8747874
Description
Flashcards by Adelia Wilder-Doctor, updated more than 1 year ago
|
Created by Adelia Wilder-Doctor
about 7 years ago
|
|
| Question | Answer |
| Sources of Measurement Error | Sampling of Items Sampling of Occasions (inconsistencies over time) Inconsistencies in judgement and scoring Differences in administrative conditions |
| Reliability expresses... | the degree of consistency |
| Error causes____ , so the greater the ____, the lower the ____ | inconsistency error reliability |
| Reliability traditionally focuses on _____ | inconsistency from measurement error |
| "true scores" are free of ____ | measurement error |
| This is the standard deviation of these multiple measures—shows how much inconsistency | Standard error of measurement (SEM) |
| What are the advantages of SEM (standard error of measurement)? | On same scale as the measurement, therefore can provide error bands for score reports. |
| What are "compositional" effects? | An alteration of trends caused by change in the make-up of the test-taking group Not the change in composition, but its effect on scores or trends |
| Why do compositional effects happen? | Happens when subgroups differ in level of performance Can be any change in trends: An increase or decrease No change, when there would have been a change without it |
| What are some examples of changes that could create compostional effects? | Changes in selectivity of SAT group Demographic change (e.g., immigration) Changed inclusion rules (e.g., LEP, students with disabilities) |
| What are some plausible explanations for score differences? | Differences in educational quality Differences in income and wealth (e.g., poverty rates) Demographic differences Differences in transience Differences in LEP rates Regional and other cultural differences Differences in score inflation (almost always ignored) |
| In traditional measures, error can have... | several sources (e.g. time, items), but is always considered random. Any given measure does not separate sources. |
| In generalizability theory, error can have... | Error can have several sources, both random and systematic Analysis separates impact of all sources of error that are considered |
| generalizability theory | Lets you specify which sources of error you want to consider (time, task, rater) |
| convergent discriminant evidence |
Want to create your own Flashcards for free with GoConqr? Learn more.