67. The Technical Secretary of the Ad hoc Working Group encouraged the Working Group to avail of the tools developed by FAO for sustainable and cost-effective data collection, citing as an example sample survey systems as explained in the Guidelines and the software package ARTFISH/ARTSER developed by FAO which is applicable for designing sample surveys for small-scale fisheries. The software has already been converted into Windows version making it user-friendly.
68. Countries in the region have continued to search for statistical data collection system that would allow obtaining information even for a small number of units. Vietnam, for example, is implementing its ALMRV project which is a detailed sampling survey on fish landings including some biological aspects.
69. Citing how sample surveys are used in their areas of work, the expert from the Mekong River Commission (MRC) indicated that sample survey has been very useful as it is a sensitive approach in collecting data from inland waters such as the Mekong River. But it could be difficult and expensive to conduct a comprehensive sampling survey in other small-scale inland water systems such as rivers, swamps, and rice fields. An alternative system such as modeling for small areas should therefore be developed to collect the needed data such as socio-economics and other biological aspects.
70. Household fish consumption is an important indicator in inland capture fisheries. In the case of the Mekong River, the data are obtained not through direct collection of the consumption data but through a special system of gathering information from the fishermen or their children or any members of the family who go out to fish. The system adopted by MRC has resulted in more accurate data and a more comprehensive collection of data up to the species level has been recorded. MRC also collects data on the ecosystem of the Mekong River while the Asian Institute of Technology (AIT) conducts surveys on catches from rice fields.
71. A case study on the collection of data on small-scale fisheries in the Andaman Sea area which includes Phang-Nga Bay, using sampling survey of villages, was also reported by the expert of the FAO Regional Office. Catch-effort and income data from three main fishing gear such as the trammel net for shrimps, mackerel gill net, and crab gill net, were collected from logbooks of fishermen. From the study, the total catch for target species, average income generated by each gear, and the income of the fishermen in the whole Andaman Sea area were generated.
72. In this connection, it was noted that sample survey data collection can be much more reliable if the sampling design takes into consideration such factors as seasonal changes, religious beliefs, cultural practices, among others. In addition, metadata (data about the data) such as seasonality, weather conditions, change in respondents should be included in the final report as this would greatly affect the accuracy of the data.
73. In the case of China, sample surveys have been used to collect data on income of fishermen or on some other socio-economic variables. However, under Chinas system of complete enumeration and bottom-up transfer of data from the grassroots level to the central government, problems related to the possibility of inaccuracies of data could arise. It was therefore suggested that a system be developed that would validate and confirm the accuracy of the data under similar circumstances.
74. In the United States, sampling survey has been used to impute data on total harvest, where the design of the survey takes into consideration the kind of data to be collected. Here, enumerators have been trained to properly collect data, taking note of the importance of validating the data before these are entered into the whole statistical system for analysis and interpretation.
75. Constant monitoring and validating of data from sample surveys are therefore very important. The possibility of adding confidence limits in the census reports to make it more accurate however, could not be applied for censuses because confidence limits are derived from sampling errors, and censuses are not subject to sampling errors.
76. In the Philippines, sample surveys are used for the collection of data from municipal and commercial fisheries. Problems were however encountered in establishing the sample survey, such as the construction of a good sample frame and appropriate sampling design, and establishing procedures for the review and validation of the whole sampling procedures. The accuracy of the sample from municipal fisheries could not be relied upon because of the difficulty in coming up with a list of municipal fishing vessels for the whole country. In such circumstance, a sampling methodology which is flexible and efficient should be developed.
77. As a user of the compiled data, SEAFDEC could optimize the usefulness of the data if these are collected using a good sample frame, and a census could be a good basis for the definition of such valid sample frame. Surveys should therefore be reliable and well-designed and based on up-to-date frame. Implementation of the surveys should be improved as this has been the main source of the inaccuracies.
78. After the discussion, it was suggested that basic courses on designing surveys be developed for the countries in the region and training of personnel involved in designing the surveys and socio-economic analysis be conducted. In addition, survey statisticians should be involved from the design phase to the implementation of the survey until the monitoring and analysis of the data collected.