Module-5:

DATA ANALYSIS
 Syllabus for Module 5
Basic data analysis: Descriptive Statistics, Univariate and Bivariate, Parametric
& Non-Parametric Tests; Null & Alternative Hypothesis, Error in Testing of
Hypothesis, Critical Region, Degrees of Freedom, One Tailed & Two Tailed Tests,
Standard Error; Procedure for Testing of Hypothesis. Parametric test, Non
parametric test (Conditions for applicability, practical applicability,
Implementation and statistical Inference of all the above tests)
Meaning of Analysis
In the context of Business Research, analysis refers to the systematic examination
and interpretation of data to extract meaningful insights that can inform decision-
making.
Data analysis
Data analysis is the application of reasoning to understand the data
that have been gathered. In its simplest form, analysis may involve
determining consistent patterns and summarizing the relevant details
revealed in the investigation. The appropriate analytical technique for
data analysis will be determined by management’s information
requirements, the characteristics of the research design, and the
nature of the data gathered. Statistical analysis may range from
portraying a simple frequency distribution to more complex
multivariate analyses approaches, such as multiple regression.
Later chapters will discuss three general categories of statistical
analysis: univariate analysis, bivariate analysis, and multivariate
analysis.
 Key Aspects of Analysis
Aspect Description
Data Collection Gathering relevant data through surveys, experiments, observations, or existing
databases.
Data Cleaning Preparing the data by removing inconsistencies, errors, and irrelevant information to
ensure accuracy.
Descriptive Analysis Summarizing the basic features of the data, providing simple summaries about the
sample and the measures (e.g., mean, median, mode).
Inferential Analysis Drawing conclusions about a population based on sample data, using statistical methods
to test hypotheses (e.g., t-tests, ANOVA).
Predictive Analysis Using historical data to make predictions about future outcomes or trends through
methods like regression analysis.
Diagnostic Analysis Examining data to understand the causes of past outcomes, often by identifying patterns
or correlations.
Prescriptive Analysis Providing recommendations based on the analysis, often through simulation or
optimization techniques.
Visualization Creating charts, graphs, and dashboards to represent data visually, making it easier to
understand and communicate findings.
 TYPES OF ANALYSIS
Types Description
Descriptive Analysis Summarizes and describes the main features of a dataset. It provides simple statistics (mean, median,
mode) and visualizations (charts, graphs).
Inferential Analysis Makes inferences about a population based on a sample. It uses statistical tests (e.g., t-tests, ANOVA) to
draw conclusions and test hypotheses.
Predictive Analysis Uses historical data to predict future outcomes. It employs techniques like regression analysis and machine
learning models.
Diagnostic Analysis Examines data to understand the causes of past outcomes. It identifies patterns and relationships to
explain why certain events occurred.
Prescriptive Analysis Provides recommendations for actions based on data insights. It often uses optimization and simulation
techniques to suggest the best course of action.
Exploratory Data Analysis Involves analyzing datasets to discover patterns, spot anomalies, and test hypotheses, often using visual
(EDA) methods.
Text Analysis Analyzes textual data to extract meaningful information, often using natural language processing
techniques.
Time Series Analysis Analyzes data points collected or recorded at specific time intervals to identify trends, seasonal patterns, or
cycles over time.
Comparative Analysis Compares different datasets or variables to highlight differences and similarities, often to assess
performance or outcomes.
Multivariate Analysis Analyzes multiple variables simultaneously to understand complex relationships and effects among them
(e.g., factor analysis, cluster analysi
 Importance of Analysis in Business Research
Informed Decision-Making: Helps managers and stakeholders make data-driven decisions.

Identifying Trends: Uncovers market trends, customer preferences, and potential areas for growth.

Performance Evaluation: Assesses business performance and effectiveness of strategies.

