Metadata

Metadata in Survey Research

In survey research, metadata plays a vital role; it is used to describe statistical data from survey conceptualization to data dissemination. Metadata can be very broad and encompasses populations, methods, survey instruments, analysis approach, results of instruments, and so on. Furthermore, it describes the information about the data, including variances, response rates, and response categories. Most survey researchers perceive this world as hierarchical; thus, a good data model must be able to handle data in a straightforward fashion as well as being able to describe data hierarchically. As a result, metadata should provide detailed information regarding variables while reflecting the data model structure.

Part of the survey process centers on raw data that are collected from respondents and converted into data sets. The entire conversion process needs data management. In general, good software packages need to deal with the data and metadata involved in the survey. Such software should be able to translate any data and metadata into any format when dealing with the survey process. Normally, data managers create metadata, and since metadata is expensive to create, a great need exists for understanding how the data would be used prior to its actual creation.

In addition, data and metadata producers both need to possess adequate communication channels between them. Metadata producers must follow certain standards. First, they need to thoroughly understand the data and be able to encode the data information. Usually, a single disk file is created for each metadata record, which describes one data set only. Second, the metadata file is arranged properly. Third, there is a need to verify the data and make sure it is adequately described.

Metadata also varies depending upon the type and context of data. In order to understand the data more completely, the following need to be addressed:

• 1.
  A definition of survey variables. Each variable must contain a valid value, a code, an identifying name, and so on.
• 2.
  Data model. This model describes relationships between the variables regarding their categories and hierarchies.
• 3.
  Route instructions. This defines order, such as how questions are asked.
• 4.
  Relationships. Whenever relationships impose restrictions on the values of variables, the restrictions must check the consistency among the collected data.