MIT-WPU | School of Health Sciences & Technology
(M.Sc. clinical psychology )
WORLD’S FIRST UNIVERSITY
FOR LIFE TRANSFORMATION
Research Methodology
Dr Heenakshi Bhansali
Faculty, School of Health Sciences & Technology
(M.Sc. Clinical Psychology)
Types of Factor Analysis
Exploratory Factor Analysis (EFA): This type of factor analysis is used
when the researcher is not sure about the underlying structure of the
variables. The goal is to explore the data and identify the number of
factors needed to explain the variance in the variables.
EFA does not assume a pre-specified number or structure of factors,
but instead identifies them based on the observed pattern of
correlations among the variables. EFA is useful for generating
hypotheses about the underlying factors that may explain the
observed patterns of correlations.
Types of Factor Analysis
Assumptions underlying EFA are:
Interval or ratio level of measurement
Random sampling
Relationship between observed variables is linear
A normal distribution (each observed variable)
A bivariate normal distribution (each pair of observed variables)
Multivariate normality
Types of Factor Analysis
A researcher is interested in understanding the factors that
contribute to job satisfaction among employees. He administers a
survey to a random sample of employees in a company, asking them
to rate their level of satisfaction with various aspects of their job, such
as pay, work-life balance, job security, opportunities for advancement,
and so on.
He then conducts an EFA on the responses to identify the underlying
factors that contribute to job satisfaction. The steps involved in EFA
include:
Types of Factor Analysis
Factor extraction: He uses a statistical technique such as principal
component analysis to extract the factors from the data. This involves
identifying patterns of covariation among the variables and grouping
them into factors that explain the most variance in the data.
Factor rotation: He rotates the factors to make them easier to
interpret. This involves adjusting the factors so that they are
orthogonal (i.e., uncorrelated) and have simpler loadings.
Factor interpretation: He interprets the factors based on the loadings
of the variables. It represents the strength of the relationship
between the variables and the factors. Variables with high loadings on
a factor are more strongly associated with that factor.
Types of Factor Analysis
EFA reveals three factors that contribute to job satisfaction: work-life
balance, opportunities for advancement, and job security. These
factors explain the majority of the variance in the data, indicating that
they are important contributors to overall job satisfaction.
Researchers can use these factors to develop targeted interventions
to improve employee satisfaction, such as more flexible work
arrangements, training and development opportunities, and
increasing job security measures. EFA is a useful tool for uncovering
underlying factors or dimensions in a data set. It can be applied in a
wide range of fields to identify latent variables that cannot be directly
measured and to develop targeted interventions to address them.
Types of Factor Analysis
Goldberg (1990) used EFA to identify the underlying factors of
personality traits using the International Personality Item Pool (IPIP), a
large set of self-report personality measures. In this study,
participants completed the IPIP, which includes 50 items that measure
five different factors of personality: extraversion, agreeableness,
conscientiousness, neuroticism, and openness.
Using EFA, Goldberg identified five underlying factors of personality
traits that corresponded to the five factors measured by the IPIP.
Types of Factor Analysis
Surgency: This factor includes items related to extraversion, such as
being outgoing, sociable, and assertive.
Agreeableness: This factor includes items related to being
cooperative, sympathetic, and compassionate.
Conscientiousness: This factor includes items related to being
organized, responsible, and dependable.
Emotional stability: This factor includes items related to being calm,
relaxed, and not easily upset.
Intellect/openness: This factor includes items related to being
curious, imaginative, and open-minded.
Types of Factor Analysis
Confirmatory Factor Analysis (CFA): In this type of factor analysis,
the researcher has a pre-specified theory or hypothesis about the
underlying structure of the variables. The goal is to confirm or reject
the theory by testing it against the data.
It assesses the dimensionality of a measure or scale, such as a survey
instrument or questionnaire. It can be used to test whether the items
in the measure are measuring the same underlying construct or factor
and whether the factor structure is consistent across different groups
or populations. It is used in the development and validation of
psychological tests and marketing research, education, and public
health.
Types of Factor Analysis
A researcher is interested in examining the validity of a measure of
depression that includes several items related to negative affect,
feelings of worthlessness, and loss of interest in pleasurable activities.
The researcher proposes that these items reflect three underlying
factors: Negative Affect, Self-Worth, and Anhedonia.
The researcher administers the measure to a large sample of
participants and then uses CFA to test whether the proposed three-
factor model fits the data. The CFA will estimate the extent to which
each item is associated with each factor and will also estimate the
relationships between the three factors.
