An invitation to use craigslist ads to recruit respondents from stigmatized groups for qualitative interviews

Craigslist.org is a website devoted to classified advertisements as well as discussion forums with locations in more than 700 sites in 70 countries. At an estimated 30 billion page views per month and more than 50 million new classified ads posted monthly, craigslist ads have the ability to reach a wide audience. Although wildly popular, no studies to date have investigated the use of craigslist ads for qualitative research study recruitment. In this research note, I offer my own experiences from 2011 to 2012 using craigslist ads (N = 77) to recruit obese respondents (N = 38) for qualitative interviews in one major metro area in the southern United States. I also discuss some advantages and limitations of using craigslist as a recruitment tool. Overall, I invite social science researchers to consider craigslist as an innovative tool to recruit respondents for qualitative research.

Recruiting respondents from stigmatized groups for qualitative interviews can be particularly challenging, and as a result, many innovative recruitment tools have been designed including snowball sampling, respondent-driven sampling, and even convenience venue-based sampling methods (Abdul-Quader et al., 2006;Goodman, 2011;Lau et al., 2011). While such innovations are important, others are recognizing the significance of Internet websites as recruitment tools for respondents from stigmatized groups (e.g. Durso and Latner, 2008;Im and Chee, 2004), and most recently, researchers have used craigslist to recruit respondents for online surveys (e.g. Grov, 2011;Ramo et al., 2010). Although wildly popular, no studies to date have investigated the use of craigslist ads for qualitative research study recruitment.

About craigslist
In 1995, Craig Newmark began 'craigslist', an email distribution list featuring events in the San Francisco Bay Area, and in 1996, craigslist became a web-based service offering multiple classified categories (Freese, 2011). In 2000, craigslist expanded to nine other US metro areas. As of 2013, more than 60 million people use craigslist each month in the United States alone. Internationally, craigslist supports 12 language platforms and is located in more than 70 countries (Craigslist.org/about, n.d.). A leading provider of free global web metrics ranks craigslist as 49th overall among websites worldwide and 8th overall among websites in the United States (Alexa.com, n.d.).
The craigslist website was initially designed to be minimalist and remains so in 2012 . Free from commercial ad-space and animated clutter, the craigslist website is devoted to classified advertisements and discussion forums. Advertising on craigslist is relatively easy. 1 Individuals can post advertisements for free in most cases. 2 Postings are divided into nine sections, which are further divided into more than 100 subsections (see Figure 1). 3

The use of craigslist as a recruitment tool: initial investigations
In 2011, I conceptualized a research project involving qualitative interviews with obese airline patrons. Because my sample was to comprise respondents from a stigmatized group, my recruitment methods had to be respectful. Others have suggested that online community discussion groups and snowball sampling via email might be effective methods of recruitment to best understand obesity stigma (e.g. Durso and Latner, 2008). For this research project, my first priority was to have respondents self-select as 'obese' to be most sensitive to the respondents (Durso and Latner, 2008). The second priority was to have respondents with diverse socioeconomic backgrounds. Finally, I was interested in utilizing an innovative recruitment method for qualitative research with stigmatized populations. Considering these priorities, I further investigated craigslist as a recruitment tool.

Craigslist sampling limitations
Recruitment-by-craigslist has four important sampling limitations to recognize at the outset. First, the sample will be a convenience sample (nonrandom and nonrepresentative sample) in a major metro area included on craigslist. Thus, rural areas and metro areas that are not included on craigslist will not be easily reached. Second, the sample population will comprise persons who have access to and utilize the Internet. Research about adult Internet users show that both men and women are equally as likely to be Internet users in the United States, although younger people are more likely to use the Internet when compared to older people with 28 percent of adult Internet users between the ages of 18 and 29 years, 40 percent aged 30-49 years, 23 percent between the ages of 50 and 64 years, and 9 percent over the age of 65 years (McDonald and Cranor, 2010). Thus, recruitment-by-craigslist may yield younger respondents when compared to other types of recruitment methods. A third limitation of recruitment-by-craigslist is that respondents must be aware of and visit craigslist.org to view the recruitment posting. With more than 50 million craigslist users per month (Craigslist.org/about, n.d.), the awareness of craigslist may be widespread, even so, individuals must visit craigslist.org and browse the sections to view the recruitment posting. As a result, only those who are actively browsing craigslist.org in the metro area in which the advertisement is posted will have the potential to be recruited for a study. The fourth limitation to recruitmentby-craigslist is that respondents must have the ability to respond to craigslist ads by email and/or phone. These four limitations must be considered as sampling restrictions at the outset of any study that will utilize recruitment-by-craigslist methods.

