105
CHAPTER 5
RESEARCH METHODOLOGY
5.1 INTRODUCTION
The present chapter is devoted to describe the method, procedure and
techniques used to achieve the objective of the study. This chapter covers
research design, determination of sample size, sampling design, questionnaire
design, administration and structure of the questionnaire, scoring of the
questionnaire, psychometric checks, reliability, validity, and primary data,
secondary data, period of the study, Frame work of analysis, statement of
hypothesis and statistical tools used for the data analysis are presented here.
5.2 Research Design
The study is descriptive in nature. It attempts to describe the, impact of
training and development programme and employees’ work related attitude such
as organizational commitment, job satisfaction and job involvement. The
stratified random sampling technique has been used to select the employees from
the selected IT companies for the study.
5.3 Data collection methods
There are two types of data collection method being used in our current study,
which are primary data and secondary data.
5.3.1 Primary data
The researcher has used both primary and secondary data. Data that
have been collected from first-hand-experience is known as primary data. Primary
data in this research are concerned with the survey instrument. The primary data
was collected from the 1000 employees who are working in IT companies at
106
Chennai. The designed questionnaire was circulated by the researcher through the
HR managers of the selected five IT companies in Chennai city.
5.3.2 Secondary Data
The primary data was supplemented by a spate of secondary sources of
data. The secondary data pertaining to the study was gathered from the web
portals of Department of Information Technology, IT Development Board, CRIS-
INFAC, CMIE, ELCOT, STPI, CII, Publications from various IT associations
like NASSCOM. The above mentioned sources were very useful in writing
introduction and industry profile chapters. Latest information was gathered from
well-equipped libraries in Chennai, Coimbatore, Bangalore, and from internet
web resources. The secondary data were used to identify the research gap through
the literature survey from various national and international journals, magazines,
periodicals, books and newspapers.
5.3.3 Period of the Study
The duration of the study was divided into four stages in a period of
four years. In the first stage, the collection of literature was done. In the second
stage, the preparation of the questionnaire and its pre-testing with limited
employees was completed. In the third stage, the data collection, processing and
analysis of data was done and in the final stage, the preparation of the thesis was
done. The primary data were collected for the period of six months i.e. June 2013
to December 2013.
5.4 Sampling Design
Sampling is a process of selecting a sufficient number of elements from a
population. It increases the probability of results obtained from sample to be
attributable to the population.
107
For this research, the IT companies are selected by a research terminology,
by selecting the top five IT companies, which come under the top 20 ranking
companies for the past five years in the IT industry as mentioned in the
NASSCOM (The National Association of Software and Services Companies)
website. As per the NASSCOM list the top five IT companies listed are TCS,
Infosys, Wipro, HCL and Tech Mahindra.
5.4.1 Target Population
According to Sekaran and Bougie (2010), population is the entire group of
people, events or things that researchers wish to investigate on. For the present
study, Chennai City of Tamilnadu State was purposively selected as it is one of
the hubs of IT companies in India. So, in this research, target population is the
total number of employees working in IT companies located in Chennai city.
5.4.2 Sampling Frame and Sampling Location
Sekaran and Bougie (2010) stated that sampling frame is the sample that is
drawn from a list of population elements that usually might be different from the
target population in actual practices.
The sampling frame for the research is the employees who are employed in
Information Technology sector. We narrow down the amount of employees by
focusing only on those who work in TCS, Infosys, Wipro, HCL and Tech
Mahindra located in Chennai city.
5.4.3 Sampling Elements
The respondents for our study are those who are employed in TCS,
Infosys, Wipro, HCL and Tech Mahindra & Mahindra IT systems located in
Chennai city. We target fairly all the employees from these IT companies, in
other words, our respondents consist of employees from all hierarchy level (i.e.
junior, middle, senior level) in the IT companies, which include the employees
working in technical and non-technical departments in IT Companies.
108
5.4.4 Sampling Technique
There are two types of sampling method which are probability sampling
and non-probability sampling. For probabilities sampling, each of the element in
the target population has an equal probability of being chosen as the sample for
the survey conducted. Probability sampling is scientific, operationally convenient
and simple in theory, and the results obtained from this method are more
generalizable toward the target population. For non-probability sampling, each of
the elements in the sampling frame does not have an equal chance to be chosen as
the sample. Admittedly this method is simpler and convenient to operate however
the results obtained cannot be confidently generalised to the population.
