What is analysis? Some tips to "become more analytical"

As a student, have you ever been told that your work was “too descriptive” or that you needed to be “more analytical”? 

The truth is, despite common prompts to make essays and research assignments “more analytical”, chances are that students are never taught what good criteria of analysis are.  Analysis is a big elephant in the room of social sciences, including for courses explicitly labelled as related to “data analysis” – such as “discourse analysis” that I teach.

The Elephant in the Room - seriykotik1970

So what is analysis and what does it take to produce good analytical work?

​What is analysis?

​Analysis is a process of transformation. Raw data and information do not have meaning per se. It is you, as researcher, who make meaning out of it, via the process of analysis.
Analysis is a creative process. What you create through analysis is a new discourse about the world, which helps people perceive the world in a different way, understand things that they did not know or did not understand before reading your research.
 
But how does it work? How does analysis make sense of the world, then?
To summarise: by bringing focus, synthesising, naming, establishing patterns and relationships that will help other people in perceiving these patterns and relationships and thus understand the world differently.
 
A good metaphor for analysis is the work of astronomers and other stargazing lovers who identified constellations. The sky is full of stars. And some people drew patterns in the sky, named these patterns and even created stories about the shapes thus formed. This way of perceiving the sky is passed down from generations to generations, taught and written in books. And people who learned them can’t help but seeing these patterns when they look at the stars.

Capricornus, Commons Wikimedia, Michelet, 2006

Here are two definitions of analysis taken from methodology handbooks:
 
“Taken literally, ‘breaking up’ something complex into smaller parts and explaining the whole in terms of the properties of, and relations between, these parts” (Robson 2011: 412)
 
“The process of bringing order to the data, organizing what is there into patterns, categories and descriptive units, and looking for relationships between them; ‘interpretation’ involves attaching meaning and significance to the analysis, explaining the patterns, categories and relationships…” (Brewer 2000: 105)
 
These definitions highlight the double process involved in analytical work:
- on the one hand, the breaking down and simplification of the inherently messy social world;
- on the other hand, the building up of patterns between selected elements to produce a new (and synthetic) way of interpreting the world.
 
In practical terms, the transcription of one hour of interview can take up to 40 pages. Thus, the transcription of 10 interviews amounts to 400 pages. You need to break down your material into manageable segments, and focus on certain elements and sacrifice others, in order to find out what is the most interesting knowledge you can produce out of this data. 

​Does that mean that everything goes?

If analysis is a creative process, does that mean that ​it is ok for social scientists to create any meaning and discourse out of their data?
​I said that analysis was a creative process, but no, that does not mean that everything goes.

Listing explicit criteria about what we need to do to produce a good analysis can help us ensure we are on tracks regarding our analytical goals.

​To sum it up:  the objective of analysis is to produce a convincing demonstration based on empirical evidence describing, explaining and interpreting a social phenomenon.
​

​Criteria of analysis 101: The ‘wheel of analysis’

In the schematic ‘wheel’ below, I unpack the definition above to make explicit a series of criteria to help you produce a good analysis:

Let’s unpack further these criteria:

• Description: What is going on? Make sure you are not jumping to conclusions before describing the evidence.
• Explanation and Interpretation: The description of the empirical material is a necessary but not sufficient dimension of analytical work. The readers need you to go one step further. Depending on your research question, it might be, for example, the reasons behind the situation, the conditions of their emergence, how actors make meaning of it in context etc… You need to go beyond simply describing what is in the data and synthesise the way you interpret and explain the situation under scrutiny by linking your material to the context, existing theory and literature.
• Specific research object: Provide an answer to the research question by selecting relevant material to analyse. ONLY keep in the final project the data that is relevant for your analysis.  You have a purpose and you need to stick to it. There will be things in your data that are interesting but that don’t directly answer your research question: you have to be ruthless and sacrifice them. Pasta is very good but if you’re baking a cake, you don’t put the pasta in the cake, otherwise you miss the point by spoiling the cake. To summarise: there may be things in your data that are very interesting, but if they are outside the scope of your research project: bye bye.
• Discourse for someone else: The analysis is not a summary of your data that you write for yourself. The analysis is something you communicate to someone else. It needs to be clear and engaging. It is an argumentative exercise. You need to convince the reader. Imagine you are trying to convince someone you know when you write.
• Demonstration:​ A very good way of producing a convincing analysis is to demonstrate the rigour of your analysis. Create something that is more than generating impressions. You need to go beyond anecdotal examples and cherry-picked cases. Aim for your demonstration to be transparent and traceable so the readers can understand how you reached your interpretation. Provide a clearly articulated account of your assumptions, procedures, and steps. Aim for your demonstration to be systematic. Develop analytical strategies and procedures with steps to be followed consistently. That means that you approach all your data (all your interview transcripts, for example) with the same framework of analysis and the same questions.

 
Each of us has natural strengths and weaknesses. For example, some of us will have no problem in expressing their argument with clarity but will struggle to establish clear boundaries to their research topic. Others will provide a rich interpretation but without demonstrating how they reached these conclusions. It is important to identify which dimensions of analysis are your weaknesses or which ones you usually tend to neglect, and work on them as a priority.

The ‘wheel of analysis’ can be used as a compass to guide your work and to make it ‘more analytical’. A key moment to use the wheel is also once you have finished your first draft. Then use the wheel as a checklist by comparing its different elements with what you have achieved so far. Are you providing an interpretation or just describing what is in your data? Did you write your analysis as a summary for yourself or as a discourse aimed at convincing an audience? Keep going back and forth between the wheel and your analysis until you improve it as to match all its dimensions!

This article is also available in Spanish and in Chinese.
You can download the pdf here.