J. Riley
Statistics Department, Rothamsted Experimental Station, Harpenden, Hertfordshire, UK
E-mail: janet.riley@bbsrc.ac.uk
The preamble to the final report of the Technical Centre for Agricultural and Rural Cooperation (CTA) study on `Strengthening biometry and statistics in agricultural research' (Riley 1998) describes the changing needs in agriculture as follows:
Traditional agricultural research has involved detailed biophysical studies in well-controlled conditions. Issues of global changes have resulted in emphasis moving to broader, but less precisely controlled, studies and surveys to assess environmental change and the effects of technology intervention upon the natural environment and the population at large.
The design and analysis of agricultural studies have necessarily become more complex: the whole research spectrum needs to be addressed from basic laboratory and station research through adaptive and applied multidisciplinary research on-farms and within communities to farmer impact assessment. Research at each of these stages can benefit from the rigour introduced by professional biometric input which ensures clear planning, appropriate information collection, information summary and interpretation and its presentation in appropriate formats to all involved stakeholders.
Biometric skills have traditionally been taught to deal with straightforward studies such as those for single-species on controlled research stations. Yet biometric needs for multidisciplinary and impact-assessment studies are greater- to cope with the informality of designs, large variability in data and variety of data types. To maintain rigour, strong biometric input is essential-professional input consisting of powerful, flexible computer-based methods. Training of agricultural researchers in the use of such methods is rare; training of professional biometricians in the use of such methods, and skills in consultancy techniques, are therefore essential to underpin the work of agricultural researchers.
Failure to achieve such flexible biometric skills in any country will result in a decline in research quality, the generation of non-representative research results, the rejection of research publications and their exclusion from the research debate. This will be detrimental to global development and the conservation and sustainability of natural resources.
This shows very clearly how biometricians must be aware of the changes they need to make to provide professional support to their scientists as they respond to changing research directions. Failure to be flexible and make these changes may well result in loss of financial support for biometric posts.
The situation is not unique to Africa. In Nepal there are three main agricultural research stations and there is no biometricians employed in them. The same is true in Papua New Guinea. The number of biometricians in the regional agricultural research network in the Caribbean has been reduced from three to one. In one of the national agricultural institutes in Colombia there were three biometricians. Now there are only two and there are threats of more cuts. In Europe agricultural institutes are in decline, many have closed and biometricians are disappearing to industry. And there are many more examples of cuts to agricultural research in many other countries. What must be done?
Institutional management must be persuaded to recognise the importance of biometrics both in the maintenance of research quality and in the decision making process. Additionally, the management structure must encourage interdisciplinary studies to maintain funds for all scientists including biometricians. This will require an understanding by the scientists and the biometricians of biometric methods for multi-component studies and skills development in this area. How is this best achieved?
Most scientists have received training in routine methods to handle well-balanced statistics designs and straightforward, well behaved analyses. To handle multi-component studies and irregular designs a greater appreciation is needed of variability in location, material, samples collected at different times, and so on. Understanding of correlation and the complexities of repeated measures demands newly-available methods to adjust for the dependencies between sets of data. Participatory research demands an understanding of qualitative data such as farmer responses and impact assessment.
In biometric training for scientists do we retain the simple design approach or do we baffle the scientists with the complexity of more advanced methods? It would be appropriate to take a middle road adapted for individual needs and abilities. This would be demanding on time and resources but would ensure that appropriate methodologies are available to the relevant disciplines.
Updating in new techniques is essential; this can be done through short courses and also through material accessible on the Internet. Awareness of new methods must be combined with a confidence building process to ensure that communication of the new techniques to scientists is unambiguous and non-technical. Writing skills are essential and often require improvement (everywhere in the world!). Biometricians need to have the same knowledge as scientists in new techniques but need to understand them in more depth so as to be able to adapt them to unusual situations and programme them efficiently.
Personal management is very important. The biometricians must be proactive and develop collaborative approaches with the scientist and take the initiative to seek funds. In particular he/she must identify research priorities and respond to the need to understand the statistical techniques required to underpin them. He/she should manage time well to ensure time to read about new methods and software. Much is available free on the Internet. An appreciation of a wide range of statistical packages and their limitations is necessary to ensure flexibility of programming for new methods.
New university syllabuses need to be designed to incorporate more flexible methods in a clear, non-theoretical way. This is essential for those commencing training as a scientist or as a biometrician, although biometricians can benefit from further theoretical training in these areas.
Short courses will be crucial for scientists, and biometricians qualified already, and may be provided by biometricians from outside the country. However, short courses for scientists may not be appropriate for biometricians as they may start at too basic a level. Two types of course should be designed:
Increased input to the training should be made by the local biometrician, until his/her confidence is such that he/she can provide it himself/herself. The biometricians from outside the country should make a gradual withdrawal from the training programme at each stage to ensure institutional development. The local biometricians should continue to monitor the uptake of the teaching material by the scientists and ensure that updates and explanations are given to encourage continued use of the material.
Short attachments to institutes in the north or to local institutes with experienced biometricians to analyse their own data, with the daily support of practising biometricians, may be more beneficial than short courses for the local biometricians. However, attachments for PhD's typically lead to too great an amount of specialisation and theory. Additionally, they cause the biometrician to be away from his/her institute for too long a period and support to the scientist may suffer.
As shown above, a good deal of initiative is needed on the part of local biometricians. This must be underpinned by multilateral support from universities, institutes, professional societies and publishers to provide access to the necessary new material. Indications are that this support is beginning to emerge.
And money is essential to support all new schemes. But there are lots of opportunities to find funds. All northern funding schemes are accessible to biometricians both in the north and in developing countries. There are large numbers of regional and national groups in developing countries that have funds unavailable to those from the north but readily available to those who reside in the region and who have the energy and enthusiasm to access them in the appropriate way. Seek and you shall find!
Riley, J. 1998. Strengthening biometry and statistics in agricultural research. Study Report. (CTA) Technical Centre for Agricultural and Rural Cooperation. Wageningen, The Netherlands. 27 pp.
