High prevalence of personality disorders among healthy volunteers for research: implications for control group bias.