One thousand sample respondents were selected by using stratified random
sampling method from the selected five IT companies where 200 sample
respondents were selected from each company and where (n) 1000 samples were
collected for conducting the study.
5.4.5 Determination of Sampling Size
A population is defined as the “total collection of individuals or objects
that forms the focus of the research” whereas the sample is “a selected part or a
subset of the population (Pretorius 1995). According to Pretorius (1995), research
is generally conducted to make inferences about the population based on the
information available about the sample, in order to make inferences from the
sample to the population.
A number of formulae have been formulated for determining the
sample size depending upon the availability of information. The researcher has
used the below mentioned formulae for calculation of sample size for an
unknown population.
Sample size n =
Where, Z = Standardized value corresponding to a confidence level of 95% =
1.96
109
S = Sample SD from pilot study of 50 sample = 0.8066
E = Acceptable error 5 % = 0.05
Hence
. .
Sample size n =
.
= 999.7435 ~ 1000
Where n = 1000
5.5 Research Instrument
5.5.1 Questionnaire Design
The researcher attempted by all means to identify the suitable structured
questionnaire developed by the eminent researchers in the chosen research topic,
even though there are enormous structured questionnaires available in the
preferred research area, but the researcher was not able to find the appropriate
questionnaire in the chosen research context in IT sector. i.e. the association
between impact of training and development programmes towards employees’
work related attitude in IT sector. In addition to that, the most of the structured
questionnaire available are not having suitable questions related to IT sector
environment. Hence the researcher has developed the new survey instrument and
also verified the reliability, validity and content validity of the designed
questionnaire after the pilot study and appropriate changes were made to improve
the quality of the survey instrument.
Moreover, the impact of training and development scale was framed on the
basis of Kirkpatrick model of training evaluation, job satisfaction scale was
developed on the basis of Frederick Hertzberg’s two factor theory, whereas the
organization commitment scale was sourced from three-component Meyer and
Allen model, and Job involvement scale was formed on the basis of Lodahal and
Kejner Model.
110
5.5.2 Administration of the Questionnaire
The questionnaire can be administered individually or in group. In order to
make the IT employees feel free, the manager’s presence was kept away.
Moreover, the respondents should remain incognito. This gave them greater sense
of security. The time generally taken for completion of one questionnaire was
thirty minutes. The purpose of the questionnaire was to measure impact of
training and development programme and employees’ work related attitude after
attaining the training and development programme in the IT sector.
5.5.3 Pilot Test
The questionnaire meant for the respondents was pre-tested with 50
employees from the selected IT Company. After pre-testing, necessary
modifications were made in the questionnaire to fit in the track of the present
study. Finally, the questionnaire was checked by the reliability test for the fifty
samples. As per the test for reliability, a high scale reliability alpha = 0.9345 for
the section pertaining to impact of training programmes, alpha = 0.9263 for the
section of job satisfaction and 0.9412 for the section of employee’s performance
and alpha = 0.9549 for the section of Organizational commitment.
5.6 Constructs Measurement (Scale and Operational Definitions)
5.6.1 Scoring of the Questionnaire
The scale against which the respondents indicated the extent of agreement
/ disagreement with reference to the characteristics of his/her organization is
defined by the following five categories.
Strongly agree -5
Agree -4
Neutral -3
Disagree -2
Strongly disagree -1
111
To find out the raw scores for each employee, the scores of all items in the
questionnaire answered by him/her were added. This gave the score of that
particular employee regarding his/her evaluate effectiveness of the training
attended and employees’ work related attitude after attaining the training
programme.
5.6.2 Scaling Technique
Nominal scale and Likert`s scale have been used in this study when the
questionnaire was developed.
5.6.2.1 Nominal Scale
Nominal scale is simply a system of assigning number symbols to
events in order to label them. Nominal scales provide convenient ways of keeping
track of people, objects and events. This scale is used for the demographic and
training details section of the questionnaire where the questions are categorized
variables.
