paf {rela} | R Documentation |
This function performs a prinxipal axis factor analysis providing the user with a set of preliminary informative estimates regarding the dimensionality and scale functioning of the submitted items. The function does not allow for rotations and is by no means considered a full factor analysis tool, but rather a supplemental module.
paf(object, eigcrit=1, convcrit=.001)
object |
Numeric dataset (usually a coerced matrix from a prior data frame) containing all items of the scale. The dataset is arranged observations (rows) by measure items (columns). |
eigcrit |
Eigenvalue criterion cut-off to be used in the iterative estimation process. By default set to use all eigenvalues greater than 1. |
convcrit |
The convergence criterion determining the number of iterations by computing difference scores between prior and estimated communalities. By default set to be less than .001 for each compared cell. |
This function is intended as a companion to the itemanal() function in this package (rela). Its primiary goal is to provide the researcher with additional information when exploring the dimensionality and reliability of a scale. Moreover, the when called the function will produce four plots: A scree plot of the eigenvalues of the original correlation matrix, a plot of the measure of sampling adequacy values for each item, a comparative plot for the initial and final communalities, and a plot series for the correlation residuals for each individual item against all other items.
Output consists of a list with the following values:
Correlation |
Correlation matrix of the submitted dataset. |
Anti.Image.Cov |
The Anti Image Covariance Matrix. |
Anti.Image.Cor |
The Anti Image Correlation Matrix. |
KMO |
The Kaiser-Meyer-Olkin measure of sampling adequacy test. |
MSA |
Individual measures of sampling adequacy for each item. |
Bartlett |
Bartlett's Test of Sphericity for covariance matrices. |
Communalities |
Initial and final communality extractions. |
Iterations |
Number of iterations needed to meet convergence criterion. |
Eigenvalues |
All eigenvalues for extracted at each iteration. |
Communality.Iterations |
All estimated communalities for each iteration submited to the convergence test. |
Criterion.Differences |
Difference scores between estimated communalities. |
Factor.Loadings |
Final item factor loadings on extracted latent components. |
Reproduced.Cor |
Factor loading reproduced correlation matrix. |
Residuals |
Correlation residuals (observed minus reproduced correlations). |
RMS |
Root mean square errors (of the correlation residuals). |
call |
Submitted arguments to the paf() function. |
Under the current version of this function/package missing data is deleted listwise. Subsequently only full cases are used in determining scale reliability.
Michael Chajewski ( http://www.chajewski.com )
Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16(3), 297-334.
Duhachek, A. & Iacobucci, D. (2004). Alpha's standard error (ASE): An accurate and precise confidence interval estimate. Journal of Applied Psychology, 89(5), 792-808.
Kim, J., & Mueller, C. W. (1978). Introduction to factor analysis: What it is and how to do it. SAGE Publications: Newbury Park, CA.
Nunnally, J. C. & Bernstein, I. H. (1994). Psychometric theory (3 ed.). McGraw-Hill: New York, NY.
Kaiser, H. F. & Cerny, B. A. (1979). Factor analysis of the image correlation matrix. Educational and Psychological Measurement, 39, 711-714.
Pett, M. A., Lackey, N. R., & Sullivan, J. J. (2003). Making sense of factor analysis: The use of factor analysis for instrument development in health care research. SAGE Publications: Thousand Oaks, CA.
library(rela) Belts <- Seatbelts[,1:7] summary(Belts) paf.belt <- paf(Belts) paf.belt Belts2 <- Belts[,-5] Belts2 <- Belts2[,-5] paf.belt2 <- paf(Belts2) paf.belt2