Factor Analysis by Diagrams PDF

Source: Pett, M.A., N.R. Lackey, and J.J.
Sullivan, Making sense of factor analysis: the use of factor analysis for instrument development in health care research. 2003, London: SAGE Publications (page 83). Developed by: Le Tan Phung, MD, MPH, PhD Candidate at QUT
ASSESSING THE CHARACTERISTICS OF MATRICES (page 83)
Visually examine the correlation matrix. Are there significant strong correlations among the item?
No Yes
Drop the poorly correlated items from the analysis. Rerun the matrix
Evaluate the Determination, |R| Is |R|=1.0?
Drop one or more highly correlated items. Check the number of subjects per item (n>10-15). Rerun
|R|=0.0
Yes
|R|=1.0
No
Factor analyses are inadvisable
The correlation matrix is an identity matrix
The correlation matrix is a singular matrix, not positive definite. Some items are too highly correlated
No 0 <|R|<1
Increase the sample or reduce the number of items. Rerun
Examine Barletts test of sphericity Is p < .05

No p > 0.05 Yes p < 0.05
The sample size is not sufficient relative to the number of items
There is a sufficient minimum sample size
Eliminate low KMO, MSA items. Rerun
No KMO < .60 MSA < .60
Examin the KMO and MSA values. Are they > .60?
KMO > .60 MSA > .60
Yes
CONGRATULATIONS! You are ready to undertake the initial factor extraction!
A good Barletts test indicates that the correlation matrix is not an identity matrix (1 diagonal and 0 off) A good KMO suggests that there is sufficient sample size relative to the number of items. A good MSA indicates that the correlations among the individual items are strong enough to suggest that the correlation matrix is factorable (page 81)
1
EXTRACTING THE INITIAL FACTORS (page 129)
EVALUATE THE CORRELATION MATRIX No Retest of Matrices Is the matrix factorable? Yes CHOOSE THE EXTRACTION METHOD
Do you want to explain total or common variance? Total Common Maximum Likelihood
PRINCIPAL COMPONENTS ANALYSIS
COMMON FACTOR ANALYSIS
Least Square
Principal Axis Factoring EXTRACT THE INITIAL FACTORS
How many factors will you retain? Percent of Extracted Variance (5%) CRITERIA FOR RETENTION Insignificant Chi-square Values
Eigenvalue > 1
Size of the Residuals
Examine the Scree Plot
Factor Interpretability and Usefulness
DECISION TREE FOR ROTATING THE FATORS (page 165)
RETURN TO THE PREVIOUS STEP
No No
Have you undertaken an initial extraction? Yes
No Have you decided on the numbers of initial factors? Yes
Do you suspect your factors are correlated?
Varimax
No ORTHOGONAL ROTATION
Yes OBLIQUE ROTATION
Direct Oblimin
|.5|
Quartimax
Orthoblique \ Promax =2,3, or 4
Equamax ROTATE THE FACTORS
How large are the factor correlations? r = .00 CHOOSE THE ORTHOGONAL SOLUTION CONSIDER ORTHOGONAL SOLUTION 0< r <|.2| r>|.5| CONSIDER DROPPING ONE OR MORE FACTORS
r=|.3|
CHOOSE OBLIQUE SOLUTION
No Have you successfully rotated the factors? Yes
CONGRATULATION! You are ready to refine and names the factors You
REFINING THE FACTOR AND EVALUATING ITS INTERNAL CONSISTENCY (p.205)
GENERATE A ROTATED FACTOR STRUCTURE MATRIX
REMOVE ITEMS FROM ANALYSIS No Is the weak-loading item important to the content area? Yes Yes
SIMPLIFY THE PRESENTATION (e.g., suppress Loadings < .40)
RESCALE ITEMS IF NECESSARY Yes
Are there items with weak loadings (<.30) on all factors?
No
Are there positive and negative loadings on the same factors?
No
RETAIN ITEMS SEPARATE FROM OTHER FACTORS
Are there items with strong loadings (<.40) on multiple factors?
Yes
PLACE MULTIPLE LOADING ITEM WITH BEST FITTING FACTOR CONCEPTUALLY
No OBTAIN AN ALPHA COEFFICIENT FOR FACTOR AND ITEM IF DELETED
Yes (2) If the alpha coefficient negatively affected by deletion of items?
Yes (1)
No RETAIN ITEMS ON FACTOR AND GO TO INTERPRETING AND NAMING THE FACTORS
INTERPRETING AND NAMING THE FACTORS (page 225)
SORT THE FACTOR PATTERN MATRIX
SELECT A FACTOR TO EXAMINE
No Are there several factors loading .60? Yes No
COMPARE GROUPED ITEMS WITH YOUR ORIGINAL CONCEPTUALIZATION
Is there a common theme to the highloading items? Yes COMPARE GROUPED ITEMS WITH YOUR ORIGINAL CONCEPTUALIZATION
Are there several groupings within the factor? Yes CONSIDER BREAKING UP THE FACTOR
Are the grouped items consistent with the original conceptualization? Yes
No
BACK TO THE DRAWING BOARD!
Is there sufficient information to name the factor? Yes KEEP THE NAME SIMPLE AND DESCRIPTIVE
No
FACTOR NAMING IS PREMATURE
A REPORT OF A FACTOR ANALYSIS SHOULD INCLUDE (page 228): 1. 2. 3. 4. 5. 6. The theoretical rationale for the use of factor analysis Detailed description of the sampling methods of participants. Descriptions of the items, including means and standard deviations. An evaluation of the assumptions of factor analysis. A justification for the choice of factor extraction and rotation methods. Evaluation of the correlation matrix: Barletts test of sphericity, Kaiser-Meyer-Olkin test. 7. Criteria for extracting the factors: the scree test, eigenvalues, percent of variance extracted. 8. Cutoffs for meaningful factor loadings. 9. The structure matrix for orthogonally rotated solutions; the structure and pattern matrices and interfactor correlations for oblique rotated solutions. 10.Descriptions and interpretation of the factors. 11.Internal consistency of the identified factors (e.g., Cronbach alpha) 12.Approach to calculation of factor-based scores. 13.Assessment of the study limitations and suggestions for future research directions. 14.
ACHIEVING A SIMPLE STRUCTURE (Harman, 1976) (Pett @ page 132): 1. Each row of the factor matrix should contain at least one zero; 2. If there are m common factors, each column of the factor matrix should have at least m zero; 3. For every pair of columns of the factor matrix, there should be several variables for which entries approach zero in one column but not in the other; 4. For every pair of columns of the factor matrix, a large proportion of the variables should have entries approaching zero in both columns when there are four or more factors; 5. For every pair of columns of the factor matrix, there should be only a small number of variables with nonzero entries in both columns.

