¹ Bebbington, A. C. and Davies, Bleddyn, ‘Territorial Need Indicators: A New Approach’, Part I, ‘The Intellectual Context’, Journal of Social Policy, 9:2 (1980), 145–68.CrossRefGoogle Scholar

² Imber, Valerie, A Classification of the English Personal Social Services Authorities, Statistical and Research Report Series 16, DHSS, HMSO, London, 1977.Google Scholar

³ See Feldstein, M. S., Economic Analysis for Health Service Efficiency, North Holland Publishing Company, Amsterdam, 1967Google Scholar; Davies, Bleddyn, Social Needs and Resources in Local Services, Michael Joseph, London, 1968Google Scholar; Culyer, A. J., Lavers, R. J. and Williams, A., ‘Social Indicators: Health’, in Social Trends, no. 2, Central Statistical Office, HMSO, London, 1971Google Scholar; and Davies, Bleddyn, ‘Needs and Outputs’, in Heisler, H. (ed.), Foundations of Social Administration, Macmillan, London, 1977 – ch. 9.Google Scholar The theory of the need judgement stated in these papers shows a family resemblance to the argument developed by Holtermann that is to be expected, given that they all draw their inspiration from the same body of micro-economic principles – see Holtermann, Sally, ‘The Welfare Economics of Priority Area Policies’, Journal of Social Policy, 7:1 (1978), 23–40.Google Scholar The above papers contain a more developed treatment of the normative theory of the cost-benefit need judgement and the need indicator than is stated in this article, as do Culyer, A. J., Needs and the National Health Service, Martin Robertson, London, 1976Google Scholar; and, among other PSSRU papers, Davies, B.. ‘The Measurement of Needs and the Allocation of Grant’, in Report of the Committee of Enquiry into Local Government Finance, Cmnd 6453, HMSO, London, 1976Google Scholar, Appendix 10. The PSSRU approach to the explanatory theory of the cost-benefit need judgement is explored in, for instance, Davies, B., ‘Causal Processes and Techniques in the Modelling of Outcomes’, in Young, K. (ed.), Essays on the Study of Urban Politics, Macmillan, London, 1975, pp. 78–105Google Scholar; and Davies, B., ‘Social Service Studies and the Explanation of Policy Outcomes’, Policy and Politics, 5:3 (1977), 41–60.Google Scholar

⁴ Another reason why social justice may demand different need judgements, in theory at least, is that welfare ‘technology’ may differ between areas. A service may be operating more efficiently and producing higher levels of benefit in some areas than it does in others. However, it is no part of our need indicator approach that local authorities or areas should be compensated for their inefficiency or lack of knowledge of the most effective methods.

⁵ There are three courses of action that a social services department might in theory adopt with regard to the substitution of its services with those of other agencies: (a) central: in which the social services department regards itself as having prime responsibility for achieving certain welfare benefits, and conducts its planning in a manner which ignores the contribution that other agencies may be prepared to make. In this case benefits resulting from the services of other agencies are simply additional; (b) residual: in which the social services department regards the services of other agencies as given, and seeks only to supplement them in such a way as to maintain the total level of welfare benefits for its clients; and (c) joint planning: in which the social services department collaborates with other agencies in an attempt to provide a mutually agreed level of benefits at the minimum social cost. The theory of the need judgement in principle permits the need indicator to be calculated under any of these forms of substitutability between agencies. Although much lip-service has been paid to the joint planning model, in practice it has varied according to the context – for example Townsend debated the ideal form of co-operation between public and voluntary agencies on the one hand and family and community on the other in the care of the elderly – see Townsend, P. and Wedderburn, D., The Aged in the Welfare State, George Bell and Sons, London, 1965.Google Scholar In practice, in recent years the domiciliary services have adopted a more ‘residual’ ideology in terms of intervening to meet the needs of the elderly with which families and the community are unwilling or unable to cope.

