Table 3.

Sex difference in percentages for examples 1 and 2.

Method	Daily newspaper reading (N = 14)					Primary school enrollment (N = 117)
	$\widehat{\beta }$	S.E.	t	p-value	95% CIs	$\widehat{\beta }$	S.E.	t	p-value	95% CIs
CLT	9.17	2.57	3.58	0.0034	[3.63, 14.71]	1.03	0.36	2.86	0.0050	[0.32, 1.75]
HC0	9.17	2.47	3.71	0.0026	[3.83, 14.51]	1.03	0.36	2.87	0.0048	[0.32, 1.75]
HC1	9.17	2.57	3.58	0.0034	[3.63, 14.71]	1.03	0.36	2.86	0.0050	[0.32, 1.75]
HC2	9.17	2.57	3.58	0.0034	[3.63, 14.71]	1.03	0.36	2.86	0.0050	[0.32, 1.75]
HC3	9.17	2.66	3.45	0.0043	[3.42, 14.92]	1.03	0.36	2.85	0.0052	[0.32, 1.75]
percBS	9.17	–	–	–	[4.55, 13.80]	1.03	–	–	–	[0.35, 1.77]
BCa	9.17	–	–	–	[5.21, 15.02]	1.03	–	–	–	[0.37, 1.81]
Winsorize	6.30	1.66	3.79	0.0068	[2.37, 10.23]	0.35	0.18	1.91	0.0601	[–0.015,0.72]
Trim	5.76	1.61	3.56	0.0090	[1.96, 9.59]	0.25	0.18	1.35	0.1820	[–0.12, 0.61]

S.E., standard error; CI, confidence interval; CLT, central limit theorem; HC, heteroscedastic consistent method; percBS, percentile bootstrap; BCa, bias corrected and accelerated bootstrap. Estimates are in the direction of male percentages minus female percentages. Winsorized and trimmed means pertain to modifying about 20% of the tail distributions; 20.43% for Example 1 and 20.51% for Example 2.