The consequence of omitting all AP-levels lying in between the extreme values is basically a loss of information compared to the original day,and a somewhat fuzzy reference level which consists of all AP-levels between the anchor levels.But as it consists of a continuous range of AP-levels,and it is equal for all analyzed attributes,the dummy regression weight can be interpreted as average expected differences in OCS as a result of very low or very high AP, compared to the expected level of OCS in case of non-extreme AP-levels.