Governance indicators need to fulfil a certain number of criteria to be useful and usable and should be:
Valid, in that they measure what they purport to measure. You should aim to have the most direct indicator that measures exactly the attribute desired, and at the correct level. In many cases, it might be necessary to use a proxy measure, which is used to substitute for information that is more difficult to get. An example of a proxy measure is the indicator percentage of births attended by skilled health personnel as a proxy for maternal mortality rates. Maternal mortality is difficult to measure, but it is widely accepted that mothers whose births are attended are much less likely to suffer maternal mortality; thus, if "attended births" data are easier to obtain, they can serve as a proxy. In cases where data availability is limited, it may be beneficial to consider proxy measures. However, these measures are not always acceptable: The main issue would be how closely related the proxy is to the original target and how similar its behaviour is. The key problem would be if efforts were then targeted at improving the outcomes as measured by proxies, rather than the original target. When using a proxy, care must be taken to repeatedly ensure it remains valid for the original target. The key question is how far removed from the item one wants to measure is the item one can measure, and to what extent the behaviour of the proxy measure follows that of the desired measure.
Clear, in that those using them can understand them. Is it clear to data collectors what data are needed and how to collect them? The effectiveness of governance indicators requires a clear idea about what will be observed and how to do so. For example, for the statements (i) Institution X is operating more effectively/efficiently; and (ii) Measurable increase in client satisfaction with ministry Y’s services to be valid as indicators, it must be defined how one will measure effectiveness, integration or satisfaction – and how data are to be identified in actual practice. An indicator lacking clarity about what and where to observe is meaningless. We need to specify exactly what should count and how it should be counted, with the exception of existing standards that are commonly known.
Sensitive toward desired changes and specific groups. It is important to assess and monitor governance from the perspective of vulnerable and marginalized groups. For example, the poor might experience governance differently from the affluent, and men might experience governance differently from women. It is therefore useful to select governance indicators sensitive to reflecting potential differences. See below for more on this issue.
Motivating, by inducing intended performance. Review candidate indicators to see whether there are ways in which tintroduction of the indicator may change incentives guiding officials' or citizens' behaviour. Strive to minimixe any danger that an indicator will create perverse incentives that can undermine aims.
Practical, in that they are affordable, accurate and available. The process of selecting indicators should start from an analysis of what is available and feasible. Indicators not yet available should be included in the assessment only if it is realistic to set up a sustainable mechanism to collect and analyze data on them. Indicators also should be cost-effective in that the costs of measurement should be proportional to the benefits.
Owned, in that they are held as legitimate in the eyes of those affected.
Balanced, in thatrarely will just one indicator be sufficient to measure an attribute of governance. For example, "access to justice" will require a number of indicators to assess and monitor whether it is improving for citizens of a country. When there exist multiple indicators for a specific attribute of governance, it is important to ensure that such indicators are balanced. Important guiding principles for putting together such a basket include: using multiple outcome indicators for a single sub-area/institution of justice, such as courts; ensuring outcome indicators cover different sub-areas/institutions of justice, such as police, courts and prison service; and building simple causal chains of input, output and outcome indicators for a single sub-area/institution of justice.
Objective, in that anyone reviewing the indicator should reach the same conclusion about progress. Every indicator will have an underlying normative assumption -- simply the assumption that more (or less) of whatever is being measured is a good thing. Care must be taken when choosing indicators to ensure that normative assumptions are valid. For example, if the indicator voter turnout is used as a measure of democracy, the assumption is that a higher turnout is better, demonstrating greater buy-in to the democratic process and interest in the result. However, voter turnout is highest where voting is compulsory, such as in Cuba, Iraq and Australia. Here are some other examples: (i) for number of persons detained without charge, the assumption is that a lower number is better and (ii) for period between detention and trial, the basic assumption would be that a shorter period is better -- but care must be taken that the period is not so short as to prevent the proper preparation of cases for prosecution and defence.
A useful guideline for developing good performance indicators for the justice sector is provided by the Vera Institute of Justice, at http://www.vera.org/publications/publications_5.asp?publication_id=207. The guideline may work equally well for developing governance indicators in other areas than justice