Web design by bayteksystems.ca
Copyright © 2019 CIETinternational. All Rights Reserved. | Terms of Use
Community intervention trials
Social audit of public services
Knowledge synthesis
Teaching and training



A central objective of CIET’s approach – usually geared for planners -- is to estimate the change in risk associated with a given potential determinant, usually a programme intervention. A household survey might produce a certain status indicator, such as that 55% of women attend prenatal care. Repeated surveys might even be able to detect a change in the associated health status, say low birth weight. The challenge is to decide whether the change in status is in reality an impact of the programme input. This is rarely a simple exercise; it requires documentation of exposure to programme interventions, and it requires that many other explanations for change be excluded. Even with modern multivariate analysis, survey data may be too shallow to provide much insight into what ought to be done about the status.


Valuable additional insights may be obtained from more qualitative approaches such as direct or participant observation, community meetings, Delphi techniques, focus or nominal group discussions and case studies. Rapid assessment procedures can deepen the knowledge base on the way things work, based on external observer opinions. On their own, these are rarely useful for measuring impact. Since qualitative techniques are based on small samples, repeating the observation or focus group discussion at another time can produce a different result by chance alone.


If the depth of knowledge generated by the qualitative studies is to be generalised, then its representativeness must withstand scrutiny in the manner of epidemiological data. Usually, qualitative techniques provide intense detail on a purposively chosen segment of society. The insights represent what goes on in those chosen segments. And if the segments chosen for qualitative study are not the same as those for quantitative study, linking qualitative appreciation to macro-level data is difficult. Typically the sample survey covers one domain (the district or country); the focus groups, participant observer reports or case studies reflect another domain.


The “sentinel sites” that characterise CIET methods are essentially survey clusters whose size has been increased to offer a representative panel of “mini-universes”.There is no sampling within a site. Each site can be followed over time by returning to it and repeating the enquiry. Like age, sex, or education, “site” can be considered an individual factor or grouped by characteristics (geography, religion, prevailing opinions, or price of basic grains) to provide a link between qualitative and quantitative data, between data from the household and data from the community or local environment.


In the evaluation of the 1998-2003 Health and Population Sector Programme in Bangladesh a detailed institutional review of facilities serving each sentinel site enabled CIET to identify two characteristics of health centres – separate toilets for women and use of screens when examining patients – which were key determinants of client satisfaction. In South Africa’s Free State province  variations in attitudes and performance of health workers in the different facilities providing antiretroviral therapy (ART) proved to be key influences on community engangement with the ART programme. In a baseline CIET survey on social vulnerability in Venezuela in 2003, a composite indicator of environmental conditions in each site was established on the basis of interviews with community leaders. Where these conditions were weakest, the risk of chronic child malnutrition as measured by height for age was more than double that in sites with better environmental conditions. This kind of analysis is not easy with typical random cluster surveys that use much smaller clusters.


Attributes of place, a specific aspect of local service, or the result of a focus group discussion – have a quite different analytical relevance when observations or focus groups are repeated in a comparable manner across a panel of communities chosen to be representative of the geographic area in question. These differences can be related to programmatic input and other factors that might be heterogeneous across different sites. The impact assessment is based on the time sequence and the heterogeneity among sites.