Yes sir.
Cooking the polling books 101.
Decrease the numbers polled and only include those geographic areas likely to give the desired result.
If someone like me can figure them out, are they stupid enough to think they are fooling anyone?
Did you ever consider the possibility that because of the holiday week, there were just a lot fewer people willing to take the time to be polled? Not everything has a malicious intent... The number of people polled when it showed Trump at 31% (358) was not that different from the number polled when it showed him at 38% (381). In fact, the sample size on the day Trump hit 43% was only 464 people.
And remember that this is a 5 day rolling poll. What happened is that Trump had a set of 3-4 VERY good polling days (highest was 47.6%), but those have now rolled off the average. Looking at the Reuters polling site the last few individual days had Trump at 32%, 27% and 26%.