Test Construction Flashcards
An examiner administers and scores the same test numerous times without deviating from the procedure in order to reduce the possibility of measurement error. This exemplifies what?
Standardization
The scores of a representative population sample on a test that an examiner compares an individual's scores to are referred to as \_\_\_\_\_\_\_\_; while they allow for comparisons on a person's performance on different tests, they do not provide the ultimate standard of performance.
Norms
A psychological test that is regarded as \_\_\_\_\_\_\_\_ is administered, scored, and interpreted independent of the subjective judgment of the examiner.
Objective
The SAT and GRE are examples of \_\_\_\_\_\_\_\_ tests, as they provide information about a person's best possible performance, while the MMPI-2 and PAI are \_\_\_\_\_\_\_\_ tests, providing information about a person's usual experience.
Maximum
performance;
typical
performance
________ tests assess the difficulty level
an examinee can attain (e.g., Information
from WAIS), ________ tests assess the
person’s response rate (e.g., Digit
Symbol from WAIS), and ________ tests
help determine whether an individual can
attain a certain level of acceptable
performance (e.g., test of reading skills).
Power;
speed;
mastery
A \_\_\_\_\_\_\_\_ occurs when an instrument cannot take on a value higher than some limit due to the measure not including enough difficult items, resulting in all high-achieving examinees getting similar scores (test is too easy); conversely, a \_\_\_\_\_\_\_\_ occurs when an instrument cannot take on a lower value and thus all low-achieving examinees get similar scores (test is too hard).
Ceiling
effect; floor
effect
In contrast to normative measures, these types of measures require individuals to use their own frame of reference to compare 2 or more desirable options and choose the one that is most preferred.
Ipsative
measures
\_\_\_\_\_\_\_\_ is the consistency of a test, or the degree to which a test provides the same results under the same conditions; \_\_\_\_\_\_\_\_ refers to the degree that a test measures what it claims to be measuring.
Reliability;
validity
A perfectly reliable test would yield every
examinees’ ________ every time it was
administered, as this would indicate the
examinees’ actual ability on whatever the
test is measuring; however, a test is
never perfectly reliable due to ________,
which is random and can be caused by
environmental noise, examinee’s mood
on testing day, and any other number of
factors.
True score;
measurement
error
The most commonly used methods of estimating
reliability of a test use a correlation coefficient,
referred to as the ________, ranging in value
from 0.0 to +1.0, where coefficients closer to 0.0
indicate less reliability and values closer to +1.0
indicate increasing reliability; the coefficient is
not squared to determine the proportion of
variability, unlike other correlation coefficients,
rather it is interpreted directly.
Reliability
coefficient
A researcher administers the same instrument to the same group of college students on 2 separate occasions; following the second administration, the researcher correlates on the first and second administrations. What type of reliability is the researcher attempting to obtain?
Test-retest
reliability (or
“coefficient of
stability”)
TRUE or FALSE: It is not recommended to use the test-retest coefficient when attempting to obtain reliability for a test that measures attributes that are unstable (e.g., mood).
TRUE: Low coefficients, in such cases, would likely be more a reflection of the attribute's unreliability rather than the test's unreliability
A researcher administers one form of a test on one day, then administers an equivalent form to the same group of people at a later date/time. What type of reliability is being sought in this example?
Alternate forms
reliability (or “coefficient
of equivalence;”
parallel-forms reliability)
When correlations are obtained among individual
test items, ________ reliability is being
assessed; the 3 methods for obtaining this
reliability include ________ (involves dividing
test into 2 parts then correlating responses from
the 2 parts), ________ (used when test items are
dichotomously scored- e.g., “true/false”), and
________ (used for tests with multiple-scored
items- e.g., “never/rarely/sometimes/always”).
Internal consistency (or "coefficient of internal consistency"); split-half; Kuder-Richardson Formula 20; Cronbach's coefficient alpha
While the split-half reliability coefficient usually lowers the reliability coefficient artificially, the \_\_\_\_\_\_\_\_ can be used to correct for the effects of shortening the measure.
Spearman-Brown
prediction formula
Measures of internal consistency are not good at assessing reliability for \_\_\_\_\_\_\_\_ tests.
Speed tests, as the
correlation would
be spuriously
inflated
Instruments that rely on rater judgments would be best to have high \_\_\_\_\_\_\_\_ reliability, which is increased when scoring categories are \_\_\_\_\_\_\_\_ and \_\_\_\_\_\_\_\_.
