You are on page 1of 109

Master of Business Administration Degree

Programme
Faculty of Commerce and Management
University of Kelaniya - 2023

Business Statistics-MBA 52063


Parametric and Non-parametric
Session 13 &14;
Performing Different Parametric and Non-parametric
Statistical Tests Using SPSS and Interpreting the Results
Dr. Anura Karunarathne,
Senior Lecturer,
Department of Accountancy,
University of Kelaniya.
Learning Objectives
After completing this module, the student will be able to:
❖Choose the right statistical test (Parametric/Nonparametric) for
Different Research in Business and Economics.

❖Compare and Contrast parametric and Nonparametric tests.

❖Performing Different Parametric and Non-parametric Statistical Tests


Using SPSS and Interpreting the Results.

❖Apply parametric and non-parametric statistical tests for business


research and decision-making.
Briefing the Previous Session…
What are the Available
Statistical Methods and
Tests for Analyzing Data in
your Research
Available Statistical Methods and Tests
Step-by-Step Process of Hypotheses Testing
1. Formulate the null hypothesis (H₀) and alternative hypothesis (H₁): The null
hypothesis represents the default position or assumption, often stating that there is no
significant difference or relationship between variables. The alternative hypothesis, on the
other hand, represents the claim or theory we want to test.
2. Select the significance level (α): The significance level determines the threshold for
rejecting the null hypothesis. Commonly used values are 0.05 (5%) or 0.01 (1%). It reflects
the level of confidence required to reject the null hypothesis.
3. Choose the appropriate test statistic: The choice of test statistic depends on the
nature of the data and the hypothesis being tested. Examples of commonly used test
statistics include t-tests, chi-square tests, ANOVA, etc.
4. Collect and analyze the data: Gather a representative sample of data relevant to the
hypothesis being tested. Perform any necessary calculations or data transformations
required for the chosen test statistic.
5. Determine the critical region: The critical region is the range of values that, if
obtained by the test statistic, would lead to rejecting the null hypothesis. The critical
region is determined based on the chosen significance level and the test statistic's
sampling distribution.
Step-by-Step Process of
Hypotheses Testing
Hypothesis testing is a statistical method used to
make inferences and draw conclusions about a
population based on sample data. It involves
several steps to ensure a systematic and reliable
analysis
Step-by-Step Process of Hypotheses Testing…
6. Calculate the test statistic: Apply the selected test statistic to the sample data to
obtain a calculated value. The test statistic measures how far the observed data deviates
from what would be expected under the null hypothesis.
7. Compare the test statistic to the critical region: If the calculated test statistic falls
within the critical region, the result is considered statistically significant, and we reject
the null hypothesis in favour of the alternative hypothesis. If it falls outside the critical
region, the result is not statistically significant, and we fail to reject the null hypothesis.
8. Calculate the p-value: The p-value is the probability of obtaining a test statistic as
extreme as or more extreme than the observed value, assuming the null hypothesis is
true. If the p-value is less than the significance level (α), the result is considered
statistically significant.
9. Interpret the results: Based on the p-value and the significance level, interpret the
results of the hypothesis test. If the p-value is less than α, we reject the null hypothesis
and conclude that there is evidence to support the alternative hypothesis. If the p-value is
greater than α, we fail to reject the null hypothesis, indicating that there is insufficient
evidence to support the alternative hypothesis.
10. Draw conclusions: Finally, based on the analysis and interpretation of the results,
draw appropriate conclusions about the population being studied. Consider the
limitations of the study and any implications or recommendations that arise from the
findings.
Parametric and Nonparametric Tests
Parametric Tests Nonparametric Tests
None Chi-square Goodness of Fit
None Chi-square for independence
Independent Sample t-test Mann-Whitney U Test
Paired sample t-test Wilcoxon Signed Rank Test
One-way between groups ANOVA Kruskal-Wallis Test
One-way Repeated measures ANOVA Friedman Test
Two-way analysis of variance(Between Groups) None
Mixed Between within Group ANOVA None
Multivariate Analysis of Variance(MANOVA) None

Analysis of Covariance(ANCOVA) None


Karl Pearson’s Correlation Spearman’s Rank Correlation
Linear Regression None
Can you select all the above
Statistical Measures/Tools
/Techniques for your Own
Research?
Quantitative Research and Hypothesis Testing
Using a Statistical Software

Step-01
Choosing the Appropriate Statistical Measures/Tests
Step-02
Performing Statistical Tests in Statistical Software;
SPSS, E-views, Minitab, SAS R, Stata…etc.
Step-03
Making Conclusion
How to Choose the
Right Statistical Test/s?

