UNFPA COUNTRY SUPPORT TEAM

The primary purpose of the UNFPA Country Support Team for the South Pacific based in Suva, Fiji, is to provide countries with high-quality technical support services to meet their needs, leading towards national self-reliance in the population field.

Among the functions of the Country Support Team towards this end, the injunction "to provide active and close backstopping to the local pool of national experts" implies more frequent interaction between CST Advisers and national counterparts than is afforded by the occasional in-country technical advisory visit.

This Discussion Papers series has been initiated by the CST, Suva, in an attempt to establish a dialogue among national population programme personnel on the multidimensional aspects of population programmes. The major objective of the series is to help in the conceptualisation and development of a more holistic programme approach.

In the area of population statistics and data collection, UNFPA has for many years supported the conduct of national population censuses and the improvement of civil registration systems. Yet there have been few reports on the impact of this support on strengthening national capacities to carry out censuses and surveys and to provide the range of data essential to population and health planning.

The present paper has drawn on the personal experiences of the CST adviser on population censuses and surveys to trace statistical development in the Pacific, to identify particular problems confronting small island nations, and to propose a way forward to meet the exploding demand for statistical information. Through its blend of fact and anecdote, it is hoped that the paper will provoke discussion among statisticians and donors on how to develop national statistical systems that reflect the Pacific character and yet at the same time acknowledge the more sophisticated and technological environment in which they will be expected to operate.

STATISTICAL SERVICES IN
THE PACIFIC:
SOME PERSONAL REFLECTIONS

abstract

1. Introduction

Quite often, statisticians from around the world meeting together with Pacific Islanders are heard to wonder aloud how statistical services in the Pacific region can be anything but excellent. After all, the Island countries are small, national statistical offices exist and seem to function, and the people are renowned for their friendliness. Surely, the argument goes, where the frequencies of events of almost any kind are low, where the scale of operations is small, and where cooperation is widespread, the collection of data must be inexpensive and uncomplicated. The reality is that for the most part the production of statistics is neither cheap nor simple: the remainder of this paper, drawing largely on my own thirty years experience in the region, will try to show why this is so.

To start with, there is no single country representing the popularized but idyllic image of the South Pacific. The region is characterized by wide heterogeneity in geography, size, politics, history and culture. A broad sub classification of the region into Melanesia, Polynesia and Micronesia is usually adopted (which I too will use frequently). But useful as they are such groupings often gloss over very important variations among countries in the same groups.

The population histories of the Pacific Islands while they reflect some common themes are for the most part the products of their own particular culture and politics. For the majority, recorded history is short and verifiable factual information dates back only to the last century: much of what is now claimed as "historic" is a colourful blending of myth and reality, painted skillfully in the art of oral tradition. There is a rich mythology in Kiribati for example that emphasizes a Samoan heritage (Census report, 1978). Modern scholars of Kiribati society generally agree that migration from Samoa to the atolls and reef islands of the central Pacific did take place some 500 to 600 years ago but there are much closer links with the Micronesians to the north (ibid). Stories of Samoan migration abound in Samoa where many believe the island of Savaii is the legendary Hawaiki, the origin of the Polynesians who were to explore the Pacific.

The earliest statistical estimates available for Polynesia are based on reports from traders, missionaries and explorers visiting the islands in the 1820's and 1830's. Many of these estimates were crude with considerable ambiguity as to exactly who were included. For much of Micronesia early estimates depend to an even greater extent on guesswork; the first reliable estimates are probably those compiled by the administering German authorities towards the turn of this century and the Japanese that followed. An even later start to data collection was made in Melanesia. In German New Guinea, for example, administrative circulars issued during 1902 and 1903 stressed the importance of collecting demographic statistics for the indigenous population, though interest in population size and composition flowed from the extensive use made of the population in road construction and other public works and from the recruitment of males as plantation labourers (van de Kaa 1971). However, despite this interest in demography, in the years that followed data were collected at infrequent intervals and for selected areas. Papua was even less well-served and the scattered data and reports on population that are available were gathered piecemeal at the initiative of individual explorers, administrators and researchers.

Early estimates of population were most often crude and theories about the pace and nature of demographic transition following contact with Europeans are based more on speculation than hard evidence. The fragmentary records that do exist throughout the Islands seem to suggest that demographic change was uneven and on the whole suggested considerable suffering by the island people. Population was stagnant for long periods and at times experienced sharp declines. Some early commentators attributed this decline to an ubiquitous inferiority complex in the face of a "superior" white civilization and culture. Rightly dismissed for its absurdity and arrogance, it is now widely recognized that early contact exposed island peoples to "shattering epidemics of unfamiliar diseases, which not only killed a large proportion of the population but, through age-selective mortality, left ...age structure distorted" (McArthur 1961).

2. Data Collection in the Early Twentieth Century

The establishment of more formal and regular statistical reporting did not really get off the ground until the early part of the twentieth century, though there were notable exceptions. In French Polynesia, McArthur reports attempts at census-taking as early as 1829 and the recording of births and deaths by some missionaries. Other Polynesian countries, including Cook Islands, Tonga and Samoa, made some attempts to establish registration of births and deaths during the latter part of the 19th century and to conduct periodic population counts. In Fiji the first official census is generally thought of as that conducted in 1879, although there are serious doubts about its accuracy.

By the turn of the century, most Polynesian countries were conducting regular population censuses, at least from the perspective of attempting full national coverage. Tonga held its first census in 1891 and successive censuses were conducted at more or less ten- yearly intervals until the early 1930's. The German authorities in Samoa conducted population counts in 1900, 1902 1906 and 1911 and instituted a system of birth and death registration; in 1917 the New Zealand administration carried out its first census. Remarkably, the early censuses show no change in the Samoan population between 1902 and 1921. Discounting the unlikely possibility that the quality of censuses was becoming progressively poor during this period, the results suggest high levels of emigration or very high mortality. While there is little doubt that Samoans were migrating as missionaries and to find work, the number involved is unlikely to be sufficiently large to account for the stationary population, strongly supporting the hypothesis that this was a period of prolonged and devastating disease epidemics.

The growing concern of administrators to know more about the populations in their charge was now clearly reflecting itself in the strengthening of data collection systems. In Polynesia, as I have already noted, regular systems of censuses and civil registration were in operation. The New Zealand administration conducted five censuses between 1917 and 1945, with hardly a break to recover from World War Two. Tonga, having held three censuses between 1900 and 1930, conducted six more during the 1930's. The German administration in Micronesia had initiated a system of data collection to present in its annual reports. Building on this initiative, the Japanese conducted regular five-yearly censuses from 1920 to 1935, followed by another in 1938. The results of these censuses, incidently, imply a rate of growth of the Melanesian population in the order of 0.4 per cent, illustrating once again how widespread were the debilitating effects of infectious diseases.

The development of statistical sources was far slower in Melanesia. In 1906, following several years of intense rivalry between the British and French over the future of the New Hebrides, the two governments drew up an agreement establishing an Anglo-French Condominium, the only one of its kind in the world! Not long after, in 1910, the first official census was conducted. Little advance had been made meanwhile in New Guinea, administered by Australia from 1920 as a mandated League of Nations territory, or in Papua, declared an Australian external territory in 1906. Indeed it was not until the 1930's that a semblance of a data collection system was put in place in these territories as patrol teams were dispatched to various areas as a means of establishing and extending administrative control.

While the contrasting situation in Melanesia and other parts of the Pacific is apparent, in retrospect a more significant point is the pace and vigour with which administrators throughout the Pacific were moving to establish population data systems. There is little doubting that judged by today's standards, the quality of these early attempts to conduct censuses or collect field data were subject to considerable variation in quality and lacked formal statistical control. But in the absence of alternative sources, they were invaluable in providing the only insights to the populations of the Pacific islands.

3. Post World War Two: Population Censuses at Last

The title to this paper promises some personal reflections and clearly I played little part in statistical development before World War Two. For most of the first two decades after the War, I was far too engrossed in my own adolescence and training in another part of the world to even know where the Pacific was. But profound changes were taking place that were to provide an opportunity for me to travel to the Pacific, that in turn would determine my entire professional life.

As the 1960's approached, the relationship between the "Western Powers" and their colonies was in transition. In the Pacific, as elsewhere, grass roots independence movements were gaining ground. Samoa was the first to achieve independence. Over the years, challenges to the New Zealand authorities had grown, especially from the matais (chiefs), who had organized themselves as the Mau movement to become the only legitimate opposition. From 1953, preparations for independence began, and was finally obtained in 1961.

