This measures the systematic difference between fulls and summed partials (if there are any fully recorded observations).
Fractional Bias = S (<Ifull> - Ipartial) / S <I>
Typically, its value is negative, ie the summed partials are bigger than the fulls, due to truncation of diffuse scattering tails on fulls (a partially-recorded observation is recorded over at least twice the angular range of a full)
Negative bias may be corrected (very approximately) by the TAILS correction
A positive bias generally indicates some serious problem with the data collection (shutter mistiming, rotation speed variation)