❖How to Choose the Right


Descriptive Measures?
❖How do you Choose the Right
Parametric or Non-parametric
Statistical Tests for your own
research?
How to Choose the Right Statistical Test?
Key Criteria
1. What is your research question asking (Purpose of your
Study)
2. The level of measurement of your variables (Nominal,
Ordinal, Interval or ratio level data)
3. Tested Results of Parametric Assumptions; the distribution of
the data(Normal/Non-normal Distribution)
4. Nature of sample (Independent/Paired) Are the samples
independent or related (same person tested twice or paired)?
5. How many groups are involved? (one, two or more groups)
The Basics of Managing SPSS Data
Files
Data Management
1) Import Data,

2) Descriptive Measures; Frequency, Measures of Central


Tendency and Dispersion

3) Graphs and Charts,

4) Select Case,

5) Cross-tabulation/Custom Table,
Questionnaire Designing in SPSS
Variable View
Covid-19 Pandemic Situation and Online Education of Undergraduates

17
Covid-19 Pandemic Situation and Online Education of Undergraduates

15. Available Telephone Connection/s

Available Telephone Connection ✓ (Tick)


1.SLT
2.Dialog
3.Mobitel
4.Hutch
5.Airtel
Other
Specify……………………………

18
Covid-19 Pandemic Situation and Online Education of
Undergraduates
16. Available Internet Connection/s
Available Internet Connection Yes(1) No(0)- Quality of the
(a) Connection*(b)
1.SLT
2.Dialog
3.Mobitel
4.Hutch
5.Airtel
Other
Specify…………………………

*1.Very poor 2. Poor 3. Average/Moderate 4. Good 5.Very Good


19
Covid-19 Pandemic Situation and Online Education of
Undergraduates

17. Do you have monthly financial support from Government


institutions / Instructions / other organizations/Persons?

Source of Financial Support Yes (1) No (0)(a) Amount Rs.(Monthly)(b)

Mahapola Scholarship Fund

Bursary
Other Specify
……………………………………………
……………………………………………
…………

20
Covid-19 Pandemic Situation and Online Education of
Undergraduates
18.Impact of Covid-19 Pandemic situation for your monthly family
income

Monthly Income(Rs.)
Income Earner Before Covid-19 After Covid-19
Father 70000 20 000
Mother
Other
Specify…………………………..
Other
Specify…………………………..

21
Covid-19 Pandemic Situation and Online Education of
Undergraduates
19. Provide your answers based on the previous semester Online
Education experience
Description 1 2 3 4 5
(a) Overall, how satisfied were you with
the course?
(b) How satisfied were you with the
content of the course?
(c) How satisfied were you with the ability
to navigate through the course?

(d) How satisfied were you with the format


of the course?
(e) How satisfied were you with the online
help features of the course?
22
Testing Hypotheses;
Parametric and Non-parametric Tests
How to Choose the Right Statistical Test? Criteria

What your research question asking (Purpose of your Study)

The level of measurement of your variables (Nominal, Ordinal, Interval


or ratio level data)

Tested Results of Parametric Assumptions; the distribution of the


data(Normal/Non-normal Distribution)

Nature of sample (Independent/Paired) Are the samples independent or


related (same person tested twice or paired)?

How many groups are involved?(one, two or more groups)


Types of Research, Hypotheses and Relevant Statistical Tests
Research Relevant Research Hypotheses Relevant Statistical Tests
Question
Descriptive No specific hypotheses Descriptive statistics (mean, median,
mode, standard deviation, etc.)

Comparative The hypothesis of difference or t-test, ANOVA, Mann-Whitney U test, Chi-


association between groups square test
Correlational The hypothesis of relationship or Pearson correlation, Spearman
association between variables correlation
Predictive Hypothesis predicting the value of Regression analysis (linear, logistic,
one variable based on another etc.)
Explanatory A hypothesis explaining the causal Experimental design (treatment and
relationship between variables control groups), ANCOVA
Exploratory No specific hypotheses, seeking to Exploratory factor analysis, cluster
discover patterns or relationships analysis
Key Research Objectives, Questions, Corresponding Research
Hypotheses, and The Statistical Tests
Relevant Research
Research Objective Research Question Hypotheses Relevant Statistical Tests
What is the current state of a Descriptive statistics,
Exploration No specific hypotheses
phenomenon? Content analysis
How does a phenomenon
Descriptive statistics, Case
Description manifest in a particular No specific hypotheses
study analysis
context?
What are the differences or
Hypothesis of difference or t-test, ANOVA, Chi-square
Comparison similarities between
association between groups test
groups?

Is there a relationship Hypothesis of relationship or Correlation analysis,


Relationship
between variables? association between variables Regression analysis

Hypothesis predicting the


Can we predict one variable Regression analysis, Time
Prediction value of one variable based
based on another? series analysis
on another
Hypothesis explaining the Experimental design,
Does one variable cause
Causality causal relationship between ANCOVA, Mediation
changes in another?
variables analysis
What new patterns or
Exploratory data analysis,
Exploration relationships can be No specific hypotheses
Data mining techniques
discovered?
Statistical Methods in Business Research
Conceptual
Framework
of a
Quantitative
Study
2. The level of measurement of
your variables
The Relation Between Data Level and Statistical Methods
Data Level Meaningful Operation Statistical
Method
Nominal; Classifying & Counting NP
Sinhalese, Muslims, Tamils
Ordinal; Classifying, Counting NP
Taste test ranking of three & Ranking
brand of soft drink
Interval; Classifying, Counting, Ranking, P
Calendar time, Monetary Addition, subtracting,
Unit, Ferenhighgt, Temperature Multiplication,& Division
of three brand of Soft Drink

Ratio; Classifying, Counting, Ranking, P


Heihgt, Weight,Revenue Addition, subtracting,
&Expenditure,P/E ratio,Stock Multiplication,
Turn Over . & Division 30
3. Tested Results of Parametric
Assumptions and Statistical
Tools/Techniques
(Normality of the Distribution of Data)
Common Assumptions in Parametric Tests

Normality
Randomness
Absence of Outliers
Homogeneity of Variances
Independence of Observations, and
Linearity
Appropriate Statistical Test

Parametric test Non-parametric test

Nominal or Ordinal Level Data


Interval or Ratio
Level Data or

+
Interval or ratio level data with a
Non-normal Distribution
Normal Distribution
Testing Key Parametric Assumptions;
Distribution of the Data?