Recruitment-by-craigslist research project design: My Obesity Study as an example
Although each research project is unique, I offer My Obesity Study as an example recruitment-by-craigslist research project design. The following discussion is not intended to be a 'how to' for recruitment-by-craigslist projects, rather, it is intended to offer an example research project design to inspire and/or inform other social science researchers about recruitment-by-craigslist methods in qualitative research. First, I discuss my experiences with designing and executing My Obesity Study, and next, I offer a discussion of limitations and advantages of recruitment-by-craigslist methods.

Craigslist ad text in My Obesity Study
The craigslist ad text in My Obesity Study (see Figure 2) is described below. I developed the text based on Lloyd and Gunter's (2009) suggestions 4 and my previous experiences using craigslist for personal nonresearch postings (i.e. selling household items).
Headline. Since the headline is often the only text individuals view when browsing through craigslist ads, the headline was designed to include as much straightforward information as possible within the 70 character limit: 'University Researcher Seeks Adult Volunteers for Interview Study, $25'. This headline included several important components: (1) university recognition as a researcher collecting data; (2) adult volunteers as the requested respondent for the study; (3) the type of study, 'interview'; and (4) the dollar amount of the recruitment incentive, '$25'. Although past research has investigated the effectiveness of headlines for advertisements recruiting job applicants in print media (i.e. newspapers, Kaplan et al., 1991), no studies to date have systematically examined how variations of craigslist headline text may contribute to the effectiveness of recruitment-by-craigslist methods for qualitative interviews.
Body. The second component 5 of the craigslist ad is the body of the text. First, proof of legitimacy must be established since most people are aware of scams elicited through fraudulent postings and misleading advertisements on the Internet (McDonald and Cranor, 2010). Although craigslist.org has responded to the myriad scams that have been enacted using craigslist ads, 6 an awareness of craigslist scams may make potential respondents reluctant to respond to craigslist ads. As a result, two strategies were utilized to demonstrate the legitimacy of this research project. First, my name, the University at which I was employed, and the details about the project were fully disclosed in the craigslist ad. This way, skeptical or hesitant potential respondents could search/locate me at the University at which I worked and could also verify the legitimacy of my name/ employment status on the University's website. 7 Second, I included the URL to my personal professional website page that described My Obesity Study. 8 Second, because there are numerous ads on craigslist, it is necessary to make your ad appealing. In My Obesity Study, because individuals may be reluctant to identify themselves as 'obese', it may be challenging to recruit them to participate in a qualitative study in which they are required to discuss their weight (Durso and Latner, 2008;Lewis et al., 2011). To combat this potential problem, a recruitment incentive was offered. 9 Recruitment incentives (typically in the form of cash or gift cards/gift certificates) are often utilized in studies involving sensitive or stigmatized subject matter (e.g. Deren et al., 1994;Roth et al., 2012). There are mixed results regarding the use of incentives for recruitment-by-Internet research. In their study, using gift certificate incentives ranging from US$5 to US$15 to recruit respondents through multiple Internet technologies, Koo and Skinner (2005) had an extremely low response rate, suggesting as their title states: 'a case study with disappointing results'. Others suggest that the use of cash/gift card incentives may actually increase response rates in recruitment-by-Internet research (Im and Chee, 2004). No studies to date have examined how incentivizing recruitment-by-craigslist for qualitative studies might be related to response rates. Third, the text should describe the logistics of the interview. When advertising for in-person qualitative interviews on craigslist, this typically implies a situation in which strangers (respondents) will meet with another stranger (researcher(s)). This can lead to anxieties and safety concerns for both the respondents and researcher(s). Furthermore, the location of the interview site can affect the rapport between the researcher and the respondent (Elwood and Martin, 2000). For example, if the interview site is the 'home turf' of either the researcher or the respondent, this can imply a power differential, thus a more 'neutral' location might be ideal. In addition, if the respondent is given some choice as to where the interview will take place, this may further empower the respondent to feel positive about the interview experience (Elwood and Martin, 2000). Finally, since respondents are usually traveling to the interview site, it may be important to choose an interview site that is located near public transportation. In My Obesity Study, interviews were conducted at public libraries of the respondents' choosing during hours of operation because I believed that public libraries were easily accessible by public transport and offered a neutral, safe, and appropriate location for interviews. 10 These details were specified in the craigslist ad to ideally alleviate any concerns or questions potential respondents would have about the logistics of the interview process. 11

Posting craigslist ads to sections/subsections in My Obesity Study
After the craigslist ad text is written, the next step is to decide the section/subsection in which to post the advertisement. After searching craigslist ads in multiple locations, it appears that ads for one-time paid research volunteer positions are predominately posted in the 'volunteers' subsection of the 'community' section. This is the location in which all ads for My Obesity Study were posted. Although no studies could be located, it would be especially informative to investigate the success of posting ads in multiple sections/ subsections to understand which generate the most responses.

