Field Ch. 18 - Exp. Factor Analysis Flashcards

1
Q

Why use Exploratory Factor Analysis?

A

to measure latent variables

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

3 uses for EFA and PCA

A
  1. understand structure of a set of variables
  2. construct a questionnaire that measures latent variable
  3. reduce data set & maintain meaningful data (identify & combine colinear variables)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

What is Kaiser-Meyer-Olkin?

A

measure of sample adequacy

.50 = adequate
.90+ = marvellous
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

How do you find sample adequacy (KMO) for individual variables?

A
  • run an Anti-image matrix
  • diagonals should be > .50
  • off-diagonals should be very small
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

What is Common Variance?

A

amount of variance shared among set of items

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What is Communality?

A

proportion of common variance found in a particular variable

  • h^2
  • between 0 and 1
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What is Unique Variance?

A

any portion of variance that’s not common

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

What are eigenvalues?

A

they represent the total variance explained by a given component

> 0 = top notch
close to or < 0 = bad news: multicollinearity

(found in ‘Total Variance Explained’ table)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What are eigenvectors?

A

they represent the weight for eigenvalues (strength of correlation between item and factor/component)

“correlation of item 1 with component 1 is…”

(found in ‘Component Matrix’ table)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

After analysis, which factors do you retain?

A
  • any factors with eigenvalue > 1
    or
  • factors to the left of the scree plot elbow
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

Why use rotation?

A

to improve interpretability

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

Why use orthogonal rotation?

A

it assumes factors are independent or uncorrelated

Varimax

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Why use Oblique rotation?

A

when factors aren’t independent and are correlated

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

SPSS step 1: outer menu choices

A

Analyze > Dimension Reduction > Factor

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

SPSS step 2: ‘Descriptives’ button

A
  • initial solution
  • coefficients
  • sig levels
  • anti-image
  • KMO & Bartletts

(less about memorizing and more about familiarity with boxes to check during analysis)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

SPSS step 3: ‘Extraction’ button

A
  • correlation matrix
  • unrotated factor solution
  • scree plot
  • max iterations = 100

(less about memorizing and more about familiarity with boxes to check during analysis)

17
Q

SPSS step 4: ‘Rotation’ button

A
  • varimax

- max iterations = 100

18
Q

SPSS step 5: ‘Scores’ button

A
  • save as variables
  • regression

(less about memorizing and more about familiarity with boxes to check during analysis)

19
Q

Kara’s interpretation notes

A
  • correlation matrix: between 0.3 and 0.9
  • KMOs greater than 0.5
  • communalities = common variance
  • eigenvalues = total variance
  • factor matrices: post-rotation has more even spread of variables among factors
  • residuals: difference btwn observed (correlation matrix) and model (reproduced correlations) should be small
20
Q

How is PCA different from Factor Analysis?

A

PCA is a linear combination of variables

Factor Analysis is a measurement model of a latent variable.