Normal Distribution
Or
Non-Normal Distribution
Normality Test

Graphical Method

Statistical Tests
Graphical Method
Normal Skewed /Non-Normal
Distribution Distribution
Statistical Tests

K-S Test,
Shapiro Tests
Jarque-Bera Test
Anderson-Darling
test …etc.
How to test the Normality

38
How to test the Normality
Hypothesis test for a test of normality
Null hypothesis: The data is normally distributed

For the skewed data, p = 0.002 suggesting strong evidence of non-


normality.
For the approximately normally distributed data, p = 0.582, so the null
hypothesis is retained at the 0.05 level of significance. Therefore,
normality can be assumed for this data set and, provided any other test
assumptions are satisfied, an appropriate parametric test can be used.
39
Data for Testing the Normality
Name Mathematics English Science
A 2 6 20
B 3 8 30
C 3 12 30
D 4 18 40
E 2 14 20
F 3 15 30
G 5 16 50
H 4 13 40
I 6 18 60
J 7 16 70
K 4 17 40
L 8 15 80
M 12 4 95
N 14 17 99
Statistical Tests
When testing the normality of a dataset, there are several
statistical hypotheses that can be used. Here are the two 1. K-S Test,
most common hypotheses: 2. Shapiro Tests
1.Null Hypothesis (H0): The data is normally 3. Anderson-
Darling Test.
distributed. 4. Jacque-Bera
This hypothesis assumes that the dataset follows a test
normal distribution.
2.Alternative Hypothesis (HA): The data is not
normally distributed.
This hypothesis suggests that the dataset does not
follow a normal distribution.
Testing Normality with SPSS
Performance of the 12 students are given below.The performance
Name Mathematics English Science
H0: The distribution is normal.
A 80 80 50
Ha: The distribution is NOT normal.
B 30 70 70
C 40 60 70
D 30 40 20
E
F
30
40
60
30
60
50
Analyze → Descriptive
G 50 50 80 Statistics → Explore
H 50 60 60
I 40 40 10
J 40 70 70
K 30 50 80
L 30 20 60
P-Value Criteria – Decision Rule
1. If P-value < 0.01 reject H0, So the distribution is NOT normal at 0.01
level of significance.

2. If 0.01 < P-value < 0.05 reject H0, So the distribution is NOT normal at
0.05 level of significance.

3. If the P-value > 0.05 No evidence to reject the Ho, So the distribution
is normally distributed.

❖It's important to note that failing to reject the null hypothesis does
not prove it to be true; it simply means that there is not enough
evidence to support the alternative hypothesis.

43
Testing Normality with SPSS
Skewness
Skewness is a measure of the asymmetry of the probability distribution of a
random variable about its mean. In other words, skewness tells you the
amount and direction of skew (departure from horizontal symmetry).
If skewness is 0, the data are perfectly symmetrical, although it is quite
unlikely for real-world data.
As a general rule of thumb:
❖If skewness is less than -1 or greater than 1, the distribution is highly
skewed.
❖If skewness is between -1 and -0.5 or between 0.5 and 1, the distribution is
moderately skewed.
❖If skewness is between -0.5 and 0.5, the distribution is
approximately symmetric.
Treatments for Non-normal Data
▪ Parametric statistics provide a correct picture only when
the metric data is normally distributed.

▪ The outliers however should be deleted first (one at a time)


if these outliers are not giving any extra information and are
just randomly appearing in the data.

▪ It's best if you collect some data, check the normality, and
decide on a transformation before you run your actual
experiment/statistical test.
Data Transformations for Normality
The parametric statistical methods assume that the variable of interest
(dependent variable) should be normally distributed. The non-normal
data can be transformed to attain normality using the box-cox
transformation or power transformation, where the central limit theorem
cannot justify the small sample size.
Following transformations suitable in different situations can be
used to convert the skewed data to the normal distribution.
➢f (x) = log(x); use it if x is positively skewed.
➢f (x) = x2; use it if the distribution of x is negatively skewed.
➢f (x) = √x; use it if x has a Poisson distribution (x =0, 1, 2, …).
➢f (x) = 1/X ; use it if the variance of x is proportional to the fourth power of
E(x).
4. Nature of the
Samples; Independent or
Related Sample
Two common types of sample nature in
business research are independent samples
and related samples.
Independent Sample
▪ Independent samples refer to cases where the
observations in one group or sample are completely
unrelated or independent of the observations in
another group or sample.

▪ This means that the individuals or entities in one sample


do not influence or have any connection with the
individuals or entities in the other sample.
Independent Sample
Independent Sample:
Research Topic: The impact of customer service on customer satisfaction

Sample 1:Customers who received customer service from Company A.


Sample 2: Customers who received customer service from Company B.

In this case, the samples are independent because the customers in Sample
1 (Company A) are unrelated and independent of the customers in Sample 2
(Company B).

