Research design

• Research design is a framework or plan for a study that is used to collect

and analyze data in a way that aims to provide answers to research
questions.

• The design helps in specifying the procedures for collecting information,

measures the variables of interest, defines the relationships among the
variables, and outlines the methods for analyzing the data.
Types of Research Design and When to Use Them

1.Descriptive Research Design

1.Purpose: To describe characteristics of an existing phenomenon.
2.When to Use: When you want to provide a detailed account of a
phenomenon.
3.Example: Studying the daily routines of high school students.

2.Correlational Research Design

1.Purpose: To identify relationships between two or more variables.
2.When to Use: When the goal is to find a connection or pattern but not
cause and effect.
3.Example: Analyzing the relationship between sleep quality and academic
performance.
1.Experimental Research Design

1.Purpose: To determine causality between variables.

2.When to Use: When the researcher wants to control and manipulate the
variables to ascertain cause-and-effect relationships.
3.Example: Experimenting to see if a new teaching method improves student
reading comprehension.

1.Quasi-Experimental Research Design

1.Purpose: Similar to experimental design, but without random assignment.
2.When to Use: When random assignment is not feasible, but the
investigation of cause-and-effect is still required.
3.Example: Studying the effect of after-school programs on student
achievement in schools where random assignment is not possible
1.Exploratory Research Design
1.Purpose: To explore a problem that is not well understood.
2.When to Use: When studying a new or poorly understood phenomenon.
3.Example: Investigating consumer attitudes towards a newly launched
product.

2.Explanatory Research Design

1.Purpose: To explain why and how a phenomenon occurs.
2.When to Use: When you want to provide a detailed explanation of a
phenomenon's occurrence.
3.Example: Understanding why certain marketing strategies work in a
particular industry.
1.Cross-Sectional Design
1.Purpose: To observe a phenomenon at a single point in time.
2.When to Use: When you want to capture a snapshot of a situation or
problem.
3.Example: Surveying the political preferences of citizens before an election.

2.Longitudinal Design
1.Purpose: To study changes over time.
2.When to Use: When the goal is to observe how variables change over an
extended period.
3.Example: Following a cohort of employees through their careers to study
the effects of mentorship.
Conclusion
• The choice of research design depends on the research question, the nature
of the data, the audience for whom the research is intended, and the
available resources.

• By carefully selecting the appropriate design, researchers can achieve the

clarity and control needed to provide meaningful answers to their research
questions.

• Different designs suit different types of inquiries, and the examples above
illustrate how the designs can be applied in various contexts.
Scales of measurement

It refer to the various ways that variables or attributes can be quantified or categorized.
They help in understanding the type of data being dealt with and, in turn, determine the
appropriate statistical techniques that can be applied. The four primary scales of
measurement are:

1. Nominal Scale
•Definition: The nominal scale categorizes data into distinct groups or categories without
any order or hierarchy.
•Example: Gender (Male, Female), Blood Type (A, B, AB, O), or Types of Cuisine (Italian,
Chinese, Mexican).
•Analysis: You can use measures like mode or percentage with nominal data, but mean and
median are meaningless here.
2. Ordinal Scale

•Definition: The ordinal scale categorizes data into distinct groups, and there is a
clear order or ranking among the categories, but the intervals between them are
not uniform or known.
•Example: Customer satisfaction ratings (Very Unsatisfied, Unsatisfied, Neutral,
Satisfied, Very Satisfied) or Educational Level (High School, Associate’s Degree,
Bachelor’s Degree, Master’s Degree).
•Analysis: You can use median or mode, but mean is usually not applicable since
the distances between the categories are not known.
3. Interval Scale

•Definition: The interval scale not only categorizes and orders the
measurements, but it also specifies the exact distances between them.
However, it does not have a true zero point (i.e., the zero does not mean the
absence of the attribute).

•Example: Temperature measured in degrees Celsius (0°C does not mean an

absence of temperature) or IQ scores.

•Analysis: You can use mean, median, and mode, as well as more complex
statistical techniques like standard deviation.
4. Ratio Scale

•Definition: The ratio scale has all the characteristics of the interval scale, but
with a true zero point. A value of zero on the ratio scale means the absence of
the attribute being measured.

