Establish First Nations partnerships
Within Australia, MPA recreational users’ KAP surveys take place on, and are related to the management of First Nations peoples’ Land and Sea Country. Therefore, First Nations peoples of the Country should be contacted during early phases of planning, and established protocols for collaborating with First Nations groups should be followed (NAILSMA & CSIRO 2020). Obtaining approval to conduct research on, or related to First Nations peoples Land and Sea Country requires sensitivity and ample time to build relationships and trust. For example, researchers wishing to collaborate with the Esperance Tjaltjraak Native Title Aboriginal Corporation are required to engage in an initial consultation meeting with representatives from the six Wudjari families (i.e., the Circle of Elders) to discuss research intentions and scope. This session is critical to gaining preliminary approval and serves as a platform for families to express concerns and expectations. It’s important to note, however, that consultation does not imply consent. Researchers should aspire to thoughtfully co-design projects with First Nations people, as opposed to having these communities be the recipient of fully-developed research plans (St John & Akama 2022, Tamwoy et al. 2022).
It is important that free, prior and informed consent is obtained, and that Indigenous Cultural Intellectual Property is protected if cultural knowledge is to be shared for data collection or co-design of specific surveys (NAILSMA & CSIRO 2020). This follows the internationally recognised CARE (Collective benefit, Authority to control, Responsibility, Ethics) principles for Indigenous data governance (Carroll et al. 2020). First Nations partnerships involved in developing and trialling this field manual have highlighted that MPA recreational users’ KAP surveys provide scalable opportunities for First Nations ranger-led research.
Develop survey instrument
Whilst it is often desirable to tailor a survey’s questions or topics to specific areas or contexts, this survey is designed to be standardised and transferable between locations. However, modules (e.g., an economic module) can be added to the end of the survey to target certain topics or enhance specific analyses.
The survey template (Appendix 1) has five sections: (1) Pre-survey, which includes interview metadata (e.g., surveyor ID, survey site), survey introduction and informed consent protocol; (2) Practices, which records details of activities the respondent has participated in on a given day; (3) Knowledge, which captures the awareness the respondent has for local jurisdictions, MPAs and their zonation; (4) Attitudes, which collects data on the attitudes and perceptions towards MPA zones; and (5) Demographics, for collecting information about the respondents.
Survey software
We recommend ESRI’s Survey123 software in combination with a light-weight weather proof tablet device, and managing the surveys using ESRI’s desktop app Survey123 Connect due to its flexibility, and extensive options to customise question parameters. Additionally, Survey123’s online platform offers real time plots and statistics for data review, which is useful for monitoring data collection. Key software features and alternative software options are detailed in Appendix 2.
The XLSform survey template provided in Appendix 3 can be uploaded into Survey123 Connect to customise a survey to a particular location.
Design sampling plan
We recommend that MPA recreational users’ KAP surveys employ a probability-based face-to-face approach at marine access points (e.g., boat ramps or shore access points). Whilst this approach can generate a gender biassed sample (Navarro et al. 2020) and is limited when asking about non-compliant behaviour (Arias & Sutton 2013), it directly targets active users of the marine environment and ensures the sampling closely reflects the targeted population (i.e., MPA recreational users). It is optimal to complement face-to-face surveys with other sampling approaches such as online open link, panel surveys or probability-based mobile phone or letter-box sampling approaches (Dillman et al. 2014, Goodrich et al. 2023) to understand sampling biases.
Probability-based sampling, where every recreational user has a known probability of being sampled, involves the development of a survey shift schedule with a known set of sampling probabilities (Cornesse & Bosnjak 2018). Weighting the survey schedule to oversample busy locations and time periods (e.g., weekends, holidays) maximises sampling efficiency, especially in remote locations.
Using a restrictive spatial and temporal sampling design (Smallwood & Ryan 2020) is a pragmatic approach to conducting MPA recreational users’ KAP surveys. However, the data collected is only truly representative of the sampling time frame, so while we encourage comparisons across MPAs, this should be interpreted attentively. To aid this interpretation it is essential that detailed information on the sampling design should be maintained and recorded in any resulting metadata products (see Appendix 1 for shift metadata template).
The sampling effort required for a robust sample size will depend on how the data are intended to be used, and the popularity of the access points sampled. As a general rule, expect an average of five interviews per four-hour shift at a high-use boat ramp or shore access point. Whilst some days may far exceed this, particularly during school holidays and in fine weather, others will have very few, if any, interviews completed. For the calculation of broad metrics (e.g., % awareness of zones), a target of 100 surveys should be completed (Navarro et al. 2021), though additional surveys will provide more robust data for detailed modelling (e.g., influence of fisher avidity). This corresponds to approximately 20 four-hour shifts.
