Monitoring Human Development Goals: A Straightforward (Bayesian) Methodology for Cross-National Indices

Abstract

We offer a straightforward framework for measurement of progress, across many dimensions, using cross-national social indices, which we classify as linear combinations of multivariate country level data onto a univariate score. We suggest a Bayesian approach which yields probabilistic (confidence type) intervals for the point estimates of country scores—a vital, and often missing, feature in cross-national comparisons. We demonstrate our approach using the United Nations Development Programme’s Millennium Development Goals (MDGs), via the Maternal and Neonatal Program Effort Index (MNPI) data (Ross et al. in Trop Med Inter Health 6(10):787–798, 2001), and Human Development Index (HDI) (2010) as examples.

Appendix

1.1 Potential Questionnaires on MDG Goals and Targets

A list of 15 questionnaires is suggested to parallel, though not exactly duplicate, the lists of MDGs and targets. These are listed in Table 2, which shows the goals and targets to which each refers. It also shows the output indicators related to each questionnaire that have been proposed in other documents. These output indicators were meant to be suggestive rather than comprehensive, presumably chosen at least partly for the availability of reliable data. What the questionnaires should address is the effort that has gone or is going into improving not only these outputs but also other outputs related to the broader goals and targets. The list in Table 2 follows the order of the MDGs.

Table 2 Proposed questionnaires on effort and the goals, targets, and output indicators they should cover

Table 3 Classification of questionnaire items

1.2 The MNPI Effort Questionnaire

A outline of a questionnaire on effort at achieving the maternal mortality target is provided here, by design of Ross et al. (2001). The data for the illustration in the paper follow this organization. We do not reproduce the entire questionnaire here.

1.2.1 Organization of the Questionnaire

The questionnaire is organized in two parts. The first, much longer part requests ratings of different features of a maternal health program. The second, short part (labeled "General background") requests relatively objective information about laws, plans, budgets, facilities, etc. relating to maternal health. All respondents are expected to answer the first part, but only a few, those more closely connected with the government maternal health program, are to be given the second part to answer. Though the two parts are somewhat different in format, they are not separated so that respondents who receive both parts will see them as a single questionnaire.

Substantively, the questionnaire covers typical project components of policy and planning, funding, service delivery, and demand generation. However, questions are not posed in this order, but start with service delivery. The purpose is to fix the respondent’s attention initially on what services actually reach women in need and can have direct effect on reducing maternal mortality. The questionnaire seeks to emphasize what is actually making a difference on the ground rather than what agreements and plans are made on paper. After asking about services in several different ways, the questionnaire moves to more general policy issues.

Questions are not necessarily grouped in categories familiar to donors. Instead, they are grouped for convenience, keeping together those with a similar frame of reference requiring answers in a similar format. Nor are questions intended as a checklist of all the specific requirements for providing proper maternal care. To keep the questionnaire at reasonable length and to avoid asking about details too fine for some respondents, the questions necessarily reflect a sampling of important best practices and dimensions of effort. To indicate how responses might be reclassified, after the data are obtained, to reflect particular issues of relevance from a planning perspective, Table 3 provides an illustration The table lists some items more than once, as reflecting more than one aspect of performance. Some items could be listed under even more categories. Subsequent empirical analysis may suggest the most useful groupings.