Though full independence for most other Pacific island protectorates and territories was at least a decade away, there were very noticeable changes in attitudes occurring as philosophical questions on the nature of colonialism and its benevolence were raised. My involvement in Papua and New Guinea (comprising the two territories administered by Australia) from 1963 and, to a considerably lesser extent the New Hebrides (now Vanuatu), illustrates how these changes extended to official views on data collection and building of statistical infrastructure.

In the early 1960's statistics on (what I shall continue to call) Papua New Guinea were almost non-existent. Less than 70 per cent of the country was considered to be under administrative control. A network of district headquarters, sub-district offices and scattered patrol posts had been established by the (then) Department of Native Affairs. Foot patrols, led by a kiap (patrol officer), endeavored to visit each village at least once a year although this was not possible for many of the remote areas. Apart from the exercise of their jurisdiction as legal officers and their assorted social and administrative tasks ranging from health and land disputes to recruitment of indentured labour, the patrol officer had one task with very important implications for statistical collection. At each village a "book" was maintained, that included essential information about each man, woman and child. Details included name, relation to household head, sex, date of birth, and, sometimes, notes on migration or other activities. The statistics compiled from the village records were defective in many ways. Coverage was incomplete, the records lacked simultaneity, national totals were always out of date, and there was insufficient detail collected to serve as a proper basis for planning. Yet they represented by far the best and most consistent set of data available, and provided the estimates of the population and its age structure presented in the Annual Report of the administration.

While the village books had serious defects from a national viewpoint, for each village in which they were maintained they were usually very accurate, and villagers were keen to ensure all household members including new-born children were recorded. Throughout much of the highland region houses are scattered among agricultural "gardens" with the result that, unlike the more nucleated villages of the coastal areas, communities are more difficult to define in spatial or geographic terms. For registration purposes clans or "lines" would be asked to assemble at designated times and meeting areas (sing sing grounds) to update the village books and meet with the kiap. In retrospect the most amazing aspect of this arrangement, given the disruption it caused to rural life, is that it worked so well.

The principal advantage of the village registration system as a source of population statistics was that it was conceived foremost as an administrative and legal system; statistics were really little more than a very useful by-product. Interestingly this informal system for collecting statistics proved far more valuable than the formal civil registration system. Official statistics of births, deaths and marriages were required under the existing Australian legislation only for Australian citizens residing in the external territories. Thus when my own daughter was born in Papua early in 1965 there was a legal requirement to register the birth. Lamentably, the few events occurring to expatriates comprised the entirety of the vital statistics presented in the United Nations vital statistics reports: no mention anywhere of the two million or so Papuans or New Guineans!

In this context the decision in 1964 to conduct feasibility studies leading to (most likely) a census in 1966 was a landmark in the statistical history of the country. It was only a few years later that the New Hebrides, similarly starved of statistics of its pi-Vanuatu population, decided to conduct a census in 1967.

Some reflection on this first population census of Papua New Guinea is appropriate since after all these years it still stands, in my view, as the most accurate conducted. With strong support from the Australian Bureau of Statistics (ABS) and an overwhelming commitment from the Papua New Guinea administration to conduct a successful census, conditions were as favorable as they could be. The existence of near-complete village books, which were updated specifically for the census, virtually guaranteed good coverage. Design and overall technical control remained the joint responsibility of the ABS and local administration; organization and supervision of fieldwork was assumed by District (now Province) Commissioners and their subordinate staff; all schools were closed during the enumeration period to release secondary schoolteachers (who were all expatriate) to serve as census supervisors and primary schoolteachers as enumerators. Census design was complex, involving as it did the selection of a sample of rural villages with elaborate rules to ensure that movement between selected villages and other areas did not contaminate coverage.

By and large the census was seen as an opportunity for festivities, especially in the more remote rural areas. From my own perspective, the census provided a rare opportunity to travel (mainly by foot) around a remarkably beautiful country, to meet many of its diverse peoples - outgoing, humorous, generous and proud - to attend marriages, funerals and traditional ceremonies and witness clan warfare. The same feelings that led to a belief that the time was right for a census in Papua New Guinea were also beginning to stir in the new Hebrides, where far more was known about the French and British residents than about the Melanesian people. With assistance from the Australian National University plans were put in place to conduct a full census in 1967.

Looking back, having assisted since that time on approximately thirty different censuses in the Pacific region, I am able to see both negative and positive aspects of that first census in PNG much more clearly. It stands as a metaphor, casting troubling images of the many paradoxes of development, not only for Papua New Guinea but for the entire region.

