METHODS OF CORRELATION,

COMPLEX DESIGNS AND

DESCRIPTIVE RESEARCH
Research methods and statistics

By Psycho Didi
 METHODS OF CORRELATION
1. Pearson's Correlation Coefficient (r) is used to measure the strength of the correlation between two quantitative variables. It ranges from
-1.00 (the strongest possible negative relationship) to +1.00 (the strongest possible positive relationship). A value of 0 indicates no
relationship.
Small correlation: r ≈ ±.10
Medium correlation: r ≈ ±.30
Large correlation: r ≈ ±.50
As Pearson's r moves closer to -1.00 or +1.00, the points on a scatterplot align more closely along a straight line. When r is 0, the points on the
scatterplot appear as a shapeless "cloud."

2. Spearman's rank-order correlation (rs) was developed by Charles Spearman (1904) to compute correlation when the data is presented in
rank order for two variables across n subjects. It is also used for inter-judge agreement when two judges evaluate the data for the same n
subjects.
Applicability:
Suitable for rank-order data.
Can be used with continuous data if Pearson's assumptions are not met.
Used to assess a monotonic relationship (i.e., a relationship that consistently increases or decreases but not necessarily at a constant
rate).
Spearman's rho is helpful in situations where the data is ordinal or where Pearson's correlation is inappropriate.
 METHODS OF CORRELATION
3. Kendall's tau, developed by Kendall (1938), is a correlation
measure based on two sets of ranks, X and Y. It ranges from -1.00
to +1.00, with values closer to ±1.00 indicating a stronger
relationship. Positive values reflect a positive relationship, while
negative values indicate an inverse one.
4. Point-Biserial Correlation (rpb) is used to assess the
relationship between a dichotomous variable and a continuous
variable. It is similar to Pearson’s correlation and is applied when
one variable is truly dichotomous, such as male-female or
experimental-control group.
5. Biserial Correlation measures the relationship between one
continuous and one dichotomous variable, assuming the
dichotomy is based on an underlying continuous distribution. It’s
useful when the dichotomous variable represents a cutoff or
threshold from a continuous scale.
 METHODS OF CORRELATION
7. The Phi Coefficient (Ф) is used to measure the
correlation between two dichotomous variables. For
example, if gender (male/female) and property ownership
(owns/does not own) are the two variables, the Phi
Coefficient calculates the Pearson’s correlation between
them, with each variable taking values of 0 or

6. Tetrachoric Correlation (rtet) measures the correlation

between two dichotomous variables that have an
underlying continuous distribution. For example, if
attitudes toward women and liberalization are measured
as positive or negative, tetrachoric correlation can be used
to assess the relationship, assuming the dichotomy
reflects an underlying continuous scale.
 COMPLETELY RANDOMIZED DESIGN (C.R. DESIGN)
A Completely Randomized (CR) Design is the simplest experimental design
based on randomization and replication. Subjects are randomly assigned to
different treatment groups, ensuring each subject has an equal chance of
receiving any treatment. It is analyzed using One-Way ANOVA and works best
in uniform environments. However, it does not control for external variables
like individual differences among experimenters.
TYPES OF CR DESIGN:
1. Two-Group Simple Randomized Design – A randomly selected sample is
divided into:
Experimental Group (receives treatment)
Control Group (does not receive treatment)
This method helps compare treatment effects but does not control
for external factors.
2. Random Replications Design – Improves accuracy by repeating
treatments across multiple groups and experimenters. This controls for
external variables and reduces random errors, making results more
reliable.
 RANDOMIZED BLOCK DESIGN (R.B. DESIGN)
The Randomized Block Design (R.B. Design) improves upon the
Completely Randomized Design (C.R. Design) by adding local
control to reduce variability. Subjects are divided into blocks based
on a relevant characteristic (e.g., IQ level) to ensure similarity within
each block. Each block contains one subject per treatment,
assigned randomly.
KEY FEATURES:
Controls for extraneous variables by grouping similar subjects.
Ensures each treatment appears equally across all blocks.
Analyzed using Two-Way ANOVA, which accounts for both
treatments and blocking factors.
EXAMPLE:
If five students (each from a different IQ level) take four versions of
a statistics test in a randomized order, this controls for factors like
fatigue or learning effects, improving the experiment’s reliability.
 LATIN SQUARE DESIGN
The Latin Square Design controls two extraneous factors, like soil
fertility and seed variation, ensuring they don’t affect treatment
results.
Key Features:
Each treatment appears once per row and column in a square
grid.
Balances both blocking factors and is analyzed using Two-Way
ANOVA.
Advantages:
✔ Controls two variables, improving accuracy.
✔ Eliminates fertility gradients for reliable results.
Limitations:
✖ Requires equal rows, columns, and treatments, reducing flexibility.
✖ Assumes no interaction between treatments and blocks.
Commonly used in 5×5 to 9×9 grids.
 FACTORIAL DESIGNS
Factorial designs are used to study the effects of multiple factors in an experiment. They are especially useful in economic
and social research where many variables influence an outcome. There are two types:
1. Simple Factorial Designs – These examine the effects of two factors on a dependent variable. Also called a "two-
factor factorial design," they can take forms like 2×2, 3×4, or 5×3. A 2×2 design includes two treatment levels and two
control levels, forming four experimental conditions. This design helps assess the main effects of treatments and
levels, as well as their interaction. The same design can compare two experimental variables or two control variables.
2. Complex Factorial Designs – Used when three or more factors are studied simultaneously, such as a 2×2×2 design
with one experimental variable (two treatments) and two control variables (each with two levels). These designs allow
researchers to analyze main effects, first-order interactions (between pairs of variables), and second-order interactions
(involving all three variables).
Advantages:
Efficiently studies multiple factors in one experiment, reducing labor.
Identifies interaction effects that single-factor studies cannot.
Provides deeper insights into variable relationships.
While factorial designs offer rich data, adding more variables increases complexity.
 SINGLE-SUBJECT DESIGN
Focuses on studying individual behavior over time, useful when group
samples are limited.
TYPES:
1. Reversal (ABA) Design: Behavior is observed (A), treatment
applied (B), then removed to check if it returns to baseline (A).
2. Multiple Baseline Design: Intervention is introduced gradually
across behaviors or conditions to assess treatment effects.
3. Changing Criterion Design: Performance expectations increase in
steps to confirm changes result from treatment.
4. Alternating Treatments Design: Rapid shifts between baseline and
treatments for quick comparison.
Analysis:
Visual analysis of graphs
Statistical methods support consistency across phases
 TIME SERIES DESIGN
This method tracks a psychological trait or behavior over time to
identify patterns and changes. It includes two main types:
Interrupted Time Series Design involves measuring behavior
before and after an event or intervention. For example, studying
depression levels before and after counseling.
Multiple Time Series Design compares treatment and control
groups over time to assess the effects. For instance, comparing
anxiety levels in different patient groups before and after therapy.

