In recent years, we have seen a sudden increase in commercial applications in machine learning and AI. Machine learning models heavily rely on data. Therefore, data scientists have invented new methods of data collection. Some of these data collection methods did not exist before the rise of the digital revolution.
However, in this article, we particularly want to revisit some of the very traditional yet useful and fundamental methods for the collection of primary data in statistics.
1. Direct Personal Interviews
The investigator personally meets concerned individuals and collects the required information from them. When the area to be covered is vast, this method may prove very costly and time-consuming. Still, this method is concerned useful for certain laboratory experiments or localized inquires. Due to the personal bias of the investigator, errors are likely to influence the results.
2. Indirect Personal Interviews
We interview the third parties or witnesses having information, whenever the direct sources do not exist, or the informants hesitate to respond for some reason or other. The reliance is not placed on the evidence of one witness only, because some of the informants are likely to give wrong information deliberately.
3. Collection through Questionnaires
The questionnaires are usually sent by mail to inquire through several pertinent questions. In questionnaires, there is a space for entering the asked information asked. The informants are requested to return the questionnaires to the investigator within a certain period. This method is cheap, reasonably expeditious, and good for extensive inquiries. However, only a small percentage of recipients respond to questionnaires if there is no incentive involved.
4. Collection through Enumerators
In this method, trained enumerators collected the information. They assist the informants in making the entries in the schedules or questionnaires correctly. If the enumerator is well trained, experienced, and discreet, then you can get the most reliable information through this method. Enumerator driven approach works best for a large scale governmental or an organizational inquiry. Private individuals or institutions cannot adopt this method as its cast would be prohibitive to them.
5. Collection through Local Sources
In this method, the agents or local correspondents collect and send the required information, using their judgment as to the best way of obtaining it, but there is no formal collection of data. This method is cheap and expeditious but gives only the estimates. It may involve local agents' bias.