Applied Measurement with jMetrik

J. Patrick Meyer

• Publisher: Routledge
• Year: 2014
• eBook ISBN:
  9781136294167
• Topic: Education
• Subtopic: Education General

jMetrik is a computer program for implementing classical and modern psychometric methods. It is designed to facilitate work in a production environment and to make advanced psychometric procedures accessible to every measurement practitioner. Applied Measurement with jMetrik reviews psychometric theory and describes how to use jMetrik to conduct a comprehensive psychometric analysis. Each chapter focuses on a topic in measurement, describes the steps for using jMetrik, and provides one or more examples of conducting an analysis on the topic. Recommendations and guidance for practice is provided throughout the book.

1
Data Management

A relational database management system (RDBMS) lies at the heart of jMetrik, and it is responsible for all of jMetrik’s data management capabilities. An RDBMS allows data and information about the data to be stored in multiple tables rather than a single flat file. This feature increases the efficiency of the program and provides a common framework for all data sets. You can think of jMetrik as a simplified interface to the RDBMS that gives you easy access to the most essential procedures for data management. You can import and export data, select subsets of data, and delete variables or an entire table. jMetrik organizes data management into three types of operations: database operations, table operations, and variable operations. This chapter begins with a description of the default RDBMS and naming conventions enforced by jMetrik. It then describes all three types of database operations and gives step-by-step instructions for data management features in jMetrik. It also describes noteworthy limitations to data management and provides recommendations for best practices.

Default RDBMS

Apache Derby is the default RDBMS in jMetrik (see http://db.apache.org/derby/). Much of the following information is specific to Derby and is subject to change if you use jMetrik with another RBDMS. Like jMetrik, Derby is a pure Java application that is compatible with any operating system. It stores data in a platform independent manner allowing you to move a database from one computer to another without affecting your ability to work with the database. Even if the Derby database is created with a Windows computer, the data are readable by a Linux or Mac OSX machine.

There is virtually no limit to the number of tables in a Derby database and virtually no limit to the number of cases (i.e. number of examinees) in a table— “virtually no limit” because the actual limit is the size of your computer’s hard drive. Almost all modern computers are built with more than enough storage. You are unlikely to reach this limit in practice. The main limitation to consider is that each table is limited to 1,024 variables. If your data file has more than this number of variables, you should separate it into multiple tables. It is also good practice to use data files with far fewer than the maximum number of variables because some jMetrik procedures will add variables to the table. jMetrik will produce an error if you try to add a variable to a table that already contains the maximum number of variables. jMetrik only works with the default RDBMS as of this writing. Functionality for other RDBMSs will be added in the future.

Database Home

Through the RDBMS you create a database that contains any number of tables, and each table may contain one or more variables. You only need one database to store all of your tables, but you may find it easier to use multiple databases to organize your data. For example, you could create one database to store all of your math test data and another database to store all of your science test data. jMetrik allows you to easily switch between databases, but you can only access one database at a time. This restriction means that you should keep all of the tables that you need for an analysis in the same database. Think of each database as an independent entity. You cannot combine data or analysis methods across multiple databases.

Databases are stored on a computer’s hard drive in a folder referred to as the database home. By default, this location is a subfolder of the user’s home folder. For example, on a Windows machine the default location is C:\Users\<user name>\jmetrik\databases. You may view the current location of databases on your computer through the jMetrik Preferences dialog. Click Edit → Preferences to display this dialog. Your databases are stored in the folder listed in the Home text field of the Database Options panel. This location will be ideal for most people, but you can change it through the preferences dialog. The only restriction is that you must have read/write permissions on the folder where the databases are stored. If you change the database home, you will be prompted to restart jMetrik.

One reason why you may want to change the database home is if you are working with a team of researchers and you are sharing a database. One team member can create the database on a shared resource to make it accessible by everyone on the team. Each person can set the shared resource as the database home and work on the same database. However, only one person can access the database at a time. jMetrik does not allow multiple people to simultaneously access the same database.

Organizing Data in jMetrik and Table Descriptions

Given that a database will contain numerous tables, you should develop a table naming strategy to help you quickly remember the contents of each table. For example, you could concatenate two-digit codes that reflect the grade, subject, and year of an academic test. With this convention, a table named g3ma11 would contain data for the grade 3 math test given in 2011, and a table named g3ma12 would contain data for the following year. jMetrik lists tables in alphabetical order. To keep related tables next to each other in the list, your naming strategy for output tables should make use of the underscore character and extend from your original table name. For example, the item analysis output table for the grade 3 math 2011 test could be labeled g3ma11_iaout. The Rasch models analysis output table could be named g3ma11_rasch. Following these conventions, analysis of subsets could extend the table name again. An item analysis of the grade 3 math 2011 test for females could be named g3ma11 _female_iaout, and the analysis for males could be g3ma11 _male_iaout. It is entirely up to you to choose a naming convention, but a systematic method for naming tables in jMetrik will help you keep the information organized and identify table contents by only using the name.

