Data analysis: learning from optical illusions

06Jun09

Long ago, at an informal workshop, I used this image to explain to young researchers the ideas behind qualitative data analysis. Here are some random thoughts from that time…

youngwomanoldlady

What do you see?

Most saw the old lady, a few who knew this puzzle saw both. And those who could not see both, after a while got flustered and peered closer and closer. So here was my first thought:

1. do not panic – increasingly, there is that desperate quest for an Insight (not insights, mind you, the Insight with a capital I, destined to take the client’s breath away) – and when no pattern seems to emerge from the data, panic sets in. I have often had young researchers comes up to me at work and say in a worried manner, but there is nothing coming out of the data… My advice to them is to take bite-sized pieces and step back…

2. step back – instead of peering closer and closer as I had mentioned earlier, some times it helps to ‘helicopter’ – stop staring at it intently and look at it as a complete picture – detachment from data sometimes helps and from a distance, things sometimes seem clearer

Word-Optical

3. different perspectives – going down one road, be prepared to back-track and look at other paths – once you see the old lady, is difficult to see the young woman – and vice-versa. some things just strike us early on in the analysis process – and sometimes simply because they are louder or appear more frequently (in this case, the word ‘optical’). And it is dangerous to hold on to them as the absolute truth. And sometimes, no, often this happens because we are not expecting – and therefore prepared – to see more than one emerging picture.

4. focus on the small bits – as I said earlier, take note of the small insights and the Insight will take care of itself – as in the case of the first image, the old woman’s eye is the young woman’s ear, her wart is the young woman’s nose, and her nose the young woman’s chin. Instead of searching for the larger picture, spotting these patterns / discrepancies (in case of data) would make it easier for the large picture to emerge…

These are a few thoughts to begin. What are your thoughts on this?

Advertisements


%d bloggers like this: