Term
|
Definition
| =(TP + FN)/Number of Decisions |
|
|
Term
|
Definition
| =(TP+TN)/Number of Decisions |
|
|
Term
|
Definition
|
|
Term
|
Definition
|
|
Term
| Positive Predictive Power |
|
Definition
|
|
Term
| Negative Predictive Power |
|
Definition
|
|
Term
|
Definition
=PPP-BR
or
=NPP-(1-BR)
or for overall accuracy:
=Hit Rate- (1-BR) |
|
|
Term
|
Definition
Monotrait-Monomethod coefficieints ()
reliability-->must be high in order for the matrix to provide useful info |
|
|
Term
|
Definition
Monotrait-Heteromethod coefficients []
High correlations provide evidence of convergent validity; aka validity diagonal
|
|
|
Term
|
Definition
Heterotrait-Monomethod coefficients <>
Low correlations provide evidence of discriminant validity.
High correlations suggest shared methods variance-->method "flavors" what it's measuring. |
|
|
Term
|
Definition
Heterotrait-Heteromethod __
Low correlations (<.3) provide evidence of discriminant validity.
High correlations (>.6)-->constructs might not really be different.
Should be higher than values in validity diagonal. |
|
|
Term
| Building a confidence interval Step 1 |
|
Definition
Calculate estimated True score
y= M + r(X-M)
M= test mean
r= reliability
X= obtained score |
|
|
Term
Building a confidence interval Step 2
|
|
Definition
Calculate Standard Error of the Estimate (SEest)
SEest= SDtest √[r(1-r)] |
|
|
Term
| Building a confidence interval Step 3 |
|
Definition
Calculate the CI band
95% CI= ±1.96(SEest) |
|
|
Term
| Building a confidence interval Step 4 |
|
Definition
| Add result from step 3 to True score, substract result from step 3 from True score |
|
|
Term
| Building a confidence interval Step 5 |
|
Definition
Look up percentiles and classification for CI band.
If M is not 100, and SD is not 15, calculate z-scores for CI band first. |
|
|
