I guess this is true for whole numbers, and in case your processor uses base-2, the set of dyadic rationals, i.e. the numbers which can be represented exactly as a fraction of the from i/2^j where i and j are integers. For example 1.5 is 3/2^1 as a fraction.
But on a processor with a different word-size, and different floating point format, the constant 1.5 might not be promoted correctly, unless you add the kind specifier suffix.