Reliability Part. 2

What is a desired alpha for preliminary validation?

• 0.60

• 0.80

• 0.90

• 0.70 to 0.80

What is a desired alpha for research use?

• 0.60

• 0.80

• 0.90

• 0.70 to 0.80

What is a desired alpha for clinical use?

• 0.60

• 0.80

• 0.90

• 0.70 to 0.80

Alia took some notes of her psychometric class regarding scale score reliability but her notes are not complete. All she wrote regarding alpha was: there are no golden rules should be used. The true response is that “it depends". As her classmate she asked you for help. So you share with her -->

(1) the reasons there is no golden rules

Evaluate responses using AI:

OFF

Answer explanation

We can have a precise alpha as being best because:

1) for index and surveys there are little correlation between items

2) For items in speed and power test the items have little correlation

3) The longer the test the more reliable it is, so there will be a longer alpha for longer tests and vice versa

4) Short test that assesses broad concepts will have a lower alpha

5) Alpha can be influenced by the spearman prophecy formula to find out what the alpha would be for a certain number of item where k= d/n (d: desired number of item and n: total amount of item)

What does scale score reliability means?

Evaluate responses using AI:

OFF

Answer explanation

It is a type of reliability that looks at how different combinations of half tests are similar in their alpha (correlation)

In the case of scale score reliability, why do we need items to be correlated to each other?

Evaluate responses using AI:

OFF

Answer explanation

Hint: think of index items

Possible answer: Because scale score looks at the correlation of item halves or possible combinations of items in relation to the whole test. If Items are not correlated than it is irrelevant. For example for an index that uses one item per construct it is irrelevant (no correlation between items)

When looking at Cronbach alpha we may want to use it to create a shorter test. How would we go about doing that?

Evaluate responses using AI:

OFF

Answer explanation

BY using the Spearman prophecy formula and have k, be k= d/n. Where d= the desired amount of item and n= the total amount of items.

Which would give us an adjusted alpha for that new test.

The following statement represents what type of source of errors in reliability?

Impact of the person grading the assessment based on the conclusion they come to.

• Test-retest

• Alternate form

• Split-half

• Scale score

• Inter-rater

The following statement represents what type of source of errors in reliability?

The same participants are assessed, using the same test and procedure, by different assessors.

• Test-retest

• Alternate form

• Split-half

• Scale score

• Inter-rater

In inter-rater reliability, ___________ is defined as arriving to the same conclusion/ratings about participants. While _______ is defined as arriving to the same classification about participants.