Risk Management: Identifies potential risks and provides insights for mitigation strategies.
 Application of Statistical Tools in Analysis
Statistical Tool Application
Descriptive Statistics Summarizes and describes the main features of a dataset using measures like mean,
median, mode, and standard deviation. Useful for initial data exploration.
Inferential Statistics Allows conclusions about a population based on sample data. Techniques include t-
tests, chi-square tests, and ANOVA. Used in hypothesis testing and confidence interval
estimation.
Regression Analysis Analyzes relationships between variables. Used for prediction and forecasting, helping
to understand how changes in one variable affect another.
Correlation Analysis Measures the strength and direction of relationships between variables. Commonly
used in market research to identify associations between factors.
ANOVA (Analysis of Compares means across multiple groups to determine if at least one group mean is
Variance) different. Useful in experiments and product testing.
Chi-Square Test Assesses the association between categorical variables. Often used in survey analysis
to understand relationships in responses.
Factor Analysis Identifies underlying relationships between variables by grouping them into factors.
Used in market research to reduce data complexity.
Cluster Analysis Segments data into distinct groups based on similarity. Used in customer segmentation
and targeting in marketing.
Time Series Analysis Analyzes data points collected over time to identify trends and seasonal patterns.
Common in sales forecasting and economic analysis.
 Exposure to Software Packages Used in Data Analysis
Software
Description Key Features Applications
Package
A widely used tool for statistical User-friendly interface,
SPSS analysis, especially in social sciences extensive statistical tests, Survey analysis, market research, social science studies.
and business research. data visualization.

A programming language and Open-source, vast libraries,

Advanced statistical analysis, data mining, and
R software environment for statistical strong data visualization
visualization.
computing and graphics. capabilities.

A versatile programming language Extensive libraries, easy

with powerful libraries for data integration with web
Python Data manipulation, machine learning, web scraping.
analysis (e.g., Pandas, NumPy, applications, strong
SciPy). community support.

A spreadsheet application that offers Pivot tables, data

Excel basic statistical functions and data visualization, add-ins for Financial modeling, data tracking, simple analyses.
analysis tools. advanced analytics.

Drag-and-drop interface,
A data visualization tool that helps
real-time data analysis, wide Business intelligence, performance monitoring, data
Tableau users create interactive and
range of visualization storytelling.
shareable dashboards.
options.
SAS A software suite used for advanced Robust analytics capabilities, Clinical trials, predictive
analytics, business intelligence, and extensive data management analytics, data mining.
data management. features, enterprise-level
support.
MATLAB A high-level language and interactive Strong matrix operations, Engineering, scientific
environment for numerical extensive toolboxes for research, algorithm
computation, visualization, and specialized analysis. development.
programming.
Stata A statistical software package used User-friendly interface, Economics, biostatistics,
for data analysis, manipulation, and comprehensive statistical social science research.
professional graphics. functions, good for panel data.
Microsoft A business analytics tool that enables Real-time data visualization, Business performance
Power BI users to visualize data and share easy integration with other analysis, reporting, data
insights across the organization. Microsoft products, interactive exploration.
dashboards.
KNIME An open-source platform for data Visual programming interface, Data preprocessing,
analytics, reporting, and integration. extensive data manipulation machine learning, data
capabilities, supports various mining.
data sources.
Descriptive Statistics
Descriptive Statistics
Descriptive Statistics provide a concise summary of the main features of a
dataset, offering a simple overview without making any conclusions about
the data's meaning or the hypothesis being tested. They are used to
describe or summarize data in a meaningful manner.
Importance of Descriptive Statistics:

Offers a foundational understanding of a dataset's characteristics.

Simplifies large quantities of data into easily understandable summaries.

Provides insights into the data distribution, variability, and central tendency.
 Significance of Descriptive Statistics in Business
Research Data analysis.
Descriptive statistics are crucial in business research as they:

Purpose Description
Summarize Data Provide a summary of the main aspects of the data.
Simplify Data Make data easier to interpret and understand.
Enable Comparison Facilitate comparison between different data sets.
Identify Patterns and Trends Reveal trends, patterns, or anomalies in the data.

Facilitate Decision-Making Assist in making informed and data-driven business decisions.

 Key Concepts in Descriptive Statistics:
Measure Category Measure Description
Mean The average value of the dataset.
Measures of Median The middle value when data is arranged in ascending or descending order.
Central Tendency
Mode The value(s) that appear most frequently in a dataset.
Range Difference between the highest and lowest values.
Variance Average of the squared differences from the mean.
Measures of
Standard Deviation Square root of the variance, indicating average distance from the mean.
Dispersion/Variabil
ity
Interquartile Range (IQR) Difference between the 75th percentile (Q3) and the 25th percentile (Q1).

Skewness Indicates the data's asymmetry around the mean. Positive skew indicates a
Measures of tail on the right, and negative skew indicates a tail on the left.
Shape
Kurtosis Describes the "tailedness" of the distribution.
Measures of Correlation Coefficient Measures the linear relationship between two variables.
Relationship
Example:

A company wants to understand the annual salaries of its employees.

