Saturday, October 2, 2010

attention bias

We are much more likely to spot something if our attention is drawn to it. There are two forms of attention bias:

* internal
* extrenal

internal attention bias

When we have a particular focus on something we tend to pay too much attention to it and this seriously skews the way we interpret and make decisions on data. For example if we buy a new car we tend to overly notice other cars with the same colour or which are the same make or model or even number plates that seem to be similar to ours. One urban myth which is caused by our internal attention bias is when an ‘infertile’ couple adopt a child. There is a totally incorrect myth that they are then much more likely to conceive themselves. There is absolutely no statistical evidence for this, it only appears that way because we will notice the fact they they conceive and put more weight on it. The fact that some ‘infertile’ couples conceive without adopting and some ‘infertile’ couples adopt and never have children of their own escapes our attention.

This example of infertile couples also happens because of:

external attention bias

In this case the news media is much more likely to bring such a story to our attention because it is ‘newsworthy’ and so our attention is drawn to it by external sources. Other external sources include our friends and other people who we know. The attention bias is compounded in these cases because other people have their own internal and external attention biases.

combating attention bias

To remove or at least decrease attention bias we can try to spot our own attention biases. We can do this by noticing when we are looking for specific things (e.g. cars with the same colour as ours) and once we realise this we can decide to look for more balanced data.

Another way of deceasing attention bias is to choose multiple focuses for our attention. For example in our car example we could look for all red cars, then all green cars etc. This will help to balance our attention biases and assist with the correct interpretation of the data on which we base our decisions.