1 / 19

CONCEPTS TO BE INCLUDED

Learn how factor analysis helps reduce a set of survey items into a limited set of factors or dimensions for creating more abstract scales. Explore factors, factor loadings, communality, eigenvalues, and Kaiser's criterion.

rogersa
Télécharger la présentation

CONCEPTS TO BE INCLUDED

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. CONCEPTS TO BE INCLUDED • Items • Factor analysis • Data reduction / scaleconstruction • Factors/dimensions • Factor loadings • Communality • Eigen value • Kaiserscriterion

  2. Factor analysis WITH more than one factor

  3. AIM Reducing a set of (survey) items to a limited set of factors or dimensions, to create a few more abstract scales • Extraction of factors • Communality or items • Eigenvalueor factors • Selectingthenumber of factors using ‘Kaiser’scriterion’

  4. EXAMPLE: THE BIG FIVE NEO-PI-R personality scales measure five domains of personality Give an assessment of normal adult personality Areused in personality assessment

  5. EXAMPLE: THE BIG FIVE Five constructs/dimensions/factors: • Openness (closed minded – intellectual curiosity) • Conscientiousness (sloppy/unreliable – highly organized) • Extraversion (reflective – attention seeking) • Agreeableness (suspicious – compassionate) • Neuroticism (emotionally stable – emotionally unstable)

  6. ITEMS MEASURING ONE CONSTRUCT: NEUROTICISM Neuroticism (emotionally stable – emotionally unstable) Items: • “Even minor annoyances can be frustrating to me” • “Sometimes I feel completely worthless” Extraversion (reflective – attention seeking) • “I really enjoy talking to people”

  7. WHAT IS FACTOR ANALYSIS? A method to check whether latent factors (a.k.a. dimensions)are able to account for the correlation between sets of items Here: neuroticism and extraversion

  8. Item 1 Concept 1 Correlation Item 2 x No correlation Concept 2 Item 3

  9. Even minor annoyancescanbe frustrating to me Neuroticism x Sometimes I feel completelyworthless x Extraversion I reallyenjoytalkingtopeople

  10. ONE FACTOR REPLACES ONLY TWO ITEMS Item 1 = a * Factor(N) + 0*Factor(E) Item 2 = b * Factor(N) + 0*Factor(E) Item 1 Concept Suppose perfect correlation = 1 x Item 2 Cannot be correlated Suppose absolutely no correlation = 0 Concept Item 3

  11. THREE ITEMS, EXPLAINED BY TWO FACTORS Two items can be replaced by factor(N), but we need a factor(E) to explain item 3 Item 1 = 1*Factor(N) + 0*Factor(E) Item 2 = 1*Factor(N) + 0*Factor(E) Item 3 = 0*Factor(N) + 1*Factor(E) Factor loadings

  12. THREE ITEMS, EXPLAINED BY TWO FACTORS Correlations are never perfect or completely absent, so factor loadings are somewhere between 0 and 1 For example: • Item 1 = 0.8*Factor(N) + 0.1*Factor(E) … • Item 2 = 0.7*Factor(N) + 0.2*Factor(E) … • Item 3 = 0.1*Factor(N) + 0.9*Factor(E) … Factor loadings

  13. MANY FACTORS, MANY ITEMS Item O1 Item O2 Item O3 … Item C1 … Item E1 … Item A1 … Item N1 Item N2 Item N3 Openness Conscientiousness Extraversion Agreeableness Neuroticism

  14. FACTOR MATRIX Squared and summed = ‘Explained variance’ of an item Is the “Communality” of an item Factor loadings

  15. INTERPRETATION OF ‘COMMUNALITY’ The extent to which all factors together are able ‘to explain’ an item: communality If communality of an item is low, relevance of this item for the factors is low (not measuring an aspect of ‘personality’) If communality is high, it belongs to the factor structure More factors  more variance explained, higher communality, BUT more difficult to interpret

  16. FACTOR MATRIX Squared and summed = ‘Explained variance’ by a factor divided by the number of items = the “Eigenvalue” of a factor Factor loadings

  17. INTERPRETATION OF ‘EIGENVALUE’ Contribution of one factor to explaining the correlations between items: an indication of the importance of the factor. If the eigenvalue of a factor is low the factor does not contribute to explaining the correlations between the items. IF a factor is able to explain LESS than one item (= Eigenvalue < 1), The factor is often seen as not relevant: Kaiser’s criterion.

  18. THIS MICROLECTURE Factor analysis as a data reduction / scaleconstructionmethod, whichyieldsfactors/dimensions. • Factor loadings • Communality • Eigenvalue • Kaiser’scriterion

More Related