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

Table 1 Final baseline survey sample size calculation

	Trial 1	Trial 2	Trial 3	Trial 4
Duration of trial recruitment	3.2 h	7 days	9 days	32 days
Total # (N) of surveys accessed	1252	4943	5412	5846
Total # of surveys removed % (of N)	30.0 (381)	34.4 (1699)	36.1 (1956)	39.5 (2310)
# (n) Reasons for removal [% of n]	381	1699	1956	2310
Blank surveys	1.0% (1)	1.9% (33)	1.9% (38)	1.5% (34)
Duplicates and suspicious data ^a	9.2% (35)	11.8% (200)	3.6% (71)	9.5% (220)
Incomplete eligible screeners	8.9% (34)	2.8% (47)	4.1% (81)	5.4% (125)
Ineligible screeners ^b	50.7% (193)	54.3% (923)	65.6% (1284)	58.8% (1359)
Participants did not pass all attention checks	27.8% (106)	27.4% (466)	22.1% (432)	21.0% (485)
Participants who indicated they did not respond honestly	2.4% (9)	1.8% (30)	2.6% (50)	3.8% (87)
Final baseline survey sample size	871	3244	3456	3536

^a Participants who accessed the survey multiple times, but only submitted one complete survey were retained once in the final sample. Those who completed the entire survey more than once were considered to have provided suspicious data and are not included in the final sample size
^b Ineligible screeners refers to those who did not proceed to the baseline survey (i.e. were <18 and reported consuming alcohol less than once per week)

ISSN: 1471-2288