week 2

Question 1

What si the KMO measure?

Answer 1

Kaiser-meyer-olkin measure. Computed as the sum of squared correlattions to the sum of squared correlations plus sum of squared partial correlations. Small values indicate the items are likely to have too little common variance for there to be underlyinng factors–higher values desirable.

Question 2

What is bartletts test of sphericity?

Answer 2

Tests null hypothesis that the correlation matrix is an identity matrix, a correlation amtrix consisting o 0 correlation between variables. Significance is what we’re after. conservative test.

Question 3

What is the determinant?

Answer 3

A test of multicollinearity in the data, similar to the other diagnostics
• We want this to be above .00001
• Provided as an odd little table note at the start of the output, often in
exponential notation

Question 4

What are the two types of reliability?

Answer 4

Test-retest reliability
• Temporal consistency
• Ideal form of reliability
• Administer a measure once, administer it again later, and
the degree to which the results are correlated shows
reliability over time

Split-half reliability
• Measure of internal consistency, the degree to which
items are informing a consistent construct
• Correlate one half of a test with the other half
• Criticised as biased by order of items
• Seldom used today as superseded by Cronbach’s alpha

Question 5

What is cronbachs alpha?

Answer 5

Statistical formula that provides the mean of all possible split-half
correlations for a given set of items
• This is commonly used today as the ideal measure of internal
consistency
• As it is influenced by the degree to which items are assessing the
same construct, it is good to obtain Cronbach’s alpha for subscales
as well as the total scale
• We want Cronbach’s alpha to be above .6 at a bare minimum,
although .8 or ideally .9 or above is better

Question 6

What is the goodness of fit test?

Answer 6

The goodness of fit test
• Provided by maximum likelihood factor analysis
• Significance indicates a poor fit, so we want it to be non-significant
• However, it is conservative and biased by large numbers of
participants (as are required for factor analysis…)
• Hence, I would never abandon an analysis because goodness of fit
was violated, although I would report that this had occurred

Question 7

What is the goodness of fit test?

Answer 7

The goodness of fit test
• Provided by maximum likelihood factor analysis
• Significance indicates a poor fit, so we want it to be non-significant
• However, it is conservative and biased by large numbers of
participants (as are required for factor analysis…)
• Hence, I would never abandon an analysis because goodness of fit
was violated, although I would report that this had occurred

Question 8

What are the five mmain souces that could inform our decision to delete items?

Answer 8

1. The actual written content of the item and its theoretical
   relevance to the underlying construct we are trying to measure
2. The extraction communality
3. The factor loading and whether cross-loading is occurring
4. The corrected item-total correlation
5. Cronbach’s alpha if item deleted