•Example: Age (0 years means no age), Weight (0 kilograms means no weight),

or Distance (0 miles means no distance).

•Analysis: You can use all statistical measures, including mean, median, mode,
standard deviation, and so on.
Why Scales of Measurement are Important

Understanding the scale of measurement for a particular variable is crucial

because it dictates what statistical methods are appropriate.

For example, it would not make sense to calculate the mean for a nominal
variable, or to rank the categories in a ratio variable.
Non-Metric Scales

1.Nominal Scale: For example, categorizing people by their

hair color (blonde, brunette, black, etc.). You cannot say that
one category is "more" or "less" than another, nor can you
perform arithmetic operations with these categories.

1.Ordinal Scale: For example, ranking movies as "poor,"

"average," "good," or "excellent." Though there is a clear
order, you can't quantify the exact difference between these
rankings.
Metric Scales

1.Interval Scale: For example, measuring temperature in

degrees Celsius. You can say that 30°C is 10 degrees more than
20°C, and such mathematical comparisons make sense.

2.Ratio Scale: For example, measuring weight in kilograms.

This scale has a true zero point, and you can make meaningful
statements such as "10 kilograms is twice as heavy as 5
kilograms."
Research Objectives with Metric Scales:

1.Evaluating Sales Performance

1.Objective: To assess the monthly sales performance across different
regions.
2.Metric Scale: Sales figures in dollars, units sold, or percentage growth.

2.Analyzing Customer Lifetime Value

1.Objective: To understand the lifetime value of customers in different
segments.
2.Metric Scale: Monetary value spent by customers over time.
3.Assessing Production Efficiency
1.Objective: To evaluate the efficiency of production processes.
2.Metric Scale: Ratio of output to input, time taken to manufacture a unit,
etc.

4.Measuring Advertising Return on Investment (ROI)

1.Objective: To measure the ROI of various advertising campaigns.
2.Metric Scale: Cost-per-click, conversion rates, revenue generated per
advertising dollar spent.
Research Objectives with Non-Metric Scales:
1.Understanding Brand Perception
1.Objective: To understand how consumers perceive a brand.
2.Non-Metric Scale: Qualitative feedback categorized into positive, neutral,
negative.

2.Identifying Market Segments

1.Objective: To identify different segments within the market.
2.Non-Metric Scale: Categories such as Millennials, Baby Boomers, or
segmentation by lifestyle like health-conscious, budget-oriented.
3.Evaluating Employee Satisfaction
1.Objective: To evaluate the level of employee satisfaction within the
organization.
2.Non-Metric Scale: Responses categorized into very satisfied, satisfied,
neutral, dissatisfied, very dissatisfied.

4.Assessing Customer Preferences for New Product Features

1.Objective: To gauge customer preferences for potential new features in a
product.
2.Non-Metric Scale: Rankings or preference order for various features without
a quantified difference between them.
A five-point scale

• A five-point scale can be either ordinal or interval, depending on how it's

structured and the context in which it's used. Here's how you can
differentiate between the two:

Ordinal Scale

• An ordinal scale represents the rank order of values without providing a

consistent degree of difference between those values. It allows you to see
how items rank relative to one another, but it does not provide information
about the absolute or relative magnitudes of differences between the ranks.
Ordinal Scale Example:
Imagine you are measuring customer satisfaction for a product using the
following five-point scale:
1.Very Dissatisfied
2.Dissatisfied
3.Neutral
4.Satisfied
5.Very Satisfied

• In this ordinal representation, the categories tell you the order of satisfaction
but not the exact or consistent differences between levels.
• You can't quantitatively say how much more satisfied a "Very Satisfied"
customer is compared to a "Satisfied" customer. The distances between these
consecutive levels are not necessarily equal, so this scale is ordinal.
Interval Scale Example:
Now, let's represent customer satisfaction in an interval scale using the same
five-point scale but with defined and equal intervals:
1.0-20% Satisfied
2.21-40% Satisfied
3.41-60% Satisfied
4.61-80% Satisfied
5.81-100% Satisfied

