The objective of FRDC project 2012/021, “Trial and validation of Respondent-Driven Sampling as a cost-effective method for obtaining representative catch, effort, social and economic data from recreational fisheries” was to trial and validate the chain referral sampling method, Respondent Driven Sampling (RDS), for obtaining representative data from specialised ‘hard-to-reach’ components of recreational fisheries. This project aimed to test this new method by undertaking a RDS survey with a population of fishers who were part of a complete licence list frame. The characteristics of the sample (e.g. age, gender) from the RDS survey would then be compared to another survey collected via random stratified sampling drawn from the licence list frame. This would allow comparison of the results between methods to determine if the RDS survey can produce a representative sample of the population.

Telephone surveys have long been regarded as a cost-effective method for large-scale population. However, in recent years, a decline in landline registration, increasing exclusive use of landlines for internet connections, and changes in population demography has led to a decline in the representativeness of the landline sampling frame on the overall population. In addition, with a limited number of recreational fisheries requiring a licence—many of which also have various exemptions— a secondary list-sampling frame is not always available to researchers to select a representative sample of fishers for a survey. These factors contribute to a degradation of the ability of scientists to yield a representative sample from the population via direct telephone polling, and highlight a need to explore new methods for more effective sampling of recreational fisheries. A trial of the RDS method is needed as it may be one of the few methods that can cost-effectively attain reliable data from specialised fisheries that lack a complete licence list frame of participants. It may also be particularly useful in situations where the participants are too rare within the wider population to be sampled in sufficient numbers using traditional survey methods.

RDS is a peer-driven recruitment process initiated by a small number (4-6) of members, or ‘seeds’, from the target population who each complete a questionnaire. On completion, each person is given a small 'initial reward' and 2-3 uniquely coded coupons to pass to eligible peers. The person is instructed they will receive a 'secondary reward' if their peers recruit to the survey. When each peer is recruited and completes a questionnaire, they are also given two coupons to pass to other eligible peers. This chain-referral process continues and produces rapidly expanding recruitment chains until the sample reaches ‘equilibrium’, whereby the proportion of population characteristics (e.g. gender, age) no longer change with further sampling.

The Tasmanian government issues a number of specialised recreational fishing licences without exemptions, which provided an opportunity to trial RDS and assess its efficacy against the known population of licence holders. These include the Tasmanian recreational set-line and rock lobster fisheries. The set line fishery is specialised in terms of the species targeted and the gear used (mainly longline). The number of licence holders is around 4000. The Tasmanian recreational rock lobster fishery again has no licence exception but is larger with around 18,000 licences issued per year.

We used the RDS method to study three populations. First, we undertook a pilot survey of a staff population at the Ecosciences Precinct (ESP), Brisbane to optimise sampling and operational procedures and validate the mechanics of the RDS method. ESP housed 827 staff at the time of the survey representing CSIRO, four government departments, and three universities. This urban and socially cohesive population experienced a range of hardships during a forced re-location to ESP from various locations around Brisbane. As such, the pilot survey of their experiences during the relocation resulted in a high level of engagement in the process. The mechanics of the survey performed as planned, with ‘waves’ of respondents being recruited from an initial seeding of 7 individuals. In total, 394 coupons were issued and 197 interviews completed. All but 10 of the respondents originated from the one seed.

Subsequently, two field trials of RDS within the Tasmanian recreational set-line fishery and rock lobster fishery were conducted. As a precursor to these trials, a workshop with recreational fisheries representatives was undertaken to explore the logistical details associated with implementing an RDS survey. The workshop was attended by international RDS experts, fisheries scientists, statisticians, a fishery manager, and recreational fishing group representatives. One key recommendation from the workshop was to undertake focus group meetings with set-line fishers to seek feedback on specific aspects of the survey method (e.g., incentive amount and type), which were undertaken in Devonport and Hobart.

The survey of Tasmanian recreational set-line fishers was conducted between November 2014 and April 2015. We developed a survey tool and database “RDS-Recfish”, for implementing RDS surveys, managing coupons and incentives. A prototype of this tool was trialled at the focus group workshops and refined following feedback on the questionnaire and survey structure. RDS-Recfish was then used to implement the first RDS survey. Initially, total of six seeds were recruited to start the survey, based on their geographic location and fishing club membership status. However, long sequence chains of RDS recruits did not occur from these seeds. From 27 recruitment coupons that were circulated by the seeds, only three fishers were recruited. A follow-up survey indicated seeds had no issues distributing coupons to other fishers, however many noted ‘obvious scepticism’ when trying to explain the research objectives to potential recruits.

