1 A clearly defined accountability framework is the foundation of performance measurement, evaluation and reporting in non-profit organizations. The concept of accountability was defined to include both the hierarchical model of mandatory accountability (the requirement to render an account for a responsibility conferred) and the voluntary disclosure of relevant information to the various internal and external constituencies of a programme or organization. For both the narrow and extended models, we then argued that a meaningful framework of accountability must include: first, shared expectations about conduct and performance; second, a shared language or currency in which fulfilment of those expectations is described; third, shared criteria defined in that currency as to what constitutes fulfilment; and, finally, a means of communicating information about conduct and performance. The balance of Chapter 1 then explored the various dimensions of accountability, and foreshadowed the major arguments of the book by emphasizing the evolution of the categories of technical information used to fulfil accountability relationships—that is, the evolution of common currencies—and corresponding criteria of conduct and performance and ways of communicating that information.
2 In Chapter 2, the concept of evaluation that underlies any accountability information system was examined in detail. Evaluation can occur at differing levels (from individual to programme to organizational to system) and can be based on inputs, processes, outputs and outcomes. Methods of evaluation involve designing the process (which requires answering the questions: why is it being undertaken, what is to be evaluated, what type of evaluation is to be used, and who will do it); choosing the data-gathering modes (quantitative and/or qualitative) and deciding on the standards to be used in interpreting the data. A key concept in evaluation is that of the ‘logic model’. Every evaluation contains one whether it is implicit or explicit. Assumptions are made about the nature of the links between inputs, processes and outcomes, and between the levels of performance they are examining.
Common fallacies in logic models were identified and the ‘ideal’ evaluation process for dealing with them was described. The problem