• In this interval representation, the five points on the scale correspond to

equal intervals of satisfaction, with each point representing a 20% increase in
satisfaction.
• The distances between consecutive levels are consistent and quantifiable.
Since the scale has equal intervals between the consecutive points, it is
considered an interval scale.
Example 1: Employee Performance Rating
Ordinal Scale:
1.Poor
2.Below Average
3.Average
4.Above Average
5.Excellent
Here, the performance ratings are ranked, but the differences between the levels
are not defined. The difference between "Poor" and "Below Average" may not
be the same as between "Above Average" and "Excellent."
Interval Scale:
1.0-20% of Targets Met
2.21-40% of Targets Met
3.41-60% of Targets Met
4.61-80% of Targets Met
5.81-100% of Targets Met

With this scale, you have defined equal intervals, making it possible to
understand the exact differences in performance between consecutive points.
Example 2: Hotel Ratings
Ordinal Scale:
1.Very Poor
2.Poor
3.Average
4.Good
5.Excellent

This rating gives an order of quality but does not specify how much better one
rating is from the next.
Interval Scale:
1.0-20 Points (Out of 100)
2.21-40 Points (Out of 100)
3.41-60 Points (Out of 100)
4.61-80 Points (Out of 100)
5.81-100 Points (Out of 100)

Here, each category represents a specific score range with equal intervals,
allowing for a more precise comparison between hotels.
Example 3: Frequency of Exercise
Ordinal Scale:
1.Never
2.Rarely
3.Sometimes
4.Often
5.Always

• This scale offers a general understanding of how frequently someone

exercises but doesn't detail the exact differences between the categories.
Interval Scale:
1.0-1 Days Per Week
2.2-3 Days Per Week
3.4-5 Days Per Week
4.6-7 Days Per Week
5.7 Days Per Week

• In this interval scale, the days of exercise are broken down into equal intervals,
providing an exact measure of how frequently someone exercises.
Movie Ratings:
1.Hated It
2.Disliked It
3.Neutral
4.Liked It
5.Loved It

• This scale tells you the order of preference but not how much more
someone liked or loved the movie.

• The intervals between the points are not necessarily equal.

Pain Level:
1.No Pain
2.Mild Pain
3.Moderate Pain
4.Severe Pain
5.Excruciating Pain

• Though it's clear that "Excruciating Pain" is more painful than "Mild Pain,"
the difference between the two consecutive levels may not be the same
across the scale.
Educational Grades (assuming equal intervals between grades):
60-69% (D)
70-79% (C)
80-89% (B)
90-99% (A)
100% (A+)

Here, each category represents an interval of 10 percentage points, with

consistent differences between consecutive points.
Customer Satisfaction (assuming defined and equal intervals):
1.0-20% Satisfied
2.21-40% Satisfied
3.41-60% Satisfied
4.61-80% Satisfied
5.81-100% Satisfied

• If you define the satisfaction levels in terms of exact percentages with equal
intervals, then it becomes an interval scale.
Calculation
Percentages:
You can also calculate the percentage of respondents for each category to
understand the distribution of opinions.

•Strongly Disagree: 5%
•Disagree: 15%
•Neutral: 20%
•Agree: 40%
•Strongly Agree: 20%
Example:
Imagine you have asked 100 respondents to rate their agreement with the
statement, "I enjoy reading books," using the five-point Likert scale
mentioned above. You might receive responses like:
1.Strongly Disagree: 5 respondents
2.Disagree: 15 respondents
3.Neutral: 20 respondents
4.Agree: 40 respondents
5.Strongly Agree: 20 respondents
Calculating Mean:
You can calculate the mean to get an average level of agreement.
1.Multiply the number of respondents for each category by the corresponding
value on the scale:
1.1×5+2×15+3×20+4×40+5×20=5+30+60+160+100=3551×5+2×15+3×20+4×40
+5×20=5+30+60+160+100=355

2.Divide the total by the number of respondents:

1.355/100=3.55355/100=3.55
The mean score of 3.55 would indicate a tendency toward the "Agree" response.
Example 1: Satisfaction with a New Product
You surveyed 50 people to gauge their satisfaction with a new product, using the following
five-point Likert scale:
1.Very Dissatisfied
2.Dissatisfied
3.Neutral
4.Satisfied
5.Very Satisfied
The responses were:
•Very Dissatisfied: 2 respondents
•Dissatisfied: 8 respondents
•Neutral: 10 respondents
•Satisfied: 20 respondents
•Very Satisfied: 10 respondents

Mean:
(1×2+2×8+3×10+4×20+5×10)/50=3.6(1×2+2×8+3×10+4×20+5×10)/50=3.6
1. To assess customer satisfaction levels with products and services.

