An alternative explanation of the cononsolvency of poly(N-isopropylacrylamide) in water-methanol solutions