Factor Analysis by Diagrams PDF

Uploaded by

Copyright:

Available Formats

Factor Analysis by Diagrams PDF

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Factor Analysis by Diagrams PDF

Uploaded by

Copyright:

Available Formats

Source: Pett, M.A., N.R. Lackey, and J.J.

ASSESSING THE CHARACTERISTICS OF MATRICES (page 83)

Evaluate the Determination, |R| Is |R|=1.0?

Factor analyses are inadvisable

The correlation matrix is an identity matrix

Increase the sample or reduce the number of items. Rerun

Examine Barletts test of sphericity Is p < .05

The sample size is not sufficient relative to the number of items

There is a sufficient minimum sample size

Eliminate low KMO, MSA items. Rerun

No KMO < .60 MSA < .60

CONGRATULATIONS! You are ready to undertake the initial factor extraction!

EXTRACTING THE INITIAL FACTORS (page 129)

PRINCIPAL COMPONENTS ANALYSIS

COMMON FACTOR ANALYSIS

Principal Axis Factoring EXTRACT THE INITIAL FACTORS

Size of the Residuals

Examine the Scree Plot

Factor Interpretability and Usefulness

DECISION TREE FOR ROTATING THE FATORS (page 165)

RETURN TO THE PREVIOUS STEP

Have you undertaken an initial extraction? Yes

No Have you decided on the numbers of initial factors? Yes

Do you suspect your factors are correlated?

Yes OBLIQUE ROTATION

Orthoblique \ Promax =2,3, or 4

Equamax ROTATE THE FACTORS

CHOOSE OBLIQUE SOLUTION

No Have you successfully rotated the factors? Yes

REFINING THE FACTOR AND EVALUATING ITS INTERNAL CONSISTENCY (p.205)

GENERATE A ROTATED FACTOR STRUCTURE MATRIX

SIMPLIFY THE PRESENTATION (e.g., suppress Loadings < .40)

RESCALE ITEMS IF NECESSARY Yes

Are there items with weak loadings (<.30) on all factors?

Are there positive and negative loadings on the same factors?

RETAIN ITEMS SEPARATE FROM OTHER FACTORS

Are there items with strong loadings (<.40) on multiple factors?

PLACE MULTIPLE LOADING ITEM WITH BEST FITTING FACTOR CONCEPTUALLY

No OBTAIN AN ALPHA COEFFICIENT FOR FACTOR AND ITEM IF DELETED

Yes (2) If the alpha coefficient negatively affected by deletion of items?

No RETAIN ITEMS ON FACTOR AND GO TO INTERPRETING AND NAMING THE FACTORS

INTERPRETING AND NAMING THE FACTORS (page 225)

SORT THE FACTOR PATTERN MATRIX

SELECT A FACTOR TO EXAMINE

No Are there several factors loading .60? Yes No

COMPARE GROUPED ITEMS WITH YOUR ORIGINAL CONCEPTUALIZATION

BACK TO THE DRAWING BOARD!

FACTOR NAMING IS PREMATURE

You might also like