Why is FA necessary? (FA Methods)
Identifies an underlying structure (factors) from a large set of correlated variables
What is a factor? (FA Methods)
A cluster of items that all measure the same idea
What assumptions are required for FA? (FA Methods)
- Normally distributed data
- Worst offenders
- Correlation of items
How do you identify for normally distributed data? (FA Methods)
SD between .5 and 1.5
How do you identify the worst offenders in a data set? (FA Methods)
z-scores
What is expected of correlation of items? (FA Methods)
- Coefficients more than .3 & less than .9
- Determinant to be above .00001
What does the Kaiser-Meyer-Olkin test measure? (FA Methods)
- Sampling adequacy
- Values range from 0 (not adequate) to 1 (very adequate)
- If below .5, rewrite questionnaire
What does a Bartletts test show? (FA Methods)
- If correlations are too small for FA
- P is significant, then FA is appropriate
How do you report Kaiser-Mayer-Olkin results? (FA Methods)
KMO test is …
How do you report Bartlett’s test? (FA Methods)
Bartletts test is non/significant [x²(df)=…, p=…]
What is factor extraction? (FA Methods)
Deciding on how many factors best capture the data
What is an eigenvalue? (FA Methods)
- The variance in all the variables accounted for by a particular factor
- Low values do not explain the data and should be removed
What are the two rules for using an eigenvalue? (FA Methods)
1) When variables < 30 and all commonalities are > .7
2) When P’s > 250 and average commonality is ≥ .6
What is a commonality? (FA Methods)
The percent of variance in a variable explained by all of the factors together (values after extraction)
What should happen if the two criteria for eigenvalues is not met? (FA Methods)
Use a scree plot
When using a scree plot, what are you looking for? (FA Methods)
- Inflection point (where the slope changes)
- Keep everything to the left of this point
What does rotation do? (FA Methods)
Optimises how the items load onto a factor (spreads variance evenly along factors)
What are the two types of rotation and when are they used? (FA Methods)
- Orthogonal (uncorrelated factors)
- Oblique (inter-correlated factors)
