This is a concise, easy to use, step-by-step guide for applied researchers conducting exploratory factor analysis (EFA) using SPSS.
In this book, Dr. Watkins systematically reviews each decision step in EFA with screen shots and code from SPSS and recommends evidence-based best-practice procedures. This is an eminently applied, practical approach with few or no formulas and is aimed at readers with little to no mathematical background. Dr. Watkins maintains an accessible tone throughout and uses minimal jargon to help facilitate grasp of the key issues users will face while applying EFA, along with how to implement, interpret, and report results. Copious scholarly references and quotations are included to support the reader in responding to editorial reviews.
This is a valuable resource for upper-level undergraduate and postgraduate students, as well as for more experienced researchers undertaking multivariate or structure equation modeling courses across the behavioral, medical, and social sciences.
Frequently asked questions
Simply head over to the account section in settings and click on âCancel Subscriptionâ - itâs as simple as that. After you cancel, your membership will stay active for the remainder of the time youâve paid for. Learn more here.
At the moment all of our mobile-responsive ePub books are available to download via the app. Most of our PDFs are also available to download and we're working on making the final remaining ones downloadable now. Learn more here.
Both plans give you full access to the library and all of Perlegoâs features. The only differences are the price and subscription period: With the annual plan youâll save around 30% compared to 12 months on the monthly plan.
We are an online textbook subscription service, where you can get access to an entire online library for less than the price of a single book per month. With over 1 million books across 1000+ topics, weâve got you covered! Learn more here.
Look out for the read-aloud symbol on your next book to see if you can listen to it. The read-aloud tool reads text aloud for you, highlighting the text as it is being read. You can pause it, speed it up and slow it down. Learn more here.
Yes, you can access A Step-by-Step Guide to Exploratory Factor Analysis with SPSS by Marley W. Watkins in PDF and/or ePUB format, as well as other popular books in Informatique & Programmation. We have over one million books available in our catalogue for you to explore.
The idea that unobservable phenomena underlie observed measurements is very old and pervasive. In fact, it may be a basic scientific principle (Hägglund, 2001). Philosophers and scientists such as Plato, Descartes, Bacon, Locke, Hume, Quetelet, Galton, Pearson, and Mill articulated these philosophical and mathematical foundations. However, it was Spearman (1904) who explicated a mathematical model of the relations between observed measures and latent or unmeasured variables (Mulaik, 1987).
Spearman (1904) described his mathematical model as a â âcorrelational psychologyâ for the purpose of positively determining all psychical tendencies, and in particular those which connect together the so-called âmental testsâ with psychical activities of greater generality and interestâ (p. 205). That is, to analyze the correlations between mental tests in support of his theory of intelligence. Spearman posited a general intelligence (labeled g) that was responsible for the positive relationships (i.e., correlations) he found among mental tests. Given that this general intelligence could not account for the totality of the test intercorrelations, he assumed that a second factor specific to each test was also involved. âThus was born Spearmanâs âtwo-factorâ theory which supposed that the observed value of each variable could be accounted for by something common to all variables (the general, or common, factor) and the residual (the specific factor)â (Bartholomew, 1995, p. 212). Spearman also assumed that mental test scores were measured with some degree of error that could be approximated by the correlation of two repeated measurements (i.e., test-retest reliability).
Exploratory factor analysis (EFA) methods were further debated and refined over the ensuing decades with seminal books appearing in the middle of the century (Burt, 1940; Cattell, 1952; Holzinger & Harman, 1941; Thomson, 1950; Thurstone, 1935, 1947; Vernon, 1961). These scholars found Spearmanâs two-factor theory over simple and proposed group factors in addition to general and specific factors. Thus, the observed value of each variable could be accounted for by something common to all measured variables (general factor), plus something common to some but not all measured variables (group factors), plus something unique to each variable (specific factor), plus error. Common variance, or the sum of variance due to both general and group factors, is called communality. The combination of specific variance and error variance is called uniqueness (Watkins, 2017). As portrayed in Figure 1.1, this is the common factor model: total variance = common variance + unique variance (Reise et al., 2018).
The contributions of Thurstone (1931, 1935, 1940, 1947) were particularly important in the development of EFA. He studied intelligence or ability and applied factor analysis to many datasets and continued the basic assumption that âa variety of phenomena within the domain are related and that they are determined, at least in part, by a relatively small number of functional unities, or factorsâ (1940, p. 189). Thurstone believed that âa test score can be expressed, in first approximation, as a linear function of a number of factorsâ (1935, p. vii) rather than by general and specific factors. Thus, he analyzed the correlation matrix to find multiple common factors and separate them from specific factors and error. To do so, Thurstone developed factorial methods and formalized his ideas in terms of matrix algebra. Using this methodology, Thurstone identified seven intercorrelated factors that he named primary mental abilities. Eventually, he recognized that the correlations between these primary mental ability factors could also be factor analyzed and would produce a second-order general factor. Currently, a model with general, group, and specific factors that identifies a hierarchy of abilities ranging in breadth from general to broad to narrow is ascendant (Carroll, 1993).
