One resource says that increasing kVp increases patient dose as the result of more Compton scatter being produced.
Most other resources state that increasing kVp decreases patient dose because there is a more penetrating x-ray beam and less photoelectric effect.
To clarify the first statement, the amount of Compton scatter over the range of photon energies used in diagnostic radiology is essentially constant (independent of kVp). There is not more Compton scatter at higher kVp. It is because the probability of photoelectric effect interactions decreases with increasing energy that therefore Compton scatter accounts for a greater
proportion of photon-matter interactions at higher kVp. Both Compton scatter and photoelectric effect contribute to patient dose. Because there is less photoelectric effect (in biologic tissues) at higher energies (at the higher end of the spectrum used in diagnostic radiology), the
probability of energy absorption is decreased at higher kVp. Lower probability of absorption means lower patient dose.
However, the kVp not only affects the
quality of photons but also
quantity. Remember that kVp is a setting on your x-ray tube. It changes not only the energy of photons produced (quality; keV, which affects probability of energy absoprtion), but also the number of photons produced (quantity). More photons means more patient dose.
The balance of quality and quantity when you change kVp, and therefore the ultimate effect on patient dose, is explained in further detail in the Radiopaedia article above. The short answer is it's complicated in the real world because you will also adjust other parameters (mAs) in conjunction to achieve a certain image quality. As
@SirGecko explained, if you increase only kVp and hold everything other parameter on the machine constant, dose increases.