EVALUATION METHODS
To assess the impact of interventions, data can be visualized using
graphs to track trends over time. Statistical methods, such as
autocorrelation adjustments, correct for patterns in closely timed data.
Regression discontinuity analysis is used to examine trends before and
after an event to identify significant changes. These methods help
measure the overall impact, account for gradual changes or delays, and
adjust for external factors.
 COHORT STUDIES
Cohort studies follow a group of people with a shared characteristic over time to
observe how different factors influence outcomes. They help identify links
between exposures and later effects.
There are two types:
Prospective Cohorts: Start with a group based on exposure and track them
over time.
Retrospective Cohorts: Begin with an outcome and look back to analyze
past exposures.
These studies require a well-defined exposure and outcome, a large sample size,
and a long follow-up period. Statistical adjustments help control confounding
variables.
In psychology, cohort studies examine how childhood experiences influence
mental health, how stress affects cognition, and how therapy impacts long-term
well-being. While they provide strong evidence of cause and effect, challenges
include participant dropout, cohort effects, and ethical concerns in long-term
studies.
 QUALITATIVE RESEARCH
Qualitative research explores the "why" and "how" of human behavior, focusing on perspectives, emotions, and motivations rather
than numerical data. It aims for a deep, contextual understanding of experiences.
CHARACTERISTICS
Naturalistic settings: Data is collected in real-world environments to capture authentic behaviors and interactions.
Inductive approach: Begins with data collection, allowing theories and themes to emerge naturally.
Holistic perspective: Considers social, cultural, and psychological factors shaping experiences.
Subjectivity & interpretation: Researchers reflect on biases and how they influence findings.
Flexibility: Methods adapt as new insights emerge, refining research focus dynamically.

