Using Mechanical Turk to recruit participants for internet intervention research: experience from recruitment for four trials targeting hazardous alcohol consumption

Table 2 Demographic characteristics of the four recruitment samples

	Trial 1 (n = 871)	Trial 2 (n = 3244)	Trial 3 (n = 3456)	Trial 4 (n = 3536)	p
US Resident, % (n)	–	–	99.0 (3421)	98.8 (3492)	0.360
Mean (SD) Age	36.4 (10.5)	35.5 (10.9)	34.5 (10.7) ^{a, b}	33.5 (10.4) ^{a, b, c}	<0.001
Male, % (n)	47.9 (417)	46.8 (1516)	42.8 (1478) ^{a, b}	43.2 (1527) ^b	0.001
White, % (n)	82.9 (722)	80.6 (2615)	79.1 (2735)	82.3 (2910) ^c	0.003
Some post-secondary education, % (n)	72.9 (635)	71.1 (2305)	72.0 (2489)	71.2 (2517)	0.618
Married / Common-law, % (n)	53.7 (468)	50.1 (1624)	48.1 (1661) ^a	47.5 (1679) ^a	0.003
Full-time employed, % (n)	70.3 (612)	64.7 (2098) ^a	59.7 (2062) ^{a, b}	57.2 (2022) ^{a, b}	<0.001
Family income less than US$20000, % (n)	27.6 (240)	17.3 (560) ^a	19.3 (667) ^a	20.5 (724) ^{a, b}	<0.001

Group differences were computed using chi-squares and one-way ANOVA tests. To determine specific group differences, post-hoc tests were performed with Bonferroni corrections; ^a significant difference compared to Trial 1; ^b significant difference compared to Trial 2; ^c significant difference compared to Trial 3

ISSN: 1471-2288