Chapter 3: Research Methodology

3.1 Introduction

Chapter three discusses the research methodology used that explains the research
strategy which outlines the way the study will be carried out and the methodology used to
complete this study in detail. The dependent and independent variable data can be found
in two ways obtained from primary and secondary data. Primary data refers to data
obtained by the researcher for understanding and solving the research problem such as
surveys, interviews, and experiments, while secondary data can be defined as existing
information that was collected by someone else earlier. As for this study, we used
secondary data for the dependent and independent variables which is using quantitative
research. It is to find the relationship between the dependent variable and independent
variables. We use time-series data collected on a yearly basis over a 30-year period from
1992 to 2021.

The remaining topic in this chapter is organized as follows: section 3.2 discusses
the research methodology, including the research approach, research design and strategy,
and the research methods. Meanwhile, section 3.3 presents the topic of data collection.
Then, the section 3.4 shows the data analyze to analyze the relationship of the oil price,
unemployment rate, lending interest rate and interest rate in impacting the inflation rate.
Finally, section 3.5 gives a summary of this chapter.

3.2 RESEARCH METHODOLOGY

3.2.1 Research Approach

Research approach is the technique or pathway in a study to get the output

information. There are two types of research approach which are quantitative and
qualitative. Qualitative approach can be described as data obtained from non - numerical
factors such as observation, interview session, questionnaires to the target population.
Meanwhile, is the numerical approach that involves a set of numbers that can be
measured and analyzed. This research method study used a quantitative approach. All the
data is being analyzed by a quantitative approach. Besides that, this study used inductive
approach. Inductive approach is the method to get final conclusion from the specific to
the general. This is meant by the general conclusion is obtained after the data analyze
process is completed. During this process, the pattern of the data is being recognized.

3.2.2 Research Design/Strategy

Research design refers to the most suitable choice of a specific data collection and
analysis method. This involves the limitation of specific targeted data. Besides that,
research design also includes the total pathway of analysis from the very beginning of the
process until the complete outcome is fully analyzed. The purpose of this research design
is to effectively address the research problem. Research methods also avoid unnecessary
problems.

3.2.3 Research Method

This research is based on quantitative secondary data that was retrieved from
different data websites such as the Department of Statistics Malaysia, Microtrends,
Ministry of Human Resources, and Economic Planning Unit Open Data. This is time-
series research where the data were extracted in a 30-year period from 1992 to 2021.

3.2.4 The Research Approach used in the Thesis

This research uses quantitative method as the research approach. A method for
testing objective hypotheses by looking at the relationship between variables is
quantitative research. To enable statistical analysis of numbered data, these variables can
be measured, often using instruments. The introduction, literature and theory,
methodology, results, and commentary make up the predetermined format of the final
written report. Similar to qualitative researchers, individuals who use this type of inquiry
make assumptions about the ability to generalize and repeat the results, adjust for
alternative hypotheses, build in bias safeguards, and test theories deductively.
3.3 Data Collection

3.3.1 Quantitative Method

This research used secondary data that have been collected from various resources
such as the Department of Statistics Malaysia, Macrotrends website, The Ministry of
Human Resources, and Economic Planning Unit website in Malaysia. It uses time-series
data collected on a yearly basis over a 30-year period from 1992 to 2021. In this research,
for a dependent variable, which is inflation, the data were collected from the Department
of Statistics Malaysia based on the consumer price index. For independent variables, oil
price and unemployment rate were collected from Macro Trends website based on Dollar
and percentage of the labor force. As for the interest rate and lending interest rate, the
data is measured based on the consumer price index from The Ministry of Human
Resources website in Malaysia.

3.4 Data Analysis

Using the econometric software called EViews, data analysis is the process of
modifying, analyzing, and cleansing raw data in order to extract useful, relevant
information that aids organizations in making educated decisions (Kelley, 2022). The
process offers helpful insights and statistics, frequently presented in charts, graphics,
tables, and graphs, which lessen the risks associated with decision-making.