⁶ A parallel approach to the planning problem, based on cost minimization subject to constraints on client numbers and resource availability, has been developed in McDonald, A. G., Cuddeford, G. C. and Beale, E. M. L., ‘Balance of Care: Some Mathematical Models of the National Health Service’, British Medical Bulletin, 30:2 (1974), 262–71.CrossRefGoogle ScholarPubMed

⁷ See for example Harris, A., Social Welfare for the Elderly, HMSO, London, 1968Google Scholar; Planning Section, London Borough of Greenwich, Old People's Homes and Sheltered Housing in Greenwich, 1972Google Scholar; Duncan, I. B., Race, D. G., MacFarlane, S. B. J. and Tate, M. J., The Care of the Elderly, Operational Research (Health Services) Unit Report, Department of Applied Statistics, University of Reading, 1974Google Scholar; Davies, R. M. and Duncan, I. B., Allocation and Planning of Local Authority Residential Accommodation for the Elderly in Reading, Operational Research (Health Services) Unit Report, Department of Applied Statistics, University of Reading, 1975Google Scholar; Brearley, C. P., Allocating Priorities in Residential Care for the Elderly, University of Birmingham Clearing House, 1975 (7), pp. 101–4.Google Scholar The Harris definition was intended to determine whether interviewees ‘ought’ to receive home help. It is the least complex of those mentioned in our article and also the earliest. It assumes that need amounts to the inability of a subject to take care of himself, and applies in the absence of anyone else who might undertake caring functions. The Greenwich definition is based on the combination of three sub-indices – physical state, mental state and social state. However, it does not link self-care needs to the availability of help, and some of those judged by the criteria to need residential care would not be recipients in today's conditions in many authorities (for example an elderly person who lives alone and is often depressed and bored, but has no other disabilities or other deprivations would not now be considered a high priority for residential accommodation). The other definitions were intended to provide a basis for an indicator of the degree of need for residential care. They were developed for use in actual allocation processes and have been subjected to a validation process against actual decisions.

⁸ These include Abel, P., Carter, K. and Luck, C. M., Care of the Elderly: The Allocation Process, Report no. 703, Institute of Operational Research, Tavistock Institute of Human Relations, 1972Google Scholar; Jackson, R. R. P. and Himatsingani, C., ‘Measurement and Evaluation of Health and Personal Social Services for the Elderly’, in Canvin, R. W. and Pearson, N. G. (eds), Needs of the Elderly, Institute of Biometry Publication no. 2, University of Exeter, 1973Google Scholar; the latter being further developed by Canvin, R. W., Balance of Care in Devon Pilot Project: The Elderly Section, Operational Research Division of the Institute of Biometry and Community Medicine Report, University of Exeter, 1976Google Scholar; Brotherton, J., Gwynne, D., Renold, J., Thursfield, P. and Tomkinson, C. B., Manchester's Old People, Report no. C120, Local Government Operational Research Unit, Royal Institute of Public Administration, Reading and Manchester, 1972Google Scholar; Wright, K. G., ‘Alternative Measures of the Output of Social Programmes: The Elderly’, in Culyer, A. J. (ed.), Economic Policies and Social Goals, Martin Robertson, London, 1974Google Scholar; Stockport Social Services Department, The Elderly: Research Report, Metropolitan Borough of Stockport, 1976Google Scholar; and Coverdale, I. L. and Negrine, S. M., The Balance of Care Project: Modelling and Allocation of Health and Personal Social Services, Operational Research Services Report, DHSS, London 1977.Google Scholar

⁹ Studies from the Balance of Care programme – see Coverdale and Negrine, op. cit.

¹⁰ These are from the County of Glamorgan (Watson, M. and Albrow, M., The Needs of Old People in Glamorgan, Glamorgan County Council, 1973)Google Scholar; the Metropolitan District of Newcastle (Newcastle City Council Social Services Department, Survey of the Elderly in Newcastle, University of Birmingham Clearing House, 1976[7])Google Scholar; and the London Borough of Croydon (Croydon Social Services Department, 1974Google Scholar, unpublished report).

¹¹ It is regrettable, although inevitable, that the more sophisticated definitions developed in the literature on the subject could not be used in the analysis because of limitations in the context of the needs surveys which we used. The four need definitions which we did use (see Tables 1–4) have been adapted somewhat from the original forms, as is explained in the text. The definitions attributed to Duncan et al. (op. cit.) and to Brearley (op. cit.) have, in their original forms, undergone validation by comparison with actual resource allocation decisions.

