Psychological Testing Questions and Answers

Answer: A

Item bias is studied by comparing (I) only, item difficulty across groups.

An item is biased when people from different groups who have the same level of the ability being measured have different chances of answering it correctly. Methods of detecting differential item functioning (DIF) such as Mantel-Haenszel and item response theory compare an item's difficulty or response curve across groups after matching on total score.

A difference in the two groups' average performance on the whole test (II) cannot show bias by itself, because the groups may genuinely differ in the ability. That is why 'Both' and '(II) only' are wrong, and 'Neither' ignores the standard method.

Answer: A

The item to add first is the one with P = 0.12.

The difficulty index P of an item is the proportion of test takers who answer it correctly, so the lower the P value, the harder the item. An item answered correctly by only 12 per cent of examinees is the hardest of the four and does most to raise the overall difficulty of the test.

P = 0.20 is the next hardest, P = 0.45 is of moderate difficulty, and P = 0.87 is an easy item that would make the test easier. The common mistake is reading P as a difficulty score where higher means harder.

Answer: A

On the official key the answer is (I) only: temporal stability.

Test-retest reliability gives the same test twice to the same group and correlates the two sets of scores. Because the items are identical, the test's content cannot contribute error, and what is left is change over the interval: the standard textbook classification (Anastasi, Kaplan and Saccuzzo) names this error source time sampling, the stability of scores over time.

Content of the test (II) is the error source of alternate-form and split-half reliability, where different items are compared, which is why options B and C fail. Individual variation and the test administrator (III, IV) are general sources of error in any testing, not the source that defines test-retest.

What could not be confirmed: some texts also treat fluctuations in the testing session as error in a retest, so the word 'only' rests on the textbook classification by type of reliability.

Answer: A

Item characteristic curve theory is also known as latent trait theory.

Item response theory models the probability of a correct answer to an item as a function of an unobserved (latent) trait, usually written theta. The S-shaped graph of that probability against theta is the item characteristic curve, and its shape is fixed by item parameters such as difficulty, discrimination and guessing. Because everything is referred to an underlying trait that cannot be observed directly, the approach was first called latent trait theory, and also ICC theory.

The standard models assume a single trait, so 'multidimensional trait theory' is wrong. Test standardisation and construct validation are procedures used in test construction, not other names for this model.

Answer: D

The most appropriate description is a standardized measure of a sample of behaviour.

Anne Anastasi's well-known definition is that a psychological test is essentially an objective and standardized measure of a sample of behaviour. A test samples behaviour, because it can observe only a limited set of responses, and it is standardized, because administration and scoring are uniform so that scores can be compared.

'A measure of a sample of behaviour' leaves out standardization. Many tests are not self-report measures, for example ability and performance tests, and tests are not limited to conscious behaviour, since projective tests aim at material outside awareness.

Answer: A

Stanine scores are used to interpret interindividual differences.

A stanine ('standard nine') is a normalised standard score on a nine-point scale with a mean of 5 and a standard deviation of about 2. It places a person's score within the distribution of the norm group, so it tells how one individual compares with others: an interindividual comparison.

Stanines are not used to trace developmental change within one person over time. They are not limited to personality tests or to ability tests: they are widely used for achievement and aptitude tests as well, so the two options with 'only' are wrong.

Answer: B

Among the options, the appropriate one is alpha reliability.

A unidimensional test measures a single attribute, so its items should correlate with one another. Cronbach's alpha is an index of internal consistency, the degree to which all the items hang together. A strong alpha is consistent with a single common dimension. Factor analysis is the more rigorous check, but of the methods listed only alpha looks at the relationships among the items within the test.

Test-retest reliability measures stability over time and alternate-form reliability the equivalence of two forms; both compare whole-test scores and say nothing about the structure of the items. Correlating the test with another test is evidence of validity, not of unidimensionality.

Answer: C

The subtests not in WAIS-IV are (3) and (4) only: picture arrangement and object assembly.

The fourth edition of the Wechsler Adult Intelligence Scale (2008) dropped Picture Arrangement and Object Assembly, which had been part of the earlier editions' performance scale, partly to reduce the weight of motor speed. It introduced three new subtests: Visual Puzzles and Figure Weights, which belong to the Perceptual Reasoning Index, and Cancellation.

Since visual puzzles (1) and figure weights (2) are new WAIS-IV subtests, every option naming either of them as missing from the scale is wrong.

Answer: D

The keyed answer is problem solving behaviour.

Situational tests place the person in a lifelike situation and observe what he or she actually does. The assessment programmes of the US Office of Strategic Services in the Second World War, for example, set candidates practical tasks such as getting a group across a stream with limited equipment or directing helpers to build a structure. These tasks present a problem and the assessors watch how it is tackled, which is why situational tests are described as measures of problem solving behaviour.

They observe overt behaviour, so they are not measures of subconscious or unconscious processes, which are the target of projective techniques. They are also not self-reports, since the person's own account is exactly what they replace.

Answer: A

The distinguishing feature of a power test is that its overall difficulty level is very high.

A pure power test gives enough time for everyone to attempt every item; what separates test takers is how far they can go on items of steeply rising difficulty, and it usually includes items too hard for anyone to solve. A pure speed test is the reverse: its items are uniformly easy, and scores depend on how many a person finishes in a short time limit.

Penalising wrong answers (a correction for guessing) can be used with either kind of test. A speed test, not a power test, normally has more items than anyone can finish. Reverse scoring belongs to attitude and personality scales with negatively worded items.