My Obesity Study live on craigslist: May 2011 to May 2012
After the craigslist ad is posted and is 'live' on the Internet, potential respondents can view the ad and contact the researcher as described in the ad. 12 Craigslist ads expire after 45 days; however, in order to be listed at the top (or close to the top) of the postings for the day, and in order to be seen most visibly by potential respondents, researchers should post multiple ads on craigslist. Craigslist allows individuals to post one ad to one section/subsection in one metro area, no more than once every 48 hours 13 (Craigslist. org/about, n.d.). This means that the maximum number of ads a researcher can post in one metro area is approximately 15 ads per month. In some metro areas, posting 15 ads per month in one section/subsection would result in an 'overkill' of ad postings since this research recruitment ad may be the only ad posted in this section/subsection over several days. Indeed, the volunteers subsection can be quite lively in larger metro areas but slower in smaller metro areas . As a result, researchers should monitor the section/subsection to see what is most appropriate as far as ad posting frequency as to not overwhelm a section/subsection with too many posts related to the research study.
In My Obesity Study, a total of 77 ads were posted on craigslist in one southern US major metro area. The 77 ad postings resulted in 38 total interviews over approximately 12 months (23 May 2011 to 28 May 2012). An investigation of the monthly ratios of interviews-to-postings suggests some informative patterns (see Figure 3). For example, December 2011 had a high ratio of interviews-to-postings (7:3). This may suggest that December is a prime month for responding to craigslist ads recruiting paid research volunteers. December might be a popular month because potential respondents may be looking for ways to earn money to purchase gifts during the holiday season. In addition, potential respondents might also have more vacation time in December, and thus, they may have more time to browse craigslist for potential volunteer opportunities. May 2012 also had a large number of interviews, which may be due to the fact that the metro area in which interviews were conducted has a large university with approximately 40,000 students, faculty, and staff. Since the school year ends in May, students, faculty, and staff may be more likely to browse craigslist for potential volunteer opportunities during the month of May. However, it is important to note that while these findings may be informative, the actual relationship between postings and interviews may be purely speculative. In other words, although interviews-to-postings ratios differed month to month, the number of postings may not actually affect the number of interviews garnered.

Flagging ads on craigslist
While posting ads can be conceptualized as a systematic endeavor (e.g. a researcher might make a plan to post an ad every 48 hours), 'flagging' can interrupt this process.
The flagging system was designed by craigslist to allow any visitors to quickly identify illegal and inappropriate ads by anonymously flagging them as misplaced, prohibited, or spam/overpost (Craigslist.org/about, n.d.). When a post is flagged, an automatic email is generated and sent to the poster, and the ad is removed. Individuals can anonymously flag posts and may continue to flag posts daily. This means that a researcher posting ads does not know who is flagging his or her posts and does not know why an individual may be flagging his or her posts as inappropriate.
In My Obesity Study, 12 ads (16%) were flagged as inappropriate. To respond to the flagging and removal of an ad, there are two strategies. First, a researcher can repost the ad. In My Obesity Study, I reposted all flagged-and-removed ads within approximately 3-4 days of the flagging. Unfortunately, this flag-and-repost system is not always successful. Sometimes, the process continues with flagged ads being reposted and then flagged again several times before one party 'surrenders'. As a result, a researcher might continue this flag-and-repost system multiple times before the ad remains 'live' on craigslist and the flagger discontinues flagging the ad. The second strategy to respond to the flagging and removal of an ad is to adjust the text in the craigslist ad. In My Obesity Study, I made a concerted effort to 'prove' my craigslist ad was legitimate by listing my credentials and including the URL to my personal professional website, which also described the research project. Even so, flaggers still flagged this ad as inappropriate. As a result, I adjusted the text on 11 November 2011. Specifically, a statement was added to the top to all postings to respond to flagging (see Figure 2, Note 1). Unfortunately, the addition of this statement did not affect the flagging of this ad. Before this statement was added, 6 out of 40 ads were flagged. After this statement was added, 6 ads out of 37 were flagged. Thus, at face value, the addition of this statement did not appease the flagger(s). 14 Such findings suggest that more research is needed to understand flagging and how to appease potential flaggers.

