Are survey responses to HBD questions honest?

In yesterday's post on the popularity of HBD, readers raised the reasonable concern that a person might be unwilling to tell an interviewer face-to-face (that's how the GSS is done) that he thinks that blacks are innately less intelligent than whites.

Fortunately, the GSS allows us to vary interview conditions in two ways: 1) was it a face-to-face interview or one done over the phone; and 2) was the interviewer white or black?

Here are the percentages who agreed that blacks are innately less intelligent:

Percent answering yes--white sample (N = 2,044)

White interviewer over the phone 3.4
White interviewer in home 8.5
Black interviewer over the phone 7.7
Black interviewer in home 4.4

Not much evidence here for the view that people are responding to the conditions of the interview. Whites should feel most comfortable giving the pro-HBD response to a white interviewer over the phone, but this condition gets the fewest affirmative responses. Myself, I would have the hardest time expressing my position to a black person, but the second-highest rate of positive response was given to black interviewers over the phone.

On the other hand, interviews over the phone are uncommon, so cell sizes are low. If we focus only on the in-home interviews, there is more support for the idea that people tend to give socially approved answers. 8.5 percent of respondents answered yes to a white interviewer at home, compared to only 4.4 percent of those responding to a black person. The difference, however, is not statistically significant. Whites might be responding to conditions, but keep in mind that only 9 percent of interviews were given by blacks.