❖The experiences of one group of customers do not influence or


depend on the experiences of the other group.
Related/Paired/Dependent Sample
Related samples involve observations that are related or
connected in some way. This can include paired or
matched samples, where each observation in one sample is
directly linked or paired with a specific observation in the
other sample.
Related/Paired/Dependent Sample
Related Sample (Paired Sample):
Research Topic:
The effectiveness of a new employee training program
Sample- 01: Pre-training performance scores of employees
Sample- 02: Post-training performance scores of the same employees.
In this case, the samples are related or paired because the observations in
Sample 2 (post-training scores) are directly linked or connected to the
observations in Sample 1 (pre-training scores) for the same group of
employees.
❖The performance scores of each employee are compared before and after
the training program.
5. How Many Groups are
involved?
Choosing Statistical Procedures
One Independent Variable Two Independent Variables
Measurement Two Levels More than 2 Levels Factorial Designs
Scale of the
Dependent Two Two Multiple Multiple
Independent Dependent
Variable Independent Dependent Independent Dependent
Groups Groups
Groups Groups Groups Groups
Two-Factor
Repeated
Independent Dependent One-Way Two -Factor ANOVA
Interval or Ratio Measures
t-test t-test ANOVA ANOVA Repeated
ANOVA
Measures
Mann- Kruskal-
Ordinal Wilcoxon Friedman
Whitney U Wallis

Nominal Chi-Square Chi-Square Chi-Square


Overall Guidelines to Selection
Appropriate Statistical Test for Your
Research;
Considering all the criteria at once
Level of Measurement and Statistical Methods
Level of Measurement and Statistical Methods
Level of Measurement and Statistical Methods
Session-14

Performing Parametric and Non-


parametric Statistical Tests Using SPSS
and Interpreting the Results
Performing Parametric
Tests
The Pathway to the Analysis of
Continuous (Interval/Ratio)
Variables
The Pathway to the Analysis of Continuous
(Interval/Ratio) Variables

Data exploration and summary statistics:


• Example: Suppose you have collected data on the annual revenue (in dollars)
of different companies. Start by exploring the dataset, checking for missing
values and outliers. Calculate summary statistics like mean, median, range, and
standard deviation to understand the distribution and variability of the revenue
variable.
Data visualization:
• Example: Create a histogram or a box plot to visualize the distribution of the
annual revenue. This can help you identify if the revenue follows a particular
pattern, such as being skewed or normally distributed.
The Pathway to the Analysis of Continuous
(Interval/Ratio) Variables

Assessing normality:
• Example: Use a Q-Q plot or perform a normality test (e.g., Shapiro-Wilk
test) to assess if the annual revenue variable is normally distributed. This is
important if you plan to use statistical tests that assume normality, such as t-
tests or ANOVA.
Correlation analysis:
• Example: Investigate the relationship between the annual revenue and
another continuous variable, such as advertising expenditure. Calculate
the correlation coefficient (e.g., Pearson's correlation) to quantify the
strength and direction of the relationship. Visualize the relationship using a
scatter plot.
The Pathway to the Analysis of Continuous
(Interval/Ratio) Variables

Hypothesis testing:
• Example: Formulate a hypothesis about the relationship between employee
satisfaction (continuous variable) and employee turnover rate. Conduct a t-test
or ANOVA to compare the mean employee satisfaction scores across different
turnover rate groups (e.g., low turnover vs. high turnover).
Regression analysis:
• Example: Perform a multiple linear regression analysis to examine how
variables like advertising expenditure, product price, and market size (all
continuous variables) affect the annual revenue. Evaluate the significance of
the regression coefficients and interpret their impact on the dependent
variable (revenue).
The Pathway to the Analysis of Continuous
(Interval/Ratio) Variables
Multivariate analysis:
• Example: Use factor analysis to identify underlying dimensions or factors that
influence customer satisfaction, based on multiple continuous variables like
product quality, customer service rating, and pricing perception.
Model diagnostics and assumptions:
• Example: Check the assumptions of regression analysis, such as linearity,
independence of errors, and homoscedasticity. Evaluate the presence of
influential observations or multicollinearity, which could affect the
interpretation of the results.
Interpretation and reporting:
• Example: Interpret the results of your analysis, explaining how variables like
advertising expenditure, employee satisfaction, or pricing perception impact
business outcomes such as revenue or customer satisfaction. Provide clear
and concise explanations, supported by relevant statistical measures and
visualizations.
Parametric Statistical Tests
Parametric Assumptions to Practical Hypotheses SPSS Procedure
Test be Satisfied Example Formation
Karl Linearity, Assessing the H0: Analyze -> Correlate
Pearson's independence of relationship There is no correlation -> Bivariate
observations, between between the two variables. Correlations
Correlation normality of variables employee job H1:
Tests satisfaction There is a correlation
scores and their between the two variables
performance
ratings
Self-Evaluation Questions-07
A manufacturing company notices an increase in the number of errors
made by its employees. To investigate the reasons for errors, 10
employees were selected and reported their level of stress, skills and
number of errors made by them as follows.
Employee Stress Scores(out of Skill Scores-X2 No. of
Name 1000)-X1 Errors-Y
A 168 53 5
B 186 18 7
C 125 75 3
D 170 45 7
E 132 70 4
F 130 81 3
G 190 48 7
H 100 88 3
I 175 55 6
J 180 41 7
P- Value Criteria – Decision Rule
1.If P-value < 0.01 reject H0, So there is a significant mean difference
among Groups and It is statistically significant at 0.01.

