Upgrade to remove ads
EDPS 514 Final S18
Terms in this set (51)
score the child should have received without any error
the true score plus error (the raw/obtained score) (raw score is meaningless by itself for the WISC or WIAT, we need to look at tables and convert to standard score)
consistency or stability of assessment results (how stable the test is in repeated uses)
want scores greater than or equal to .80
.90 for IQ tests
How to improve reliability
increase the length of the test & increase # of items to ensure you are sampling the content
extent to which a measure varies from one use to another
Test re-test Reliability
measures the stability of a test over time by administering the test to the same group of people (within 30 days). Time frame eliminates practice effects
degree to which different raters give consistent estimates of the same behavior
two forms administered in the same session (or slight delay) to the same group
extent to which a measure is consistent within itself
measure extent to which all parts of the test contribute equally to what is being measured (internal reliability)
measurement of scale reliability and discovers how closely related a set of items are as a group
similar to coefficient alpha but used for measures with dichotomous choices. A score of greater than 90 indicates a homogeneous test.
does the instrument measure what it is supposed to measure?
EX: a test of intelligence should measure intelligence, not memory
Is the content appropriate?
Whether the test appears to measure what it claims to at face value. Type of content validity.
Extent to which a test relates to underlying specific theoretical constructs/traits. Type of content validity
Criterion related validity
the relationship to other measures
degree to which a test relates to an existing similar measure? Type of criterion-related validity.
degree to which a test accurately predicts later performance on a related criterion. Type of criterion-related validity.
Factorial Validity (internal structure)
degree to which covariance among judged traits resembles the actual or true covariation of observable behaviors underlying these traits. The clustering of correlations of responses by groupings of items in a questionnaire.
Why use multiple types of reliability?
So that you have greater confidence in an instrument.
Standard Error of Measurement (SEM)
estimates how repeated measures of a person on the same instrument tends to be distributed is the person's true score (mean of this distribution is the person's true score). Similar to standard deviation of distribution of scores if it was measured infinite times.
Confidence Interval (CI)
range of values in which the true score falls somewhere in between. Set the CI to create the interval (95%). higher confidence=wider interval.
Relationship between SEM and CI
Use SEM to create CI
Exploratory Factor Analysis (EFA)
statistical method used to uncover the underlying structure of a relatively large set of variables (technique within factor analysis). Often used for new tests which helps guide the confirmatory analysis. Important because the factor structure that emerges determines how the test will be interpreted.
Confirmatory Factor Analysis (CFA)
Statistical technique used to verify the factor structure of a set of observed variables. CFA allows researchers to test the hypothesis that a relationship between observed variables and their underlying latent constructs exists.
Woodcock Johnson Cognitive (EFA & CFA)
WJ achievement test does not have adequate structural validity. It was never subjected to separate EFA or CFA. It was extrapolated from the broader WJ IV which itself has problems as the approach to CFA and EFA was criticized
EFA & CFA of WJ IV Cognitive Article
there were considerable problems with choices within both EFA & CFA when analyzing the full WJ IV battery. They used two EFA procedures but did not present the results, did not include rudimentary EFA stats, and problems with CFA procedures (examined only 3 competing models, results showed poor model fit on several stats). Results: WJ IV should be 4 factor test (not 7). EFA indicates WJ IV cognitive is a measure of general intelligence as it accounts for a majority of the subtests' total and common variance. LMAO good luck, idk this one either.
Eye of the beholder issue when rating child behavior
Parents can be biased or unwilling to disclose personal information, child may act differently in various settings, makes ratings more subjective. You should compare ratings from various sources and observe yourself when possible.
Who are the best raters of children's behavior?
teachers; better at recognizing externalizing behaviors than parents
Importance of adhering to standardization administration procedures
standardization procedures are set rules on how test should be given. Each examinee needs to read the instructions the same way because this can affect results and how the test taker interprets the test. Not reading the instructions properly makes the test & scores invalid.
How should one interpret scores where standard protocol was not followed?
Make sure it's properly addressed (?)
a bell-shaped curve that represents the probability of a student receiving a particular score. Performance compared to others
Normal Curve (reference notes)
be able to label your chart!
Giftedness (normal curve)
standard score of 130 (~3% of the population)
Clinically significant (normal curve)
T-score of 70 or higher (symptoms more severe than 97.5% of the population)
Standard score of 70 or less (2 SD below mean)
relevant for standard scores. 25th - 75th percentile. 90-110 standard score is the average range.
VMI & IQ
sufficient VMI skills are a prerequisite for IQ testing as weak VMI skills may contribute to lower than average IQ scores
Correlations between instruments
close to 0 = weak/no relationship between instruments
close to 1 = stronger relationship between instruments
r= .10-.29 small (weak)
r=.5-.1 large (strong)
p must be significant
IQ test sufficient in determining Intellectual disability?
No, was prone to classifying African American students incorrectly (Larry P VS Riles). We need to look at adaptive behavior
Correlation between IQ and behavior
we want to ensure that these tests are related but not the same assessment
Vineland (adaptive behavior)
communication, daily living skills, socialization, motor
ABAS-III (adaptive behavior)
communication, community use, functional academics, home living, health and safety
EFA Class Notes
internal validity or factorial validity. If it is a new instrument/significantly revised instrument, use to guide before confirmatory analysis
EFA correlation matrix
Factor structure that determines how the instrument should be interpreted
Is EFA appropriate?
KMO & Bartlett's test of sphericity
EFA & CFA important statistics
variance, communality, omega, H
CFA Class Notes
specify parameters, guided by theory, may be adjusted, fit stats to comparison models
Post-hoc tinkering via specification of crossloadings, specification of correlated residuals, and constraint of variances and other parameters to 0 or 1 to "force" the model to fit or to achieve superior fit statistics
YOU MIGHT ALSO LIKE...
Research and Measurement Methods
RELIABILITY AND VALIDITY
OTHER SETS BY THIS CREATOR
EDPS 535 Study Guide
School psychology comps
EDPS 513 Class Notes
EDPS 513 Terms and Important Concepts