Using updated version of sample size calculator V2- 6th September 2017, and given information as below :

Scenario:

Vaccine dose of interest is PENTA3

No. of strata =100 (districts =`1000)

Expected coverage = 85%

CI no wider than +/-5%

Design effect = 4 (10 childern per cluster with ICC of 1/3 for RI)

Households to be visited per team = 5

Non - response rate = 15% = response rate 85%

Answers :

Overall sample size and details :

Total households visited: 621,177

Total households per stratum: 6,212

Number of cluster per stratum: 59

Total number of clusters: 10,600

What do you think of the estimated sample size?

It is definitely a very large sample but good representative sample of the population. This sized survey needs a huge budget, time, skilled staff , human resources for planning and management for good quality outputs .

How feasible will it be to conduct the survey?

In the given context, it is less feasible to conduct a survey involving 621,177 households as compared to an effective sample size of 264.

What are the tradeoffs in terms of time , money and quality of survey implementation?

In terms of time and money tradeoffs the survey needs to be reconsidered in terms of levels and sample size. As we have learnt from the presentation on statistical calculation (by Dale) that cumulative effect of districts would be pooled for provincial level and provincial for the national level so provinces sample may be considered rather than 100 districts. Precision is also another factor to be considered. They can vary it from +/-5 to a higher value in other to get less number of sample as it will be an estimation. On the other hand placing in good quality assurance mechanisms, well monitored training and less human resourced for a smaller sample size can remarkably reduce the cost and time needs for the survey.