What are the important issues to consider when using secondary data?

Question

Say you want to conduct a research study that observes people engaged in a generic activity. Rather than record this evidence oneself, would it be wise to try to gather recorded evidence that was part of other research studies? I am asking about gathering raw evidence, to re-analyze for the purposes of a new study, not leveraging/citing data/conclusions in other research studies. For example, if another research study recorded 1000 hours of people doing general enough activities, would I be able to ask the owner of that study to share their recorded evidence with me, so that I might use it for my own analysis?

In other words, what are some of the issues to consider when using secondary data?

If it isn't common, why not? What are the pitfalls of using evidence gathered from other studies, to drive my own research?

Answer 1

It is not uncommon to use data from other studies. The caveats deal with data quality. Even if the data is provided with error bars, you can still not necessarily rely on it being completely accurate. Typos can occur and other errors could enter data sets by mistake. I have seen such cases personally, data that when analysed yields different results than presented in a paper.

If you use a data set that comes from a paper, you should make sure it at least yields the same results as the published paper, since that is what has been officially concluded. This does, however, not mean all is good so I would always threat data with varying degrees of healthy scepticism. The problem is particularly bad if you rely on the data to draw far reaching conclusions on your own. I can see situations where this could be life threatening and situations where it would lead to sending research off in the wrong direction but also, most often no major issues. So depending on how you rely on the data for vital conclusions, you need to adept your critical scrutiny of the data.

This may seem like I forebode a catastrophe but the key point is just to not take everything published at face value.

Answer 2

Reusing data gathered in previous research works save a lot of time, energy and may be money. However, there are some subtle details you need to keep in mind when using such practices.

1. Make sure the author providing the data knows what he is talking about.
2. Make sure the data are valid and genuine.

As I heard from one of my professors, citing results of another author does not mean that you would not be held responsible in case of false/inaccurate results.

Update: Following the comment about raw data I think such data can be trusted, unless proved otherwise. Also, it is better to go for raw data (data sets) used widely thus reducing the chance there would be something terribly wrong with the data.