Factor Analysis assumptions

?
What type of data should we have to run a factor analysis?
Interval or ratio data
1 of 13
When can we assume our data is normally distributed?
If both skew and kurtosis are between the range of -2 and +2
2 of 13
What is the linearity assumption of factor analysis?
The variables (e.g. questions) should be linearly related to each other.
3 of 13
What is the rule for multicollinearity with factor analysis?
no extreme multicollinarity- we want items to be related - but not so much they are effectively the same- look at correlation matrix (shouldn't be .8 or above)
4 of 13
What is another method for looking for multicollinearity?
Check by looking at the determinant - tests whether the variables are highly correlated - determinant shouldn't be less than .00001
5 of 13
What is the correlation rule for factor analysis?
We want some variables to be correlated- if different variables are part of the same factor then they should correlate to some degree- check correlation matrix (shouldn't be below 0.3)
6 of 13
What is another method for testing for correlations?
Bartletts test of Sphericity - tests correlations against an extreme - tests whether we have an identity matrix (all correlations between variables are 0)- needs to be significant p<.05 (indicating variables are correlated)
7 of 13
What is the sample size rule for factor analysis
5-10 pps per items OR ideally 300+ pps
8 of 13
What is the Kaiser- Myer - Olkin
Measure of sampling adequacy - an index of comparing magnitudes of the observed correlation coefficients to the magnitudes of the partial correlation coefficients - if FA want partial correlations be small, as we should have underlying factors.
9 of 13
What does the KMO vary between?
varies between 0 and 1 - needs to be >.5 for factor analysis to be a good idea ( but the higher the better)
10 of 13
What is the point of inflection?
Point at which the shape of the curve changes direction and becomes horizontal
11 of 13
/How do you compare a 2 factor with a 3 factor solution?
The default in SPSS is to extract as many factors as there are eigenvalues above 1- but you can also specify how many factors to extract
12 of 13
How do you decide on the number of factors?
look at the factor loadings for each item - and which items load on each factor- does the factor now contain items which don't make theoretical sense?
13 of 13

Other cards in this set

Card 2

Front

When can we assume our data is normally distributed?

Back

If both skew and kurtosis are between the range of -2 and +2

Card 3

Front

What is the linearity assumption of factor analysis?

Back

Preview of the front of card 3

Card 4

Front

What is the rule for multicollinearity with factor analysis?

Back

Preview of the front of card 4

Card 5

Front

What is another method for looking for multicollinearity?

Back

Preview of the front of card 5
View more cards

Comments

No comments have yet been made

Similar Psychology resources:

See all Psychology resources »See all Visual System resources »