During February 1965, Mr. Hubert Renson, the U.N. Fisheries Development Adviser, with the agreement of the Ministry and UNTAB requested that the methods used to collect statistics on fish production be reviewed by the National Statistical Office. A short time later, Mr. Renson and I visited the Chief Fisheries Officer, Dr. Birkenmeir, in Fort Johnston, and discussed the fisheries statistics program with him. The program, as it was explained to me then, includes a complete census of fish production and of fishing equipment for the industrial and semi-industrial fishermen and a sample census of production and fishing gear of traditional fishermen. The Census of industrial and semi-industrial fishermen probably provides accurate information on that part of the industry (about 30 per cent of the total at that time). The survey design used to collect information on the traditional sector of the industry had, in my judgement, several deficiencies:
The sample of traditional fishermen was non-random and fishermen on the smallest beaches were excluded from the universe.
A few very large beaches were in sample with certainty and the remaining sample beaches were the largest of the smaller beaches.
The catches were counted on the certainty beaches on about 18 days of the month while counts were made on the remaining beaches one day of the month. All the fishermen on the smaller sample beaches were included in sample, but the subsample of those on the certainty beaches were selected on a non-random basis.
The catches were counted by species. A dozen fish of any one species was assumed (based on a few sample weighings) to weigh an average weight. The total production of fish in tons was calculated by multiplying the average weight of each species times the estimated dozens of that species.
When the number of fishermen docking at sample beaches declined during the migration of the fish another non sample beach was substituted for it.
Estimates were being made for one area along the beach even though no fishermen were in sample for that area.
Estimates of amount of gear owned and efficiency of use of gear were based on an out dated census of fishermen.
There was no systematic checking of the fisheries assistants' work.
As part of the review I recommended that the sample design be changed to a probability sample of beaches and fishermen and that steps be taken to correct the deficiencies listed above. Mr. Renson strongly recommended that scales be used to weigh the sample catches. We both felt that regular checking of the fisheries assistants' work would be especially helpful in producing more accurate statistics. Before any of our recommendations were considered Dr. Birkenmeir's contract with the Malawi Government expired and he did not return to government service.
The present Chief Fisheries Officer, Mr. R.B.Williamson, has proposed a plan which includes many of our recommendations and which takes a fresh look at the whole problem of collecting fish statistics in the traditional sector. The written description of the plan leaves out much important procedural detail and a few changes have been made since it was written.
At subsequent meetings attended by Mr. Renson, Mr. Williamson and myself the details were agreed upon and the overall survey design was accepted as being reasonably sound.
In the remainder of this memo I have described the proposed plan in enough detail that a set of operating procedures can be derived from the description. Such procedures will need to be reviewed and revised after the first month or two of operation.
OBJECTIVES
The objectives of the program are to obtain monthly estimates for the following statistical parameters in the traditional fishing industry:
Total gear available to the fishermen times maximum possible use of gear by type of gear. (Total available effort.)
Mean efficiency of use of available gear by type.
Pounds catch per unit effort by type of gear.
Total production of Fish in Tons by species of fish and type of equipment.
The estimates will be made separately for a number of small areas on Lake Chilwa, Lake Malombe and Lake Malawi and for the nation as a whole. Of the parameters to be estimated the first three listed are equally important for the management of the fisheries as are good estimates of total production.
COLLECTING THE DATA
The first step in the data collection process will be a Census of fishermen and of fishing gear available for use by the traditional fishermen. An attempt will be made on a bimonthly basis to list each fisherman along the three lakes and record the amount and type of his gear. The Census will be conducted by the fish rangers or the senior fisheries assistants.
Further information on the fishermen's activities will be collected from a sample survey conducted throughout the month. The catches and usage of equipment from randomly selected fishermen on randomly selected days of the month will be recorded. Summaries of the records, together with the Census results, can be used to provide the necessary estimates.
The steps in the selection of the sample will be as follows:
All very large fishing beaches will be in sample with certainty. (It would appear that a cut-off for “large” would be roughly beaches with 25 or more fishermen.)
A sample of all other beaches will be selected at the rate of 1 in 4.
Fisheries assistants areas will be formed by grouping sample beaches which are geographically near each other so that approximately equal workloads are in each area.
A table will be prepared which tells each fisheries assistant on which sample beaches he should appear on what sample days. The table will randomly assign days of the month to sample beaches so that each certainty beach is in sample for six days of the month (in clusters of three days) and each non certainty beach is in sample for four days of the month (in clusters of two days).
