Concern was raised about two issues concerning the use of surveys: (1) the potential of selective coverage in survey samples, and (2) the difficulty in studying small, high-risk populations using national-level survey data. And it was mentioned that the ability to perform data analysis was a problem in some countries. It was pointed out that individuals involved in survey execution often are not trained in data analysis. It was therefore suggested that those organizations conducting surveys and training individuals to help with data collection in a country also train them to analyze the data.
CONCLUSION
During the wrap-up session, John Ross recapitulated what he thought were the most important ways to improve evaluation techniques that were brought up during the meeting:
-
Linking different data sets together and watching for opportunities to link;
-
Establishing more survey panels to interview respondents over an extended period of time following an intervention;
-
Paying careful attention to proper sampling frames in designing experiments;
-
Watching for computerization opportunities and studying how the field can make better use of computer capabilities;
-
Increasing efforts to monitor the logistical flow of contraceptives from the time they leave the original warehouse to their actual receipt by clients;
-
Developing and using geocoded data bases and networking among researchers to increase attention to the newly emerging software opportunities;
-
Continuing work to develop good clinic-based service statistics;
-
Stressing the differences in flexibility and complexity between what small, manageable programs can take on and what large national programs can manage;
-
Monitoring administrative changes continuously instead of relying only on “snapshot” surveys;
-
Emphasizing local perspectives and needs, to help local people understand the data that they generate and increase their own use of it;
-
Training middle-level research staff more extensively than is now done, including the use of geographical information systems;
-
Using focus groups and exit interviews more fully;
-
Exploiting past research and the available literature fully before investing heavily in a new study.
Ross also addressed some topics that were not discussed. The most important was that researchers in the field of family planning do not need to choose between doing no experiments and doing huge experiments. He reminded the participants that smaller-scale experiments, at low cost and with short time frames, are important options.