If they are available from secondary sources, the task becomes easier for the researcher. If not, they have to necessarily resort to collection of primary data. So long as the requirements are confined to quantitative information, they can be collected easily. At the same time, if qualitative information is required, the problem gets compounded.
In particular, when a researcher is interested in capturing the behavioral pattern of the chosen samples, task would not be as easy as collective quantitative information. Given the fact that the researcher cannot enter into the minds of respondents, at the best what goes on in the minds of respondents can be captured by designing appropriate questions / statements and eliciting their responses.
While the subject matter has been debated over the years, there is a consensus amongst researchers on the usefulness of measurement scales. These scales are tools which sorts or rates or ranks the respondents view points on the pre-determined criteria. They have been used widely in the areas that call for using psychometric exercises. The behavior of respondents thus gets captured by noting the respondents responses on a scale, and these responses when codified become the data for further analysis.
Once the researchers decide what kind of measurement scales are to be used, it is equally important to determine if the tools used reliability and validity. Of course, a researcher could use a number of measurement scales, but choosing the right scale is a challenge. Only those scales which are reliable and valid alone will have to be chosen for measurement purposes.
When a tool used for measurement produces similar results on repeat, that tool could be considered as reliable. In fact, reliability of a tool is nothing but the degree to which what has been measured yields consistent result each time it get used. Moreover, if what get measured remains free of error, the tool used to measure them is considered as reliable.