0

I have noticed several polisci or economics papers in the past few years such as this or this that address important research questions on political behavior

but in political environments where data on public opinions have been historically scarce or unreliable (e.g. China, Russia, etc).

However, my question here is that I noticed that the authors emphasize the data used in the paper as a major contribution to the literature (e.g. because it was collected from online textual data and the effort it takes to clean and prepare the data for the analysis).

As a junior social science researcher, I was wondering if it makes sense to write a paper where one emphasizes that a main contribution to the literature is the originality and uniqueness of the data used? Specifically, I have been trained in my PhD program that it is the questions and results that make a paper 'good', rather than the uniqueness of the data, but then I am noticing more and more political science papers that may study a question already addressed in democratic countries, but the contribution is that it's currently studied in an autocratic country like China.

1
  • If there is no data available, how would one address questions and come up with results? And if getting that data takes years and years of effort, well, how long can one delay a student's graduation?
    – Jon Custer
    Aug 16 at 14:05

2 Answers 2

4

I will disagree with @buffy's answer that data itself is "boring" (I'm paraphrasing her answer here :-) ).

There are many cases where the data itself is the main product of research. For example, you find a well-preserved tree trunk from thousands of years ago and you do an isotope analysis to infer the climate year-by-year in that tree. The methodology for this is well-established, there is nothing new about this. But you get data that was simply not available before and that others cannot easily get either without finding their own well-preserved tree. So the data is the important part.

Or you have access to a unique resource that allows you to do an experiment that is not easily repeatable by others. Say, a large telescope. Or a big computer. In principle, everyone can take pictures with a telescope, and there is nothing about the data that needs to be described regarding how exactly you used the telescope. But if you come up with pictures of 100 black holes at centers of galaxies and can do some simple statistics about them, then the data is the important part, not the methodology.

In other words, data can absolutely be the important part of a publication. The key piece is whether it is data that is easily generated by others, or whether you have a unique resource that makes it difficult for others to reproduce things.

(I will note that this is no different from software. Traditionally, specific software you wrote was not a topic you could publish on. But over time, people have realized that software that consists of 100,000s or millions of lines of source code is a resource that can not easily be reproduced: Nobody has the time to re-write such packages even if the underlying algorithms are well documented. As a consequence people have started writing papers that describe the implementation of these algorithms in software, and I tend to think that these are worth-while publications to write.)

4
  • 2
    The question OP actually wrote seems to be about about something slightly different - they seem to be asking about applying a previously used analysis to a new dataset, rather than publishing a paper about new data itself. That said, I think this answer likely addresses the question OP meant to ask...
    – Bryan Krause
    Aug 15 at 15:19
  • @BryanKrause I took my cue from the following sentence: "I was wondering if it makes sense to write a paper where one emphasizes that a main contribution to the literature is the originality and uniqueness of the data used?" Aug 15 at 17:10
  • 1
    Yes, though the examples they give do not seem to actually be about the type of unique/original data that you're talking about where someone might actually write the paper principally about the interesting data. Instead, they seem to be just basic incremental steps in their field, less about a "well-preserved tree trunk from thousands of years ago" and more about "Eken et al did this study of oak tree growth, but if we want to know more about tree growth more generally I'm going to use their methodology to study maples". I think OP is incorrectly labeling this "data as research contribution".
    – Bryan Krause
    Aug 15 at 17:17
  • 1
    I'm chemist and would like to point to another type of useful data: Evaluating whether a given substance can be used as calibration standard and giving reference values for the measurements. Such a substance would preferrably not even be rare - so other labs can easily acquire such a substance/sample and then calibrate their equipment against the provided literature values. Aug 17 at 17:21
-1

My view is that the data itself wouldn't be a research contribution, but the "research methodology", would be if it is an advance over standard practice. That methodology is (a) subject to analysis and (b) potentially reusable in other studies.

The data itself is interesting in so far as it is used to answer the initial question(s), but of more limited value otherwise. It is, in particular, difficult to validate without reference to the methodology that gathered and processed it. There might be tidbits of information that might be tweaked out of it, but, IMO, that would be anecdotal evidence of whatever is found without an analysis of the methodology.

So, other things than the research question and conclusions might be considered a "novel" contribution to research, but the data, not so much per se.

And note that the methodology in such situations must be analyzed lest it introduce biases that affect claimed outcomes. That goes doubly for reuse of the data.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .