Table 3.
Comparison of recruitment methods.
| Recruitment metric | Recruitment method | |||||||||
|
|
News feed | Clinic | Snowball sampling | Other methods | ||||||
| Days of action, n | 20 | 27 | 48 | 271 | ||||||
| Direct incremental marketing costs | ||||||||||
|
|
Money spent, US $ | 328.85 | 672.40 | 0.00 | 0.00 | |||||
|
|
Impressions, n | 28,274 | 40 | N/Aa | N/A | |||||
|
|
Cost per impression, US $ | 0.012 | 16.81 | 0.00 | 0.00 | |||||
|
|
Unique viewers, n | 11,738 | 40 | N/A | N/A | |||||
|
|
Cost per unique viewer attracted, US $ | 0.028 | 16.81 | 0.00 | 0.00 | |||||
|
|
Inquirersb, n (% of unique viewers) | 274 (2.3) | 32 (80.0)c | 13d | 7d | |||||
|
|
Cost per inquirer attracted, US $ | 1.20 | 21.01 | 0.00 | 0.00 | |||||
|
|
Completers of webform, n (% of inquirers) | 32 (11.7) | N/A | N/A | N/A | |||||
|
|
Responsive volunteerse, n (% of inquirers) | 20 (7.3) | 20 (62.5)c | 12 (92.3)f,g | 5 (71.4) | |||||
|
|
Cost per responsive volunteer attracted, US $ | 16.44 | 33.62 | 0.00 | 0.00 | |||||
|
|
Eligible volunteersh, n (% of responsive volunteers) | 8 (40.0) | 2 (10.0)i | 8 (66.7)g | 2 (40.0) | |||||
|
|
Cost per eligible volunteer attracted, US $ | 41.11 | 336.20 | 0.00 | 0.00 | |||||
| Other costs, US $ | ||||||||||
|
|
Start-up | 189.76 | 32.38 | 32.38 | 16.19 | |||||
|
|
Contacting and explaining study to inquirers | 86.40 | 92.80 | 35.10 | 18.90 | |||||
|
|
Screening responsive volunteers for eligibility | 162.00 | 162.00 | 97.20 | 40.50 | |||||
| Total costs: cost per eligible volunteer enrolled, US $ | 95.88 | 479.79 | 20.59 | 34.92 | ||||||
aN/A: not applicable; this value was not traceable.
bDefined as person who clicks (news feed advertisement) or requests more information from the research team (other recruitment methods).
cGreater than news feed by chi-square (inquirers: χ21=919.8, P<.001; responsive volunteers: χ21=72.1, P<.001).
dThe percentage cannot be calculated because the number of unique viewers (ie, the denominator) was not traceable.
eRefers to proportion of clickers volunteering (called conversion in the literature) (3-way P<.001).
fGreater than news feed by Barnard test (P<.001).
gGreater than clinic by Barnard test (response rate: P=.048; eligibility: P<.001).
h3-way P=.003. Not significant for news feed versus snowball sampling (Barnard test P=.23).
iLess than news feed by Barnard test (P=.03).