jMetrik Features
Features and options included with each method of analysis are briefly described below. This list is not exhaustive but the main features are highlighted.
Descriptive Statistics
- Minimum
- First quartile
- Median
- Mean
- Third quartile
- Maximum
- Standard deviation
- Interquartile range
DIF Analysis
- Currently based on Mantel-Haenszel procedure
- Effect size computation (ETS Delta or standardized mean difference)
- ETS DIF classification
- Nonparametric IRT option charateristic curves illustrate DIF for each item and the test characteristic curve
Frequency Analysis
- Frequency and proportion endorsing a response
- Cumulative frequency and cumulative relative frequency
Graphs
- Pie chart
- Bar chart
- Histogram
- Line chart
- Scatterplot
- Kernel density estimates
- Wright Map
- Nonparametric ICCs
- IRT plots (ICC, TCC, Information, Standard error)
IRT Scale Linking and Score Equating
- Handles IRT linking for binary and polytomous items
- Supported linking models include any combination of Rasch, 2PL, 3PL, PCM, RSM, GRM, or GPCM
- Mean/mean, mean/sigma, Haebara, Stocking-Lord procedures
- Fixed common item parameter estimation for Rasch, partial credit, and rating scale models
- Robust z statistic for item parameter drift
- Item parameter correlations
- Save transformed person and item values to database
- IRT true score equating using any combination of Rasch, 2PL, 3PL, PCM, RSM, GRM, or GPCM
Item Analysis
- Proportion endorsing each response option
- Option standard deviation
- Flexible scoring allows user to assign score value to each option
- Partial credit and multiple correct answers allowed
- Point-biserial correlation
- Biserial correlation
- Test score descriptive statistics (see list of statistics above)
- Reliability methods include Guttman’s lambda 2, Cronbach’s alpha, Feldt Gilmer coefficient, Feldt-Brennan coefficient, and Raju’s beta
- For a given method of reliability, the reliability and standard error of measurement if item deleted is optionally provided
- 95% confidence interval provided for Cronbach’s alpha
- Standard error of measurement
- Conditional standard error of measurement at cut-score
- Huynh’s decision consistency and raw agreement indices and standard errors
Item Response Theory
- Rasch, partial credit, and rating scale models
- Joint maximum likelihood estimation
- MLE, MAP, and EAP estimation of examinee ability
- Missing data allowed and handled by ignoring it or scoring it as zero
- Save item parameter estimates to database
- Save person parameter estimates to database
- Item fit and scale quality statistics
- Item map graph
New in Version 4.0!
- Marginal maximum likelihood estimation of item parameters
- 4PL, 3PL, and 2PL item response models for binary items
- Generalized partial credit model and partial credit model for polytomous items
Test Scaling
- Sum score, average score, percentile ranks, Kelley regressed scores, and normalized scores
- Linear transformations of scale scores
- Save scale scores to database