Okay, let's break down factor analysis, covering its assumptions, methods, rotation, and
interpretation.
What is Factor Analysis?
Factor analysis is a statistical method used to reduce a large number of variables into fewer,
uncorrelated variables, also known as factors (Factor Analysis, 2025). These factors capture the
common variance among the original variables (A Practical Introduction to Factor Analysis:
Exploratory Factor Analysis, 2025). It's primarily used for data reduction and simplification, aiming to
identify underlying relationships between measured variables (A Practical Introduction to Factor
Analysis: Exploratory Factor Analysis, 2025).
Assumptions of Factor Analysis
While factor analysis is a powerful tool, it relies on several key assumptions:
1. Linearity: Assumes that the relationships between variables are linear.
2. Multivariate Normality: Variables should be normally distributed.
3. Adequate Sample Size: A sufficiently large sample size is needed for reliable results. There is
no strict rule, but some suggest at least 10 participants per variable.
4. Correlation Between Variables: There should be significant correlations between the
variables to justify factor analysis. If correlations are low, it may not be suitable.
5. Data Level: Factor analysis is usually applied on interval or ratio scale data.
Methods of Factor Analysis
There are different approaches:
Exploratory Factor Analysis: Used when there is no prior hypothesis about the structure of
the data. The goal is to explore the data and identify the number of factors and the variables
that load onto each factor (A Practical Introduction to Factor Analysis: Exploratory Factor
Analysis, 2025).
Confirmatory Factor Analysis: Used to test a specific hypothesis about the factor structure.
The researcher specifies the number of factors and which variables load onto each factor,
and then the analysis tests how well the data fit this model.
Factor Extraction Methods
Principal Component Analysis: Aims to explain the total variance in the data (A Practical
Introduction to Factor Analysis: Exploratory Factor Analysis, 2025). It assumes that there is
no unique variance, meaning all variance is common variance (A Practical Introduction to
Factor Analysis: Exploratory Factor Analysis, 2025).
Common Factor Analysis: Aims to explain the common variance among the variables (A
Practical Introduction to Factor Analysis: Exploratory Factor Analysis, 2025). It acknowledges
the presence of unique variance. Common factor analysis is considered more appropriate
when the goal is to identify underlying latent factors.
Rotation
Rotation is a crucial step in factor analysis that simplifies the factors and makes them more
interpretable (Darton, 1980). Rotation methods maximize the loading of each variable on one of the
factors while minimizing its loading on the other factors (Darton, 1980).
Orthogonal Rotation: Assumes that the factors are uncorrelated. The most common
orthogonal rotation method is Varimax (Forina et al., 2005).
Oblique Rotation: Allows the factors to be correlated. Common oblique rotation methods
include Direct Oblimin and Promax.
Interpretation
Interpreting factors involves examining the factor loadings, which indicate the strength and direction
of the relationship between each variable and each factor. Variables with high loadings on a factor
are considered to be representative of that factor. The researcher then names the factor based on
the variables that load highly on it (2013).
PCA and EFA/CFA (Exploratory/Confirmatory Factor Analysis) are all dimensionality reduction
techniques, but they have different goals and assumptions (Castro-Schilo, 2021; Principal
Components Analysis or Exploratory Factor Analysis, 2023).
Principal Component Analysis
Goal: PCA aims to explain the total variance in the data (Principal Component Analysis vs.
Exploratory Factor Analysis, 2023). It transforms the original variables into a new set of
uncorrelated variables (principal components) that are linear combinations of the original
variables (Xu et al., 2020). The first principal component accounts for the largest amount of
variance, the second accounts for the second largest, and so on (Principal Component
Analysis vs. Exploratory Factor Analysis, 2023).
Focus: PCA concentrates on explaining the diagonal elements of the covariance matrix
(Principal Components Analysis or Exploratory Factor Analysis, 2023).
Assumption: PCA assumes that all variance is common variance, meaning there's no unique
variance (A Practical Introduction to Factor Analysis: Exploratory Factor Analysis, 2025).
Exploratory Factor Analysis
Goal: EFA aims to explain the common variance among the variables in the data (A Practical
Introduction to Factor Analysis: Exploratory Factor Analysis, 2025). It identifies underlying
latent factors that account for the correlations among the observed variables.
Focus: EFA focuses on explaining the off-diagonal elements of the covariance matrix
(Principal Components Analysis or Exploratory Factor Analysis, 2023).
Assumption: EFA acknowledges the presence of unique variance, meaning each variable has
a part that is not explained by the common factors (A Practical Introduction to Factor
Analysis: Exploratory Factor Analysis, 2025).
Confirmatory Factor Analysis
Goal: CFA is used to test a specific hypothesis about the factor structure (Wolf et al., 2013).
The researcher specifies the number of factors and which variables load onto each factor,
and then the analysis tests how well the data fit this model.
Relationship to EFA: CFA is often used after EFA to confirm the factor structure that was
discovered in the exploratory phase (Aichholzer, 2014).