5.6.2.2 Likert`s Scale
In a Likert`s scale, the respondent is asked to respond to each of the
statements in terms of several degrees, usually five degrees of agreement or
disagreement. At one extreme of the scale there is strong agreement with the
given statement and at the other, strong disagreement, and between them lie
immediate points. Five point Likert`s scale (5- Strongly Agree, 4- Agree, 3-
Neutral, 2-Disagree, 1- Strongly Disagree) was used for all dimensions except
demographic profile and training details section which consists of its own choice.
5.7 Standard Measurements of Variables
This study utilized six section questionnaires viz. Personal details, impact
of training programme, job satisfaction, organizational commitment, job
involvement and opinion towards training based on the most recent or
112
contemporary and appropriate theories related to each exogenous variables and
endogenous variables in order to achieve the purpose of the study.
5.7.1 Personal details
The first section of the questionnaire included a demographic profile
and training details based on the purpose of the demographic and training
questions to identify the respondents’ demographic characteristics training details.
These parameters included; Age, Sex, Marital Status, Educational Qualification,
Department, Designation, Total work Experience, Salary, Methods Used for
Training, Number of Promotions Received, Need for Additional Training
Programmes, Number of Training Programmes Attended, Training Practices
Reduces Human Cost, Management Uses Latest Technologies for Training,
Training and Developmental Activities Maintain Employee Retention Rate.
5.7.2 Impact of Training Programme
The second section of the survey questionnaire consisted of twenty five
items. It’s for measuring the impact of training programme attended by the
employees. A five point Likert type scale (1 – Strongly Disagree, 2 – Disagree, 3
– Neutral, 4 – Agree, 5 – Strongly Agree) was used to measure the perceived
level of impact of training programme among the IT employees.
5.7.3 Job Satisfaction Scale
The third section of the survey questionnaire consisted of twenty items.
The aim was to measure the level of job satisfaction of employee’s. A five points
Likert type scale (5 – Highly Satisfied, 4 – Satisfied, 3 – Neutral, 2 – Dissatisfied,
1 – Highly Dissatisfied) was used to evaluate the level job satisfaction among the
IT employees.
113
5.7.4 Organizational Commitment Scale
The fourth section of the survey instrument consisted of fifteen items.
The aim of the scale was to measure the psychological state that characterizes the
employees’ relationship with the organization. A five points Likert type scale ((1
– Strongly Disagree, 2 – Disagree, 3 – Neutral, 4 – Agree, 5 – Strongly Agree)
was used to measure the level of organizational commitment among employees’
those who are working in IT at Chennai.
5.7.5 Job Involvement Scale
The fifth section of the survey instrument consisted of twenty five
items. The aim of the scale was to measure the immediate improvement in the
knowledge, skill and ability to carry out job related work and its relationship with
the organization on their performance. A five points Likert type scale (1 –
Strongly Disagree, 2 – Disagree, 3 – Neutral, 4 – Agree, 5 – Strongly Agree) was
used to measure the level of job involvement those who are working in IT
Company at Chennai.
5.8 Data Processing
After collecting all the questionnaires from the respondents, data
processing step is then taken before any analysis being implemented. The data
preparation process consists of checking, editing, coding, and transcribing.
Meanwhile, all the unusual responses are identified.
Before checking all the collected questionnaires, we counted and
numbered every questionnaire to assure the required amount of questionnaires is
being returned by the respondents. After that, we checked the all data that we had
collected to ensure that all the questions inside the questionnaire were being filled
up completely by the respondents without any omission. Any incomplete
questionnaire found would then be taken out. Next, we did coding, data entry,
editing, and data transformation for all the remaining survey questionnaires.
114
5.8.1 Coding
This step is taken in order to key the data into the Statistical Package for
Social Sciences (SPSS) system. We assigned the code to each participant’s
response. For instance, in Section A- Personal Particulars under department, we
assigned “1” to technical department and “2” to non-technical. While for the
question of work experience, we assigned “1” to “3” for all the responses.
Apart from that, we attributed “1” for male and “2” for female. For the
questions in Section B, D and E, started the code from “1” to “5” for all the
responses, whereas“1” symbolizes strongly disagree and “5” symbolizes strongly
agree. In Section C Job Satisfaction, we started the code from “1” to “5”, here “1”
symbolizes highly dissatisfied and “5” symbolizes highly satisfied.