2. To analyze the performance of the sales team across different regions.

3. To understand the relationship between employee engagement and

productivity.

4. To explore how social media engagement influences sales.

5. To investigate the relationship between brand loyalty and purchase behavior

6. To understand how training programs affect sales performance

7. To gauge consumer preferences for potential new features in a product.

Impact ,Effect – Regression (bivariate)
Relation- Correlation
3. productivity.- (Hours worked, tasks completed, revenue generated
per employee)
employee engagement-(Employee engagement levels categorized as
highly engaged, engaged, or disengaged)

4. social media engagement - (Social media metrics like followers,

engagement rate, click-through rate)

5. purchase behavior-Frequency of purchases, average spending per

customer
brand loyalty-Brand loyalty levels such as high, medium, low.

6. Sales - Pre- and post-training sales figures, growth in sales.

Training - Training satisfaction levels categorized into very satisfied,
satisfied, unsatisfied.
Impact ,Effect – Regression (bivariate)
Relation- Correlation
Type I and Type II errors

• Type I and Type II errors are concepts related to hypothesis testing in

statistics. When you're conducting a hypothesis test,

• you start with a null hypothesis (usually denoted as H0 ) that represents a

statement of no effect or no difference, and an alternative hypothesis
(usually denoted as H1 or Hα ) that represents what you want to prove.
Type I Error
• A Type I error occurs when you incorrectly reject a true null hypothesis. In
other words, you conclude that there is an effect or difference when there
actually isn't one. This is often referred to as a "false positive.“
Example of Type I Error:
Suppose you are a quality control manager in a factory, and you are testing
whether a batch of products is defective (Alternative Hypothesis) or not (Null
Hypothesis).
•Null Hypothesis (H0): The batch is not defective.
•Alternative Hypothesis (Hα ): The batch is defective.

• If you reject the null hypothesis and claim the batch is defective when it's
actually not, you have made a Type I error.
Type II Error
A Type II error occurs when you fail to reject a false null hypothesis. In other
words, you conclude that there is no effect or difference when there actually is
one. This is often referred to as a "false negative."
Example of Type II Error:
Using the same factory example, suppose the batch is indeed defective, but
your test fails to detect this.
•Null Hypothesis (H0): The batch is not defective.
•Alternative Hypothesis (Hα ): The batch is defective.

• If you fail to reject the null hypothesis and claim the batch is not defective
when it actually is, you have made a Type II error.
Summary:
•Type I Error (False Positive): Rejecting a true null hypothesis.
• Example: Declaring a batch defective when it's not.
•Type II Error (False Negative): Failing to reject a false null hypothesis.
• Example: Failing to detect that a batch is defective when it is.

Example 1: Medical Testing

Null Hypothesis (H0 ): A patient does not have a specific disease. Alternative
Hypothesis (Hα ): A patient has a specific disease.
•Type I Error: The test indicates the patient has the disease when they actually
do not (False Positive).
•Type II Error: The test indicates the patient does not have the disease when
they actually do (False Negative).
Example 2: Criminal Trial
Null Hypothesis (H0 ): The defendant is innocent.
Alternative Hypothesis (Hα ): The defendant is guilty.
•Type I Error: Convicting an innocent person (False Positive).
•Type II Error: Acquitting a guilty person (False Negative).

These examples highlight the consequences and considerations in different

contexts. Type I and Type II errors represent different kinds of mistakes, and the
severity and implications of these mistakes can vary widely depending on the
situation.

In some cases, a Type I error might be more costly, while in others, a Type II error
might be more concerning. Understanding these errors helps guide decisions
about the level of evidence needed and the balance of risks in a particular testing
or decision-making scenario.