nearly significantP = 0.0502
If self- and non-self- reports were tested separately for the effect of study characteristics (one characteristic at a time), a significant effect was found only in self-reports for year when survey was conducted (k = 7, b = −0.1425±0.0519, P = 0.006) and a nearly significant effect was found again in self-reports for survey delivery method (k = 7, b = −1.2496±0.6382, P = 0.0502) Sensitivity analysis Self-report admission rates varied between 1.65% -following the removal of Kalichman and Friedman (1992) [35] - and 2.93% -following the removal of Martinson et al.