¹² It would be tempting to draw a parallel between our ‘moderate’, ‘considerable’ and ‘intensive’ need target groups and the classification by Isaacs, and Neville, in The Measurement of Need in Old People, Scottish Health Service Studies no. 3, 1975Google Scholar, of need amongst the elderly into ‘long interval’, ‘short interval’ and ‘critical interval’, relating to the urgency with which help must be supplied when it is required, in order to maintain a basic living standard. However, the bases for our categories are not the same as those of Isaacs and Neville; it is an open question how far the groupings-based definitions of need used here would overlap with one based on the judgements which the Isaacs and Neville method demands. It was not considered feasible to use Isaacs's and Neville's method with the data bases studied here.

¹³ The American literature has been much concerned with the accuracy of synthetic estimators; see Gonzales, M. E. and Wakeberg, J., Estimation of the Error of Synthetic Estimators, Conference of the International Association of Survey Statisticians, Vienna, 1973Google Scholar, unpublished; and Erickson, E. P., ‘A Regression Method for Estimating Population Changes of Local Areas’, Journal of the American Statistical Association, 69 (1974), 867–75.Google Scholar Useful reviews include Gonzales, M. E. and Hoza, C., ‘Small Area Estimation with Application to Unemployment and Housing Estimates’, Journal of the American Statistical Association, 73 (1978), 7–15Google Scholar; while more specifically in the welfare field is Levy, P. S. and French, D. K., Synthetic Estimates of State Health Characteristics based on the Health Interview Survey, Vital and Health Statistics Series 2, no. 75, U.S. National Centre for Health Statistics, U.S. Government Printing Office, Washington, D.C., 1977Google Scholar.

The approach nearest to ours in the American synthetic estimation literature is that of Heumann, L. F. in Identifying the Housing and Support Service Needs of the Semi-Independent Elderly: Towards a Descriptive Planning Model for the Areas Agencies on Aging in Illinois, Housing Research and Development, University of Illinois, 1977.Google Scholar The study was based on the first of the two kinds of synthetic estimation technique, that is, the one based on the cross-tabulation of area census data into cells that are mutually exclusive and between them inclusive of the whole population. Heumann concluded in applying his method that ‘It is virtually impossible to synthetically estimate the proportion of elderly at the neighbourhood level who suffer from functional disabilities’ – ibid. p. 93. For instance it would be possible to use his complete matrix of proposed applications in the synthetic estimation of needs for only one county in the State of Illinois. One clear merit of Heumann's study is that it consolidates need judgements for housing, domiciliary health and other domiciliary support services. Secondly, it considers, but does not fully integrate ‘sub-area quality’, discussing the development of an index of the ‘location and quality of independent housing’. Thirdly, it discusses the need to take account of variations in the administrative overhead time and the total time of service delivery personnel required to deliver an hour of service, arguing that this would vary territorially and between client groups. However, Heumann did not use data to quantify the argument. In other ways his work is less advanced than our own; in particular it does not discuss the sensitivity of his synthetic estimates to variations in need judgements.

¹⁴ Table 5 illustrates a method by which a prediction equation is developed for membership of each target group. This method was used with Definitions 3 (Table 3) and 4 (Table 4), but a slightly different approach was used for Definitions 1 (Table 1) and 2 (Table 2), which, it will be noted, are based on a need ‘score’. It was found that particular theoretical distributional forms (in one case the negative binomial distribution) fitted well to the sampling distributions of scores – distributions which only varied with respect to one parameter when comparisons between surveys were made. Hence the size of each target group can be determined from an estimate of the mean score in that area alone, and hence we could make do with a single prediction equation rather than one for each group.

Statisticians will be interested in the degree of success with which we were able to predict the target group membership of individuals from the census-style items illustrated in Table 5. This of course varied with surveys and with different definitions. For Definitions 1 and 2 this may be measured by the square of the multiple correlation coefficient of the regression equation predicting the need score. For Definitions 3 and 4, which use a linked set of four prediction equations, we use as a single summary measure ‘1 – Wilk's lambda’, which measures the proportion of the total variation between individuals on these variables that can be explained as variations between target group centroids (see for example Tatouska, M. M., Multivariate Analysis, Section 6:4, John Wiley and Sons, New York, 1971).Google Scholar These figures are as follows (expressed in percentage form):

¹⁵ The rule by which the elderly living in non-private accommodation, apart from those in local authority care, are assigned to target groups in the present prototype need indicator is as follows: it is assumed that 1.4 per cent of the elderly population of each local authority have needs which are being met by hospital services, and will be in the ‘no need’ category so far as the personal social services are concerned. The remainder in non-private accommodation are assumed to be distributed between the target groups in the same proportions as are the elderly in private households, and therefore it is merely necessary to scale up slightly the estimated target group membership of the latter group. We are aware that these assumptions are crude, perhaps the weakest adopted anywhere in our procedure. For example many of those who are self-supporting in nursing homes have physical conditions which put them well within the margin of entry to local authority residential care. The assumption being made is that the needs of the majority are being met by the establishment in which they are living.