Institutional Review Board involvement and ethical considerations in recruitment-by-craigslist research
In My Obesity Study, the university's Institutional Review Board (IRB) reviewed and approved this recruitment method; however, several steps were taken to protect respondents. First, informed consent was established by having each respondent sign and date an IRB-approved informed consent sheet that indicated the detailed parameters of the project, confidentiality information, the voluntary nature of the study, and contact information for the researcher and a local behavioral health center should the respondent need to discuss his or her feelings further. Two signed copies of the informed consent sheet were required (one for the respondent to keep and one for the researcher to file). Second, after data collection was complete, respondent identification information was deleted, and all data were stored in a locked cabinet to preserve the confidential nature of the study.
Even though My Obesity Study was IRB approved, there may also be a few additional ethical concerns to consider in recruitment-by-craigslist research for stigmatized populations. First, the wording and language of a craigslist posting should be sensitive in nature. For example, in My Obesity Study, the mere use of the word 'obesity' might have been too stigmatizing for some. Thus, ethical considerations regarding the text for the craigslist posting should be paramount. Second and related, because recruitment-by-craigslist research for stigmatized populations requires potential respondents to self-select into a stigmatized group, this experience could be negative for some, even when an incentive is offered. Third, although public libraries offered a safe and neutral location for interviews, there may still be privacy concerns with others being able to overhear the interview content. Overall, studies of the ethical parameters of the use of the Internet for research are still in their infancy, and more work is certainly needed to guide future research (Bakardjieva and Feenberg, 2001).

Craigslist and My Obesity Study
Overall, recruitment-by-craigslist was successful in My Obesity Study. I was able to interview 38 respondents from one southern US major metro area over approximately 12 months. In addition, the use of craigslist allowed me to successfully recruit respondents from a stigmatized group in a way that allowed the respondents to self-select into this stigmatized category, ideally helping respondents to feel more at ease about the interview. I was also able to recruit respondents with diverse backgrounds. 15 Finally, I was able to utilize an innovative recruitment method for research with stigmatized populations. The results from this endeavor have produced a wealth of qualitative interview data that will be used for multiple projects. Furthermore, I hope my use of recruitmentby-craigslist may inspire future work.
While My Obesity Study was a successful exploration of recruitment-by-craigslist methods, it is important to note that this is only one example of this type of study, and therefore, results may not be generalizable to other research using recruitment-bycraigslist methods. Specifically, there are seven highly specific qualities of My Obesity Study worth noting. First, My Obesity Study involved recruitment for a qualitative study with adults, thus it is unknown how recruitment-by-craigslist for those under the age of 18 years and for studies recruiting for quantitative research might work. Second, recruitment for My Obesity Study took place in one US metro area. As a result, it is unknown how recruitment-by-craigslist might work in other US metro areas and in other countries. Third, the text of the craigslist ad used in My Obesity Study was unaltered throughout the study (with the exception of the proof of legitimacy statement). Future studies might vary the text of craigslist ads and proof of legitimacy statements to see which ads garner the most respondents to help inform future studies about successful recruitment text. Fourth, the craigslist ad used in My Obesity Study was posted in one section/subsection (community/volunteers). As a result, it is unknown how posting ads in other sections/ subsections might affect responses to craigslist ads in recruitment-by-craigslist research. Fifth, this study used a recruitment incentive; thus, other recruitment-by-craigslist research without a recruitment incentive would certainly complement this study. Sixth, in My Obesity Study, respondents self-selected into a stigmatized group and were required to discuss their weight with a stranger/researcher. The sensitivity of weight-related conversations may preclude some individuals from self-selecting to be respondents for My Obesity Study. As a result, it is unknown how recruitment-by-craigslist might work with nonstigmatized populations. Finally, in My Obesity Study, respondents were required to have access to transportation to meet for an in-person interview in a public library. Thus, research about recruitment-by-craigslist for telephone and/or Skype interviews and for interviews to take place at sites other than public libraries would be informative.