5.8.2 Data Entry (Transcribing) and Editing
Once all the questions’ responses had eventually been coded, we began to
enter all the data into the SPSS database. This process is also known as
transcribing (Malhotra, 1993). Before running the reliability test, we carried out
the editing tasks towards all the responses. We attempted to detect and correct the
problems, such as illogical, inconsistent, or illegal responses.
Illogical response is the response that is given by the respondent which
looks significantly different from others’ responses. Sometimes, this respondent is
known as outlier. While, inconsistent responses happened when the respondents’
responses that is incoherent with other information provided. Also, it is possible
that the inconsistent responses are caused by bias. As a result, we need to edit the
inconsistent responses provided by the respondents.
Meanwhile, illegal codes are values that are not indicated in the coding
instructions provided. The best mean to solve this problem is through the use of
computer to generate frequency distribution and then look for the illegal codes.
115
5.8.3 Data Transformation
Additionally, we also carried out data transformation after data entry and
editing. Data transformation is a data coding variation, which is the process of
altering the original numerical representation of a quantitative value to another
value (Sekaran and Bougie, 2010). The data transformation was not required in
this research, because all the questions are in positive forms.
5.9 Data Analysis
According to Zikmund et al. (2010), data analysis is defined as the
reasoning application which helps the researchers to understand the data that have
been collected. The purpose to implement data analysis is to examine and model
the data by assigning facts and figure to answer research problem. Also, it
highlights the useful information by recommending assumptions to take
advantage of the collected data in order to solve some specific problem, such as
addressing the research problem.
The computer software that has been applied to analyze the collected data
is SPSS.SPSS provides us many types of analysis that is very helpful in our
current research. Typically, there are three types of analysis that is required in our
research, which are descriptive analysis, scale measurement, and inferential
analysis
The core of the study is “employees’ work related attitude towards the
training and development programme in IT sector at Chennai city”, hence the
study centers on the dependent variable viz., employee work related attitude and
its relationship with the related independent variables such as organizational
commitment, job satisfaction and job involvement. The study also attempts to
find the impact of training and development programme by using Kirkpatrick four
level of evaluation model.
116
5.9.1 Descriptive Analysis
Descriptive analysis is used in order to clarify and describe the
characteristics of the variables of interest in a situation (Sekaran and Bougie,
2010). Besides, Zikmund defined descriptive analysis as the elementary
transformation of data in a way that illustrate the fundamental characteristic, such
as central tendency and variability. Generally, mean, median, mode, variance,
range, and standard deviation are widely applied in describing the descriptive
statistic. The advantage of using descriptive analysis is that it helps to summaries
the sample and measure. It also forms basic quantitative data analysis with simple
graphics analysis.
In this research, Descriptive analyses were done for the responses derived
from ‘Section A’ Personal details of the respondents by using tabulation method
with frequency and percentage.
5.9.2 Scale Measurement- Reliability Test
5.9.2.1 Psychometric Checks
As mentioned earlier, a structured questionnaire developed by the
researcher was used as the instrument for data collection for the study. Items
selected for the constructs were mainly adopted from prior studies to ensure
content validity. However, the instrument was validated for the main study with a
size of 1000 respondents.
5.9.2.2 Reliability
Reliability, also called consistency and reproducibility, is defined in
general as the extent to which a measure, procedure, or instrument yields the
same result on repeated trials (Carmines & Zeller, 1979). It can be used to assess
the degree of consistence among multiple measurements of variables (Hair,
Anderson, Tathman, & Black, 1998). The internal reliability of the measurement
models was tested using Cronbach’s alpha and Fornell’s composite reliability
(Fornell and Larcker 1981). The Cronbach’s reliability coefficients of all
117
variables should be higher than the minimum cut-off score of 0.70 (Nunnally
1978; Nunnally and Bernstein, 1994). The questionnaire meant for the
respondents was pre-tested with 50 employees from the selected IT Company.
After pre-testing, necessary modifications were made in the questionnaire to fit in
the track of the present study. Finally, the questionnaire was checked by the
reliability test for the fifty samples. As per the test for reliability, a high scale
reliability alpha = 0.9345 for the section pertaining to impact of training
programmes, alpha = 0.9263 for the section of job satisfaction and 0.9412 for the
section of job involvement and alpha = 0.9549 for the section of organizational
commitment.
