The effect of nonresponse bias in undergraduate student surveys