Inter-rater (interscorer); mutually exclusive (a particular behavior belongs to a single category); exhaustive (categories cover all possible responses/behaviors)
The \_\_\_\_\_\_\_\_ estimates the amount of error to be expected in an individual test score and is used to determine a range, referred to as a/an \_\_\_\_\_\_\_\_, within which an examinee's true score will likely fall.
Standard Error of
Measurement;
confidence
interval
What is the
formula for the
standard error of
the measurement?
SDx√1-rxx (SDx = standard deviation of test scores; = reliability coefficient)
What is the probability that a person's true score lies within a range of plus or minus 1 standard error of measurement (SEM) of their obtained score? How about plus or minus 1.96 (2) SEM? And finally, plus or minus 2.58 (2.5) SEM?
68% of the
time; 95% of
the time; 99%
of the time
TRUE or FALSE: Hypothetically, a test with a reliability coefficient of +1.0 would have a standard error of measurement of 0.0.
TRUE: A test
with perfect
reliability will
have no error
The standard error of measurement is \_\_\_\_\_\_\_\_ related to the reliability coefficient (rxx) and \_\_\_\_\_\_\_\_ related to the standard deviation of test scores (SDx).
Inversely;
positively
What reliability
coefficient, when
practical, is the
best to use?
Alternate-forms
Classical test theory states that an observed score reflects \_\_\_\_\_\_\_\_ plus \_\_\_\_\_\_\_\_.
True score
variance;
random error
variance
Methods of recording behaviors include \_\_\_\_\_\_\_\_ recording (elapsed time that behavior occurs is recorded), \_\_\_\_\_\_\_\_ recording (number of times behavior occurs is recorded), \_\_\_\_\_\_\_\_ recording (rater notes whether subject engages in behavior during given time period), and \_\_\_\_\_\_\_\_ recording (all behavior during an observation session is recorded).
Duration;
frequency;
interval;
continuous
Simply put, \_\_\_\_\_\_\_\_ refers to the degree a test measures what it purports to measure.
Validity
A depression scale that only assesses the affective aspects of depression but fails to account for the behavioral aspects would be lacking what type of validity?
Content validity, which refers to the extent to which test items represent all facets of the content area being measured (e.g., EPPP)
TRUE or FALSE: Content validity assessment requires a degree of agreement between experts in the subject matter, thus it includes an element of subjectivity.
TRUE: Tests should also correlate highly with other tests that measure the same content domain
In contrast to content validity, \_\_\_\_\_\_\_\_ occurs when a test appears to valid by examinees, administrators, and other untrained observers; it is not technically a type of test validity.
Face
validity
A personality test that effectively predicts the future behavior of an examinee has what type validity?
Criterion-related validity, which is obtained by correlating scores on a predictor test to some external criterion (e.g., academic achievement, job performance)
Criterion-related validity is assessed using a/an \_\_\_\_\_\_\_\_ to determine the relationship between the predictor and the criterion; for interpretation this value can be squared, producing the "\_\_\_\_\_\_\_\_," which indicates the proportion of variability in the criterion that is explained by variability in the predictor.
Correlation
coefficient;
coefficient of
determination
The process of \_\_\_\_\_\_\_\_ validation involves the predictor and the criterion being collected at the same time, providing information regarding a test's usefulness for predicting a given current behavior; \_\_\_\_\_\_\_\_ validation involves a waiting period between collection of predictor scores and criterion data, providing information regarding a test's usefulness for predicting future behavior.
Concurrent;
predictive
When interpreting a person's predicted score on a given criterion measure, the \_\_\_\_\_\_\_\_ will determine within what range of scores their actual score will likely fall.
Standard
Error of
Estimate
The standard error of measurement constructs a confidence interval around an examinee's \_\_\_\_\_\_\_\_ score (using a reliability coefficient), while the standard error of estimate does the same for an examinee's \_\_\_\_\_\_\_\_ score (using a validity coefficient).
Obtained;
predicted
Interviewees are given an aptitude test (predictor) to predict work success (criterion), with hiring contingent on achieving a certain minimum score, called a/an \_\_\_\_\_\_\_\_ score. The manager then rates performance on work tasks, an indication of success, and only those who score above a certain \_\_\_\_\_\_\_\_ are deemed successful.