Descriptive Statistics Application in Business Situation:
1. Mean: They calculate the average salary to be $50,000.
2. Median: The middle salary when all are arranged in order is
$48,500.
3. Mode: The most frequently occurring salary is $45,000.
4. Range: The difference between the highest and lowest salaries is
$40,000.
5. Standard Deviation: Salaries have a standard deviation of $7,000,
indicating variability around the mean.
.CSV File on Consumer Behaviour
Practical Class in SPSS
Practical class in SPSS
 SPSS Descriptive Analysis
Step Action Description
Open the SPSS software to
1. Open SPSS Launch SPSS software
begin analysis
2. Import Data File → Open → Data → Select CSV file Import the CSV file into SPSS
Perform descriptive analysis
on selected variables (e.g.
3. Descriptive Analyze → Descriptive Statistics →
Age,
Statistics Frequencies or Descriptives
Environmental_Concern,
Eco_Product_Knowledge)
Run the analysis to generate
4. Run Analysis Click OK
descriptive statistics output
Output:
Univariate and Bivariate
Univariate and Bivariate
Type of Analysis Definition Importance
Provides a detailed description
of each variable,
Analysis of a single variable to
Univariate Analysis understanding its distribution,
describe patterns within it
central tendency, and
dispersion
Offers insights into
relationships, correlations, or
Analysis of two variables to
causal patterns between two
Bivariate Analysis determine relationships or
variables, providing a deeper
associations between them
understanding of data
dynamics
 Univariate Analysis
Univariate Analysis Definition Techniques Example

Central Tendency Describe the data Mean, Median, Mean age of

Measures and find patterns Mode employees: 32

Dispersion Describe the spread Range, Variance, Age range: 21-50,

Measures of the data Standard Deviation Range: 29

Histogram: Majority
Visualize the data Histograms, Bar
Graphical Methods of employees in 25-
distribution Charts, Pie Charts
35 age bracket
Bivariate analysis
Bivariate analysis is a statistical method used to determine the
relationship between two variables. It provides a simple and
straightforward way to study the correlation and potential causation
between two factors. By comparing two sets of data, researchers can
understand the strength and direction of the relationship, if any.
Types of Bivariate Relationships:

Positive Relationship: Negative Relationship:

As one variable As one variable
increases, the other increases, the other
also increases. decreases.

No Relationship:
Changes in one
variable do not predict
changes in the other.
Example
A retail store wants to understand the buying behavior of its customers.
They have data on customer age and the amount spent on their last
visit.

Analysis Type Application Example

Customer Age: Histogram, Mean,
Median; Amount Spent:
Univariate Analysis Analyze individual variables
Frequency Distribution, Standard
Deviation
Analyze relationships between Age vs. Amount Spent: Scatter
Bivariate Analysis
two variables Plot, Correlation Analysis
Correlation Coefficient (r):
A numerical measure of the strength and direction of a linear relationship
between two variables. It ranges from -1 to +1.

Positive Correlation: Negative Correlation:

A coefficient close to A coefficient close to -
+1 indicates a strong 1 indicates a strong
positive relationship. negative relationship.

No Correlation: A
coefficient close to 0
indicates little to no
linear relationship.
Causation vs. Correlation:

Correlation does not

imply causation. Just
because two variables are
correlated doesn't mean
one caused the other.
Methods:

Scatter Plots: Pearson's Correlation

• Graphical representation to Coefficient:
visualize the relationship between • A statistical test used to measure
two variables. the degree and direction of the
• Each dot on the scatter plot linear relationship between two
represents a single observation. quantitative variables.

Cross-tabulations:
• Used for categorical data.
• Shows the frequency distribution
of the categories of two variables.
 Examples:
Analysis Type Example Description
Upward trend suggests
Advertising Spend vs. Sales positive relationship
Scatter Plot
Revenue between advertising spend
and sales revenue
Pearson's r value of 0.7
indicates strong positive
Pearson's Correlation Attendance vs. Final Grades
correlation between
attendance and final grades
Higher percentage of
Gender vs. Product females prefer Product A,
Cross-tabulation
Preference while males are evenly split
between products
Limitations:

As noted, correlation does

Bivariate analysis only
not imply causation. Other
examines two variables at a
studies or multivariate
time. It doesn't account for
analyses might be necessary
other potential influential
to draw definitive
factors.
conclusions about causation.
Conclusion:

Bivariate analysis is a crucial tool in

business research, allowing
researchers to identify potential
relationships between two variables.
However, while it's a powerful tool,
it's essential to remember its
limitations, especially concerning
inferring causation.
ere is the information in a table format:

Practical: For the last consumer behaviour data set:.

