Pitfalls in the Measurement of the Return to Education: An Assessment Using Swedish Data