Term
| Standardization sample, AKA normative sample |
|
Definition
| Define your target population and obtain a representative sample from each of the target groups. May be appropriate to have multiple standardization samples, considering age, race, geographic regions, etc. |
|
|
Term
|
Definition
| Simple random sampling, systematic sampling, stratified random sampling, cluster sampling |
|
|
Term
| Sampling techniques- simple random |
|
Definition
| every member of the population has an equal chance of being sampled |
|
|
Term
| Sampling techniques- systematic sampling |
|
Definition
| pick every 5th, 10th, 20th, etc |
|
|
Term
| Sampling techniques- stratified random sampling |
|
Definition
| population is divided into subgroups/strata based on demographics |
|
|
Term
| Sampling techniques- cluster sampling |
|
Definition
| Used when target population is large and when it is not feasible to list all individuals. Clusters are selected and participants are selected from each cluster (difference from stratified is that this is regional, not demographic) |
|
|
Term
|
Definition
| Developmental and within group norms |
|
|
Term
|
Definition
| age norms, grade norms, etc. This allows us to determine whether an individual's test score is similar to, below, or above the average of others at the same age or grade level |
|
|
Term
|
Definition
| make comparison for one person against the rest of the group. Use standardized/derived scores to provide meaning (percentile, z scores, t scores, deviation IQ, SAT GRE |
|
|
Term
|
Definition
(B/N)x100
B=number of observations with lower values
N= total number of observations |
|
|
Term
|
Definition
In psychology, reliability refers to stability of scores over time and consistency of a test over time. Is my score at time one the same as time two? Are all of my items assessing the same construct? A reliable test will measure each person in approximately the same way every time.
Higher reliability= more we are confident that scores are due to true differences |
|
|
Term
| Sources of random error that could affect reliability |
|
Definition
| testing conditions, energy level of test taker, motivation of test taker, etc |
|
|
Term
| 3 main measures of reliability to know |
|
Definition
| test-retest, equivalent forms, internal consistency (split half, inter item) |
|
|
Term
|
Definition
Gives coefficient of stability. Correlation of scores on the same test taken on two separate occasions. Assume that true ability doesnt change, but errors do.
Advantages- good for measuring stable constructs
Disadvantages- People may change from time one to time two. Carryover effects. Requires two administrations |
|
|
Term
| Equivalent forms reliability |
|
Definition
Gives coefficient of equivalence, also known as parallel forms reliability. Accomplished by administering one test, waiting, and giving alternate form to same group.
Advantages- less time between administrations. Fewer carryover effects, because we are not using the same Q's.
Disadvantages- must develop two tests instead of one. Still requires two test administrations |
|
|
Term
| Internal consistency reliability |
|
Definition
| gives us information on whether items on test are measuring the same thing. |
|
|
Term
| Internal consistency reliability- split halves |
|
Definition
give the test to one group, then split the items up and correlate scores on both halves
Advantages- only one administration solves carryover problem
Disadvantages- not confident that even and odd items are equal. by splitting test in half, we lose information. Doesnt tell us about stability of test scores over time.
Our reliability coefficient here is underestimated, therefore we correct with the Spearman Brown formula |
|
|
Term
| Internal consistency reliability- inter item correlation |
|
Definition
Instead of splitting the test in half, we treat each item as its own equivalent form. More exact than split halves and no correction is needed.
Two different ways to calculate- coefficient alpha and KR-20
Adv- same as split halves
Dis- doesnt tell us about stability of scores over time |
|
|
Term
| Reliability coefficient of .90 |
|
Definition
| 90% of variance observed is due to true ability |
|
|
Term
| Acceptable reliability level |
|
Definition
|
|
