Abstract
Data from the General Social Survey, 1972–1996 were examined to see whether index reliability changes when the don't know respondents are included. On the abortion and civil liberty attitude indexes, reliability remained the same regardless of whether the don't knows were included or excluded. Reliability actually increased somewhat when the don't know respondents were included on two cognitive indexes measuring vocabulary knowledge and reasoning performance. When don't knows are excluded by list-wise deletion, attrition occurs very disproportionately among the least educated and oldest groups. Since there is no trade-off of increased reliability for decreased representativeness, list-wise deletion of cases is hard to justify.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