Concluding remarks: the future of craigslist
This research note offered one example of a study using recruitment-by-craigslist methods for qualitative interviews. As craigslist becomes more popular over time, bloggers (e.g. Blog.craigslist.org, n.d.) suggest many innovations to craigslist, such as refining the flagging process, including more sections/subsections, expanding to more metro areas and countries, incorporating more fees, and creating/updating smartphone applications. These potential changes to craigslist may also affect the ways craigslist can be used to recruit respondents. Overall, I invite other researchers to consider using recruitment-bycraigslist methods for qualitative interviews.

Funding
The research project discussed in this article was financially supported by the Vice President for Research of the University of Oklahoma. 'for sale', and 'personals' sections, although images were not included in My Obesity Study. 6. Craigslist has been host to several scams related to car purchases (Craigslist.org/about/scams, n.d.;Freese, 2011). Their website (Craigslist.org/about/scams, n.d.) indicates 'common-sense rules' to sidestep scammers when using craigslist. In addition, the highly publicized murderer, Philip Markoff, media-named 'The Craigslist Killer' may also make individuals hesitant to browse/respond to craigslist ads (Freese, 2011). 7. While I did include enough information in the craigslist ad for respondents to search/locate me on my University website, it is unknown how many potential respondents did so. As Bakardjieva and Feenberg (2001) note, potential respondents utilize the Internet to investigate researchers. A few respondents I interviewed mentioned that they did view my personal professional website and/or my University website and that this made them feel more comfortable contacting me for the interview, although the vast majority did not mention this in our meeting. 8. The inclusion of the URL to a website further describing the volunteer activity has been strongly recommended by craigslist specialists to best elicit responses to volunteer ads ). Google analytics results from my personal professional website showed that during the time of data collection, the page describing My Obesity Project (which was the URL included in the craigslist posting) had an average of 22 page views per month with the least number of views in October 2011 (6 page views) and the most number of views in November 2011 (42 views). Google analytics results also showed that there was an average of 9 referrals per month from craigslist to my personal professional website during data collection ranging from 0 referrals in May 2011, January, March, and April 2012 to 28 referrals in November 2011. 9. A grant from the Vice President for Research at the University of Oklahoma supported the purchase of gift cards for this project. 10. Besides the proximity of most public libraries to public transport, public libraries were the chosen location for qualitative interviews because they are away from the 'home turf' of the researcher (i.e. University offices), and they typically offer 'group study' areas that are relatively quiet but also appropriate for comfortable conversation, as opposed to other public establishments that have a great deal of background noise (i.e. coffee shops). In addition, most libraries have multiple seating areas in private 'nooks' that allow for privacy without being too private, which may feel unsafe for two strangers (researcher and respondent). Overall, the public libraries worked very well for the interviews, and I would certainly conduct another project using public libraries as interview sites. 11. In addition, in email/phone correspondence, I let the respondents know that I would wait just inside the front doors of the library holding a sign with my name and the logo of my University so that I would be easily recognized. This was done to ensure that respondents would be identifying me, and I would not presume to identify them. This may also alleviate any concerns about my classification of respondents as 'obese' and ideally allows respondents to feel more at ease about their weight and the interview process. 12. The vast majority of respondents contacted me via a gmail email address I included in the craigslist posting, although a few located me through the University and called or emailed me through my University contact information available on the University website and my personal professional website. I was able to schedule interviews with most people who emailed me about the project through approximately two email exchanges. I also provided my personal cell phone number in email correspondence so that respondents could call/text me if they needed to reschedule or cancel the interview. Most respondents were very reliable and courteous about the process; however, I did have three respondents who did not show up to the scheduled interview and did not respond to follow-up correspondence to reschedule the interview. 13. If an individual attempts to post the same ad more frequently than once every 48 hours, craigslist will send an automated email indicating that this is a violation of craigslist policies and the ad will not be posted. 14. Anecdotal evidence suggests that specifically requesting that an ad not be flagged in a posting may actually exacerbate flagging (see Lloyd and Gunter, 2009). 15. Overall, the sample was diverse: 61 percent of respondents were White (14% biracial/ethnic, 11% Black, 7% American-Indian, and 7% other race) and the average age of respondents was 35 years (ages 19-29 (34%), ages 20-39 (36%), ages 40-49 (21%), ages 50-59 (7%), and ages 60+ (2%)). About half of respondents indicated an overall household annual income of US$30,000-US$59,000, and respondents reported varied types of employment (18% medical field, 18% clerical/administrative, 16% student, 16% teacher/education, 14% stay-at-home mom/nanny, 14% sales, and 4% unemployed/disabled). All 38 respondents were talkative and informative. They reported a wide range of experiences, and their interviews have provided the data for a book proposal currently in preparation.