What can I learn from correspondence analysis?
Correspondence analysis reveals the relative relationships between and within two groups of variables (or data points from your survey). It’s a particularly helpful analysis technique when you want to measure similarities / differences between brands, and the strength of brands in terms of their relationships with different attributes.
You can of course obtain these findings from a set of data tables, but by using correspondence analysis we can create a visual brand map which enables us to compare brands at a glance (which adheres to one of the key principles of good data visualisation).
Brands are most strongly associated with the attributes closest to them on the map. If brands are situated close to each other, it means they have a similar image or profile in the market. From brand maps we can see which brands have a unique position, see which brands are competing in the same space, and also identify any unclaimed spaces in the market.
Example brand map
How do I interpret brand maps?
There are some quirks to the interpretation of brand maps, which make them a little trickier than they first appear….
When comparing brands, we can look at the distance between them – brands closer together on the map are similar, brands further away from each other are more differentiated. The same is true when we want to compare attributes – in our example trustworthy and honest are close together, and therefore very similar attributes. Makes sense.
But when we want to look at the relationship between brands and attributes, things become a little bit more complicated. To do this, we need to pay attention to how far away they are from the centre of the map, and the angle between the brand and attribute in question:
- The longer the line between the centre of the map and the brand/attribute, the more differentiated something is
- The smaller the angle between the lines of the brand and attribute, the more closely they are associated
In the example above, we’ve shown lines for Brand E and the attribute old-fashioned – in this example you can see that both lines are relatively long and the angle is relatively small, indicating a strong, differentiating relationship (although not a very good one!).
Finally, it’s important to remember that correspondence analysis only shows us the relative relationship between and within brands and their attributes, it does not show us which brands score most highly for which attributes – you still need the data tables for that. For example, although Brand A is really close to ‘trustworthy’ on our map, all brands actually score quite low for this attribute – it just so happens that Brand A is the best of a bad bunch. As such, it’s important to use brand maps as the starting point for further analysis, or as a summary tool once the finer details have been explored.
Hopefully you’ll uncover much more positive results about your brand that the examples we’ve selected above – there’s only one way to find out…
