Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
6 Cards in this Set
- Front
- Back
Reliability
|
Reliability of an item is the ratio of the true score to the total (true + error) score variance
The greater the error the lower the reliability |
|
Internal Consistency Reliability:
Split Halves Reliability |
* Split scale in half (randomly) such that each part of the scale has half of the items.
* Compute the total score for each half based on the items. * Calculate the correlation between the two total scores. This is the split half reliability Cronbach’salpha is equivalent to the average correlation of ALL possible splits |
|
Internal Consistency Reliability:
Cronbach’sAlpha |
Internal Consistency Reliability:
Cronbach’sAlpha *The proportion of a scale’s total variance that is attributable to a common source * α = 1-error variance = 1 –ratio of unique variance to total variance * 0 <= α <=1 |
|
Cronbach’sAlpha Derivation
|
Cronbach’sAlpha
Adjusting for number of terms: alpha= k*(Sum of all covariancesijfor i≠j) -------- k-1* (Sum of all covariances i≠j)+ Sum of all variances) |
|
Reliability Guidelines
|
*
Alpha should be .70 or higher (.60 for exploratory); however many researchers insist on a stringent cut-off of .80 (Robinson et al;
Clark and Watson) - Exceeding .80 is of little utility to internal consistency or content validity * Inter-item correlations of - .30 or better (Robinson et al) - .15-.50 in general and .40-.50 for narrowly defined constructs * Number of items for narrowly defined constructs (i.e. 4-7 items) |
|
Composite Reliability
|
Theta-Delta
Should be > .70 (Hair et al) |