In addition to a naming convention, jMetrik allows you to provide a written description of each table to help you keep track of table contents. jMetrik automatically provides a minimal description for any table created in jMetrik. You can edit this information and provide more details about a table or record other notes about the table for later reference. The description is limited to 1,000 characters, and you can edit it at any time through the Table Descriptions dialog. To start the dialog, click Manage → Table Descriptions. Select the table in the list on the left side of the dialog, and then add or edit a description in the text area. Press the Submit button to save your description. You can continue writing descriptions by selecting another table. Press the Submit button after adding or editing a description. Press the OK button when you are done to close the dialog and return to the main program. Make heavy use of table descriptions because your database will eventually contain hundreds of tables. Table descriptions will help you keep track of them and make it easier to recollect table contents.

The following sections discuss three types of data management operations that you can perform: database operations, table operations, and variable operations.

Database Operations

After you install jMetrik, you must create a database to hold your data. You only need to create one database, but you can create others if it helps to organize your data. To create a database, click Manage → New Database, and type a name for the database in the dialog. Click the Create button to produce the database, or click Cancel to return to the program.

TABLE 1.1 Naming conventions injMetrik

Database Structure	Permitted Characters	Case Sensitive^a	Maximum Length
Database	A–Z, 0–9,_	No	System
Table	A–Z, 0–9,_	No	120
Variable	A–Z, 0–9,_	No	20

^a Although names are not case sensitive, you should only use names with all lowercase letters.

The name you choose for a database must comply with the following conventions and recommendations. Database names are case insensitive, meaning that the name “mydb” and the name “myDB” are equal in the eyes of the program. jMetrik will convert all database names to lower case. You can use the letters A through Z and the digits 0 through 9 in a database name. White spaces and punctuation marks are prohibited from database names. jMetrik will attempt to remove white spaces and punctuation marks from a database name before creating a database. It will inform you of any changes it makes to the database name. For the default RDBMS, there is virtually no limit to the length of a database name. However, good practice dictates the use of short yet descriptive database names. Database names longer than 120 characters are cumbersome to read in jMetrik and are not recommended. Table 1.1 summarizes database, table, and variable naming conventions.

Creating a database adds files to your computer and makes the database available to jMetrik, but it does not open the database. You must open a database before you begin using it. Click Manage → Open Database, and select the database that you would like to open in the dialog. Click the OK button to open the database. Once it is open, you will see the database name appear in the middle section of the status bar, and a list of existing tables will be displayed in the list on the left side of the interface. If the database is new and no tables have been created, nothing will be listed in the interface. You must import data into jMetrik to create a table and make data available to the program.

A database will remain available to the program until it is deleted. Once you no longer need a database, you can delete it by clicking Manage → Delete Database to display the Delete Database dialog. Select the database you would like to delete from the list of available databases, and then click the Delete button. You will be warned before any action is taken. If you accept the warning, your database will be permanently deleted from the system. Be very careful when you choose to delete a database. It is not possible to recover the database after it has been deleted.

Table Operations

Table operations in jMetrik give you control over importing, exporting, and deleting tables in the database. There is also an operation for viewing and editing table descriptions. To begin using a data file, you must first import it into jMetrik.

Importing Data

You can import a comma, tab, colon, or semicolon delimited file that contains no more than 1,024 variables. A delimited file may have variable names in the first row that uniquely identify each column of data. Variable names must be unique values that are 20 characters or less. They are limited to the characters A through Z, the numbers 0 through 9, and the underscore character “_”. White spaces and punctuation marks are not allowed in variable names and will be automatically removed by jMetrik. As a recommended practice, use an underscore to represent a space in your variable name. For example, you could represent the variable “student id” as “student_id.” Variable names are case insensitive, meaning that the variable “student_ID” and the variable “student_id” are viewed as the same value by the program. However, it is recommended that you use all variable names with letters in the same case (i.e. all lowercase or all uppercase). If your delimited file does not include variable names in the first row, jMetrik will create unique generic names (e.g. V1, V2, V3) and add them to the data table during import.

To import data into an open database, click Manage → Import Data to display the Import Data dialog (not shown). Type a name for your table in the first text field. Table names must be unique to a database, and they are limited to the letters A through Z, the digits 0 through 9...

Table of contents

• Cover
• Title
• Copyright
• Dedication
• CONTENTS
• List of Figures and Tables
• Preface
• Acknowledgments
• 1 Data Management
• 2 Item Scoring
• 3 Test Scaling
• 4 Item Analysis
• 5 Reliability
• 6 Differential Item Functioning
• 7 Rasch Measurement
• 8 Polytomous Rasch Models
• 9 Plotting Item and Test Characteristics
• 10 IRT Scale Linking and Score Equating
• Appendix
• References
• Index