Posts in data analysis

When you think you got it

junio 24th, 2011 Posted by blog, data analysis, empirical research, empirical-inductive approach, The sorcerer's apprentice, theoretical-deductive approach No Comment yet

So you have been preparing the pilot study for your translation process research project with passion: You have been weighing the pros and cons of every single method and tool of data collection—you just want the best one, of course. You have been choosing the source texts meticulously, screening them from many different angles. You have been defining your experimental subjects and recruiting people, some of whom you finally managed to convince to participate in the test.

Then came the D-day: You carried out the experiment, or, properly speaking, you let your subjects carry it out—the subjects patiently translated with Translog (; they also filled out some questionnaires. And, afterwards, evaluators had a look at their translations.

This is the moment when you start feeling that you made it, that the first big step is done, your first tentative data collected; and it’s true, but it’s also true that there is a much steeper step awaiting you now:

The analysis of data

(“Night on Bald Mountain” might do as a soundtrack effect here).

You may feel some kind of dizziness or trepidation in the face of the amount of data you have collected. So, now what? What’s next?

First of all: Stay calm and don’t despair!

There are two ways out of this trial:


  1. You could have a look at the materials you collected, those based on a theory you have been working out before. The theory offers you one or various perspectives onto your data (theoretical-deductive approach).

Or else

  1. You could let your data speak first and let the facts emerge and grow, and adapt your interpretation and your theory to the outcome (empirical-inductive approach).

If you were looking for something well defined, if it was an experimental setting, if you just wanted to (dis)prove some theory or theoretical point, if you —or your dissertation director— are not willing (!!) to modify the theory, then option 1 seems to better fit your needs. If you are doing descriptive research, if everything is foggy and you don’t trust the rosy & complex notions you have been using, then option 2 might get you further down the way.

The problem is, your research project may fall somehow in between, and anyway data collected are simply overwhelmingly rich. So, before choosing one way or the other, you may want to ask yourself the following questions (if you haven’t done so already):

What was I looking for?

Which elements in the data are useful for my purpose? Which are not?

How could I check what I wanted to see/know/measure?

When you have found the answers, start selecting your material. Focus on the data you really need for your study aim(s) and leave those aside you don’t need for this project; maybe later you can use them in another project, so don’t think they are worthless. Do NOT dispose of anything. Umberto Eco said that one of the main problems in writing a dissertation is chopping off side branches, I mean, mmm, reducing the scope of your goals to a size that can be managed in a few years. The times are over when a PhD research project was the crown of a whole career, welcome are now dissertations that let you prove you can do high quality research.

In any case, in translation process research you nearly must resort to triangulation, to cross-referencing qualitative and quantitative data, both to improve intersubjective agreement within your scientific community and to avoid the distortion effects of each single method.

So just take a deep breath now and keep going. Just do not go into the light. Research weather is always foggy.


Further reading

Shuttleworth, M. (2009). What is the scientific method? URL:
Wang, J. & Khosravi Sereshki, H. (2010). How to implement ITIL successfully? Jönköping. Chapter 2.2. URL:

Recent Comments