2. If 0.01 < P-value < 0.05 reject H0, So there is a significant mean
difference among Groups and It is statistically significant at 0.05.

2. If P-value > 0.05 There is no evidence to reject the H0 and It is not


statistically significant. So there is no significant mean difference among
Groups.

❖It's important to note that failing to reject the null hypothesis


does not prove it to be true; it simply means that there is not
enough evidence to support the alternative hypothesis.
68
Parametric Statistical Tests
Recommended
Parametric Assumptions to Assumption Practical
Test be Satisfied Testing Methods Example Hypotheses Formation
H0:
µ = µ0
Shapiro-Wilk test, Determining if the (the population mean is
Random
Single Anderson-Darling average customer equal to the benchmark
sampling, satisfaction score is
value)
Sample-t test, or visual significantly different
Normality of
test inspection of Q-Q from a benchmark H1:
data
plots value
µ ≠ µ0
(the population mean is
different)
SPSS Procedure:
Analyze -> Compare Means -> One Sample t-test
Company ROA
Company A 0.10
Company B 0.15
Company C 0.12
Company D 0.11
Company E 0.09
Company F 0.13
Company G 0.14
Company H 0.16
Company I 0.08
Company J 0.10

Sample Data Sheet Company K


Company L
0.05
0.04
Company M 0.07
Company N 0.06
Company O 0.03
Company P 0.05
Company Q 0.09
Company R 0.08
Company S 0.06
Company T 0.07
Parametric Statistical Tests
Recommended
Assumptions to be Assumption Testing Practical Hypotheses
Parametric Test Satisfied Methods Example Formation

Comparing the H0: µ1 = µ2


Independent Shapiro-Wilk test, mean customer
Independent Anderson-Darling test, (the means are
random samples, satisfaction
Samples t- or visual inspection of equal)
normality of data, Q-Q plots; Levene's test scores between
test H1: µ1 ≠ µ2 (the
equal variances or Bartlett's test two different
product groups means are different)

SPSS Procedure:
Analyze -> Compare Means -> Independent Samples t-test
The independent samples t-test

The purpose of using the independent samples t-test in


research is to compare the means of two independent groups
and determine if there is a statistically significant difference
between them. It allows researchers to investigate various
factors and their impact on financial outcomes.
Key Assumptions of the Independent Samples t-
test
❖Independence:
▪ This assumption is typically assumed in study design, and It relies on appropriate sampling or
experimental design.
❖Normality:
▪ Assume that the distribution of the data within each group follows a normal distribution.
▪ To test the assumption of normality, you can use the Shapiro-Wilk test or the Kolmogorov-
Smirnov test in SPSS.
▪ These tests assess whether the distribution of the dependent variable in each group
significantly deviates from a normal distribution. If the p-value is greater than 0.05, the
assumption of normality is met.
Key Assumptions of the Independent Samples t-test

❖Homogeneity of Variance: To test the assumption of


homogeneity of variance, you can use Levene's test in
SPSS. This test compares the variances of the dependent
variable between the two groups. If the p-value is greater
than 0.05, the assumption of homogeneity of variance is
met.
Research : Example

Title
"The Effect of Corporate Governance on Firm Performance"
Research Objectives:
❖To examine the relationship between corporate governance practices and firm
performance.
❖To determine if firms with stronger corporate governance structures exhibit better
financial performance compared to firms with weaker governance structures.
Research Problem & Questions: Example
Research Problem:
❖ The research aims to explore whether corporate governance practices have an
impact on firm performance.

Research Questions:
❖Is there a significant difference in financial performance between firms with
strong corporate governance structures and firms with weak governance
structures?
❖Do firms with stronger corporate governance structures exhibit better financial
performance compared to firms with weaker governance structures?
Research Hypotheses

❖ Null Hypothesis H0: There is no significant difference in financial


performance between firms with strong corporate governance structures
and firms with weak governance structures.
❖ Alternative Hypothesis H1:
Firms with strong corporate governance structures will have better
financial performance compared to firms with weak governance
structures.
Variables Identification and Measurements
Independent Variables: Corporate governance (strong or weak)

Dependent Variable: Firm performance

• Moderator Variables: None specified in this example.

Measuring Each Research Variable:

• Corporate Governance: Use established corporate governance indices or


scoring systems to assess the strength of governance practices in each firm.

• Firm Performance: Measure financial performance using indicators such as


return on assets (ROA), return on equity (ROE), or earnings per share (EPS).
Company Governance ROA
Company A Strong 0.10
Company B Strong 0.15
Company C Strong 0.12
Company D Strong 0.11
Company E Strong 0.09
Company F Strong 0.13
Company G Strong 0.14
Company H Strong 0.16
Company I Strong 0.08
Company J Strong 0.10

Sample Data Sheet Company K


Company L
Weak
Weak
0.05
0.04
Company M Weak 0.07
Company N Weak 0.06
Company O Weak 0.03
Company P Weak 0.05
Company Q Weak 0.09
Company R Weak 0.08
Company S Weak 0.06
Company T Weak 0.07
SPSS Procedures for Analyzing Data using Independent Samples T-test
1. Launch SPSS software and open the dataset containing the variables of interest.
2. Select "Analyze" from the menu bar, then choose "Compare Means" and "Independent-Samples T Test."
3. Move the dependent variable (ROA) to the "Test Variable(s)" box.
4. Move the independent variable (Governance) to the "Grouping Variable" box.
5. Click on the "Define Groups" button and specify the values or categories for the two groups (e.g., strong
and weak governance).
6. Click "OK" to run the analysis.
7. SPSS will generate output, including descriptive statistics, t-test results, effect size measures, and
confidence intervals.
8. Interpret the results by examining the significance level (p-value) of the t-test and effect size measures
to determine if there is a significant difference in firm performance between the two groups.
Company Dividend Policy Stock Returns
Company A Dividend-paying 5%
Company B Dividend-paying 10%
Company C Dividend-paying 8%
Company D Dividend-paying 6%
Company E Dividend-paying 9%
Company F Dividend-paying 7%
Company G Dividend-paying 3%
Company H Dividend-paying 2%
Company I Dividend-paying 4%
Company J Dividend-paying 7%

Sample Data Sheet Company K


Company L
Non-dividend
Non-dividend
2%
1%
Company M Non-dividend 3%
Company N Non-dividend 5%
Company O Non-dividend 4%
Company P Non-dividend 2%
Company Q Non-dividend 1%
Company R Non-dividend 3%
Company S Non-dividend 6%
Company T Non-dividend 4%
Parametric Statistical Tests
Recommended
Parametric Assumptions to be Assumption Testing Practical Hypotheses
Test Satisfied Methods Example Formation
H0:
Analyzing the
Shapiro-Wilk test, impact of µ1 = µ2 = µ3 = ... =
Independence of Anderson-Darling different µk
One-way observations, test, or visual leadership styles (the means are
ANOVA normality of data, inspection of Q-Q on employee job equal)
equal variances plots; Levene's test satisfaction H1:
or Bartlett's test across multiple
At least one mean is
departments
different

SPSS Procedure:
Analyze -> Compare Means -> One-Way ANOVA
ANOVA Test
❖The analysis of variance (ANOVA) test is a statistical method used to
determine whether there are any significant differences between the
means of two or more groups.

❖The purpose of using ANOVA in research, particularly in finance, is


to compare means across multiple groups. It helps determine if there
are significant differences in the dependent variable among different
categories or levels of the independent variable(s). ANOVA is used
when there are three or more groups to compare.
Testing Assumptions

• Independence: This assumption is typically assumed to be met unless


there is a specific reason to suspect dependence between observations.
• Normality: You can use the Shapiro-Wilk test or the Kolmogorov-
Smirnov test in SPSS to check the normality assumption. In SPSS, go to
"Analyze" > "Descriptive Statistics" > "Explore." Select the dependent
variable and click "Plots." Under "Normality plots," choose the desired
test.
Testing Assumptions

• Homogeneity of variances: To test this assumption, you can


use Levene's test in SPSS. Go to "Analyze" > "Compare Means" >
"One-Way ANOVA." Select the dependent variable and click
"Options." Tick the box for "Homogeneity of variances test."
Performing ANOVA Using SPSS-Example-01
To perform an ANOVA test for your research on the impact of
corporate governance levels on financial performance, you would
need a datasheet that includes the firm's governance level (strong,
moderate, weak) and their corresponding return on assets (RoA)
values.
Company Governance Level Return on Assets
Company 1 Strong 0.12
Company 2 Strong 0.15
Company 3 Strong 0.09
Company 4 Strong 0.14
Company 5 Strong 0.11
Company 6 Strong 0.13
Company 7 Strong 0.10
Company 8 Strong 0.08
Company 9 Strong 0.07
Company 10 Strong 0.11
Company 11 Moderate 0.08
Company 12 Moderate 0.07
Company 13 Moderate 0.06
Company 14 Moderate 0.09
Company 15 Moderate 0.05
Company Governance Level Return on Assets
Company 16 Moderate 0.07
Company 17 Moderate 0.06
Company 18 Moderate 0.08
Company 19 Moderate 0.09
Company 20 Moderate 0.07
Company 21 Weak 0.03
Company 22 Weak 0.02
Company 23 Weak 0.01
Company 24 Weak 0.04
Company 25 Weak 0.06
Company 26 Weak 0.03
Company 27 Weak 0.02
Company 28 Weak 0.01
Company 29 Weak 0.04
Company 30 Weak 0.06
P- Value Criteria – Decision Rule
1.If P-value < 0.01 reject H0, So there is a significant mean difference
among Groups and It is statistically significant at 0.01.

2. If 0.01 < P-value < 0.05 reject H0, So there is a significant mean
difference among Groups and It is statistically significant at 0.05.

2. If P-value > 0.05 There is no evidence to reject the H0 and It is not


statistically significant. So there is no significant mean difference among
Groups.