Predictor
cutoff;
criterion cutoff
Scoring above both the predictor and criterion cutoff points produces \_\_\_\_\_\_\_\_; scoring above the predictor cutoff point but below the criterion cutoff point produces \_\_\_\_\_\_\_\_; scoring below the predictor cutoff point but above the criterion cutoff point produces \_\_\_\_\_\_\_\_; and scoring below both the predictor and criterion cutoff points produces \_\_\_\_\_\_\_\_.
True positives (valid acceptances); false positives (false acceptances); false negatives (invalid rejections); true negatives (valid rejections)
Some factors contributing to a low validity coefficient include the validation group being \_\_\_\_\_\_\_\_ or the predictor and/or criterion being \_\_\_\_\_\_\_\_.
Homogenous;
unreliable
When a test has a different validity coefficient for one group compared to another, the variables affecting validity are called \_\_\_\_\_\_\_\_ variables; when this is the case, the test is said to have \_\_\_\_\_\_\_\_.
Moderator;
differential
validity
This is the process whereby an already validated test is re-validated with a different sample of people than the original validation sample.
Cross-validation
What term is used to describe the reduction that occurs in a criterion-related validity coefficient after cross-validation?
Shrinkage
The greatest shrinkage occurs when the original validation sample is \_\_\_\_\_\_\_\_, the original item pool is \_\_\_\_\_\_\_\_, the number of items retained is \_\_\_\_\_\_\_\_ relative to the items in the item pool, and/or item are not chosen based on \_\_\_\_\_\_\_\_ or \_\_\_\_\_\_\_\_.
Small; large; small; previously formulated hypothesis; experience with the criterion
\_\_\_\_\_\_\_\_ is one way a predictor might end up looking more valid than it actually is, which occurs when predictor scores themselves influence any person's criterion status (e.g., manager is aware that factory worker did well on predictor, this knowledge positively influences manager's ratings on criterion performance).
Criterion
contamination
How is criterion
contamination
prevented?
Criterion raters should have no prior knowledge of examinees' predictor scores
Theorized psychological variables (e.g., personality, intelligence) that are abstract and not directly observable are referred to as \_\_\_\_\_\_\_\_, hence \_\_\_\_\_\_\_\_ provides an indication of the degree to which an instrument measures or correlates with such variables.
Construct;
construct
validity
A newly developed test of personality has a high correlation with the MMPI-2 and a low correlation with the Wechsler Memory Scale, indicating the test has both \_\_\_\_\_\_\_\_ validity and \_\_\_\_\_\_\_\_ validity, respectively.
Convergent;
discriminant/divergent
- both are forms of
construct validity
TRUE or FALSE: The only time a low correlation coefficient provides evidence of high validity is when discriminant validity is indicated due to there being a low correlation between 2 tests that measure different constructs.
TRUE: In all other cases, high validity is indicated by a high correlation coefficient
What complex procedure for assessing convergent and discriminant validity requires the assessment of 2 or more traits (e.g., personality, depression) by 2 or more methods (e.g., self-report, peer rating)?
Multitrait-multimethod
matrix
When using the multitrait-multimethod
matrix, ________ validity is indicated
when tests that measure the same traits
are highly correlated, even when different
methods of measurement are used;
conversely, ________ validity is indicated
when tests that measure different
constructs are minimally correlated, even
when the same method of measurement.
Convergent;
discriminant
The ________ coefficient is a reliability
coefficient, as it indicates the correlation between
itself and the measure; correlations between two
measures that measure the same trait using
different methods are called ________
coefficients; correlations between two measures
that measure different traits using the same
method are called ________ coefficients; and
correlations between 2 measures that measure
different traits using different methods are called
________ coefficients.
Monotrait-monomethod;
monotrait-heteromethod;
heterotrait-monomethod;
heterotrait-heteromethod
When assessing validity using the multitrait-multimethod matrix, convergent validity is indicated when there is a high \_\_\_\_\_\_\_\_ correlation, while discriminant validity is indicated by a low \_\_\_\_\_\_\_\_ correlation and further confirmed by a \_\_\_\_\_\_\_\_ heterotrait-heteromethod correlation.