4. Independence and Dependence: Or Who Needs Statistics Anyway?

The driving force behind the push for improved statistics was undoubtedly the administering authorities, who with the best intentions, were responding to the need to leave viable systems behind. In PNG and Vanuatu, as elsewhere in the Pacific, the senior posts in the statistical offices had always been filled by expatriates. The perceived needs to improve basic statistics provided one of the best arguments for continuing to employ overseas staff beyond independence. After all, how could local staff be expected to fill these technical posts? Some years down the track they might, the argument went, but not now.

While I think this sentiment was widespread, the political will to localize national statistical services was one of the important elements in determining the real pace of statistical development. A second was the state of education and the potential of national staff to be trained speedily and mature as professional statisticians. By the 1970's significant differences were appearing in the responses of the fledgling Pacific island nations to the needs to manage their own statistical services. The independent Polynesian countries of Samoa and Tonga, while continuing to second senior staff from overseas, were able to fill most staff positions with local personnel. The Micronesian countries, under the trusteeship of the United States, provided career opportunities to expatriate staff, with few openings at senior level for Micronesians. In any case the centralization of the TTPI administration in Saipan was not able to cater for the future fragmentation into small island nations.

The situation in Melanesia was the most complex. Nationhood in this ethnocentric sub- region was an outward looking concept for most where social welfare and politics were seen as very much parochial issues. Education systems were poorly developed and hardly geared to meeting the technical needs of a modern statistical office. I recall in my early days in PNG the first ever graduate returning from Australia to wide acclaim. In Vanuatu, the uncanny mix of Anglo-French education was confusing at best and hardly conducive to developing the high-level skills required for the national statistics office. Fiji provided an exception and was ready to replace expatriate with national staff on achieving independence.

Staffing was not the only problem faced in the attempt to develop national capabilities. As the new nations looked beyond their traditional donors in search of development assistance, they found that funds for statistical activities were available if they could help meet the data needs of the donor communities. Given the problems that national planning agencies were having in finding national staff with the qualifications and background to define their own tasks, to articulate their data requirements and to find resources to support their priority programmes, it was virtually impossible to resist these outside pressures.

Methodological and technological developments in the field of statistics in a perverse way compounded the problems of dependence. Notions of excellence in statistics related largely to performance at international gatherings and the technical quality of papers and, not surprisingly, the range, innovation and presentation of data. Little heed was paid to the importance of strengthening national skills in basic statistics. As a consequence, there was an even greater incentive to recruit international staff to bolster the reputation of the statistics office at home and abroad. Advances in computer processing was a very important cog in this chain of dependence.

5. New Technologies to the Rescue

In many countries national statistics offices have been at the vanguard in exploiting possibilities offered by the new automated technologies. Computing, as even the name suggests, lends itself well to statistical applications. Moreover, each generation of computer is more compact, more robust, and, most importantly of all, cheaper. One might think that the potential power to process and analyze statistics offered by the microcomputer and the access to statistical skills provided through packaged software, would help finally to break the unhealthy chain of dependence referred to in the previous paragraph. Herein lies a paradox.

Undoubtedly the countries to gain most from computing are the most developed. From the highly scientific to the most humdrum, computer-assisted applications have found a place. Man has reached the moon, designed airports, diagnosed the sick, run offices and homes, all with the assistance of computers. National statistics offices have been able to harness some of the enormous power on offer to speed and improve efficiency of data processing and tabulation. In some countries computer- assisted applications include sampling, quality control, coding, management, mapping, publishing, to name a few. Improved user services have been established providing automated access to data through a computer terminal, telephone or other media.

Unfortunately the small statistical offices have been slow to exploit the potential of computers. This is not to say that computerization has not spurred statistical development or contributed to improving statistical services. But relative to the larger or more developed countries, small island countries have achieved only limited success. Why is this? Part of the answer lies in the tardiness in introducing data processing to the Pacific, a tardiness linked to dependence on more advanced countries.

A review of the processing of censuses during the past few decades will illustrate the point. I have inferred that Polynesia in many respects has developed its statistical systems in more favourable circumstances than elsewhere in the Pacific. Yet in Samoa the three censuses held prior to 1986 were processed in USA, New Zealand and Fiji. As a result until very recently there was no data processing capability within the national statistics office. The 1986 census was thus the first to be processed nationally (in fact, processing was undertaken by a commercial company). Not surprisingly the decision to process the census in Samoa in 1986 brought with it a new clutch of problems and challenges. But it left behind a strong desire to develop computing in-house, leading to the establishment of a small facility in time for the 1991 census.

