45063
| Frage | Antworten |
| Procedural aspects affecting Reliability | Conditions under which a test is taken; Inconsistent test administration practices; Poor test-scoring practices: Inappropriate norm groups applied; Internal state of test-taker: Experience level of test-taker. |
| Reliability - Stability of results of tests | Test-retest stability (i.e. repeating test - how long? generally >= 3 months); Carry-over effects (questions and responses remembered); Practice Effects (methodology honed); Attrition (failure to attend second sitting) |
| Reliability - Consistency 1 (of results of tests) | Split-half testing (either sequentially, or by odd/even numbered questions); Spearman-Brown formula to correct for error of artificially reduced reliability, due to resulting lower effective sampling size |
| Reliability - Consistency 2 (of results of tests) | Kuder-Richardson Reliability aka "Coefficient Alpha". Assesses inter-item consistency by considering: a) content sampling adequacy, & b) heterogeneity of domain being sampled |
| Reliability - Equivalence of results (parallel form) 1 | Equivalence between different versions of the same test. Danger of "Bloated Specifics" (Cartell, 1973) - reduced criterion coverage by manipulation having several items of essentially same theme, giving artificially raised impression of reliability - so now less valid! |
| Reliability - Equivalence of results (parallel form) 2: other considerations | Parallel versions of same test today facilitated by computer cloning. Can be useful, because in "milk-round" unfair advantages eliminated, as everyone will have done several! Also note equivalence and "adverse impact", discrimination against any particular population categories. (e.g. blacks and cognitive ability tests.) |
Möchten Sie mit GoConqr kostenlos Ihre eigenen Karteikarten erstellen? Mehr erfahren.