Launch the Item Analysis Platform

Launch the Item Analysis platform by selecting Analyze > Multivariate Methods > Item Analysis.

Figure 11.4 Item Analysis Launch Window 

For more information about the options in the Select Columns red triangle menu, see Column Filter Menu in Using JMP.

Y, Test Items

Assigns two or more columns to be analyzed. The columns must be numeric, continuous, and contain only 0s and 1s.

Tip: Use Cols > Recode if you need to recode your data to 0s and 1s. See Recode Data in a Column in Using JMP.

Freq

Assigns a frequency variable to this role. This is useful if your data are summarized.

By

Produces a separate report for each level of the By variable. If more than one By variable is assigned, a separate report is produced for each possible combination of the levels of the By variable.

Model

Specifies the desired model from the following options:

Logistic 2PL

The 2-parameter logistic model.

Logistic 3PL

The 3-parameter logistic model.

Logistic 1PL

The 1-parameter logistic model with a Rasch parameterization.

Logistic 3PL Model Details

If you select Logistic 3PL for Model, you are prompted to enter a penalty for the guessing parameters after you click OK. For a 3PL model, the default value of the penalty is zero. However, you can enter a non-zero penalty for the c parameters (the guessing for each item). This penalty is similar to the type of penalty parameter that you would use in ridge regression. The penalty is on the variance of the estimated guessing parameters. The use of the penalty has the following benefits:

• Stabilizes the estimation of model parameters.

• Speeds up computations.

• Reduces the variability of the guessing parameter across items at the expense of some bias.

Large values of the penalty force the guessing parameters to zero while smaller values help reduce the variability of the guessing parameter across items. A value of zero can be used for no penalty.

Data Format

The Item Analysis platform requires a data table that contains a row for each individual and a column for each item. The item columns must be numeric and contain only 0s and 1s to indicate incorrect or correct responses, respectively. The MathScienceTest.jmp sample data table illustrates the required data format for an item response analysis of 1,263 individuals responses to 14 test questions.