Types of Factor Analysis
The results of the CFA show that the three-factor model fits the data
well, indicating that the measure of depression has good validity. The
researcher can then use the factor scores as a way of summarizing
participants' levels of Negative Affect, Self-Worth, and Anhedonia,
and examine how these factors are related to other variables of
interest, such as anxiety, stress, or social support.
CFA is a powerful tool for testing the validity of hypothesized factor
structures in psychology and can help researchers better understand
the underlying dimensions of complex constructs such as depression,
personality traits, or intelligence.
Extraction methods
Extraction methods are used to identify the underlying factors that
explain the variance in the observed variables. There are several
extraction methods and the choice of method will depend on the
research question, the nature of the data, and the goals of the
analysis. Some of the most common extraction methods:
Principal Component Analysis (PCA) finds the linear combinations of
the variables that account for the most variance. These linear
combinations, known as principal components, are used as the factors
in the factor analysis.
Extraction methods
Principal Axis Factoring (PAF): PAF is an extraction method that is
similar to PCA, but it takes into account the commonality of the
variables. Communality is the proportion of the variance in a variable
that can be explained by the underlying factors.
Unweighted Least Squares (ULS): ULS is an extraction method that
minimizes the sum of squared residuals between the observed
variables and the factor scores. It is useful when the researcher wants
to estimate the factor structure while assuming that the variables are
uncorrelated.
Extraction methods
Maximum Likelihood (ML): ML is an extraction method that is used
when the data is assumed to be normally distributed. It estimates the
parameters of a model that describes the relationship between the
observed variables and the latent factors.
ML is useful when the researcher wants to test a specific hypothesis
about the underlying factor structure.
Rotation Methods
Rotation is an important step in factor analysis, which is a statistical
technique used to identify the underlying factors that explain the
correlations among observed variables. The rotation method is used
to simplify and interpret the factors that were extracted from the
observed variables.
After the initial factors are extracted using an extraction method such
as Principal Component Analysis (PCA), the factor loadings are often
complex and difficult to interpret. Rotation methods aim to simplify
the factor loadings by rotating the factor axes in order to achieve a
simpler, more interpretable factor structure.
Rotation Methods
There are two main types of rotation methods: orthogonal rotation
and oblique rotation.
Orthogonal rotation methods, assume that the factors are
uncorrelated, and therefore rotate the factor axes to maximize the
variance of each factor's loadings on the observed variables. Varimax
rotation is the most commonly used orthogonal rotation method.
Oblique rotation methods, allow the factors to be correlated and
rotate the factor axes to maximize the amount of variance explained
by each factor, while also allowing for the correlation between
factors.
Higher order factor analysis
Higher-order factor analysis is a statistical technique used to
investigate the underlying structure of a set of variables. In this
technique, a higher-order factor is created to account for the
correlation among lower-order factors. This approach can be
particularly useful when dealing with complex data that has multiple
levels of latent variables.
The nature of higher-order factor analysis is based on the assumption
that there is an underlying hierarchy of factors that affect the
observed variables. The lower-order factors are thought to represent
specific aspects of the phenomenon being measured, while the
higher-order factor represents a more abstract and general construct.
Higher order factor analysis
The characteristic of higher-order factor analysis is that it allows for
the examination of the relationships between multiple levels of latent
variables, which can provide a more comprehensive understanding of
the underlying structure of the data.
This approach also allows for the assessment of the hierarchical
structure of the latent variables, which can help to refine theories and
models of the phenomenon being studied.
Higher order factor analysis
For example, let's say you have a set of variables that measure
different aspects of intelligence, such as verbal ability, spatial
reasoning, and memory. A first-order factor analysis would identify
these individual factors and how they relate to each other. However, a
higher-order factor analysis would go one step further and identify a
broader factor that explains the relationships between these
individual factors, such as "general intelligence.“
In essence, higher-order factor analysis is a way to simplify and
summarize complex relationships among variables by identifying
overarching patterns and underlying factors that can help to explain
these relationships.
Higher order factor analysis
Exploratory higher order factor analysis: This approach is used to
identify the underlying structure of a set of variables and to
determine the relationships between the lower order factors and the
higher order factor.
Confirmatory higher-order factor analysis: This approach is used to
test a specific theoretical model that includes a higher-order factor
and lower-order factors. The aim is to determine whether the
observed data fit the model.
Higher order factor analysis
Bi-factor higher order factor analysis: This approach is used when
there is a need to separate the general factor from the specific
factors. In this approach, a higher-order factor is used to represent
the general factor, and the lower-order factors are used to represent
the specific factors.
Second Order Factor Analysis: In this type of higher-order factor
analysis, the lower-order factors are directly related to the higher-
order factor, and the relationships between the lower-order factors
are not modeled.
THANK YOU
S. No.124, Paud Road, Kothrud, Pune,
411038
22