3.4.1 Descriptive Test

A statistical method and analysis type called descriptive analysis aids in

accurately describing, displaying, and summarizing data. It allows you to draw
conclusions about the range of your data and find patterns among variables. In essence,
descriptive statistics provide simple summaries of the sample and data measures to assist
in describing and evaluating the properties of a certain data collection. Two frequently
used measures in descriptive statistics are the coefficient of determination and the
measure of variability. Three averages—the mean, median, and mode—are used to
measure central tendency, whereas measures of variability include standard deviation,
ranges, variance, quartiles, and skewness.
3.4.2 Diagnostic Test

A diagnostic test is one that is used to identify the situation or its underlying
cause. It aids in determining the primary cause. A diagnostic test may be performed as
part of a physical examination to identify an illness or to determine what is causing
symptoms. A diagnostic test may also be used for other purposes, such as identifying
your strengths and weaknesses. Additionally, it is used to instruct students, direct their
path of study, and promote their participation in the learning process.

3.4.2.1 Normal Distribution Test (Jarque-Bera Test)

A test for normality is the Jarque-Bera test. Many statistical tests, such the
t test and the F test, make normality one of its presumptions. The Jarque-
Bera test is typically conducted before to one of these tests to ensure
normality. Due to the unreliability of other normality tests when n is big, it
is typically employed for huge data sets (Glen, 2022). This test is design to
test the compatibility of the skewness and kurtosis in order to observe if it
matches a normal distribution. A normal distribution has a skew of zero
and a kurtosis of three, which indicates how much data is included in the
distribution's tails and how "peaked" it is. Running the test does not
require the data's mean or standard deviation.

3.4.2.2 Serial Correlation LM Test

The Durbin-Watson test, sadly, has several drawbacks. As previously

stated, it can only be applied when the serial correlation is first-order,
when the equation has a constant, and when a lagged dependent variable is
not present. The inconclusive zone of the Durbin-Watson test is another
downside, especially given that it grows as the number of independent
variables rises.

The Lagrange Multiplier (LM) test, a well-liked substitute for the Durbin-
Watson test, examines serial correlation by evaluating how well the lagged
residuals explain the residual of the original equation in an equation that
also contains all the explanatory variables of the original model. If the
lagged residuals are significant in explaining this time’s residuals (as
shown by the chi-square test), then we can reject the null hypothesis of no
serial correlation (Studenmund, 2014). A generalized Lagrange Multiplier
testing strategy that can be used to solve several econometric issues
includes the LM serial correlation test.

Procedure:

1. From the estimated equation find the residuals, for equation with
two independent variables, it would look like:

2. Use the lagged residuals as well as all the variables on the right
side of the original equation as independent variables in an
auxiliary equation with these residuals as the dependent variable:

3. Calculate Equation using OLS, and then use the following test
statistic to determine whether is true:

where, in both the auxiliary equations, N is the sample size and R2 is the
uncorrected coefficient of determination.

LM has a chi-square distribution with one degree of freedom for big

samples (the number of restrictions in the null hypothesis). If LM is
greater than the critical chi-square value from Statistical Table B-6, then
we reject the null hypothesis that α3 = 0 and conclude that there is serial
correlation in the original equation. Note that even though α3 tends to be
positive in economic examples, this is a two-sided test (Studenmund,
2014).

3.4.2.3 Heteroskedasticity

When the variance of the residuals is uneven throughout a range of

measured values, this is referred to as heteroskedasticity (CFI Team,
2022). Heteroskedasticity causes the residuals of a regression analysis to
 have an uneven scatter (also known as the error term). A fan or cone-
shaped pattern on a residual plot suggests the presence of
heteroskedasticity. Because regressions using ordinary least squares (OLS)
presume that the residuals are taken from a population with constant
variance, heteroskedasticity is viewed as an issue in statistics. The results
of the study may be invalid if there is an unequal spread of residuals
because this indicates that the population involved in the regression has an
unequal variance. In general, heteroskedasticity usually occurs more in
cross-sectional model rather than time-series model. The test used to test
the presence of heteroskedasticity is the White test.

The hypothesis is set as below:

H0: There is no existence of heteroskedasticity

H1: There is existence of heteroskedasticity

The null hypothesis is rejected if auxiliary regression is lower than

5% level significance (Jhangiani & Chiang, 2015). The white test is
significant and there is no presence of heteroskedasticity.

3.5 Summary

To sum up, this segment incorporates a particular statement on research design,

data gathering technique, sampling design, variable, theoretical framework, data
processing, data analysis, and hypothesis formulation. All the methodologies that were
used have been described distinctly. The test will be overseen using EViews and the
outcome for each procedure will be presented in the subsequent segment.