4.1 The qualities of the statistics produced related to their use
If the purpose of collecting the data and the users for whom the statistics are intended have been kept in mind throughout the collecting process then the actual details of presentation and publication should raise few problem.. This is not to say that it is unimportant to spend time in ensuring clear and informative presentation - on the contrary - but the audience and level of detail required should be sufficiently well specified by now to make the construction of tables and figures a straightforward design problem. Published tables will not be the only product required of the data collecting system and it is in the field of processing, tabulation and storage of statistics in varying degrees of elaboration that the major advances due to computer have taken place recently. The methods of applying these to the collection of fishery statistics have yet to be developed fully. It should soon be possible to publish only the most widely used tabulations and to produce the more detailed, statistics used for example by scientific working groups when they are required..
The processing and analysis of the data collected in all stages of the fishery, statistical system is not a simple matter of adding together all the results. The methods of processing and analysis depend directly on the methods of sampling and recording used and there must therefore be the closest possible integration between the two. It is quite useless collect data which cannot be processed because the facilities do not exist or because they are too complicated or incomprehensible to the processor. The importance of timeliness in producing statistics has already been stressed on several occasions, but it is worth mentioning again here because timely processing makes for easier checking of the data back to its source. Also the cooperation given by fishing companies and fishermen may be greatly in. creased by processing quickly and giving back the information in a form which they find useful (see Figure 4).
There are many ways in which fisheries statistics can be published, ranging from the weekly records of fish prices and landings at different ports and markets, which are published in the fishing industry newspapers, to the annual summaries of world trade and catches in each country, published by FAO. Once a format for a statistical publication has been decided on it should be kept to as closely as possible from year to year so that people can get used to it and know where to find things. The units and categories should be clearly specified every time, with details of the methods of processing and raising, if these could make an important difference. Any changes from year to year in the units used, e.g., from "landed weight" to "nominal catch" or in the categories or methods of processing should be clearly indicated.
Figure 10a: Catch and effort reporting form used in the northwest Atlantic
Figure 10b: Catch and effort reporting form used in the northeast Atlantic
Data processing, whether manual or automatic, may be divided into five steps:
1. Editing
2. Coding
3. Raising and estimation
4. Tabulation
5. Presentation
The completed forms, log books or questionnaires mast be edited before they can be processed further. Editing consists of (a) checking that they are complete and that no essential information is missing and (b) trying as far as possible to verify the accuracy of the contents. This may involve simply checking that the data given lie within certain specified limits or it may require a considerable background of knowledge of the data being reported.
Source documents should be designed for the minimum of coding, particularly if they are used directly for punching and automatic processing. Any coding which is needed after they have been filled in should be done at the time of editing. The reason for punching directly from the source document and for reducing the amount of coding or transcription to a minimum is that each extra procedure takes time and introduces further possibilities of mistakes.
Once the editing is complete the data can be used to estimate population characteristics. For example it may be necessary to determine what proportion of the catch was measured and to calculate factors for raising the sample to the total catch. If this type of procedure is followed as a routine, carried out max ally, then worksheets should be used which break the calculation down into easily handled parts. The worksheets should have checking procedures tilt into them so that any errors are detected at an early stags. The estimates obtained from the worksheets are next transferred to detailed tabulations which may he further summarised for the final presentation.
The introduction of ADP should reduce the actual handling of data to a minimum. This will free staff from much of the drudgery involved in processing, but good background knowledge is still required for editing the data before it is punched. In the initial stages of implementation of an ADP system it is often necessary to introduce more checks than usual on the output, because the intermediate checks will be carried out by the programme itself.
The scope for improvement in fisheries statistics by the use of ADP, particularly if it incorporates data storage and retrieval, is undoubtedly very great but progress is at present slow. Several national and regional bodies use ADP for compiling and publishing their routine statistical tables and there are plans for data storage systems to be used by scientists carrying out regional assessments. There is in fisheries a large gap between what is technologically feasible in terms of data storage, handling and manipulation by means of computers and what has so far been achieved. A great deal of time at international working groups is spent in processing and summarising the available data, before the actual assessment work can even be started. The main problems in setting up a fully automated data processing, storage and retrieval system are:
1. The quality of data received at present from national statistical offices varies a great deal and needs careful checking before it is used.
2. The coding arid classification systems used must be standardised between the national agencies and the regional body and also between regional bodies, since a country may be reporting to several regional bodies.
These are not particularly complex problems, but they do require a high degree of international cooperation and a far higher level of priority and expenditure than they receive at present. Since it is likely that ADP systems will be set up in many countries within the next decade, the need to resolve these problems is urgent if the national systems are to be made compatible.
At the national level the decision about what method of processing to adopt will depend on the volume and complexity of fisheries data to be processed and on the availability of computers and programming skill. The quality and degree of sophistication of the products will be increased arid personnel can be freed for less mechanical jobs, but automation is not a substitute for a reliable data collection system.