¹⁶ Allocation studies such as those by Abel Carter and Luck in 1972 and the Borough of Stockport in 1976 (see footnote 8) have suggested that perhaps between one-fifth and one-quarter of the elderly in long-term residential care might be capable of returning to the community. These elderly people are divided between the ‘considerable’ and ‘intense’ need target groups on the basis of the proportions of the elderly in private households in the local authority concerned who fall into the four target groups, in such a way that, nationally, 80 per cent of those in residential care are assumed to be in the ‘intense’ need group. Exact details of this method are available from the authors of this article.

¹⁷ See Davies, Bleddyn and Knapp, Martin R. J., ‘Hotel and Dependency Costs of Residents in Old People's Homes’, Journal of Social Policy, 7:1 (1978), 1–22.CrossRefGoogle ScholarPubMed

¹⁸ See Davies, Bleddyn and Coles, Oliver, Labour Markets, Service Operating Characteristics, and the Unit Wage Costs of Helps and Organisers, PSSRU, University of Kent, 1979.Google Scholar

¹⁹ The standard unit costs for providing meals and day centre places to the elderly, excluding capital contributions, debt charges and income, based on 1975/6 DHSS revenue outturn statistics, are as follows:

Note that the standard unit cost of meals is slightly adjusted to all for provision by voluntary agencies and meals in day centres.

²⁰ See footnotes 7 and 8.

²¹ Total expenditures (manpower and running costs only) for services to the elderly, from DHSS revenue outturn statistics for 1975/6, were:

²² A full set of the tables referred to in the text and a complete specification of their derivation are available from the authors of this article.

²³ As footnote 22.

²⁴ By the policy paradigm is meant the entire constellation of beliefs, values, assumptions about cause and effect, appreciations about the techniques of intervention and beliefs about the consequences of alternative actions which are widely shared within the community of actors who influence outputs in the political and bureaucratic processes of service delivery. Although it is generally vague, containing innumerable contradictions, and although it very rarely leads to consensus judgements, it is something recognized by most, if not all of the actors of political and bureaucratic systems. Like Kuhnian scientific paradigms, policy paradigms are ideologies – social constructs. The policy paradigm provides the basis for social action. The concept ‘policy paradigm’ was developed by T. S. Kuhn in the United States in the early 1970s. Parallel concepts were being developed independently by British writers at much the same time. See for instance Sharpe, L. J.'s development of the concept ‘operating ideology’ in his essay, ‘Instrumental Participation and Urban Government’, in Griffiths, J. A. G. (ed.), Policy and Administration, Allen and Unwin, London, 1976Google Scholar; and Young, K. G., ‘Values: The Policy Process’, Policy and Politics, 5 (1977), 1–2.Google Scholar At the same time one of the writers was developing the concept of the policy paradigm – see Davies, , ‘The Measurement of Needs and the Allocation of Grant’Google Scholar; and Davies, , ‘Social Service Studies and the Explanation of Policy Outcomes’Google Scholar. The three writers together recognized the relationship between the concept in one form or another and other concepts like ‘needs’ in Davies, B. P., Newton, K., Sharpe, L. J. and Young, K. G., ‘Urban Policy Studies: The Prospects of Convergence’, Policy and Politics, 5:3 (1977).Google Scholar

²⁵ The usual argument is that this rationale can only be provided by developing a theory of causal processes which serves to relate the indicator to the unobservable factor. See for example Land, K. C., ‘On the Definition of Social Indicators’, The American Sociologist, 6 (1971). 322–5Google Scholar; Land, K. C. and Felson, M., ‘A General Framework for Building Dynamic Macro-Social Indicator Models’, American Journal of Sociology, 82 (1976), 565–604Google Scholar; and Bunge, M., ‘What is a Quality of Life Indicator?’, Social Indicators Research, 2 (1975), 75.Google Scholar

²⁶ Imber, op. cit.

²⁷ See footnote 22.