Table 3. Estimated biases for different observation years and outcome categories (using unrounded estimates of the regression coefficients from Table 1).

		Period Before Survey
Obs. year	Direct	< 5 Years	5 - 10 Years	10+ Years
1954	Yes	−1.19
	No	−0.74
1961	Yes	−1.06	0.01	0.24
	No	−0.61
1970	Yes	−0.89	0.18	0.40
	No	−0.44
1980	Yes	−0.70	0.37	0.59
	No	−0.25
1994	Yes	−0.44	0.63	0.85
	No	0.01
1999	Yes	−0.35	0.73
	No	0.11
2004	Yes	−0.25
	No	0.20