Procedure in SPSS Step-by-Step Instructions

1. Go to Analyze → Descriptive Statistics → Frequencies. <br> 2. Select a

categorical variable (e.g. Gender, Income_Level,
Frequencies
Willingness_to_Pay_More). <br> 3. Click OK to see frequency distribution
(count and percentage for each category).

1. Go to Analyze → Descriptive Statistics → Descriptives. <br> 2. Select scale

variables (e.g. Age, Eco_Product_Purchase_Frequency,
Descriptives
Environmental_Concern). <br> 3. Click Options to choose statistics to view
(e.g. mean, median, range, standard deviation). <br> 4. Click OK to get output.
Practical: Bivariate Analysis
Bivariate analysis examines the relationship between two variables. You
can analyze both categorical and continuous variables to understand
their relationships through correlation, cross-tabulation, or comparison
of means.
Procedure in SPSS:

a. Cross-Tabulation (for categorical variables):

•Go to Analyze → Descriptive Statistics → Crosstabs.
•Select two categorical variables like Gender and Income_Level.
•You can also click on Statistics and check Chi-Square to test for independence between
the two variables.
•Click OK to generate the cross-tabulation and Chi-Square test (if requested).
b. Correlation (for scale/continuous
variables):

Go to Analyze → Correlate → Bivariate.

Select continuous variables like Age, Environmental_Concern, and
Eco_Product_Purchase_Frequency.
Choose Pearson for correlation and click OK.
c. Comparing Means (for one categorical and
one continuous variable):
• Go to Analyze → Compare Means → Means.
• Select a continuous variable like Age or
Eco_Product_Purchase_Frequency as the dependent variable.
• Select a categorical variable like Gender or Income_Level as the
independent variable.
• Click OK to see the average value of the continuous variable for each
category.
Example of Interpretation:

Bivariate Analysis: A correlation

between Environmental_Concern
Univariate Analysis: From Age, you
and Willingness_to_Pay_More
might find that the average
might show a Pearson correlation
respondent is 30 years old with a
of 0.45, indicating a moderate
standard deviation of 5.6.
positive relationship between the
two.
Parametric & Non-Parametric Tests

In statistical hypothesis testing,

the decision of which statistical
test to use depends largely on
the data’s level of measurement,
variance, and distribution. Tests
are categorized as either
parametric or non-parametric.
Parametric Tests:

Tests that make assumptions about the parameters of the population

distribution from which the sample is drawn.
Assumptions:

The data is normally distributed (follows a bell curve).

The data is interval or ratio level.

Homogeneity of variance (equal variances) among the groups

being compared.
 Popular Parametric Tests:
Statistical Test Purpose Example

Comparing mean scores of

T-Test Compares means of two groups male and female students in
a math test

Comparing mean test scores

ANOVA (Analysis of Compares means among more
of students taught by three
Variance) than two groups
different teaching methods

Studying relationship
Examines relationship between between years of
Linear Regression
two continuous variables experience and salary in a
company
 Non-Parametric Tests:
Tests that don't make stringent assumptions about the population parameters. They are often used for ordinal or
nominal data.
Assumptions: Fewer assumptions compared to parametric tests. No need for normal distribution or equal
variance.
Popular Non-Parametric Tests:

Statistical Test Purpose Example

Determines association between two Examining relationship between
Chi-Square Test
categorical variables gender and smoking status
Compares differences between two
Comparing sales rankings between
Mann-Whitney U Test independent groups
two sales teams
(ordinal/continuous, non-normal)

Non-parametric ANOVA (compares Comparing job satisfaction ratings

Kruskal-Wallis Test
differences among >2 groups) across three company departments

Examines relationship between two Studying relationship between rank in

Spearman’s Rank Correlation
ordinal variables class and extracurricular activities
When to Use Which?

Data Type & Distribution: No Assumption

If data is interval or ratio Fulfilment: If the
level, and the distribution assumptions of parametric
is normal, then opt for a tests aren't met, choose a
parametric test. non-parametric test.