Key Differences Summarized
Feature PCA EFA CFA
Test a hypothesized factor
Goal Explain total variance Explain common variance
structure
Assumes all variance Acknowledges unique Tests a specific model of common
Variance
is common variance and unique variance
Use Data reduction Identifying latent factors Confirming factor structure
Exploratory, used when Theory-driven, used when
Prior No prior assumptions
relationships are not well- relationships are hypothesized in
Knowledge about structure
defined advance
In short, PCA is a data reduction technique, while EFA and CFA are used to explore and confirm the
underlying factor structure of a set of variables (Principal Components Analysis or Exploratory Factor
Analysis, 2023).
Okay, let's break down Exploratory Factor Analysis with an example:
What Exploratory Factor Analysis Measures
EFA is used to discover the underlying structure of a set of variables (Exploratory Factor Analysis,
2020). Specifically, it:
Identifies Latent Factors: It helps to find a smaller set of underlying, unobserved variables
(factors) that explain the relationships among a larger set of observed variables (2013).
Reduces Data Complexity: It simplifies complex data by reducing the number of variables
(Plucker, 2003).
Explores Relationships: It is useful to look for relationships within a dataset (Harwani &
Sihite, 2019).
Construct Validation: EFA can be used to evaluate the validity of constructs.
Data Requirements for EFA
1. Sample Size: A sufficiently large sample size is needed for reliable results. There is no strict
rule, but some suggest at least 10 participants per variable.
2. Correlation Between Variables: There should be significant correlations between the
variables to justify factor analysis (Hu et al., 2015).
3. Level of Measurement: EFA is typically applied to data measured on an interval or ratio
scale.
Example from Psychology: Measuring Attitudes Towards Exercise
Let's say you want to understand the underlying attitudes people have toward exercise. You create a
survey with several questions (variables) related to exercise beliefs and behaviors.
Variables:
1. I enjoy exercising.
2. Exercise is important for my health.
3. Exercise makes me feel good.
4. I find exercise boring.
5. Exercise is too tiring.
6. I feel guilty when I don't exercise.
7. Exercise is a good way to socialize.
8. Exercise helps me manage stress.
9. I am too busy to exercise.
Dummy Data:
Let's imagine you collect data from 200 participants. The data for each variable would be on a scale
(e.g., 1-5, strongly disagree to strongly agree). Here's a snippet of what the data might look like:
Participant Enjoy Exercise Important for Health Feels Good ... Too Busy
1 4 5 4 ... 2
2 2 3 2 ... 4
3 5 5 5 ... 1
... ... ... ... ... ...
200 3 4 3 ... 3
Running the EFA:
You would enter this data into a statistical software package (like SPSS, R, or others mentioned
previously). You would then run an EFA, specifying the variables to include.
Findings, Results, and Discussion
1. Number of Factors: The EFA might reveal two underlying factors:
o Factor 1: Positive Attitudes Towards Exercise
o Factor 2: Barriers to Exercise
2. Factor Loadings: The analysis would produce a factor loading matrix, showing how strongly
each variable relates to each factor. For example:
Variable Factor 1: Positive Attitudes Factor 2: Barriers
I enjoy exercising. 0.82 -0.10
Important for health 0.75 0.05
Exercise makes me feel good 0.88 -0.02
I find exercise boring -0.08 0.79
Exercise is too tiring -0.01 0.85
I am too busy 0.12 0.68
3. Interpretation:
o Factor 1: Variables like "I enjoy exercising," "Exercise is important for my health," and
"Exercise makes me feel good" load highly on this factor. This suggests that this
factor represents a general positive attitude and belief in the benefits of exercise.
o Factor 2: Variables like "I find exercise boring," "Exercise is too tiring," and "I am too
busy to exercise" load highly on this factor. This suggests that this factor represents
common barriers or obstacles that prevent people from exercising.
Discussion:
The EFA has helped you understand that people's attitudes toward exercise can be broadly grouped
into two categories: positive attitudes and perceived barriers (Harwani & Sihite, 2019). This
information could be used to:
Develop Interventions: Design interventions that target these specific barriers to increase
exercise adherence.
Further Research: Conduct further research to explore the factors that influence these
attitudes.
Tailor Messages: Create more effective health messages that resonate with people's specific
attitudes and concerns about exercise.
This example demonstrates how EFA can be used in psychology to explore the underlying structure of
attitudes, beliefs, and behaviors (Plucker, 2003).
Yes, factor loadings are definitely used in Exploratory Factor Analysis.
Factor loadings represent the correlation between each observed variable and the
underlying factors (A Practical Introduction to Factor Analysis: Exploratory Factor Analysis,
2025). They indicate the strength and direction of the relationship between the variable and
the factor.
Researchers often use a cutoff value to determine which variables are significantly related to
each factor (Howard, 2015). For instance, some suggest not interpreting factor loadings
below 0.30 (Ondé & Alvarado, 2018).
By examining the pattern of factor loadings, you can interpret what each factor represents
(Getting Started with Exploratory Factor Analysis, 2022). Variables with high loadings on a
particular factor are considered indicators of that factor.