When measuring deflection on such a thin test piece are you getting a less than optimal result because of the extra resin on the surfaces ? An accurate test would require you to cut the piece to size, to allow aggregate to be evenly distributed right through, but this would weaken the material too.