Each of the examples is between 4 and 6% off, in a race that would be harder to determine (since there’s at least 3 or 4 candidates per primary). In each of those examples the other candidates were within a percentage or 2 off, which places the polls within the margin of error. They most likely are anomalies, since the discrepancies happen to random candidates (it could be fraud, but the other candidates’ percentages being accurate in those primaries makes it unlikely).
This contrasts the Democratic primaries, such as in Ohio, where both Clinton and Sanders are around 5% off. Also, the trend is different too — instead of it skewing towards one candidate, or every candidate except Trump, the Republican differences affect each of them randomly. So, I concluded that those are anomalies, since 8 of the 13 primaries I have data for are spot on, and in the others, only one candidate’s data is off each time, making it not outside the margin of error. If I miscalculated and one or two of them is outside the margin of error, they most likely still are anomalies, but it would be worth looking into.