Monotrait-heteromethod;
heterotrait-monomethod;
low
\_\_\_\_\_\_\_\_, often used to assess the construct validity of a test or tests, involves reducing a larger set of variables into fewer classified sets of variables based on the construct that is primarily "picked-up" by each measure; each variable is correlated with every other variable, creating a \_\_\_\_\_\_\_\_
Factor
analysis;
factor matrix
The main purpose of factor analysis is to reveal how many and to what degree underlying constructs, also called \_\_\_\_\_\_\_\_ due to the fact that the analysis does not directly intend to measure them, can account for scores on a larger number of tests.
Latent
variables
In a hypothetical factor analysis, the factor matrix indicates a correlation coefficient of .68 between the depression subscale of the MMPI-2 and Factor II. What term is used to describe the correlation between the depression subscale and Factor II?
Factor loading, which refers to the correlation between a given test and a given factor (e.g., the depression subscale loads .68 on Factor II); it can be square to determine proportion of variability
\_\_\_\_\_\_\_\_ determines the proportion of variance of a test that is attributable to the factors; it is the sum of squared factor loadings.
Communality (h-squared) - not the case when oblique rotation is used
The amount of variability in a test that can be explained by whatever traits are represented by the factors is referred to as \_\_\_\_\_\_\_\_, while variance that is specific to the test and not explained by the factors is referred to as \_\_\_\_\_\_\_\_.
Common variance (represents communality); unique variance (represents specificity)
In a factor analysis, these values indicate the amount of variance in all the tests accounted for by the factor; they are analyzed to determine whether or not the factor is accounting for a significant amount of variability in the tests.
Eigenvalues
(or explained
variance)
If a factor analysis is performed on 8 tests, what is the largest the sum of the eigenvalues can be?
Since the sum of the eigenvalues can be no larger than the number of tests included in the factor analysis, the answer is 8
A procedure that facilitates factor matrix interpretation is \_\_\_\_\_\_\_\_, which involves re-dividing the test's communalities so that a clearer pattern of loadings emerges.
Rotation
Two general rotation strategies include \_\_\_\_\_\_\_\_ for factors that are uncorrelated (independent of each other) and \_\_\_\_\_\_\_\_ for correlated factors; the decision as to which one is used is based on the researcher's theoretical assumptions.
Orthogonal;
oblique
When construct validity is being assessed using factor analysis, a high correlation between a test and a factor the test is expected to correlate highly with is referred to as what?
Factorial
validity
While factor analysis assumes variance in a variable is composed of \_\_\_\_\_\_\_\_, \_\_\_\_\_\_\_\_, and \_\_\_\_\_\_\_\_, principle components analysis assumes variance is composed of \_\_\_\_\_\_\_\_ and \_\_\_\_\_\_\_\_.
Communality; specificity; error; explained variance; error variance
Factor is to factor analysis as \_\_\_\_\_\_\_\_ or \_\_\_\_\_\_\_\_ is to principal components analysis.
Principal
component;
eigenvector
What method might a researcher who is interested in developing a taxonomy (classification system) of different personality characteristics use?
Cluster
analysis
In \_\_\_\_\_\_\_\_ analysis, only interval and ratio data can be used and researchers typically have an a priori hypothesis about what traits a set of variables measure; by contrast, \_\_\_\_\_\_\_\_ can be performed using any type of data (interval, ration, nominal, ordinal) and is not designed for studies where the researcher has an a priori hypothesis.
Factor analysis;
cluster analysis
TRUE or FALSE: A reliable test is not always a valid test, though a valid test must be a reliable test.
TRUE: Reliability is a necessary but not sufficient condition for validity
The \_\_\_\_\_\_\_\_ coefficient is less than or equal to the square root of the \_\_\_\_\_\_\_\_ coefficient; it cannot be any higher, thus the latter sets a \_\_\_\_\_\_\_\_ on the former.
Validity;
reliability;
ceiling (or
upper-limit)
A researcher discovers a test has low reliability; however, she is interested in what the validity coefficient of the predictor would be if both the predictor and the criterion were perfectly reliable. What formula would she use?
Correction
for
attenuation
What is the correlation between the factors in a factor analysis where an orthogonal rotation is used?
By definition,
the correlation
would be 0.0
What is used to determine which test items will be retained for the final version of a test and to ensure that a test is both reliable and valid from the start?
Item
analysis
The ________
the p-value, the
________ the
item.
Higher (lower);
less difficult
(more difficult)
The percentage of examinees that answer an item correctly is referred to as a/an \_\_\_\_\_\_\_\_, which is abbreviated \_\_\_\_\_\_\_\_; most test developers prefer items with a \_\_\_\_\_\_\_\_ value at or around \_\_\_\_\_\_\_\_.