The second field trial of RDS involved the Tasmanian recreational rock lobster fishery. Based on the findings from our set-line study, we adapted our methods to increase the likelihood of developing long recruitment chains that expanded into the general population of fishers. This involved dramatically expanding the number of initial seeds to 41 fishers over multiple waves of recruitment, seeding across potential barriers to recruitment—namely geography and gear type—and undertaking personal briefings of seeds and a follow up survey to better understand psychological aspects of the recruitment process. While our follow up survey indicated that most seeds had passed on their coupons, only five eligible fishers were recruited from the 135 coupons distributed.

While there appeared to be no issue with distribution of the coupons by the seeds the following key mechanic of the method, which required the coupon recruited fisher to make a phone call on their own initiative back to the researcher, rarely occurred. A level of psychological inertia was not overcome by these fishers, as they were not sufficiency motivated to make this call. We think that in additional to the generous monetary reward offered, another strong non-monetary incentive may also have been required, such as was the case in the ESP study, to improve the survey response rate.

There has been widespread success of RDS in a range of highly connected hard-to-reach populations (e.g., illicit drug users) in densely populated urban settings. We think that in addition to fishers not be motivated by the solidarity of stigmatisation - they are after all participating in a legal and licenced activity with strong cultural roots – potentially their low frequency of social interactions may have been a further impediment to their motivation to make contact with the researcher.  In other RDS research including our office block EPS study close, repeated close social interactions with seeds can provides ‘peer pressure’ or ‘group-mediated social control’ to encourage participation in the survey.

Another potential impediment was the choice of contact technology. In addition to declines in land-line use there has been a further recent shift towards text based communication by the general population. The survey was dependent on voice phone calls, and paper coupons. While other methods, such as SMS, could be used to distribute coupons codes, fishers still needed to ring a phone number and leave a voice message. An option to establish communication via various on-line text forms (i.e. social media) may have improved the response. 

A further possibility for the failure, particularly of the set line case study, was scepticism among fishers that the use of research survey data will be used as a justification for implementing management measures to limit their fishing opportunities. Such negative attitudes towards research have the potential to spread through the social networks of fishers to inhibit RDS recruitment. However, in the rock lobster fisher study, there was strong support for the science aims of the work both by seeds and during the follow up survey.

Finally, we were not able to test the representativeness of the data as we could not get the mechanics of the RDS method to work for our two case study fisheries. However a simulation of the RDS methods suggested that differential recruitment by seeds of fishers can lead to substantial bias and this bias cannot be detected from the RDS sample alone.

Despite the comprehensive preparation and collective efforts of our team, international RDS experts, recreational fisheries survey design experts, fishery managers and recreational fishing advisory members, RDS did not function as anticipated in two distinct recreational fisheries trials. Through the field trials, the simple act of calling the project’s freecall telephone number appeared to present the greatest impediment to recruitment from the many fishers who accepted a coupon from their peers to participate in the surveys. Our method may not have also accounted for other specific psycho-social factors that created impediments to recruitment. Further work focusing on the motivations of fishers to participate in research surveys, their preferred communication technology, their psychological responses to incentive types, and the social inertia that needs to be overcome to recruit one’s peers, may guide researchers to continue to adapt interview methods for recreational fisheries research.

Without a highly motivated population of socially closely connected fishers, RDS does not appear to be cost-effective method for obtaining representative catch, effort, social and economic data from recreational fisheries.

Future trials of similar methods for surveying recreational fisheries may consider using other types of survey administration that do not require direct voice contact with staff (e.g. self-administered surveys online via social media) may result in more recruitment. However, such methods need careful consideration and testing prior to use since they may introduce a suite of poorly understood sampling biases that compromise the representativeness of the sample.

A repeat of previous economic surveys of the recreational rock lobster fishery, based on a representative sample of the licence frame, could provide an interesting assessment of high value placed on landing lobsters.