Compensating for noncoverage of nontelephone households in random-digit-dialing surveys: A comparison of adjustments based on propensity scores and interruptions in telephone service