In Tonga the situation was no better. With a history of dependence on outside agencies and advisers, development of an in-house computing facility has been slow. Problems in processing (among other things) led to unreasonable delays in producing results in both 1976 and 1986. Indeed, even with the help of a regional agency to process the data for the 1986 census, the results took five years to produce. Only now has a fragile computer facility been set up.

Experiences in smaller Polynesian countries introduce the more fundamental issue of scale. It is ironic that in two of the smallest countries in Polynesia, Tuvalu and Niue, computers were being used quite successfully by the mid 1980's. The reason for this was the interest in computing shown by the incumbent statisticians. There is a remarkable similarity in the two cases. In both countries a single statistician ran the statistical service. While neither had a formal tertiary qualification, both had demonstrated an interest in and aptitude for computing. In Niue where there has been, fortunately, little change in personnel, the competence remains; in Tuvalu, with considerably greater turn-around, it has proven more difficult to sustain capability. The real issue is that the very small countries are exceptionally vulnerable to the whims of governments in making appointments and in staff movements, including overseas migration. This vulnerability results largely from the lack of depth in statistical capability in the national statistical office.

Elsewhere in the Pacific, even among the "larger" countries, computing in statistical offices has taken time to grow firm roots. In the TTPI the 1980 census was conducted by the United States Bureau of the Census. Completed census forms were sent to Washington for processing, with the familiar consequence that the opportunity to build competence in the island countries was lost. The record since then is mixed, but reinforces the argument that development of national skills has been slow. Palau, with the assistance of an adviser, conducted and processed its own census in the mid 1980's. In the 1990's, it again solicited the assistance of the US Bureau, although it appears to be developing at least some in-house capability. The Federated States of Micronesia, also formerly part of the TTPI, following unsatisfactory results in the US censuses of 1970 and 1980, made more successful attempts of its own in 1973 and the late 1980's. For the planned 1994 census processing will largely be done in-country, but not without the assistance of a full-time adviser.

Melanesia, containing the largest countries in the region, might have avoided some of the problems of the very small nations. But still, the situation is not particularly healthy. Papua New Guinea has in fact a long record of using electronic data processing equipment. The 1966 census was processed in Australia, but the national Bureau of Statistics was quick to see the advantages in developing its own capability. All censuses since 1971 have been processed at home. Yet throughout this period, the data processing unit has been assisted by overseas experts. Even now there are many who still believe that recruitment of foreign advisers in data processing in the best way to prepare for the next census. Both PNG and Fiji have in fact developed strong computing centres, within the formal structure of the Bureau in the case of PNG, and outside in the case of Fiji. But, possibly as a result of the existence of these centres, both have been slow to respond to the need in statistical offices to develop computing skills among subject-matter statisticians. This is a point that I shall return to. Vanuatu illustrates the risks involved in (the probably inevitable) linking of statistical development to donor assistance. Of all Pacific countries, Vanuatu has been best served as a recipient of aid in the form of computers and computer experts. Yet its statistical service remains among the least developed in the region and it is doubtful to my mind whether much in the way of permanent competence has been established even in the field of data processing. In the Solomon Islands, the 1991 census was processed in Holland, leaving the Pacific statistical office without a proper computing facility and wasting an excellent opportunity for national staff to gain hands-on experience.

Unfortunately the availability of packaged software has not provided the panacea that many hoped it would. Certainly the ideas and expertise encapsulated in a computer programme that is itself inexpensive and portable, do seem to offer skills at the touch of a keyboard. But as always there is a rub (or rubs!). First and foremost vendors and developers want you to buy their products. Once purchased, they would like you to continue to buy their products. To do this they need to lock consumers into, yes, a dependent relationship. The hardware purchased only supports a limited range of software; up-grades and advice are only available to valued(?) customers. Alas, PIC's are small and able to assert very little influence on a global market. Added to which the national statistical offices, ever dependent on aid, at times have little say in the equipment they get. As a result the hardware configurations and the software used are not always well-suited to the job.

