Miner does not use 1.88 factor in calculations, using it is a bad idea because it is theoretical factor and real implementation may be not so good. Therefore miner just counts all solutions found, and time that was spent. If algorithm implementation has a bug that causes 1.8 factor, it will cause less number of solutions and less calculated hashrate. So this idea is wrong.
I still think it's related to pool. What pool do you use? I will put all my polaris cards to it and check 2-3 daily hashrates to see hashrate deviations.
flypool
I mined on flypool for a long time with other miners (SA5, Optiminer), and the pool hash rate was equal to my raw hash rate at the miner. So, unless something has changed on the pool recently, I wouldn't think it's their calculations causing the discrepancy.