Lecture 3 - Factor Analysis

Question 1

What test might be used in order to know which items on a questionnaire contribute reliably to the overall score?

Answer 1

Cronbach’s alpha

Question 2

What is the first and second actions to be carried out when looking at item-total correlations?

Answer 2

Recode negative items, then delete items with the highest Cronbach’s alpha if deleted.

Question 3

What is the ideal value for Cronbach’s alpha?

Answer 3

0.7 and above.

Question 4

Why do we need factors?

Answer 4

• Responses from individual questions are too detailed and specific, making them vulnerable and unreliable. Individual questions can also be inter-related.
• Overall score is very simplistic and too broad/course. Has no context or detail.

Question 5

What is the benefit of using overall scores to assess someone’s responses on a questionnaire?

Answer 5

They are reliable and standardised.

Question 6

What is the benefit of using responses on individual questions to assess responses to a questionnaire.

Answer 6

They measure subtle differences between participants.

Question 7

What are orthogonal factors?

Answer 7

Factors that are completely unrelated.

Question 8

What are completely unrelated factors called?

Answer 8

Orthogonal factors

Question 9

What are the 5 features of factor analysis?

Answer 9

1. Communalities
2. Variance explained
3. Scree plot
4. Factor loadings (initial and rotated)
5. Naming the factors

Question 10

What are communalities in relation to factor analyses?

Answer 10

The extent that items contribute to the variance explained in the factor solution.

Has to be as close to 1.0 as possible. if low, it does not contribute much to the understanding of the underlying factors.

Question 11

What is variance explained in relation to factor analysis?

Answer 11

The amount of variance in responses that a factor is responsible for. As close to 100% as possible

Question 12

What is a scree plot in relation to factor analysis?

Answer 12

Representation of the factors that we extract.

Question 13

What are factor loadings in relation to factor analyses?

Answer 13

Correlations of each item with the factor.

Question 14

What is the order of tasks when running a factor analysis? (In terms of extracting factors)

Answer 14

1. Carry out the factor analysis with factors that have eigenvalues equal to or greater than 1 (/based on the scree plot as well).
2. If necessary re-do analysis with increased/decreased number of factors, if it explains more of the variance.

SPSS performs varimax rotation

3. Look at items loading on each factor and name the factors.
4. Check quality of the analysis - does it make sense. If not, re-run with different number of factors to see if the overall solution can be improved.

Question 15

Factor names should be what?

Answer 15

Longer than a single item’s name.

Question 16

What is the purpose of factor names?

Answer 16

To help readers understand your interpretation of the factors.

Question 17

The rotated component matrix is transformed into what?

Answer 17

Factor loadings table

Question 18

The factor loadings table comes from information in what?

Answer 18

The rotated component matrix.

Question 19

What are the 4 steps needed to make a new questionnaire from scratch?

Answer 19

• Item pool –> which questions may be asked
• Pilot testing –> do the questions ask what we want to ask?
• Reliability test –> is the scale consistent?
• Validity checks –> is it measuring what we want it to measure?

Question 20

Varimax rotations find factors with what?

Answer 20

The least amount of overlap.