A far more serious problem, however, is the lack of training among national staff in specialized application or systems engineering. There are a huge number of courses available for specific software packages. But in many instances participants from Pacific countries are sent to courses with no particular brief on what training they should be seeking nor with any thought on the relevance or applicability of the software to the Pacific environment. With limited experience in its use and almost none with systems design or maintenance, it is rare that a participant on return home from these courses is willing or able to install and use the new application. This raises several issues of vital concern to the Pacific statistical offices:

1. How do they obtain information on the availability and suitability of software packages?

2. How do they develop national skills to support particular applications?

3. How, especially in the absence of national skills, do they obtain cheap, speedy and inexpensive advice on all of the many problems they encounter?

6. Institutionalizing Statistics: Putting Pacific People First

A colleague of mine, himself a Pacific islander, relates a story of an adviser who visits the Pacific only to complain about the laziness and incompetence of the local population. His successor, taking time to sit and listen to local people, can in contrast report only of his wonder at their knowledge and creative abilities. In a short period he learns much of local agriculture, fishing and boat-building. The purpose of relating this story is not to down-play the contribution that international donors have made to the development of statistics in the region. Rather I would wish to emphasize the importance of continuing to take the challenges and opportunities offered in the region very seriously. But I do believe there are important lessons to be learned from the Pacific that should guide future development initiatives. A constraint to development has been, as I have shown, an unhealthy reliance on foreign expertise. The current position vis-a-vis the role of computer technology discussed above, exemplifies this point well. Despite the excellent performance of many Pacific Islanders on data processing courses, by and large there is still far to go in developing strong in-house facilities and national capabilities.

In the complex and emotive area of national versus overseas specialists, I believe a greater sense of the long term needs of the statistical offices is required. It is easy for both the donor and recipient to propose the recruitment of an adviser to remedy an apparent deficiency in statistics. But history sits firmly on the side of caution. Samoa, Cook Islands and Tonga are exemplary in their judicious use of technical assistance, even if the preference for national personnel has at times resulted in some loss in quality or timeliness of statistics. All substantive posts in these offices are filled by national staff. In Samoa, two United Nations Volunteers assist in data processing and cartography: other short-term advisory services have been provided for the agricultural and population censuses. In Tonga, the top posts are now filled by national staff: resident experts in economic statistics and short-term visits from advisers in demographic statistics have provided technical support.

Other statistical offices are still recovering from years of misplaced benevolence. In PNG until the appointment of the incumbent, no native Papua New Guinean had ever been appointed to the top post. Worse from the perspective of long term planning, almost all second tier posts were filled with foreigners. When at last the national staff went on strike to protest the situation, and a Papua New Guinean was appointed, the problems created through lack of foresight were just beginning. The new statistician had little experience in top management and no preparation for the job. In an ironic way the situation now called loudly for an input by sensitive and competent international staff to assist through a difficult period. This went unheeded. Similar failures to balance short-term needs with longer-term development of national capabilities can be found throughout the Pacific. In the FSM, for example, following the completion of an expatriate's contract as head of the statistics section in the Planning and Statistics Office, the few local (and junior) staff remaining in the section were snapped up by other departments, with the result that no statistical organization existed at all. After some turbulent years, the first National Statistician was appointed, and he has continued to serve in the position. A parallel situation with a similarly successful and stable outcome can be observed in Kiribati.

The problems of transition are epitomized in the recent history of Vanuatu. Up until 1987 the statistical office had a long unbroken tradition of appointing a French or British national as the Principal Statistician. At that time the most recent appointment from the United Kingdom had been met with some disquiet, many, including myself, believing that the acting statistician, a national, was perfectly competent to do a good job. The effect of appointing an overseas person was to lower the morale of the office and to bring about the resignation of the acting Statistician. Clearly the time was ripe to prepare for the next change at the top. For the 1989 census, UNFPA agreed to support a national post of Census Commissioner (at equal rank to the Principal Statistician) on the understanding that the appointee would serve as the Principal Statistician elect, to be appointed to that post within three years. With near-certain knowledge that he would be appointed, the Census Commissioner (with the help of advisers) was able to use his time well to prepare for the transition. In my view his successful term as Principal Statistician justified the elaborate preparation.

The situation today is a far cry from the recent past. Long-term observers, among whom I include myself, comment at regional and international gatherings firstly that all or almost all PIC's are represented by national staff and secondly on the incredible growth in confidence as these largely young statisticians present an island perspective on statistical development. These comments, I am sure, sound singularly unsurprising to people from other parts of the world.

