Data, Pseudopanel

views updated

Data, Pseudopanel

Among all sociological research designs, longitudinal studies are the best choice if the purpose of research is to examine changes over time. Crosssectional studies observe a phenomenon at only one point in time, and causal relationships between independent and dependent variables in crosssectional studies are unclear because the time sequences cannot be established. There are three special types of longitudinal studies usually adopted by researchers: trend studies, cohort studies, and panel studies. Of them, panel studies are the most difficult for researchers who are collecting data because they have to interview the same sample of subjects repeatedly. Trend studies and cohort studies can provide changes of attitudes or behaviors of a population, but they cannot detect changes in a subject, only overall changes of the studied samples.

Panel studies can eliminate such a problem by examining the same subjects over time; typically researchers find that overall changes remain stable (the number of subjects who change their responses from A to B is balanced by those who change their responses from B to A). Although this is a great advantage of panel studies, it has its shortcomings. Panel studies are costly in both time and money (due to the high number of interviewers and the expenses associated with interviews and traveling), especially in a large-scale survey across several regions or states. Moreover panel studies often have a common problem: sample attrition. Researchers must revisit these subjects for a long-term research project, and subjects may fall out of contact. In order to retain most of the advantages and avoid the disadvantages of panel studies as well as the limitations researchers sometimes encounter in obtaining sufficient supports for panel studies, pseudopanel studies may be the second-best choice.

Pseudopanel studies may take different formats. In one a researcher asks respondents about past and current attitudes or behaviors and then treats the data as if they represented different time points. Another format uses simple logic to infer the time order of two or more variables. For example, it is more likely that people have tried cigarettes before cocaine because smoking is more socially acceptable than using cocaine; criminological studies have supported this conclusion. Researchers use established causal relationships from prior studies to assume similar relationships in their current studies when adopting crosssectional study design.

Although the pseudopanel study avoids some of the problems or difficulties of the panel study, it is not recommended unless a panel study design is not accessible. Pseudopanel study is vulnerable to its own problems. For one, people’s faulty memories are hard to detect and not uncommon for many reasons. Subjects may sincerely forget things that occurred in their pasts due to poor memory, or they may lie about their pasts for various reasons. In addition the use of logical inferences depends on the researchers’ knowledge of the research topic. An expert might have a strong knowledgeable background to make robust inferences, but still that might not rule out the potential errors of pseudopanel study design.

SEE ALSO Census; Data, Longitudinal; Integrated Public Use Microdata Series; Methodology; Methods, Quantitative; Panel Studies; Research, Longitudinal