Social Research Glossary
Citation reference: Harvey, L., 2012-19, Social Research Glossary, Quality Research International, http://www.qualityresearchinternational.com/socialresearch/
This is a dynamic glossary and the author would welcome any e-mail suggestions for additions or amendments. Page updated 3 June, 2019 , © Lee Harvey 2012–2019.
|A fast-paced novel of conjecture and surprises|
Factor analysis is a procedure that reveals whether data exhibits underlying relationships that may enable the data to be reduced to a smaller set of component factors.
An example is when the ninety-six items of the Eysenck's Personality Inventory Test are found to be describable by the extraversion, neuroticism and lie-scale factors.
In social science factor analysis tends to be used at an exploratory stage as an aid to organising potentially large numbers of 'causal factors' into related groups. It operates by suggesting, on the basis of an assessment of the available data, a set of 'shadow' factors that map out the underlying relationship. These 'shadow' factors generated by the method provide clues as to the nature of some underlying concepts that may underpin the data. These emerge through an examination of the patterns of relationships that are related to each of the emergent 'shadow' factors.
An example is the determination of three underlying factor (extraversion, neuroticism and lie) for the ninety-six items of the Eysenck's Personality Inventory Test.
However, what factor analysis doesn't do when identifying, statistically, underlying groupings, is to provide names for the groupings, this has to be done by an examination of the material and through a theoretical analysis.
Factor analysis is, though, a general term that encompasses a number of different approaches to determining factors. However, there are three crucial steps to factor analysis and each of these may be conducted in one of two distinct alternatives (although within each dichotomy there several variations).
The first stage of factor analysis is the preparation of the correlation matrix (using either R or Q type factor analysis). The second stage is the extraction of the initial factors (using either defined or inferred factors). The third stage is the rotation to a terminal solution (using either orthogonal or oblique solutions).
Any combination of these may be adopted, so 8 basic approaches are possible irrespective of varieties within the dichotomies (e.g., R type, defined, oblique).
Factor analysis is not a simple process for working out the relationships between a mass of theoretically unconnected data. Factor analysis requires a selection of suitable variables to be analysed. It involves various computational techniques that assume certain properties of the data (such as normal distributions) and the nature of the relationships between variables and factors. These are usually linear relationships. Factor analytic procedures are still developing and there are no definitive approaches. In short, factor analysis in social research cannot be anything other than indicative. It involves a lot of effort (even with a computer) for only marginal insights, in most cases.
Colorado State University (1993–2013) defines factor analysis as :
A statistical test that explores relationships among data. The test explores which variables in a data set are most related to each other. In a carefully constructed survey, for example, factor analysis can yield information on patterns of responses, not simply data on a single response. Larger tendencies may then be interpreted, indicating behavior trends rather than simply responses to specific questions..
Colorado State University, 1993–2013, Case Studies available at http://writing.colostate.edu/guides/guide.cfm?guideid=90 , accessed 1 February 2013, still available 3 June 2019.
accessed 1 February 2013,
still available 3 June 2019.
copyright Lee Harvey 2012–2019
copyright Lee Harvey 2012–2019