Restrictive spatial and temporal sampling design of probability-based face-to-face MPA recreational users’ KAP surveys is outlined in five steps below:
Step 1: Identify the relevant MPA access points that you wish to sample
This may include boat ramps, beach launching locations (for boat-based recreational users) and/or shore access points. It is often impractical to sample all possible sites; therefore local knowledge (e.g., that of First Nations rangers or relevant management agency) can be used to identify a minimum set of heavily visited access points. This will normally exclude remote sites, and their recreational users which may differ systematically from those at popular sites (e.g., part of their motivation for recreation could revolve around solitude). If excluded it should be highlighted that the sample does not represent remote users of the MPA.
Step 2: Create a sampling schedule
A sampling schedule describes when surveys will take place. To design a sampling schedule the primary sampling unit needs to be defined; this could be a full day or can be divided into two shifts per day (e.g., AM and PM) in which case sampling would occur from the combination of all days and shifts. In designing a cost-effective sampling schedule an important consideration is whether survey staff are local to the study region or must travel there from further afield.
Local survey staff:
Using local survey staff is ideal as this allows flexibility in the sampling schedule; shifts can occur across a few months rather than having to be clumped into “field trips”. Employing local First Nations Rangers is a suitable way of ensuring staff are local.
To develop a sampling schedule with local staff the first step is to identify the time period across which you wish to sample. This could be the entire year, or a select few months (e.g., when weather is favourable for boating), with the choice being a trade-off between representativeness and practicality. It is important to capture both school holiday and non-school holiday periods to capture a representative sample of both locals and visitors. With a target period identified, a sampling schedule should be created by randomly selecting days from the target period.
Remote survey staff:
Having non-local staff is challenging as the sampled days must be clumped into field trips. From a practical point-of-view the survey schedule would then consist of every day (or shift) in the field trip. To capture a representative sample of locals and visitors, we suggest timing field trips to capture both school holiday and non-school holiday periods. Multiple field trips at different times of the year are also desirable to increase representativeness.
Step 3: Assign access points
Once a sample schedule has been developed, access points can be assigned to samples. For boat ramp surveys we recommend selecting just one access point per shift for shifts approximately 3-4 hours in length. This ramp can be randomly selected or selected using a weighted random sampling process to oversample busy locations to maximise data.
For shore-based surveys we recommend a transect design, in which the sites are ordered geographically into a route with fixed arrival and leave times for each site. To construct this schedule each shore site should be assigned an on-site time (we recommend spending more time at busy sites). Start points and route direction (i.e., north-south or south-north) should be randomised. If there are more sites in the route than can be sampled in a single shift we recommend random (or weighted random) sampling to select a fixed number of sites for a given day.
Step 4: Expecting the unexpected
It is common for bad weather (or other circumstances) to mean that there are zero or very few individuals to sample at a given site. Local knowledge should be used to guide decisions on when shifts should be cancelled (e.g., if wind speed exceeds a threshold). For quiet days that aren’t cancelled a threshold can be used to identify when it is worthwhile changing site. For example if there are less than five trailers a contingency site should be sampled instead.
Step 5: Inverse probability sampling
Where a weighted random sampling plan has been used to oversample priority sites or time periods (e.g., boat ramps, school holidays), or when there have been unexpected changes to the sampling plan, inverse probability weights should be used in analyses to correct for sampling bias and increase representation across sampling time frame (see Appendix 4 for details on calculating inverse probability weights).
Obtain approvals
Any research with or about people, including MPA recreational users’ KAP surveys should be subject to a formal review by a human ethics review committee or agency specific body, who can assess a survey against nationally accepted standards. Researchers without access to a formal review committee should consider partnering with researchers who do (NHMRC 2007).
We recommend that MPA recreational users’ KAP surveys be conducted anonymously, without collecting personally identifying information (e.g., full name, address). Doing so minimises risks to respondents, and is likely to maximise response rates. However, if personally identifying information is to be collected, the protocols of collecting, handling and storing this data must adhere to relevant privacy laws (NHMRC 2007).
We also recommend that researchers obtain informed consent for the re-use of anonymous survey responses in human ethics agreements and that this is clearly outlined to potential participants. This would allow data to contribute to national syntheses, maximising data impacts and minimising the need for repeat surveys and associated survey burden on respondents.
In Australia, researchers require approvals to conduct research relevant to, or on, agency managed land or waters, including First Nations peoples (see Establish First Nations partnerships above).
Surveyor training
Prior to fieldwork, surveyors must undergo training on ethical data collection procedures, participant recruitment methods, and have an understanding of the survey’s purpose and broader research context to inform respondents adequately. They should understand community attitudes, especially those that might trigger conflict (e.g., lack of support for a proposed MPA), and be briefed on handling antagonistic responses.