From the list prepared during the bimonthly Census a sample of fishermen from the beaches in sample with certainty will be selected at the rate of 1 in 6 by the fish ranger.
All the fishermen on the non certainty beaches will be in sample including those docking there temporarily.
Interviews will be conducted only with those fishermen selected in 5 above who actually dock at the sample beach for the reasons given in 11.
The fisheries assistants will complete a questionnaire form each day for each sample fisherman showing the amount and type of equipment owned, the amount and type of equipment used that day, the weight of each of the major species of fish he caught, and any other information needed by the Department of Fisheries.
For each fisherman selected into sample from the list for the beach but not docking at that beach on the sample day it will have to be determined if he was fishing and landed on another beach that day and with how much effort or not fishing at all.
The questionnaire forms containing the information should be sent to headquarters as soon as possible after the collection of the data.
A number of fishermen who will be selected into sample from the list for the
certainty beaches will not in fact dock at their census beach. It will be
impractical to try to find these fishermen at the place where they actually
dock and weigh their catches. It will also be impractical for most certainty
beaches to ignore the census list and select the sample from the fishermen who
actually dock there. Therefore, additional steps will be taken periodically to
show that the effect of omitting the catches of these fishermen is such that
the ratio 
is not seriously different than if every sample catch was included.
This can be shown as follows:
Their proportion of the total number of sample fishermen from certainty beaches should be smaller than 25 per cent.
The coefficient of regression for pounds of fish on amount of available effort should be approximately the same for the group of fishermen who leave the beach (based on the observation of the fishermen while on the beach) as for those who remain there.
QUALITY CONTROL OF THE DATA COLLECTION
Each fisheries assistant's work will be checked every month if possible but will be checked no less than every two months. The fish ranger or senior fisheries assistant will observe the fisheries assistant as he performs every step of his work, will check his scales and other equipment for accuracy and will discuss with the fisheries assistant any deficiencies which he observes in his work.
A fisheries manual will be written and each assistant will be trained in its use and will be required to keep the manual with him at all times while on duty. Regular retraining will be given each assistant at least once a year when facilities for the training become available. In addition, an assistant will be sent for extra retraining if the fish ranger who supervises him thinks it necessary.
MAKING THE ESTIMATES
Using the symbols given, the Formulae from which the estimates of the parameters in the order they are used on page 2 can be made are shown below:
This formula can be read:
| (Total available effort | - inactive available effort) | (Sample pounds caught) (Effort used in sample to catch Ps) |
Where:
UT is the total amount of any one type of equipment from the bimonthly census times the maximum possible use of the equipment.
US is the total effort available to the fishermen selected from the census list for the sample beaches.
Uk is the total effort available to the inactive fishermen selected from the list.
Us is the amount of effort used by sample fishermen both those whose catches were weighed and those whose catches were not.
Ps is pounds of fish of any one species caught with any one type of equipment.
Usi is the amount of effort used to catch Ps pounds of fish.
TABULATION OF THE DATA
At first the tabulation will, as at present, be carried out by hand by clerical officers in the Department of Fisheries. When the electronic data processing unit of the National Statistical Office (N.S.O.) is set up, however, the tabulation of the data will be shifted to this unit, thereby minimising tabulating errors and permitting the Department of Fisheries to concentrate more on minimising the efforts in the data collection phase of the operation. Overall responsibility for providing accurate fish statistics will remain with the Department of Fisheries with the N.S.O. merely providing a tabulating service.
TIMING
The plan will be introduced at Lake Chilwa immediately and will be extended to all fishing areas as soon as staff is available for training the fisheries assistants.
REMARKS ABOUT THE PLAN
Generally, the plan is sound. Much of it is based on the system in use in Zambia with appropriate modifications for Malawi. If the bimonthly equipment Census is carefully carried out and the fisheries assistants are properly supervised, good estimates of the four parameters are possible. It is important that the plan be put into effect over the whole traditional fishing industry as soon as possible and that the controls for non-sampling errors be set up promptly.
If quarterly estimates were adequate for the users of the data cost of the program could be considerably reduced. The sample size would need to be only about one-third as large and the equipment census could be taken less often. More time and effort could be spent on controlling the quality of the data collecting with the result that yearly statistics would probably be improved. The needs of the Department of Fisheries should be reviewed every so often and this possibility considered.
Zomba, 26th May, 1966
F. O'Quinn
National Statistical Office
P.O.Box 333
Zomba
