The point I with to make is that the map that is created shows LSOAs next each that are either end of the deprivation scale. This can be seen in the map below;
Now I need to transfer the IMD scores from LSOA to my grid squares. There are 4760 LSOAs in London (5 more in the City of London) and there are 26116 grid squares so many grid squares fall completely within a LSOA: these in my calculations are allocated the IMD score of the LSOA it falls within. Others fall across two all more LSOAs: these in my calculations are allocated the average of the LSOA scores the grid square intersects. The maps below show these intersection counts, the maximum is 9.
The result of this is what my colleague Dan called naive smoothing; meaning a gradual transition from less to more (or visa verse) deprived areas. Whether this is a more realistic and intuitive mapping of deprivation I will leave to another occasion. This method obviously introduces fuzziness as the calculation assumes that a grid square that intersects LSOA does so with an equal area in each which is obviously not the case. The resulting map are shown below.
I have also gone up the scale from LSOA to Wards with IMD scores. As LSOA fall entirely inside the 624 wards no advantages/problems of naive smoothing. The maps are below;