A variety of books on factor analysis have been published. Some presented new methods or improved older methods (Cattell, 1978; Harman, 1976; Lawley & Maxwell, 1963). Others compiled the existing evidence on factor analysis and presented the results for researchers and methodologists (Child, 2006; Comrey & Lee, 1992; Fabrigar & Wegener, 2012; Finch, 2020a; Garson, 2013; Gorsuch, 1983; Kline, 1994; Mulaik, 2010; Osborne, 2014; Osborne & Banjanovic, 2016; Pett et al., 2003; Rummel, 1970; Thompson, 2004; Walkey & Welch, 2010). In addition, there has been a veritable explosion of book chapters and journal articles explicitly designed to present best practices in EFA (e.g., Bandalos, 2018; Beaujean, 2013; Benson & Nasser, 1998; Briggs & Cheek, 1986; Budaev, 2010; Carroll, 1985, 1995a; Comrey, 1988; Cudeck, 2000; DeVellis, 2017; Fabrigar et al., 1999; Ferrando & Lorenzo-Seva, 2018; Floyd & Widaman, 1995; Goldberg & Velicer, 2006; Hair et al., 2019; Hoelzle & Meyer, 2013; Lester & Bishop, 2000; Nunnally & Bernstein, 1994; Osborne et al., 2007; Preacher & MacCallum, 2003; Schmitt, 2011; Tabachnick & Fidell, 2019; Watkins, 2018; Widaman, 2012; Williams et al., 2010).
Conceptual Foundations
As previously noted, EFA is based on the concept that unobserved or latent variables underlie the variation of scores on observed or measured variables (Bollen, 2002). Alternative conceptualizations have been described by Epskamp et al. (2018). A correlation coefficient between two variables might exist due to: (a) a random relationship between those two variables, (b) one variable causing the other, or (c) some third variable being the common cause of both. Relying on the third possibility, EFA assumes that the correlations (covariance) between observed variables can be explained by a smaller number of latent variables or factors (Mulaik, 2018). âA factor is an unobservable variable that influences more than one observed measure and which accounts for the correlations among these observed measuresâ (Brown, 2013, p. 257).
Theoretically, variable intercorrelations should be zero after the influence of the factors has been removed. This does not happen in reality because no model is perfect, and a multitude of minor influences are present in practice. Nevertheless, it is the ideal. As described by Tinsley and Tinsley (1987), EFA
is an analytic technique that permits the reduction of a large number of interrelated variables to a smaller number of latent or hidden dimensions. The goal of factor analysis is to achieve parsimony by using the smallest number of explanatory concepts to explain the maximum amount of common variance in a correlation matrix.
(p. 414)
The simplest illustration of common variance is the bivariate correlation (r) between two continuous variables represented by the circles labeled A and B in Figure 1.2. The correlation between variables A and B represents the proportion of variance they share, which is area Aâ˘B. The amount of overlap between two variables can be computed by squaring their correlation coefficient. Thus, if r = .50, then the two variables share 25% of their variance.
With three variables, the shared variance of all three is represented by area Aâ˘Bâ˘C in Figure 1.3. This represents the general factor. The proportion of error variance is not portrayed in these illustrations, but it could be estimated by 1 minus the reliability coefficient of the total ABC score.
With multiple variables, the correlation structure of the data is summarized by EFA. As illustrated in Figure 1.4, there are 1 to n participants with scores on 1 to x variables (V1 to Vx) that are condensed into a V1 to Vx correlation matrix that will, in turn, be summarized by a factor matrix with 1 to y factors (Goldberg & Velicer, 2006).
EFA is one of several multivariate statistical methods. Other members of the multivariate âfamilyâ include multiple regression analysis, principal components analysis, confirmatory factor analysis, and structural equation modeling. In fact, EFA can be conceptualized as a multivariate multiple regression method where the factor serves as a predictor and the measured variables serve as criteria. EFA can be used for theory and instrument development as well as assessment of the construct validity of existing instruments (e.g., Benson, 1998; Briggs & Cheek, 1986; Carroll, 1993; Comrey, 1988; DeVellis, 2017; Haig, 2018; Messick, 1995; Peterson, 2017; Rummel, 1967; Thompson, 2004). For example, EFA was instrumental in the development of modern models of intelligence (Carroll, 1993) and personality (Cattell, 1946; Digman, 1990) and has been extensively applied for the assessment of evidence fo...
Table of contents
Cover
Half Title
Title
Copyright
Contents
List of Figures
Preface
1 Introduction: Historical Foundations
2 Data
3 SPSS Software
4 Importing and Saving Data
5 Decision Steps in Exploratory Factor Analysis
6 Step 1: Variables to Include
7 Step 2: Participants
8 Step 3: Data Screening
9 Step 4: Is Exploratory Factor Analysis Appropriate?