5.9.2.3 Validity
A scale is said to be valid if it measures correctly what it is expected to
measure. In other words, a scale is valid only when it is real and correct. The
validity of a questionnaire relies first and foremost on reliability. If the
questionnaire cannot be shown to be reliable, there is no discussion of its validity.
Researchers use different methods of establishing the validity of the instrument
which they have developed. They are: content validity, convergent validity,
discriminate validity and nomological validity. In the present study the content
validity was established. It is given in the following section.
5.9.2.4 Content Validity
For the content validity, a thorough review of the literature was conducted.
As mentioned earlier, all items of the constructs have been drawn from well-
established studies to ensure content validity. The questionnaire was also
reviewed by a panel of experts i.e. Senior IT Professionals and Human resource
managers working in the IT sector and academicians. The changes suggested by
the panel members were incorporated to improve both the content and clarity of
the questionnaire. The instrument was tested through two stages. In the first stage,
two English faculty members reviewed the instrument to ensure the clarity of
118
items and the accuracy of the language. In the second stage, a panel of experts
was selected to establish face and content validity of the instrument. The panel of
experts consisted of six individuals, two senior IT Professionals, two human
resource managers of the IT companies, who had earlier participated in the
instrument development and two senior academicians.
5.10 HYPOTHESES DEVELOPMENT
5.10.1 Demographic variables and impact of training and
development:
According to (Mohinder Chand, Ankush Amhardar, 2010; Allan bird, Susan
Heinbuch and Roger Dundar, 1993; Dr.Ludy Balatbat, 2010; Yuhafan Diana
H.wa and Connie.K.Haley, 2011) all this study has been conducted with
demographic variables to investigate the impact of training and development
programme.
H0: There is no association between demographic variables and training
and development practices support business goals of the organization.
H1: There is an association between demographic variables and training
and development practices support business goals of the organization.
5.10.2 Demographic variables and work related attitude of employees:
There is a closest connection between demographic variables and
work related attitude of employees in the study conducted by Fiona Edgar
& Alan Geare (2004) where the authors found that employee demography,
especially gender, ethnicity and employment sector, does influence
employee attitudes. According to Dr. Nasser S. Al-Kahtani (2012) Study to
explore relationship between demographic variables and work related
variables among employees which focused a positive relationship between
them. It has been found by many researchers across different time periods
119
that demographic variables and have an impact on work related attitude of
employees (Sarath, P & Raju, S, 2013; Christine M. Riordan and Lynn
Mcfarlane Shore, 1997; James E.Martin, Robert P. Michel, 1999) the results
revealed that there is a strong influence of demographic variables on job
satisfaction, organizational commitment, job attitudes.
H0: There is no significant difference between demographic variables with
respect to employees’ work related attitude towards training and
development programme.
H1: There is a significant difference between demographic variables with
respect to employees’ work related attitude towards training and
development programme.
5.10.3 Job involvement and Organizational commitment:
There is a closest connection between job involvement and organizational
commitment noted in a study by E.J. Lumley, M. Coetzee, R. Tladinyane & N.
Ferreira (2011), where the authors found an association between job involvement
and organizational commitment. This shows that there is a positive relationship
between job involvement and organizational commitment.
H0: There is no association between level of job involvement and level of
organizational commitment.
H1: There is an association between level of job involvement and level of
organizational commitment.
5.10.4 Job satisfaction and Organizational commitment:
According to M Sheik Mohamed, M Mohideen Abdul Kader and H. Anisa (2012)
the study investigate the relationship between organizational commitment and job
satisfaction which shows positive inter- relationship. It has been found by many
120
researchers across different time period (Ali Erbasi, Tugay Arat, Osman Unuvar,
2012; Jomon Joy, Dharwad Joseph Thomas R, 2014; Paul Ayobami
Akanbi, Kehinde Adeniran Itiola(2013)) the correlation between organizational
commitment and job satisfaction. More over existence of a close connection
among job satisfaction, organizational commitment, this is investigated by Md.
Sahidur Rahman, Rana Karan and Md. Iftekhar Arif, 2014).