7. Molding Statistical Services to Pacific Needs

A common thread in these transitional years is the struggle to develop truly national statistical offices. Somehow in the process of creating momentum to achieve this end, the users of statistics have been forgotten. Who then are the users and why have they remained so silent?

During the past decade, I have been involved in six or seven country reviews of national statistical services. Most striking is the widespread belief by potential users of statistics that the national statistical offices are not doing a proper job. Among the major complaints are that priorities are distorted, statisticians do not listen to users, statistics are out of date and of poor quality, reports are unintelligible and other data are inaccessible. None of these criticisms are entirely fair but equally statisticians should be concerned that so many leaders and decision- makers feel the way they do.

The most important users are those involved in development planning and in implementing the various policies and programmes that must be put in place to achieve plan objectives. Often the more outspoken users are the academics and donor representatives who are better able to articulate their data needs and offer support in their collection and processing. I asked why the (more important) users have remained so silent. Some statisticians assume that users know what they want and that they will make most use of what they are given. Both assumptions are false, at least in a sufficiently large number of instances to raise questions. The development of the national statistical offices has not occurred in isolation. Planning agencies have just as dourly struggled to grow and to reflect the Pacific perspective. As a result a relatively large number of planners are coming to terms with their new responsibilities, still working out what is involved and how statistics might help. They have remained silent in the dark, but are now moving out of the shadows. A surge in demand from national users has only just begun.

One example will suffice to illustrate the extent of the problem. In Samoa, recognizing the need for better dialogue with users, the Department of Statistics agreed to outpost statistical officers to various key ministries. Within a very short period all these outposted officers were offered promotion in the substantive ministries, reflecting their value as statisticians to fledgling planning offices, but at the same time causing a serious loss of human resources to the statistics office. The solution, understandable in the circumstances, was to discontinue the service, hardly in the best long-term interests of Samoa.

There can be little doubt that the greatest challenge for national statistical offices will be the training of national staff. But resources for statistical development will always be tight and more cost-effective ways of meeting user needs will have to be devised. For small Pacific countries, the resources needed to conduct regular censuses and surveys to meet all the needs of users would very quickly prove prohibitive. Under the present system, the smallest countries can do little more than the population census and, possibly an agricultural census. These two major undertakings consume most of the resources of the statistical offices during fieldwork and processing, but at almost any time account for much of the intellectual work of the senior staff.

More sample surveys may offer a partial solution to meeting the growing need for data on a host of topics of relevance to national development planning, and in my view have been dismissed too easily. An obsession with sampling errors will of course lead to a conclusion that in many of the small PIC's, sample surveys are inefficient. However a more important variable is the ability to control the quality of fieldwork with far too few trained staff and meagre financial resources. In many cases the non-sampling errors that result from poor design, supervision and overly-ambitious objectives easily outweigh considerations of sampling error.

However, quite clearly neither censuses nor surveys provide a complete answer and the small countries must look elsewhere for a solution. In the field of population for example it is apparent that apart from Fiji, no Pacific country has really conducted a successful demographic survey. Population censuses most certainly provide some data, though of insufficient detail to meet in full the needs of planners. Civil registration could provide the continuous data required for monitoring of health and population programmes, but in too many cases coverage is defective.

The statistical systems required in the Pacific must therefore look beyond traditional censuses and surveys, which can be prohibitively expensive and time-consuming. They must take account of the peculiar resource constraints facing small island nations. Statistical offices have preciously few trained staff, suffer from high turnover, and are too small in most instances to train or need specialists. Yet at the same time they must respond to the growing demand for data as users expectations and sophistication grow. To meet all these requirements, Pacific countries will need to enlarge the concept of a statistical system to ensure that information that is of potential use in planning or fulfilling the goals of national development, regardless of its source, is so valued that it is fed into the system to be shared by users. Such an enlarged view of the statistical system will require changing perceptions about where statistics come from and how they are compiled.

8. The Statistical Grid: A Glance at the Future

During the past months, I have sat on various forums with my United Nations colleagues listening to cries for improved statistics to develop plans of various kinds and to monitor projects and programmes. Thus as UNDP launches its Equitable and Sustainable Development Programme, it will necessarily need to build into the strategies for human development a provision for monitoring progress. In compiling The Human Development Report for the Pacific it will equally need to provide a comprehensive range of data describing the current situation and identifying problem areas, sufficiently accurate and broad to serve as a benchmark for future monitoring. In a similar way the UNIFEM Pacific Mainstreaming Project on women will need to establish a statistical framework describing the situation and status of women in Pacific countries and establish a system for monitoring change. A host of other agencies including UNICEF, WHO and the South Pacific Commission are anxious to establish procedures to measure the health of infants, children and mothers and to monitor the movement towards fulfilling the various global and national goals and declarations.

