The limits of econometric analysis based on secondary data

Most of empirical economic research is carried out under the traditional neoclassical paradigm that fails to recognize the importance of distinction between primary and secondary data in research design. Primary data is collected directly from a source such as through interviews or surveys. It is original research conducted for a specific purpose, and the questions researchers ask are tailored to the study to elicit data to assist the study. Primary data provide up-to-date information that is reliable and more accurate than secondary data. However, it is time consuming and costly, and can be affected by sample selection bias.

In contrast, secondary data are gathered from sources which have already conducted primary research for other purposes. Sources of secondary data include official statistics from government archives and various agencies, scholarly journals, literature review articles, universities, and others. When conducting secondary data analysis, the quality of data must be assessed by determining the original purpose of the data and whether valid research methods were used, checking the author’s credentials and experience in the field, and cross-referencing data when numbers do not add up. Secondary data can be used to generate a hypothesis for primary data analysis, as well as provide a baseline to compare the primary data to; it is a complement to primary data. Although secondary data are quicker to obtain than primary data and less costly, in both time and money, the main demerit is that data collection methods may vary so comparability of the data may be compromised if researchers use discretion to choose which information to collect or omit, resulting in inaccuracies in the data. In this paper, we will discuss general weaknesses and limitations of research design that are based on secondary data. Particularly, we will explain what conclusions, if any, can be reached when using secondary data in an economic research.