Sample Size: Non-

parametric tests can be
more suitable for smaller
sample sizes.
Limitations:

Non-parametric tests are often

considered less powerful and
Parametric tests are sensitive to
might not detect a difference or
violations of their assumptions. If
effect unless it's strong. However,
the assumptions aren't met, the
they're more versatile, especially
results can be misleading.
when data doesn't meet
parametric test assumptions.
Conclusion:

Both parametric and non-parametric tests offer valuable tools for

statistical hypothesis testing. The choice between them should be
driven by the nature of your data and the specific assumptions that can
be met. When in doubt, visualizing your data (e.g., using histograms or
QQ plots) and conducting assumption checks can guide the selection
process.
Null & Alternative Hypothesis and
Errors in Testing of Hypothesis
 Null & Alternative Hypothesis:
Null Hypothesis (H₀): A Alternative Hypothesis (H₁ or
statement of no effect, no Ha): A statement indicating the
difference, or no relationship. presence of an effect,
It represents the default or difference, or relationship. It
status quo scenario that represents what a researcher
researchers aim to test against. wants to prove.
Explanation:
In hypothesis testing, researchers start by assuming that the null
hypothesis is true. They then test this assumption using data. If there is
sufficient evidence against H₀ in favor of H₁, the null hypothesis is
rejected.
Examples :

Product Testing:
• H₀: The new product design does not increase sales.
• H₁: The new product design increases sales.
Employee Training:
• H₀: The new training program has no effect on employee
productivity.
• H₁: The new training program improves employee productivity.
Errors in Testing of Hypothesis:
Errors in Testing of Hypothesis:
When testing hypotheses, there's always a risk of making incorrect conclusions.
These mistakes fall into two main categories:
Concept Definition Explanation
Probability of committing a Type I error,
Type I Error (α) Rejecting the null hypothesis when it is actually true commonly set at 0.05 (5% chance of
wrongly rejecting H₀)

Probability of correctly rejecting the null

Failing to reject the null hypothesis when the alternative
Type II Error (β) hypothesis when the alternative is true,
hypothesis is true
increases as Type II error decreases

Explanation

Significance Level Commonly set at 0.05, meaning there's

Probability of committing a Type I error
(α) a 5% chance of wrongly rejecting H₀

Probability of correctly rejecting the null hypothesis when Increases as the chances of a Type II
Power (1 - β)
the alternative is true error decrease
 Examples in Business:

Scenario Type I Error Type II Error

Concluding the campaign

New Marketing Believing the campaign had no effect
increased sales when it had no
Campaign when it actually increased sales
effect

Discarding a batch of products

Quality Control in Passing a batch of products as quality
believing them to be defective
Manufacturing products when they are actually defective
when they are fine
 Relevance to Business Research Methods:

In business research, hypothesis testing is essential for decision-making. Here's why

understanding these concepts is crucial:

Benefits of Understanding Errors Description

Testing hypotheses enables businesses to make decisions based on evidence
Informed Decision-making
rather than intuition.

Understanding errors helps allocate resources effectively, avoiding wastage

Resource Allocation
on non-profitable investments (Type I error).

Setting appropriate significance levels and understanding error risks enables

Risk Management
businesses to manage decision-related risks.

Explaining error risks to stakeholders sets the right expectations and builds
Stakeholder Communication
trust when presenting findings.
Conclusion:

Understanding the null and alternative hypothesis, along with the

potential errors in hypothesis testing, is foundational in business
research methods. It equips researchers and decision-makers to draw
valid conclusions from data and make informed decisions that can
profoundly impact business strategies and outcomes.
Scatter Plot
One Tailed & Two Tailed Tests
One Tailed & Two Tailed Tests
In hypothesis testing, the tails of a test refer to the regions of the
distribution where the test statistic falls, leading to the rejection of the null
hypothesis.
• One Tailed Test: Tests the statistical significance in one direction (either
greater than or less than) from a particular value.
• Two Tailed Test: Tests the statistical significance in both directions (both
greater than and less than) from a particular value.
Importance of One Tailed & Two Tailed Tests:
• Helps researchers decide the direction of the hypothesis based on prior
knowledge or research objectives.
• Determines how the critical region (where we reject the null hypothesis) is
set up.
One Tailed vs. Two Tailed Tests:

Directionality:

One Tailed: Assumes the effect is only in a specific direction (either positive or negative).