❖It's important to note that failing to reject the null hypothesis


does not prove it to be true; it simply means that there is not
enough evidence to support the alternative hypothesis.
89
Group 1 Group 2 Group 3 Group 4
11.61 9.78 14.40 14.65
14.37 6.59 14.27 12.28
13.61 15.14 11.13 12.08
13.35 8.78 15.46 5.37
8.62 9.68 13.53 14.30
8.90 14.59 11.81 5.84
Case No-05 13.48 8.97 12.64 7.21
19.64 8.59 14.52 19.13
13.89 12.21 14.37 21.04
13.48 6.23 9.61 13.80
13.41 9.72 15.38 19.18
Sales in Rs’’ Mn 10.82 13.50 12.14 26.59
18.99 4.79 13.50 17.44
among 4-groups 11.73 8.73 12.68 21.21
13.07 6.95 10.69 9.42
8.85 8.61 14.12 11.11
9.67 5.67 12.09 11.13
17.68 13.82 11.45 15.72
16.80 14.56 9.81 18.49
16.33 5.37 13.96 9.81
Parametric Statistical Tests
Recommended
Assumptions to be Assumption Testing Practical Hypotheses
Parametric Test Satisfied Methods Example Formation
Evaluating the
effectiveness of H0: µd = 0
Shapiro-Wilk test, (the means are
Paired a new training
Anderson-Darling equal)
Paired observations, program by
test, or visual
Samples t-test normality of comparing pre- H1: µd ≠ 0
inspection of Q-Q
differences and post-
plots (the means are
training test
different)
scores

SPSS Procedure:
Analyze -> Compare Means -> Paired Samples t-test
Paired Samples t-test
❖The paired samples t-test is a statistical test used to determine whether there
is a significant difference between the means of two related variables
in a sample.

❖It is appropriate when the same group of participants is measured


twice, or when two related variables are measured within the same
participants. The test compares the mean difference between the paired
observations to zero.
Testing each assumption using SPSS
❖Normality: You can test the normality assumption by examining the distribution of the
differences graphically using a histogram or a Q-Q plot. Additionally, you can use statistical tests
such as the Shapiro-Wilk test or the Kolmogorov-Smirnov test to assess normality.
❖Independence: In a paired samples t-test, independence is assumed because each pair of
observations comes from the same individual or unit. However, if there is any form of
dependency or clustering in your data, you should consider using alternative statistical methods.
❖Homogeneity of variances: To test this assumption, you can use Levene's test for homogeneity
of variances in SPSS. The test examines whether the variances of the differences between paired
observations are equal across all levels of the independent variable.
Parametric Statistical Tests
Title:
The impact of Corporate Governance on the Financial Performance
of Listed Companies in the CSE in Sri Lanka.
Set up Hypotheses:
Null hypothesis (H₀): There is no significant difference in the financial
performance (RoA) of listed companies between the pre-and post-
Covid-19 pandemic situations.
Alternative hypothesis (H₁): There is a significant difference in the
financial performance (RoA) of listed companies between the pre-and
post-Covid-19 pandemic situations.
Pre-Pandemic Post-Pandemic
Company
RoA RoA
A 0.28 0.04
B 0.28 0.08
C 0.16 0.05
D 0.21 0.03
E 0.14 0.05
To perform the paired samples t-test for the F 0.19 0.01
G 0.23 0.04
research on the financial performance (RoA) H 0.16 0.01
I 0.19 0.09
between the Pre and Post-Covid-19 pandemic J 0.21 0.02
K 0.16 0.11
situations of listed companies in the Colombo L 0.19 0.09
M 0.25 0.12
Stock Exchange (CSE) in Sri Lanka, you would N 0.27 0.06
O 0.18 0.13
need a data table that includes the performance P 0.21 0.06
Q 0.16 0.11
data of equal no. of firms for both periods R 0.28 0.14
S 0.21 0.12
(n=25). T 0.26 0.11
U 0.23 0.06
V 0.18 0.11
W 0.21 0.12
X 0.24 0.05
Y 0.23 0.06
Steps to Perform the Paired Samples t-test:

▪ Step 1: Define the null hypothesis (H0) and the alternative hypothesis
(Ha)
▪ Step 2: Calculate the difference between the Pre and post-pandemic
financial performance for each company.
▪ Step 3: Calculate the mean (M) and standard deviation (SD) of the
differences
Steps to perform the paired samples t-test
▪ Step 4: Use the paired samples t-test formula to calculate the t-value:
t = (M - μ) / (SD / √n)
Where μ is the hypothesized mean difference (usually 0 for a
paired t-test), n is the number of paired samples (25 in this case),
M is the mean of the differences, and SD is the standard of the
differences.
▪ Step 5: Determine the degrees of freedom (df) for the t-distribution,
which is equal to (n - 1).
▪ Step 6: Determine the critical t-value or p-value based on your desired
level of significance (e.g., 0.05).
Steps to Perform the Paired Samples t-test

Step 7: Compare the calculated t-value with the critical t-value or


p-value to make a decision:
▪ If the calculated t-value is greater than the critical t-value or the p-value is less than
the chosen significance level, reject the null hypothesis (H0) and conclude that
there is a significant difference between the Before(Pre) and After
(Post)pandemic financial performance.

▪ If the calculated t-value is less than the critical t-value or the p-value is greater than
the chosen significance level, fail to reject the null hypothesis (H0) and conclude
that there is no significant difference between the pre and post-pandemic
financial performance.
P- Value Criteria – Decision Rule
1.If P-value < 0.01 reject H0, So there is a significant mean difference
between Two pairs( Pre and Post) and It is statistically significant at
0.01.

2. If 0.01 < P-value < 0.05 reject H0, So there is a significant mean
difference between Two pairs( Pre and Post) and It is statistically
significant at 0.05.