Each of these agencies will, if they have not already done so, in turn approach the national statistics offices in the Pacific countries to provide as much information as they can relating to the various area of interest. Just as certainly each in turn will bemoan the fact that, apart from the population census, so little is available and probably add that a special survey is needed. Some, though not all, will accuse the statistical office of lack of interest or lack of cooperation.

In some countries agencies such as the World Bank and the International Monetary Fund, not finding what they want, actually provide funds to conduct a survey. While this might contribute to national development and produce useful statistics, it has in several instances done little more than overstretch the limited and valuable national resources and distorted national priorities.

The direction in which the national statistics offices will need to move to provide the wide and flexible service that is being asked of it, will almost certainly be away from the high dependence on censuses and surveys. We are approaching a new age of information technology and, rather than fear it, we in the Pacific must learn how to exploit it to our advantage. It is hard to imagine any government on non-government agency not collecting at least some records relating to the administration of their programmes. THESE RECORDS ARE VALUABLE and should provide the raw data for the statistical system. Consider some of the advantages:

• they cover almost every conceivable area of human endeavour

• since they were collected for some other purpose, as a source of statistics they are inexpensive

• they draw statistical users and producers together in a common system unlike censuses and surveys, they provide continuous records providing statistics for any defined time interval

• they cover all programme participants

Of course this scenario is a little idealistic. In reality, setting up a comprehensive system will not be easy. A number of problems both administrative and technical will need to be resolved. Let us consider some of these:

• many of the records are badly kept/are incomplete

• concepts and def~nitions do not accord with statistical standards

• some agencies will be secretive or uncooperative

• agencies are more interested in their own programmes/not trained in statistics

The recognition of both the potential and the problems to be overcome helps considerably in putting together national strategies for improving statistics. The most important feature must be the need to begin immediately the move towards long-term goals. The two most important pivots of any national strategy are l) the need to identify the most important and easiest areas for starting the expansion of the statistical system and 2) to train staff in both the statistical office and the substantive agencies in the requirements of the system.

As a beginning, improvement of statistics might be restricted to a few sources since the efforts to cooperate and reorient will demand a great deal of commitment. While it is always important, emphasis on the national information/statistical network will add to the need to establish a high-level National Statistical Committee. This Committee will give direction to the statistical office and, most importantly will clear the way with potentially un-cooperative "partners". The composition of the Committee is thus crucial and should draw on the major departments represented at least at assistant secretary level. It helps to have a Minister as chairman!

Training is a key to success. Already international statistical agencies are giving more stress to what are loosely termed "administrative records" and training institutions are beginning to provide some instruction in applied statistics courses. But these moves fail to appreciate fully the importance of a change in emphasis to the small and fragile statistical offices in the Pacific. More training will be required at home and every opportunity must be taken for the users and producers of statistics (who will become less distinguishable) to meet together to discuss national and sectoral needs. Statisticians will need more training in design and standardization of forms and will need to spend far more of their time in the substantive agencies. In time as their contribution is recognized and as the statistical system becomes more attuned to national needs it is hoped that the Public Service Commissioners will work to ensure the statistical organization is appropriate for the task. Again, strong Statistical Committees would help.

To provide a proper basis for planning statistical services, government statisticians will need to sketch the course to be followed by the pathway to the future, along which, together, both users and producers will march. Momentum should be provided by donors who will also need to make a commitment to strengthening statistical infrastructure, recognizing that this will mean taking a longer run view on development, working together in a more coordinated way, and above all, ensuring that objectives and the strategies employed to achieve them reflect the needs of the Pacific.

While some of these approaches are novel, mostly they are already happening. What is important now is to speed and support the process in the hope that in a few years from now rather than continue its identification with poor statistical systems, the Pacific Island countries can look with pride at what they have achieved. We will nod agreement with our fellows from other parts who remind us that in the Pacific island countries there are few events, the scale of operations is small and cooperation is widespread, collection of data must be inexpensive and uncomplicated.