Two Tailed: Does not assume a specific direction; the effect can be either positive or negative.
• Critical Region:
One Tailed: The critical region is entirely in one tail, either the right or the left.

Two Tailed: The critical region is split between both tails.

• Type I Error (α):
One Tailed: All α is in one tail.

Two Tailed: α is divided between two tails (commonly α/2 in each tail if α = 0.05).
 Example:

A company has developed a new drug they believe increases recovery speed from the flu.
They want to test if the new drug's mean recovery time differs from the old drug's mean
recovery time of 5 days.
Two Tailed Test Hypothesis:
• 0H0​: μ = 5 (The mean recovery time is equal to 5 days)
• Ha​: μ ≠ 5 (The mean recovery time is not equal to 5 days)
One Tailed Test Hypothesis (if they believe the new drug is faster):
• H0​: μ ≥ 5 (The mean recovery time is 5 days or more)
• Ha​: μ < 5 (The mean recovery time is less than 5 days)
Procedure for Testing of
Hypothesis
Hypothesis Testing
The Hypothesis Testing process involves making an initial assumption, collecting
evidence, and then statistically determining if the evidence supports or contradicts
the initial assumption or hypothesis.
Importance of Hypothesis Testing:

Provides a structured framework for making decisions based on data.

Aids in validating or refuting assumptions, theories, or models.

Facilitates objective and systematic analysis.

 Procedure for Testing of Hypothesis:
Step Description Details
1. Formulate H₀: No effect or no difference, H₁: Presence
Define null and alternative hypotheses
Hypotheses of an effect or difference
2. Choose Set the probability of rejecting H₀ when it's
Common choices: 0.05, 0.01, 0.10
Significance Level (α) true
Choose a statistical test based on data
3. Select Test Statistic type, distribution, sample size, and study e.g., t-test, ANOVA, chi-squared test
design
4. Compute Test Calculate the test statistic value using
Statistic collected data and relevant formulas
5. Determine Critical Find the value corresponding to α, defining
Value the rejection region for H₀
Reject H₀ if test statistic falls in rejection
6. Make a Decision Compare test statistic to critical value(s)
region (two-tailed or one-tailed)
Accept or reject H₀ based on test statistic
7. Draw a Conclusion
and critical value
 Example:
A shoe manufacturer believes that introducing eco-friendly materials will increase their
average monthly sales from the current 5,000 pairs.
Hypothesis Testing Process:
Step Description Details
H₀: Monthly sales ≤ 5,000 pairs, H₁: Monthly sales >
1. Formulate Hypotheses Define null and alternative hypotheses
5,000 pairs
2. Significance Level Set the significance level α = 0.05

3. Test Statistic Choose a statistical test One-sample t-test (assuming normal distribution)

Using sales data for months after introducing eco-

4. Compute Test Statistic Calculate the test statistic
friendly materials
5. Determine Critical Using a t-distribution table for α = 0.05 and degrees of
Find the critical value
Value freedom

6. Decision Compare computed t-value to critical t-value Reject H₀ if computed t-value > critical t-value

Determine if there's statistical evidence that monthly

7. Conclusion Draw a conclusion based on the decision
sales increased after introducing eco-friendly materials
Parametric Tests
 Parametric Tests
Parametric Tests are statistical tests that make certain assumptions about the parameters of
the population distribution from which the samples are drawn. Commonly, they assume that
the data follows a normal distribution and that samples have equal variances and are
independent.
Importance of Parametric Tests:

Generally more powerful and yield more accurate results when their assumptions
are met.

Suitable for making inferences about population parameters, such as means or

proportions.

Widely applicable to various types of business research questions.

 Common Parametric Tests:
Statistical Test Purpose Description
t-test (Student's t-test) Compare means of two Compares means from the same group at different
groups times (Paired Sample t-test) or means from two
different groups (Independent Sample t-test)

ANOVA (Analysis of Variance) Compare means of three Assesses the impact of one or more factors by
or more groups comparing means at different levels of the factor(s)

Regression Analysis Estimate relationship Estimates the relationship between a dependent

between variables variable and one (simple regression) or more (multiple
regression) independent variables
Pearson's Correlation Measure linear Measures the linear relationship between two
Coefficient relationship between two continuous variables
variables
Z-test Compare sample mean to Used to compare a sample mean to a population
population mean or mean, or to compare proportions, especially when the
proportions sample size is large
Example:

A company introduced a training program aiming to enhance employee

productivity. They recorded productivity scores (out of 100) for a
sample of employees before and after the training.
Parametric Test Application:
To determine if the training had a significant effect on productivity
scores, they could use a Paired Sample t-test to compare the means
before and after training.
Non-Parametric Tests
 Non-Parametric Tests
Non-Parametric Tests, often referred to as distribution-free tests, are statistical tests that do
not make specific assumptions about the population distribution. They are used when the data
doesn’t meet the normal distribution criteria or when working with ordinal or nominal data.
Importance of Non-Parametric Tests:

Useful when data violates assumptions of parametric tests.

Applicable to ordinal and nominal data, which can't always be addressed by

parametric tests.

More robust against outliers or skewed data.

 Common Non-Parametric Tests:
Statistical Test Purpose Description
Mann-Whitney U Test Compare distributions of two Alternative to independent sample t-test
(Wilcoxon Rank-Sum Test) independent samples when normality is violated

Wilcoxon Signed-Rank Compare two related samples or Non-parametric version of paired sample t-
Test repeated measures test
Kruskal-Wallis H Test Compare distributions of more than Non-parametric alternative to ANOVA
two independent samples
Spearman's Rank Measure strength and direction of Non-parametric version of Pearson's
Correlation Coefficient association between two ranked correlation coefficient
(Spearman's rho) variables
Chi-Squared Test of Test association between two Determines if frequencies observed align
Independence categorical variables with frequencies expected by chance
Example:

A retailer wants to understand if there's any significant difference in customer

satisfaction between three brands of a product they sell. They collect ordinal data
based on a survey where customers rank their satisfaction as "Poor," "Average," or
"Excellent."
Non-Parametric Test Application:
To compare the satisfaction rankings across the three brands, they could use the
Kruskal-Wallis H Test.
Relevance and applicability of One Tailed & Two Tailed Tests
• One-tailed and two-tailed tests are both statistical tests that Relevance and applicability of one-tailed and two-tailed tests
can be used to draw inferences in business research. in drawing inferences in business research:
However, they differ in terms of their assumptions and how
• One-tailed tests are more powerful than two-tailed tests,
they are used to interpret data.
but they are also more likely to produce false positives. This
• One-tailed tests are used when the researcher has a specific means that a one-tailed test is more likely to reject the null
hypothesis about the direction of the effect. For example, a hypothesis even when it is true.
researcher might hypothesize that a new marketing
• Two-tailed tests are less powerful than one-tailed tests, but
campaign will increase sales. In this case, the researcher
they are also less likely to produce false positives. This
would use a one-tailed test to test the hypothesis that the
means that a two-tailed test is less likely to reject the null
new marketing campaign has a positive effect on sales.
hypothesis even when it is false.
• Two-tailed tests are used when the researcher does not
• The choice of whether to use a one-tailed or two-tailed test
have a specific hypothesis about the direction of the effect.
depends on the research question and the researcher's
For example, a researcher might hypothesize that a new
assumptions about the data. In general, it is more
product will affect sales, but they are not sure whether the
conservative to use a two-tailed test, as this reduces the risk
effect will be positive or negative. In this case, the
of false positives. However, if the researcher has a strong
researcher would use a two-tailed test to test the hypothesis
theoretical basis for believing that the effect will be in a
that the new product has a significant effect on sales,
specific direction, then a one-tailed test may be more
regardless of the direction of the effect.
appropriate.
 Examples of how one-tailed and two-tailed tests
can be used in business research:
Scenario Hypothesis Type of Test Direction of Effect
Test effectiveness of new New marketing campaign One-tailed test Positive effect
marketing campaign increases sales

Test effect of new New product has a Two-tailed test Either positive or
product on sales significant effect on sales negative effect

Test effectiveness of new New training program One-tailed test Negative effect
training program reduces employee
turnover
Choice between Parametric and Non-Parametric Tests is
influenced by the nature of the data in Business Research.

The choice is influenced by the nature of the data.

Parametric tests are suitable for normally distributed data and

require interval or ratio data.

Non-parametric tests are used when data is ordinal or nominal

and when assumptions of parametric tests are not met.
Data Analysis

Read notes related to

Practical class conducted
in class on statistical
analysis using data file in
SPSS 25 software