Item difficulty
index; p; p;
.50
The rule-of-thumb for item difficulty on a
test is that the optimal difficulty level of
test items should be approximately
halfway between 1.0 (i.e., everyone is
correct) and the level of success
expected by chance alone. That known,
what is the optimal item difficulty level of
a multiple choice test with 4 options (e.g.,
EPPP)?
p = .625, which means there is a 62.5% chance of guessing the correct answer to an item
According to Anastasi, the p-level expresses item difficulty in terms of an \_\_\_\_\_\_\_\_ scale, as conclusions cannot be made about the differences in difficulty between items, only that certain items are easier/harder than others.
Ordinal (difficulty
level are rankings,
according to
Anastasi)
The degree to which a test item differentiates among test-takers in terms of the behavior the test is designed to measure is called \_\_\_\_\_\_\_\_ and can be assessed by calculating a/an \_\_\_\_\_\_\_\_, which is abbreviated as "\_\_\_\_\_\_\_\_."
Item
discrimination; item
discrimination
index; D
An item on a measure of anxiety would have good \_\_\_\_\_\_\_\_ if low-anxiety examinees consistently answered it differently than high-anxiety examinees.
Discriminability
(item
discrimination)
An item's \_\_\_\_\_\_\_\_ level places a ceiling on its \_\_\_\_\_\_\_\_ index; higher levels of discriminability are associated with \_\_\_\_\_\_\_\_ levels of difficulty.
Difficulty;
discrimination;
moderate
TRUE or FALSE: The reliability of a test will decrease as the mean discrimination index (D) increases.
FALSE: There is a direct correlation between test reliability and mean D
A graphical depiction of both item difficulty and item discrimination is called a/an \_\_\_\_\_\_\_\_; analysis based on \_\_\_\_\_\_\_\_ is derived from these.
Item
characteristic
curve (ICC); item
response theory
What are the 2 technical properties of an item characteristic curve that are used to describe it?
Item difficulty
and item
discrimination
Item response theory assumes (1) performance on an item is related to the estimated amount of a/an \_\_\_\_\_\_\_\_ being measured by the item, and (2) \_\_\_\_\_\_\_\_ (an item should have the same characteristics regardless of the sample of people taking the test).
Latent trait;
invariance of
item
parameters
The computerized selection of test items for individual examinees is referred to as what?
Computer
adaptive
assessment (or
testing)
What item difficulty level is associated with the maximum level of differentiation among examinees?
.50, indicating half
answered correctly
and half answered
incorrectly
What factor
most affects an
item’s difficulty
level?
Characteristics
of examinees
What type of interpretation indicates where the examinee stands in relation to others who have taken the same test?
Norm-referenced
interpretation
Providing a general indication as to the progression a person has made along the normal developmental path, \_\_\_\_\_\_\_\_ norms include \_\_\_\_\_\_\_\_ and \_\_\_\_\_\_\_\_.
Developmental;
mental age;
grade equivalent
scores
What is the
calculation
for ratio IQ?
(mental
age/chronological
age) x 100
A 20-year-old performs as well on a test as the average 10-year-old. His mental age is \_\_\_\_\_\_\_\_ and his ratio IQ is \_\_\_\_\_\_\_\_.
10-years-old;
50
Indicating the grade level a person's performance is equivalent to, \_\_\_\_\_\_\_\_ are typically used in the interpretation of educational achievement tests.
Grade equivalent scores (e.g., Wide Range Achievement Test, 4th Ed [WRAT-4])
TRUE or FALSE: When using developmental norms, scores obtained by people of different age groups are not comparable.
TRUE: This is due to the fact that standard deviation is not accounted for
Including percentile ranks and standard scores, \_\_\_\_\_\_\_\_ norms compare examinee scores to those of the most nearly comparable standardization sample.
Within-group
Z-scores, t-scores, stanine scores, and deviation IQ scores are all examples of \_\_\_\_\_\_\_\_, which express a raw score's distance from the mean in terms of standard deviation.
Standard
scores
Identify the mean (M) and standard deviation (sd) of: z-scores, t-scores, stanine scores, and deviation IQ scores.
Z-score (M = 0, sd = 1); T-score (M = 50, sd = 10); Stanine (M = 5, sd = about 2); Deviation IQ (M = 100, sd = 15)