HO: There is no association between level of job satisfaction and level of
organizational commitment.
H1: There is an association between level of job satisfaction and level of
organizational commitment.
5.10.5 Job involvement and Job satisfaction:
According to Muhammmad Ahsan and Naeem Ullah (2014) examines the
relationship between human attitudinal and behavioral factor of job involvement
and job satisfaction which shows positive relationship. It has been found by many
researchers (Dr. Nazir Ahmad Gilkar, and Javid Ahmad Darzi, 2013; Beeler,
Jesse D; Hunton, James E; Wier, Benson, 1997; Mughees Uddin Siddiqui, 2014;
Anita Sharma, 2014) that there is an association between job involvement and job
satisfaction.
H0: There is no association between level of job involvement and level of
job satisfaction.
H1: There is an association between level of job involvement and level of
job satisfaction.
5.11 Statement of Hypothesis
On the basis of the review of literature given in chapter 2 and theoretical
framework described in Chapter 3, the following hypotheses were developed:
121
The following hypotheses are formulated for the present study:
H1: There is no association between the demographic variables and the
impact of training and development programmes.
H2: There is no association between demographic variables and work related
attitude of employees towards training and development programme in IT
Sector.
H3: There is no association between level of job involvement and level of job
satisfaction.
H4: There is no association between level of job involvement and level of
organizational commitment.
H5: There is no association between level of job satisfaction and level of
organizational commitment.
H6: There is no significant difference between mean ranks towards factors
affecting impact of training and development programmes.
H7: There is no significant difference between mean ranks towards factors
influencing job involvement.
H8: There is no significant difference between mean ranks towards factors of
job satisfaction.
H9: There is no significant difference between mean ranks towards factors
affecting organizational commitment.
H10: There is no significant difference between mean ranks towards factors
influencing employees’ work related attitude.
H11: Employees’ work related attitude is not having positive impact with
various dimensions like impact of training and development programmes job
122
satisfaction, organizational commitment and job involvement among the IT
employees.
5.12 Statistical Tools for Data Analysis
The requirement and importance of statistics is escalating, especially in
social sciences and management research. It is important to recognize an
appropriate statistical design which brings solutions to the entire research
hypotheses. Statistics are the tools used to check our facts about the data.
Tools for data analysis includes
Frequency distribution
Chi square
t test
ANOVA
Friedman test
Correlation
Multiple regression analysis
Factor analysis
Structural equation modeling
5.12.1 Frequency distribution
A frequency table is a simple way to display the number of occurrences
of a particular value or characteristic.
5.12.2 Chi-square
A Chi-square is a statistical measure used in the context of sampling
analysis for comparing a variance to a theoretical variance. As a non-parametric
123
test, it can be used to determine if categorical data shows dependency or the two
classifications are independent. It can also be used to make comparisons between
theoretical populations and actual data when categories are used. Thus, the chi-
square test is applicable in large number of problems. The test is, in fact, a
technique through the use of which it is possible for all researchers to (1) test the
goodness of fit (2) test the significance of association between two attributes, and
(3) test the homogeneity or the significance of population variance.
5.12.3 Independent samples t test
The independent samples t test allows researcher to evaluate the mean
difference between two populations using the data from two samples. This test is
used in situations where a researcher has no prior knowledge about either of the
two populations being compared. The general purpose of the independent samples
t test is to determine whether the sample mean difference obtained is a real
difference between the two populations or simply the result of sampling error.
5.12.4 ANOVA
Analysis of variance procedures are powerful parametric methods for
testing the significance of differences between sample means where more than
two conditions are used, or even when several independent variables are involved.
ANOVA makes it feasible to appraise the separate or combined influences of
several independent variables on the experimental criterion (Mouton & Marais
1990). ANOVA test was therefore used to identify whether there is a statistical
significant difference between the demographical variables and impact of training
and development programme, employees work related attitude.
124
5.12.5 The Friedman test
The Friedman test is a test for comparing three or more related samples
and which makes no assumptions about the underlying distribution of the data.
The data is set out in a table comprising n rows by k columns. The data is then
ranked across the rows and the mean rank for each column is compared.
