LEARN MORE...

Home
Documents
March 2021
9
Florida international university psy 4304 exam 2 review

Florida international university psy 4304 exam 2 review

Content type

User Generated

Uploaded By

rivir88

Pages

Showing Page:

1/28

Psychological Testing

___

Exam 2

MILLER CH 6: RELIABILITY/PRECISION

● Define reliability/precision, and describe three methods for estimating the

reliability/precision of a psychological test and its scores.

Reliability: consistency of test scores. All test contain some error.

More consistency of test scores, more reliability/precision

Results of the statistical evaluation of reliability > reliability coefficient

● Measurement Error: variations on the measurements of the room

● Reliable Test: one we can trust to measure e/person in approx. the same way in every

way it’s used.

Psychological tests don’t have the same reliability as physical measurements.

Reliability Coefficient: (3 categories) The method chosen depends on the test itself and the

conditions under which it is administered.

1. Test-retest (TRT) method: same group, two occasions. Scores are then compared using

correlation. The time interval might be few hours to several years, the longer the time the

more TRT reliability will decline. Dif. in administration will introduce error.

○ Ex. Personality Assessment Inventory (PAI) Interval M = 24 days, 75 normal

adults

○ Practice Effect (cons): when test takers benefit from taking the test the first time,

which enables them to solve problems more quickly. TRT is appropriate when test

takers are NOT likely to learn something or when interval time is long enough to

prevent practice effect.

2. Alternate forms(AF) method: two forms of the same test, same people. Time interval is

as close as possible, usually on the same day. Parallel test might not be truly equivalent.

Much easier to dev. For well-defined characteristics (i.e. math ability) not for personality

traits. No practice effect.

○ Ex. Test of Nonverbal Intelligence, ed. 4 (TONI-4) Does not req. Language,

answers are given by nod, blink, or point.

○ Order Effect: changes in test scores resulting from the order in which tests were

taken.

3. Internal consistency method: (1 test, 1 group) measure of how related the items (or

group of items) on the test are to one another, how they are measuring the same/similar

attribute. Requires a large sample.

○ Split-half method: divides the tests into ½, equivalent in length and content, &

compares the set of ind. scores on the first ½ to the second ½. Book: use random

Showing Page:

2/28

assignment to place the Q inside the test. Professor: odd/even number method, 1-

50/51-100 method. Shortening the test reduces reliability.

Note:

Homogeneous tests = measures only 1 characteristic or trait.

Heterogeneous tests = measures more than 1 characteristic or trait. Internal consistency might

be lower, a solution is separating the test by traits.

○ Ex. PAI scale has 5 options, they used the Crombach 𝛂.

4. Scorer reliability: an ind. can make mistakes in scoring, 2 or more should score the

test.Conducting studies of scorer reliability for a test ensures that the instructions are clear

so that multiple scorers arrive to the same results faster.

○ Interscorer agreement: amount of consistency among scorers’ judgment/

○ Intrascorer reliability: each scorer internal consistency.

○ Ex. Wisconsin Card Sorting Test (WCST) tests preservation 7 abstract thinking.

Also Bailey Scales of Infant & Toddler Dev. far exceeded accepted guidelines.

● Describe how an observed test score is made up of the true score and random error,

and describe the difference between random error and systematic error.

Classical Test Theory: any test score that a person obtains can be considered to be made of

two parts.

X(obtained/observed score) = T(true score) + e(error)

1. True Score: a measure of an attribute that the test measures, it cannot be really known. If

an ind. took a test infinite # of times & his scores were averaged.

2. Random error (RE): dif. b/t a person’s actual score X and his true score T. In a single

examination it’s impossible to know its effects. B/c it’s random, over infinite measures its

M = 0. This is why making a test longer, reduces the influence of error.

a. Systematic error(SE): a single source of error that increases the T by the same

amount in every measure. Difficult to identify. Unlike random error, SE does not

lower the reliability of a test. Practice and order effect can add systematic &

random error to test scores.

Reliability Coefficient (RC) (Pearson product moment correlation)

We use correlation to provide an index of the strength of the relationship b/t two sets of scores

(symbol r

, always two subscripts of the same letter)

● Differentiate between the KR-20 and coefficient alpha formulas, and understand

how they are used to estimate internal consistency

Adjusting Split-half Reliability

The # of questions relates to reliability, since the test is split in two, RC can be adjusted with the

Showing Page:

3/28

End of Preview - Want to read all 28 pages?

Access Now

Unformatted Attachment Preview

Psychological Testing ___ Exam 2 MILLER CH 6: RELIABILITY/PRECISION ● Define reliability/precision, and describe three methods for estimating the reliability/precision of a psychological test and its scores. Reliability: consistency of test scores. All test contain some error. More consistency of test scores, more reliability/precision Results of the statistical evaluation of reliability > reliability coefficient ● Measurement Error: variations on the measurements of the room ● Reliable Test: one we can trust to measure e/person in approx. the same way in every way it’s used. Psychological tests don’t have the same reliability as physical measurements. Reliability Coefficient: (3 categories) The method chosen depends on the test itself and the conditions under which it is administered. 1. Test-retest (TRT) method: same group, two occasions. Scores are then compared using correlation. The time interval might be few hours to several years, the longer the time the more TRT reliability will decline. Dif. in administration will introduce error. ○ Ex. Personality Assessment Inventory (PAI) Interval M = 24 days, 75 normal adults ○ Practice Effect (cons): when test takers benefit from taking the test the first time, which enables them to solve problems more quickly. TRT is appropriate when test takers are NOT likely to learn something or when interval time is long enough to prevent practice effect. 2. Alternate forms(AF) method: two forms of the same test, same people. ...
Purchase document to see full attachment