Use of Internet Longitudinal Survey

Referring to University Student Monitors

Abstract

One of the recent developments in quantitative survey methods is the use of panel（longitudinal）survey data. This method allows researchers to accurately capture individual changes; however, the survey costs, including labor, time, and management costs, are too high for the survey to be conducted easily. As a measure to reduce these costs, in this paper, we discuss the use of a longitudinal survey using registered monitors owned by survey companies, calling it an “internet longitudinal survey.”

Monitors that voluntarily respond to recruitment by survey companies are nonprobability samples; thus there is no guarantee that they are representative of the entire population. Since the youth, such as university students, are likely to make large coverage and nonresponse errors in probability sampling, it is recommended that internet monitors which enable capturing individual changes with lesser errors are used to supplement the survey data. However, it is known that youth monitors are more likely to drop out during a panel survey than monitors of other age groups. Therefore, when conducting an internet longitudinal survey using youth monitors, these attritions need to be dealt with.

This paper introduces an internet longitudinal survey for university student monitors to test the educational credentials hypothesis in job opportunities. In order to deal with the attrition problem, we constructed a weight variable from the predictive analysis of the continuous response rate in the second wave and tested the hypothesis using the weighted data. The results show that the hypothesis is robust, both with and without weight correction.