5.12.6 Correlation
Degree and type of relationship between any two or more quantities
(variables) in which they vary together over a period; for example, variation in
the level of expenditure or savings with variation in the level of income. A
positive correlation exists where the high values of one variable are associated
with the high values of the other variable(s). A 'negative correlation' means
association of high values of one with the low values of the other(s). Correlation
can vary from +1 to -1. Values close to +1 indicate a high-degree of positive
correlation, and values close to -1 indicate a high degree of negative correlation.
Values close to zero indicate poor correlation of either kind, and 0 indicates no
correlation at all. While correlation is useful in discovering possible connections
between variables, it does not prove or disprove any cause-and-effect (causal)
relationships between them.
5.12.7 Multiple regression analysis
Multiple regression analyses the common and separate influences of
two or more variables on a dependent variable (Kerlinger 1986), and it is used to
establish the extent to which various differing variables add to predict another
variable (Guyatt et al 1995). Multiple regression was therefore used to study the
dependent variable (Job Involvement) is statistically significance on the variance
in independent variables such as organizational commitment, overall impact of
training and development and job satisfaction.
125
5.12.8 Factor analysis
A factor is an underlying dimension that accounts for several observed
variables. There can be one or more factors, depending upon the nature of the
study and the number of variables involved in it. Factor analysis involves many
terminologies which are presented in this subsection for better understanding of
the related techniques.
Correlation coefficients matrix is the original observations between
different pairs of input variables. Factor loadings matrix representing the
correlation between different combinations of variables and factors. Communality
is the sum of squares of the factor loadings of the variable ‘i’ on all factors. Eigen
value is the sum of squares of the factor loadings of all variables on a factor.
After obtaining factor loadings, one should examine whether the factor
loading matrix possesses a simple structure. If a factor loading matrix has a
simple structure, it is easy to make interpretations about the factors. If there is no
simple structure, then the n – dimensional space of the factors should be rotated
by an angle such that the factor loadings are revised to have a simple structure
which will simplify the process of interpretation of the factors. Such rotation is
called rotation of factors. A simple structure means that each variable has a very
high factor loading (as high as 1) on one of the factors and very low factor
loading (as low as 0) on other factors. The communalities of each variable before
and after factor rotation will be the same. The popular methods of rotation of
factors are varimax method and promax method. Varimax method of factor
rotation employs orthogonality between different pairs of factors axes. This
means that the angles between different pairs of factors axes are 90° even after
rotation. The promax method employs oblique rotation. This means that the
angles between different pairs of factors axes are not 90° after rotation. Both the
techniques aim at better interpretations.
126
5.12.9 Structural Equation Modeling (SEM)
An explanation for the use of Structural equation modeling with
AMOS and methods to asses construct validity and reliability for all measures is
addressed in this study. The research or model describes the causal relationship
among reaction to training programme, skills acquisition, behavioural change,
effect of training, organizational commitment, Job Involvement and job
satisfaction. These paths are related to with causal processes. Thus the Structural
Equation Modeling (SEM) approach was necessary in order to examine these
variables. The data analysis was carried out by means of SPSS (statistical package
for the social science, version 20) and AMOS 20 (analysis of movement structure,
version 20) software packages for windows.
Structural Equation Model (SEM) with AMOS 20 software provided
several indicators to asses’ fit. The confirmatory factor analysis showed the
acceptable model fit by including the normal fit index (NFI), the comparative fit
index (CFI), the root mean square error of approximation (RMSEA), hypothesis
model (Research model) based on the research hypothesis and review of the
theoretical and empirical literature, the hypothesis model was examined in this
study. There are two latent variables; impact of training and development and
employees’ work related attitude to explore the cause and effect relationship
among these variables in the hypothesized model.
SEM is a statistical methodology with a confirmatory approach to
analyze multivariate data (Byrne 2001). The general SEM model is composed of
two sub models; a measurement model and a structural model. James et al (1982)
recommended the measurement model testing first, followed by full structural
model testing.
Statistical significance for all analysis was set at less than 0.05. The
measurement model identifies relationships between the observed and latent
127
variables. By means of CFA, the measurement model provides the link between
scores on an instrument and the constructs that they are designed to measure.
Hence structural model identifies the causal relationships among the latent
variables and specify that particular latent variables directly or indirectly
influence certain other latent variables in the model (Byrne 2001).
