458
JBT 1	.332	.200	.130	.762
JBT 2	.227	.294	.147	.762
JBT 3	.254	.358	.133	.678
Extraction Method: Principal Component Analysis.  Rotation Method: Varimax with Kaiser Normalization.
a. Rotation converged in 6 iterations.

Therefore the selected factors are:

Component
1		2	3		4
IWB 6	HPHRP 2			POF 6	JBT 1 JBT 2

Figure 3: Scatter plot matrix 

In order to determine the variables that pass the reliability test, the Cronchbach’s test is conducted for both when the item is deleted and when it is kept. 

From the statistics, there is a high percentage of consistency between the variables since the Cronchbach statistic ranges from 0.939 to 0.942 which is relatively higher than the o.5 minimum that indicate reliability. 

From the tables below, the number of items used in the Cronchbach test are six and the alpha statistic is greater than 0.7, where the Innovative work behavior has 0.922, the High performance HR practices has 0.865, the Person- Organization fit has 0.899 and lastly the employee performance has a Cronchbach alpha of 0.922.

Table 9

Reliability Statistics
Cronbach's Alpha	Cronbach's Alpha Based on Standardized Items	N of Items
.865	.863	6

Table 10

Reliability Statistics
Cronbach's Alpha	Cronbach's Alpha Based on Standardized Items	N of Items
.889	.890	6

Reliability Statistics
Cronbach's Alpha	Cronbach's Alpha Based on Standardized Items	N of Items
.922	.923	6

Reliability Statistics
Cronbach's Alpha	Cronbach's Alpha Based on Standardized Items	N of Items
.872	.873	6

In the test of normality, the p-value for both Kologomorov-Smirnoff and Shapiro-Wilk were 0.000 < 0.05 at a significance level of 95%. These statistics indicate normality of the dataset. The degrees of freedom for the test of normality is 216  

To prove that the dataset has a normal distribution, the fitted model residuals are not widely spread from the residual line. 

The F-statistic to test the null hypothesis that the variances are significantly equal to zero is 7.037 with a p-value of 0.000 <0.05 hence we fail to reject the null hypothesis and conclude that the variance are constant for the residual variables. 

Residuals Statisticsa
	Minimum	Maximum	Mean	Std. Deviation	N
Predicted Value	2.24	6.08	4.06	.545	216
Residual	-2.462	2.761	.000	.680	216
Std. Predicted Value	-3.332	3.717	.000	1.000	216
Std. Residual	-3.466	3.887	.000	.957	216
a. Dependent Variable: JBE 2

The residual errors have a mean of zero. Therefore the error term is homoscedastic with equal variances and mean zero.

Some variables show a Variance inflation factor of above 5 up to 10 and a tolerance rate of less than 0.2 indicating presence of multicollinearity in variables such as IWB 6 

H ₁: High-Performance HR Practices is positively related to Innovative Work Behaviors

H _a: High-Performance HR Practices is not positively related to Innovative Work Behaviors 

The Chi-Square statistic to test the null hypothesis against the alternative hypothesis is 0.000 at 95% confidence interval which is less than 0.005. We therefore fail to reject the null hypothesis and conclude that Performance HR Practices are positively correlated with innovative work behaviors.

H ₂: Person-Organization (P-O) Fit mediates the relationship between High-Performance HR Practices and Innovative Work Behaviors

H _a: Person-Organization (P-O) Fit does not mediate the relationship between High-Performance HR Practices and Innovative Work Behaviors 

The Chi-Square value is 0.000 which is greater than 0.005, we therefore fail to reject the null hypothesis and conclude that Person-Organization (P-O) Fit mediates the relationship between High-Performance HR Practices and Innovative Work Behaviors.

H ₃: IWB positively related to job performance

H _a: IWB is not positively related to job performance 

Table 13  

Scalar Estimates (Group number 1 - SEM)

Maximum Likelihood Estimates 

For the confirmatory factor analysis there were no negative Eigen values, i.e. the factors are extracted for Eigen >1 

The goodness of fit (GFI) Index has a value of 0.738 which is near 1 hence indicating a good fit while the root mean square residual (RMR) is 0.440 for the SEM model which is greater than 0 indicating a not so perfect fit, whereas the

RMR, GFI 

The standardized RMR (SRMR) is 0.276 which near zero indicating a good model fit. Additionally, the Comparative fit index (CFI) is 0.432 while the Tucker-Lewis Index (TLI) is 0.458  

The model has a high chi-square statistic of 536.047 indicating that the model is a good fit, which according to David (2016) “a model with a chi-square of more than 500” the chi-square is always statistically significant.

CMIN 

Following the results and subsection discussion, it is noted that several factors are key in influencing employee job performance. Such factors may include Innovative Work Behaviors as well as Person-Organization (P-O) Fit. Pett and Sullivan (2003) in an article on management argue that the factors that influence job attitude and satisfaction are: Questions (HPHRP 1 to HPHRP 6) measures: High-Performance HR Practices

• Probability of promotions
• Working environment
• Labor market behavior

Factors such as working environment, tend to affect the innovative nature of workers be it consciously or subconsciously. For instance, a worker in a company with adequate technological access may tend to be more engaged in innovative work behaviors than that with limited resources which ultimately affects the employee job performance. From the analysis, Person-Organization (P-O) Fit mediates the relationship between High-Performance HR Practices and Innovative Work Behaviors, i.e. Personal factors and organizational factors such as working environment and management tend to be the determining factor as to which extent a worker may engage in innovative worker behavior. Elsewhere, the results indicate that there is a relationship between, High-Performance HR Practices and Innovative Work Behaviors. High performance HR practices such as motivational incentives which may include prospects of promotion, good management- worker relationship etcetera, influence the workers motive to take part in innovative work practices. Ultimately, innovative work practices are therefore seen to be positively related to employee job performance, i.e. better engagement in innovative work leads to better employee performance.

References

Tavakol, M., &Dennick, R. (2011).Making sense of Cronbach's alpha. International journal of medical education, 2, 53.

Zhu, M., & Ghodsi, A. (2006).Automatic dimensionality selection from the scree plot via the use of profile likelihood. Computational Statistics & Data Analysis, 51(2), 918-930.

Williams, B., Onsman, A., & Brown, T. (2010). Exploratory factor analysis: A five-step guide for novices. Australasian Journal of Paramedicine, 8(3).

Gilem, K., & Gilem, V.(2013). Structural equation modeling. [online]. Available from: https//:www.statisyicalsolutions.com/sem 

Pett, M. A., Lackey, N. R., & Sullivan, J. J. (2003). Making sense of factor analysis: The use of factor analysis for instrument development in health care research.Sage.