CFA

Question 1

b parameter

Answer 1

item attractiveness

Question 2

communality

Answer 2

proportion af variance explained by the factor/model predicted variance

(factor loading x factor variance)/(factor loading x factor variance+residual variance)

squared factor loading

Question 3

uniqueness

Answer 3

proportion of variance not explained by the factor

1 - factor loading x factor variance
——————————————–
factor loading x factor
variance+residual variance

1 - squared factor loading

Question 4

identification

Answer 4

option 1 - in the factor
- mean of factor = 0
- variance of factor = 1

option 2 - in the loading
mean of factor = 0
fix 1 factor loading to 1

Question 5

standardized factor loadings

Answer 5

correlation between factor and variable

square this to get the explained variance (communality)

Question 6

standardized residual variances

Answer 6

uniqueness

Question 7

EFA and CFA differences

Answer 7

EFA
> factor structure is unknown
> number of factors is systematically altered
> all items load on all factors

CFA
> number of factors is known from theory
> loadings are derived from theory/expectations

Question 8

parameter estimation

Answer 8

1. principal factoring
   > kaiser, scree, parallel
   + no distributional assumptions
   + no improper solutions
   - no explicit falsifiable model
2. ML
   > X^2 goodness of fit, RMSEA
   > option to only use covariance matrix
   + explicit model
   + falsifiable
   - sometimes improper solutions
   - multivariate normal distribution of the data

Question 9

identification

Answer 9

1. scaling the latent variable
   - results in m^2 restrictions
2. statistical identification
   > k can’t exceed M
   > in EFA this can happen if the number of factors in the model is too large
   > df = M - k

Question 10

rotation

Answer 10

1. orthogonal
   > factors remain uncorrelated
   > varimax
2. oblique
   > facotrs are correlated after rotation
   > promax

Question 11

identifying in the common factor variance vs factor loadings

which parameters are affected and which are not

Answer 11

affected
1. factor variance
2. factor loadings

not affected
1. model predicted covariances
2. model predicted variances
3. residual variances
4. X2 statistic
5. intercepts

Question 12

falsifiable and unfalsifiable structure

Answer 12

unfalsifiable
1. GLB
2. Cronbach’s alpha
> by only considering these, it is not clear how to interpret the reliability, as the underlying structure of the true (parameter estimates) is unknown

falsifiable
(one-)factor model