Lecture 10: Factor Analysis & Short forms | Psychometric Theory

0.0(0)
Studied by 0 people
call kaiCall Kai
learnLearn
examPractice Test
spaced repetitionSpaced Repetition
heart puzzleMatch
flashcardsFlashcards
GameKnowt Play
Card Sorting

1/62

encourage image

There's no tags or description

Looks like no tags are added yet.

63 Terms

1
New cards

Factor analysis can only be applied to what type of scoring?

Nondichotomous scoring

2
New cards

What is factor analysis?

A statistical tool used to mathematically determine which items are associated with latent constructs.

3
New cards

What is a latent construct?

A construct that cannot be measured directly; all psychological constructs are latent.

4
New cards

Are factors and latent constructs the same thing?

Yes

5
New cards

Why is factor analysis considered powerful?

It statistically determines where uncertain items belong, reducing guesswork in item development

6
New cards

What does factor analysis help determine about test items?

Item quality without relying on guesswork

7
New cards

What were the three example subconstructs of anxiety discussed in class?

Emotional, physical, and cognitive components

8
New cards

What are examples of emotional anxiety symptoms?

Fearfulness, worry, nervousness, and tearfulness.

9
New cards

What are examples of physical anxiety symptoms?

Increased blood pressure, nausea, muscle tension, rapid breathing, sweating, increased heart rate, dizziness, and possibly fatigue

10
New cards

What are examples of cognitive anxiety symptoms?

Rumination, hypervigilance, and decreased concentration/focus

11
New cards

In the lecture example, why was fatigue important?

Factor analysis determines whether fatigue belongs in the emotional, physical, or cognitive factor

12
New cards

What are the three necessary conditions to run a factor analysis?

  • The factor structure represents what we know about the construct.

  • The factor structure can be replicated.

  • The factor structure is clearly interpretable


13
New cards

What sample size did the professor emphasize for factor analysis?

At least 200 participants; below 200 may produce a corrupted factor structure

14
New cards

What does EFA stand for?

Exploratory Factor Analysis

15
New cards

When is Exploratory Factor Analysis used?

When there is no prior hypothesis about the factor structure.

16
New cards

What is Exploratory factor analysis

Let factors emerge naturally from the data.

17
New cards

When is EFA especially helpful?

When developing a test for the first time

18
New cards

What does CFA stand for?

Confirmatory Factor Analysis

19
New cards

What is the purpose of CFA?

To test whether data fit a pre-existing theoretical model.

20
New cards

What is required before conducting CFA?

An a priori specification of the number of factors

21
New cards

What key phrase distinguishes CFA from EFA?

CFA confirms predetermined hypotheses, whereas EFA allows factors to emerge

22
New cards

Why should researchers initially create more items than needed?

Because factor analysis will identify items that should be removed

23
New cards

What did the professor emphasize as the key advantage of having more items?

More items tap more aspects of the construct, including aspects not previously considered

24
New cards

What are facets?

Homogeneous clusters of items that map onto larger factors; they are subconstructs

25
New cards

According to the lecture, facets and subconstructs are ________.

synonymous

26
New cards

What are the Big Five personality factors (NEOAC)?

Neuroticism, Extraversion, Openness, Agreeableness, and Conscientiousness.

27
New cards

Why is the Big Five called a five-factor model?

Factor analysis identified five major domains and their facets.

28
New cards

Why can't dichotomous response formats be used for factor analysis?

They can seriously disturb the correlation matrix

29
New cards

What happens to factor loadings if dichotomous items are used?

Factors may fail to load properly and the matrix may become redundant.

30
New cards

What response format is recommended for factor analysis?

Likert/rating scales with 5-7 response options.

31
New cards

Why are Likert scales preferred for factor analysis?

They capture more variance

32
New cards

Why is variance important in factor analysis?

Variance is required to conduct factor analysis

33
New cards

What type of sample is needed for factor analysis

A heterogeneous sample

34
New cards

Why is heterogeneity important?

The sample must represent all dimensions of the construct and the full normal curve.

35
New cards

Why might undergraduates be a poor sample for studying depression?

They may not adequately represent individuals with unipolar depression.

36
New cards

What is meant by a hierarchical factor structure?

An overall factor with lower-order subfactors or facets.

37
New cards

What is one major benefit of factor analysis?

It identifies items relevant to a construct while reducing guesswork.

38
New cards

How can factor analysis help discover new constructs?

It may reveal factors researchers had not previously considered.

39
New cards

What major criticism exists regarding factor analysis?

Factors may not map onto a measurable real-world criterion

40
New cards

How does factor analysis improve psychometric properties?

It helps remove, revise, or add items to improve internal consistency reliability.

41
New cards

What should all good tests demonstrate?

Strong internal consistency

42
New cards

Why should test items discriminate between groups?

To distinguish people who possess a trait from those who do not

43
New cards

Why should redundant items be removed?

They add inefficiency and duplicate measurement error.

44
New cards

Why are researchers creating short forms of tests?

  • more efficient

  • less time consuming

  • easier for examiners and examinees


45
New cards

What is the first major objection to short forms?

They may not provide enough information for a proper assessment

46
New cards

What is the second major objection to short forms?

Their validity is often inadequately examined

47
New cards

What key question should always be asked about short forms?

Are they as accurate as the full-length version?

48
New cards

Why can't reliability and validity from a long form automatically be applied to a short form?

Reduced content coverage may compromise reliability and validity

49
New cards

If a long form has poor validity, what does that imply about a short form?

A short form is unlikely to have good validity either.

50
New cards

What important step do many short-form developers fail to conduct?

Content domain checks

51
New cards

What commonly happens to reliability coefficients when tests are shortened?

They decrease significantly.

52
New cards

What reliability value did the professor note many short forms fell below?

.65

53
New cards

What must researchers do after creating a short form?

Rerun factor analysis to determine whether the same factor structure remains

54
New cards

Why must factor analysis be rerun on a short form?

You cannot assume the same factors exist after removing items

55
New cards

What sample should be used to validate a short form?

A new, independent, heterogeneous sample.

56
New cards

What comparison should be made before using a short form clinically?

Determine whether it classifies clinical populations as accurately as the long form

57
New cards

What is the danger of relying on short forms?

Diagnostic inaccuracies that can harm patients.

58
New cards

According to the professor, are short forms psychometrically as sound as long forms?

Generally no; thoroughness is often sacrificed.

59
New cards

What is Exploratory factor analysis (EFA) 

Discovers the underlying factor structure of a dataset when you have no a priori hypotheses about how many factors exist or which variables load onto them 

60
New cards

What is Confirmatory Factor Analysis (CFA) 

Tests and verifies a specific, pre-defined theoretical model on a new dataset 

61
New cards

Reasons for conducting a factor analysis 

  1. Developing and Identifying Hierarchical Factor Structure 

  2. Improving psychometric properties of a test  

  3. Developing Items that Discriminate btwn samples  

  4. Developing more unique items – decreasing redundancy in item construction 


62
New cards

2 primary objections to short form development 

  1. Short form cannot give the lvl of info required for appropriate assessment

  2. Short forms developed without careful examination of new forms validity / accuracy


63
New cards

General problems for short forms

  1. assumption that reliability and validity from long from automatically applies to short form

  2. Assumption that new shorter form requires less validity evidence