2. If P-value > 0.05 There is no evidence to reject the H0 and It is not


statistically significant. So there is no significant mean difference between
Two pairs( Pre and Post).

❖It's important to note that failing to reject the null hypothesis does
not prove it to be true; it simply means that there is not enough
evidence to support the alternative hypothesis.
99
Parametric Statistical Tests
Recommended
Assumptions to be Assumption Testing Practical Hypotheses
Parametric Test Satisfied Methods Example Formation
H0:
Studying the There is no effect
impact of of the training
Shapiro-Wilk test,
Independence of different methods over
Anderson-Darling
observations, training
Repeated test, or visual time
normality of data, methods on
ANOVA inspection of Q-Q H1:
sphericity employee
plots; Mauchly's test
assumption performance There is an effect
for sphericity
over three-time of the training
points methods over
time
SPSS Procedure:
Analyze -> General Linear Model -> Repeated Measures
Parametric Statistical Tests
Recommended
Parametric Assumptions to Assumption Practical Hypotheses
Test be Satisfied Testing Methods Example Formation
Investigating the H0:
Independence of Shapiro-Wilk test,
combined effect There is no interaction
observations, Anderson-Darling of product quality effect between the two
Two-Way normality ofTests
data, test, or visual and customer independent variables
ANOVA equal variances, inspection of Q-Q service on
interaction effect plots; Levene's test or customer H1:
assumption Bartlett's test
satisfaction in There is an interaction
different regions effect

SPSS Procedure:
Analyze -> General Linear Model -> Univariate
Parametric Statistical Tests
Parametric Assumptions to be Practical Hypotheses SPSS
Test Satisfied Example Formation Procedure

Examining the H0:


Linearity, impact of
Linear independence of advertising The independent Analyze->
variables have no
Multiple observations, expenditure, Regression->
pricing, and effect on the
Regression normality of dependent
Linear
product quality on
variables variable
sales performance
Purpose of Using Regression analysis-
Example
▪ The purpose of using regression analysis in this research is to
examine the relationship between sustainability report
disclosure and company value in Sri Lankan listed companies.
▪ It aims to investigate the moderating effect of profitability on the
relationship between sustainability report disclosure and
company value.
▪ Regression analysis will help in understanding the impact of
sustainability report disclosure on company value and how this
relationship is influenced by profitability.
Assumptions of Regression Analysis

• Linearity: The relationship between the dependent variable and


independent variables is linear.
• Independence: Observations in the dataset are independent of each other.
• Normality: The residuals follow a normal distribution.
• No multicollinearity: The independent variables are not highly correlated
with each other.
• Homoscedasticity: The variance of the residuals is constant across all
levels of the independent variables.
Assumption Testing Procedures
▪ Linearity: Create scatterplots to visually inspect the relationship between the dependent variable and
each independent variable. You can use SPSS's "Graphs" menu to plot the variables against each other.

▪ Independence: Independence is often assumed in regression analysis, especially when data is


collected through random sampling or experimental design.

▪ Normality: After running the regression analysis, examine the histogram or probability plot of the
residuals to assess normality visually. You can use SPSS's "Graphs" menu to create these plots.
Additionally, you can perform formal tests for normality such as the Shapiro-Wilk or Kolmogorov-
Smirnov tests using SPSS's "Explore" or "Nonparametric Tests" options.
Organizing the Data Sheet
Company Sustainability Report Company Value Profitability
Disclosure
Company A 0.75 100,000 0.10
Company B 0.60 80,000 0.08
Company C 0.85 120,000 0.12
Company D 0.70 90,000 0.09
Company E 0.80 110,000 0.11
Company F 0.65 85,000 0.07
Company G 0.70 95,000 0.09
Company H 0.75 105,000 0.10
Company I 0.90 130,000 0.13
Company J 0.80 115,000 0.11
Company K 0.70 95,000 0.09
Company L 0.75 105,000 0.10
Company M 0.85 125,000 0.12
Company N 0.75 100,000 0.10
Company O 0.70 90,000 0.09
Company P 0.80 110,000 0.11
Company Q 0.65 85,000 0.07
Company R 0.70 95,000 0.09
Company S 0.75 105,000 0.10
Company T 0.90 130,000 0.13
Research Hypotheses

❖Hypothesis 1: There is a positive relationship between sustainability report


disclosure and company value in Sri Lankan listed companies.

❖Hypothesis 2: There is a positive impact of sustainability report disclosure on the


company value in Sri Lankan listed companies.

❖Hypothesis 3: The profitability of companies moderates the relationship


between sustainability report disclosure and company value.
SPSS Procedures for analyzing data using Regression
1. Import the dataset into SPSS.
2. Select "Analyze" from the menu and choose "Regression" and then "Linear."
3. Move the dependent variable (Company Value) into the "Dependent" box and the
independent variable (Sustainability Report Disclosure) into the "Independent(s)" box.
4. Click on the "Covariates" button and move the moderator variable (Profitability) into the
"Covariate(s)" box.
5. Specify any additional options, such as confidence intervals or diagnostic statistics.
6. Click "OK" to run the regression analysis.
7. Examine the regression coefficients, p-values, and standard errors to assess the significance
and direction of the relationships.
8. Interpret the results, considering the research objectives, hypotheses, and practical
implications in the context of accounting and finance.
Q&A
Session

You might also like