PRINCIPLES OF QUALITATIVE RESEARCH

Empathy & Reflexivity: Researchers strive to understand participants' perspectives while critically reflecting on their own
biases and influence on the research.
Trustworthiness: Ensured through credibility, transferability, dependability, and confirmability, using strategies like
triangulation and member checking.
Iterative Analysis: Data is continuously reviewed and refined, allowing deeper insights and evolving interpretations.
Rich Description: Detailed, context-sensitive accounts enhance understanding and relevance of findings.
 TYPES
Narrative Research: Focuses on individual stories and experiences to understand personal perspectives. Data is collected through
interviews, letters, or diaries, and analyzed for recurring themes or meanings.
Phenomenology: Aims to understand the essence of a specific phenomenon by analyzing participants' experiences. Through in-
depth interviews or written materials, researchers explore commonalities and variations in how individuals perceive the
phenomenon.
Grounded Theory: An inductive method that generates theories from systematically collected data. Researchers use interviews,
observations, and document analysis, applying coding and constant comparison to identify patterns, leading to the development of
a theory.
Ethnography: Involves immersion in a group’s daily activities to observe and participate. Data is gathered through field notes,
interviews, and document analysis to provide a holistic understanding of cultural practices and social dynamics.
Case Study: Examines a specific instance, event, or individual in its real-life context. Researchers gather data from multiple
sources, such as interviews, observations, and documents, to create a detailed understanding of the case. This approach is used to
explore complex phenomena or evaluate interventions.
Focus Groups: Involves guided discussions with a small group of participants to explore shared experiences, opinions, or attitudes
about a topic. Researchers analyze group interactions, discussions, and emerging themes to understand collective perspectives.
Delphi Method: A structured approach where experts provide input on a topic through multiple rounds of questionnaires.
Responses are analyzed and refined in each round to reach a consensus, making it useful for forecasting and decision-making.
 TYPES
Participatory Action Research (PAR) –
What is it? A collaborative research approach that involves participants in
identifying issues, analyzing data, and implementing solutions to create social
change.
Who developed it? Developed by Kurt Lewin in the 1940s, later expanded by
Paulo Freire and others focusing on empowerment and social justice.
4 Steps of PAR:
a. Planning – Identifying problems and co-developing research strategies.
b. Action – Implementing interventions or changes based on findings.
c. Observation – Collecting data and assessing the impact of actions.
d. Reflection – Analyzing results and refining the process for continuous
improvement.
Why is it used?
Empowers communities to address their own issues.
Promotes collaboration between researchers and participants.
Leads to practical, real-world solutions.
Encourages social justice, advocacy, and policy change.
Used in education, healthcare, community development, and social work.
 PURPOSE OF QUALITATIVE RESEARCH
Qualitative research aims to generate theories, refine existing ones, and provide actionable insights. It
enhances understanding of human experiences, social dynamics, and cultural contexts while
complementing quantitative research.
Generating new theory: Develops novel frameworks by identifying patterns and relationships in rich
data.
Refining existing theory: Expands or challenges theories by revealing complexities and context-
specific nuances.
Providing practical insights: Informs policy, practice, and interventions by exploring lived
experiences.
Enhancing empathy and understanding: Fosters cultural sensitivity and inclusivity through diverse
perspectives.
Supporting mixed-methods research: Complements quantitative research by generating hypotheses
and providing contextual explanations.
 DESCRIPTIVE RESEARCH
Descriptive research is an exploratory method that describes a population, circumstance, or phenomenon. It
answers "what," "where," "when," and "how" questions but not "why" questions. It does not manipulate
variables or establish cause-and-effect relationships. Descriptive research is useful when the aim is to
understand characteristics, frequencies, trends, and categories without manipulating variables. It often
precedes more complex research and is helpful when limited knowledge exists on a topic.
Examples include:
What changes have occurred in urban gardening in Mumbai over the last two decades?
What are the differences in climate change perceptions between coastal and inland farmers in the
Philippines?
CHARACTERISTICS OF DESCRIPTIVE RESEARCH
Quantitative or Qualitative: Can involve both approaches for data gathering and analysis.
Observational: Non-invasive; researchers observe and report without impacting the subject.
Cross-sectional: Provides a snapshot of different sections of the same group.
Springboard for further research: Data guides future research methods.
1. Which of the following is NOT a type of correlation typically used in statistical
analysis?

a) Pearson’s Correlation
b) Spearman’s Rank-Order Correlation
c) Kendall’s Tau Correlation
d) Biserial Linear Correlation
2.:Which correlation method is best suited for assessing the relationship
between two continuous variables with assumptions of normality and
linearity?

a) Point Biserial Correlation

b) Kendall’s Tau
c) Pearson’s Correlation
d) Tetrachoric Correlation
3. Which of the following statements best describes the purpose of Time
Series Design in research?
a) It focuses on comparing different independent groups over time.
b) It analyzes the correlation between two variables measured over time in the
same group.
c) It involves measuring variables at only one point in time for comparison.
d) It assesses the long-term effects of an intervention on a single group across
different periods.
4. Which of the following is the main feature of the Latin Square Design?

a) Each treatment appears only once across both rows and columns.
b) It requires only one factor to be studied, unlike factorial designs.
c) Each participant is exposed to all treatments in random order.
d) It involves two independent variables interacting to determine the outcome.
5. Which of the following is a core principle of Participatory Action Research
(PAR)?

a) The researcher remains completely neutral, without influencing the research

process.
b) Data is collected solely by the researcher to ensure unbiased results.
c) It involves collaboration between researchers and participants throughout
the research process.
d) The participants only provide feedback at the conclusion of the study.