Table 3 Main sources of penalty for 82 mortality surveys, Darfur, Sudan, June 2004 - December 2008
Mortality surveys penalized for the criterion (%) | |
|---|---|
Selection of basic sampling unit | |
1) No cross-checking of the source for population size | 50 |
2) Household selection in cluster sampling done by the "spin the pen" method | 77 |
3) Further household selection by proximity with no interval between households | 74 |
4) No definition of household | 58 |
5) No description of revisit strategy | 50 |
6) Number of nonresponding households not reported | 98 |
Mortality data collection | |
1) No evidence of the use of a structured questionnaire including explicit questions and sequences of questions | 83 |
2) No evidence that questionnaire/tally sheet was prepiloted before data collection | 55 |
3) If multipurpose survey, no evidence of mortality questions at the beginning of interview and before anthropometric measurements | 100 |
4) Aggregated questionnaire/tally-sheet only | 87 |
5) No use of a calendar with salient dates for estimating date of deaths | 68 |
6) No evidence of field supervision, including direct observation of interviews by investigators | 85 |
7) No evidence that data collection forms were checked at the end of each day by investigators | 94 |
8) No evidence that the interviewers explained that the survey was not part of a registration effort and would not affect relief allocation for the household | 87 |
Analysis of mortality data | |
1) Standard errors or confidence intervals were not reported in the results | 58 |
2) No evidence that adjustment of standard errors for clustering was done, when relevant | 13 |