Multilog: generating the syntax file

In this example, the generation of syntax that includes the reading of an external criterion is illustrated. For a complete discussion of the problem, please click here.

The first step in creating a new syntax file using the syntax wizard is to select the New option from the File menu to activate the New Analysis dialog box.

The type of problem and the name and location of the new syntax file are defined using the New Analysis dialog box. As we wish to score (estimate ) in this run, the MLE or MAP Computation option is selected in the Select type of problem group box. This selection corresponds to the SCORE option on the PROBLEM command.

The location in which the new syntax file is to be stored is specified next. By default, the folder in which MULTILOG has been installed will be displayed. This can be changed by either typing an alternative path in the Folder Location field or by using the Browse button to the right of this field. Finally, the name of the syntax file is entered in the File name field. In this case, we want to create the syntax file knee.mlg in the (default) mlgwin folder. Click OK to continue with the syntax specification.

The Fixed Theta dialog box is now displayed, allowing you to include the reading of a fixed value with the data. Click the radio button next to the Yes option to add the CRITERION option to the PROBLEM command. Then click Next to go to the Input Data dialog box.

The Input Data dialog box is used to provide information on the position and contents of the raw data file. By default, the Data file name will be assumed to be in the same folder and to have the same file name as the new syntax file. This may be changed by either correcting the entry in this field or by using the Browse button () to the right of this field.

The variable format statement describing the contents of this file must be entered in the Data Format field. The data are in the format:

40 10.5 2112111111112111111111111111111111
 33 1 1.0 3113211111112122111111111111111111
 33 1 2.0 4333211111113122111111111111111111
 29 1 3.0 4543211111113122111111011111111111

As the data file contains individual data, an identification field is required as the first entry in the variable format statement. The format statement shown below reflects the position of the examinee identification field (10A1), the 34 item responses (34A1) and the criterion (F4.0). MULTILOG will read the chronological age of each individual and use that as a starting value for the iterative modal estimation procedure. The "T" format is used to tab to the correct positions of the respective fields. The first ten characters on each record, which are read as an identification field, are also used to assign a value to the NCHARS keyword on the PROBLEM command through the Input Parameters dialog box (see later in the example).

Select the Individual item response vectors option from the Type of data group box. The selection made in this case will add the INDIVIDUAL option of the PROBLEM command, while the entry in the Data file name field will be used in conjunction with the DATA keyword on the same command. The variable format entered in the Data format field will be echoed to the generated syntax file.

The parameters for the 34 indicators are in a file called knee.par. This file was produced by MULTILOG as a consequence of the inclusion of >SAVE among the commands in a (previous) calibration run. As the parameters for the SAVE and other optional commands cannot be set using the syntax wizard, these commands will be added after generating the syntax file. Instructions concerning this can be found at the end of this example. Having completed the Input Data dialog box, click Next to go to the Input Parameters dialog box.

The problem uses 34 items, and this is indicated by setting the value in the Number of items field to 34. Data from 13 examinees are available, and this is specified using the Number of examinees field. Finally, the NCHAR keyword on the PROBLEM command is set to 10, as previously indicated in the variable format statement description, using the Number of characters in ID field.

Entries in this dialog box correspond to the following MULTILOG keywords on the PROBLEM command:

Click Next to go to the Test Model dialog box.

A graded model is used here, and is specified by clicking the radio button next to the Graded model option in the Test model group box. This corresponds to the GR option on the TEST command.

The "test" has varying numbers of re-sponse categories for the 34 indicators, which are entered in the NC list on the >TEST command. As all items are used, the All check box in the Use column of the Test Items group box is clicked, and the number of categories is set by item in the Category column as shown below. Once the number of categories for each item has been indicated, click OK to go to the Response Code (non-binary data) dialog box.

All possible responses in the data are entered in the Response Code field. The corresponding correct response codes are entered in the Correct response codes group box. On each line, the number of entries permitted corresponds to the number of items specified in the Input Parameters dialog box.

Once the response codes (123450) are entered in the Response Code string field, these codes appear as the first column of the Correct response code group box. For each response code and each item, a category number is entered. Permissible values are 1, 2, ? NCAT, where NCAT denotes the total number of categories for a given item. In any row, a "0" indicates that the response code value is excluded from the analysis. A valid entry for item 1, for example, is Code 1 = 5, Code 2 = 2, Code 3 = 3, Code 4 = 4, and Code 5 = 1. The entry specifies that data values equal to 1 is assigned to the 5-th category of item 1, while data value 5 is assigned to the first category.

Start by entering the correct responses for the first code, and press the Enter key on you keyboard when done to proceed to the next line of the window. Note that, if an attempt is made to specify response codes not in agreement with previous selections, no value will appear in this box. Only when valid codes are entered, will the results be displayed. Once all codes have been entered, click OK to go to the Project Settings dialog box. Entries in the Response Code dialog box will appear after the END command in the generated syntax file.

The Project Settings dialog box displays a summary of all selections made up to this point. To go back to any of the previous dialog boxes, the Back button may be used. To generate the syntax, click Finish. Syntax generated using the wizard is now displayed in the main MULTILOG window. Before running this problem, the following (optional) commands are added to the syntax in this window by using standard Windows editing functions:

>START ALL, FORMAT, PARAM='KNEE.PAR';
>SAVE;

The START command is used to override the default starting values for all the item parameters and enter others, in this case from the file knee.par.

Click the Run option on the main menu bar to start the analysis. Once the analysis has been completed, the output generated may be viewed using the Output option on the same menu bar. The output file will then be displayed in the main window, and the Window option may be used to switch between syntax and output files.