© May 2005

This appendix serves to supplement the commentary in The Drum Beat (Issue #302) [1] by Jane Bertrand, dated June 6, 2005.

The available evaluation textbooks often provide useful glossaries of terms that help the health professional or student of programme evaluation to grasp the lexicon of terms used in this field. However, many people are unclear how one set of terms (e.g., formative-process-summative evaluation) relates to another (e.g., input-process-output-outcome). Are they totally different? Do they overlap?



Figure 1 [2] presents the most widely used terms in programme evaluation and is designed to shed light on this issue of overlapping terminology. Specifically, it outlines what different types of evaluation actually measure and how one set of terms relates to another. The figure consists of a series of six horizontal bars, which are meant to be read (1) left to right, to get a sense of chronological order, as in the case of formative, process, and summative evaluation, and (2) vertically, to compare the terms on one bar to those in the same "column" on another, with the aim of identifying how one set of terms (formative-process-summative) relates to another (inputs, process, outputs, outcomes). For example, the bar on "programme evaluation" runs the full width of the chart, indicating that it is an umbrella term that covers the full range of different types of evaluation. The same is true of the term "monitoring and evaluation" (M&E). The different bars of the chart are as follows.



What is measured? The first bar on the chart (in yellow) describes "what is measured" in terms that do not require specialised knowledge of evaluation. These include level of funding, activities completed, service statistics, and changes in behaviour, to name a few.



Formative, process, and summative evaluation. The second bar - in blue - relates the terms "formative, process, and summative" to "what is measured" in bar 1. For a fuller description of the terms, see The Drum Beat #302 (June 2005).



Monitoring inputs, outputs, and outcomes. The third bar in lighter blue indicates how the terms "input, output, and outcome" link to the previous two bars. One might question why "process" doesn’t get its own box on the chart. Indeed, we often hear mention of "process indicators." In this chart, we have instead used the terms "functional outputs," which in fact quantify the activities conducted (e.g., number of persons trained, number of booklets produced, number of schools enrolled, number of community meetings held). This contrasts to "service outputs," which relate to measures of quality of care and access to services, in those programmes where they apply. For example, "length of waiting time" would be one service output. "Number of VCT clinics per 100,000 population would be another example of a service output, in this case measuring access. The third type of output measures service utilisation: number of clinic visits, number of VCT tests performed, number of ORS packets distributed. Box 1 [3] provides more detailed definitions of input-process-output-outcome.



Figure 1 [2] attempts to shed some light on the term "monitoring," which is one of the most widely used terms in the field of programme evaluation (see the light green bar). Indeed, monitoring is a very broad term that can apply to tracking many aspects of the programme, from inputs used to start the programme to long-term outcomes such as fertility or mortality. The term "monitoring" generally implies some type of tracking or measurement over time, without benefit of causal attribution. For example, government and donor agencies may track the level of contraceptive prevalence in a given country over a 20+ year period, with a strong notion that their programmatic inputs have contributed to the continued rise in contraceptive prevalence. However, this type of monitoring does not take into account and can not rule out other factors (e.g., changes in economic conditions, urbanisation, literacy levels) that may also influence this change in prevalence over time.



Impact assessment. The light green bar distinguishing monitoring (tracking change) from impact assessment (i.e., did the intervention in question cause the observed results?) In Figure 1 [2], we set apart - as if in a class by themselves - those evaluation methodologies that allow for causal attribution. As described in the Drum Beat Commentary, these include experimental designs (rarely feasible when evaluating large scale communication programmes), quasi-experimental designs, or post-test only cross-sectional studies using advanced statistical techniques (e.g., propensity scoring, structural equation modeling). These designs yield causal inferences that are probabilistic or plausible, depending on the rigor of the design (Habicht et al, 1999).



Programme performance. The fourth bar involves "programme performance" (in pink). This term relates to the different types of monitoring (inputs, outputs, and behavioural outcomes), but stops short of long-term outcomes. The reason is that factors other than the programme also contribute substantially to these changes (e.g., in fertility, mortality, and morbidity), such as the state of the economy or the social status of women.



M&E and Programme Evaluation. The final two bars - "monitoring and evaluation" (peach-colored bar) and "programme evaluation" (light turquoise) run the full width of the page, indicating that both are umbrella terms used to describe any aspect along with continuum. Regarding M&E, some have made the distinction that monitoring refers to the act of tracking outputs or outcomes, whereas evaluation explains why the change occurred. However, this distinction is by no means universally accepted. It is probably more truthful to say that within the field of evaluation, there is no common agreement where monitoring ends and where evaluation starts, especially since "programme evaluation" is often used to describe the full gamut of evaluation activities.



The underlying premise to the framework in Figure 1 [2] is a causal chain: that inputs are used to create and implement a series of activities (processes) that produce measurable outputs and lead to outcomes. Missing from this causal chain of events is some sense of context that will influence the process. For example, a donor agency may give ample funding for the planning and implementation of an HIV/AIDS programme, resulting in an elaborate mass media programme and community-based activities. Yet this may or may not lead to a change in HIV/related behaviour or avert the transmission of HIV, depending on competing motivations and barriers to behaviour change that are deeply embedded in the socio-cultural environment.



It is not particularly useful to the field for one researcher or set of researchers to play "word police" regarding the terminology used in programme evaluation. On the other hand, health professionals and students trying to understand the basic terminology used in this field may find the overlap of concepts and the different classifications for types of evaluation somewhat confusing. Figure 1 [2] represents the attempt of one researcher to bring greater understanding of the range of terms use in connection in the programme evaluation.

References

Bertrand, Jane T. (2005). "Evaluating Health Communication Programmes," The Drum Beat, Issue 302 [1], The Communication Initiative.

Bertrand, Jane T. and Gabriela Escudero. (2002). Compendium of Indicators for Evaluating Reproductive Health Programs. MEASURE Evaluation Project, Carolina Population Center, University of North Carolina at Chapel Hill, MEASURE Evaluation Manual Series, no. 6.

Habicht, J.P., C.G. Victora, and J.P. Vaughn. (1999). Evaluation designs for adequacy, plausibility and probability of public health programme performance and impact. International Journal of Epidemiology 28:10-18.