1..a How many strata does the Harmonia survey will have?

10 strata which relates to the 10 provinces

1.b List and briefly describe at least two advantages and disadvantages of having these many strata compared to having more, for example 100 (one for each district)? (MANDATORY)

The first advantage is that it reduces the number of people needed to interviewed, thereby reducing costs and time, resulting in the project being more doable within the given timeframe. Results are more quickly obtained too.

The second advantage is that it simplifies the report so that policy makers and other decision makers can quickly glance at the results and immediate see how it compares to their expectations and to the existing routine statistics. By providing a cross-sectional comparison of the main regions of the country, it enables you to look at the regional differences in uptake across the programmes so that you can see which region has the highest percentage uptake and which has the lowest. You can then focus in (or look deeper) at the provinces with the lowest uptake for the explanations of lower coverage/uptake. This would target resources.

A disadvantage is that by reporting on a regional/province level it masks any variations within a province, for example differences between rural and urban areas, ethnic communities, socio-economic groups or geographical variations within a province.

A second disadvantage is that by sampling on a province level, you risk under-representation of some societal groups or individuals. This will result in not having a representative picture of true uptakes in these groups or communities.

2. How much will the sample size decrease or increase (overall) if you were to design a survey with and with +/-8% precision per strata or +/- 3% precision per strata (as opposed to +/-5% precision per strata used now). Just provide the ‘Total Completed Interviews Needed’. Think about the importance that the desired precision has on overall sample size. (OPTIONAL)

+/-5% precision per strata: 10498

+/-8% precision per strata: 4645 – 5845 fewer interviews needed

+/-3% precision per strata: 26282 – 15784 more interviews needed

3.a What kind of data do you need to collect to complete a table like the one below? (MANDATORY)

Breakdown of demographics including age, sex, ethnicity, religion and numbers of members per household as well as by province

3.b How does this table relate to potential selection bias and what cautions should you have when interpreting the vaccination indicators if response rate is not 100% and/or several households could not be interviewed? (MANDATORY)

The table should help you to identify the characteristics of those households that were more likely to response and those who refused. This can help you to identify who’s been under-represented and you may be able to weight your vaccination variables so that they are more representative of the population that are under-reported.

4.a What variables do you need to collect or define in the analysis to estimate a [weighted] vaccination coverage with the first dose of diphtheria-tetanus-pertussis containing vaccine (DTP1) and with the third dose of diphtheria-tetanus-pertussis containing vaccine (DTP3) among children aged 12-23 months in each stratum by: a) ‘documented evidence of vaccination (home-based record – HBR OR facility-based record – FBR)’, b) ‘by recall’, and c) ‘by documented evidence of vaccination + recall’. [Do NOT include the data you need to calculate weights]. (MANDATORY)

I would include the following to see if there is variation in uptake by gender, urban/rural area, socio-economic status, ethnicity, religion, socio-economic status, education level of mother, household income level.

4.b How does this table relate to potential information bias and what cautions should you have when interpreting the vaccination indicators in surveys with “low” percentage of documented evidence? (MANDATORY)

There may be recall bias – parents thinking the child was vaccinate when they weren’t – and there may be social desirability bias, whereby the responders give the responses they think the interviewer wants to hear.

4.c What variables do you need to collect or define in the analysis to estimate a weighted percentage of zero dose children aged 12-24 months in each stratum by documented evidence of vaccination (HBR or FBR), by recall, and by documented evidence of vaccination + recall. (OPTIONAL)

Using census data I would weight by sex, racial identity/ethnicity.

4.d What would you do with records without information about DTP1 (HBR and FBR did not have a date or tick mark for DTP1 and the mother/caregiver could NOT remember whether the child had ever received a vaccine against diphtheria-tetanus-pertussis and the question was left blank? Explain why. (OPTIONAL)

This is missing data. I think I would not quantify and ignore and exclude from the analysis as we can’t assume either way that they were vaccinated or not vaccinated and such assumption would affect the uptake rates. However, it there were many of them, it may reduce the sample size for that variable and affect statistical power. It may be possible to utilize missing data analysis methodologies (e.g. sensitivity or intention to treat styles).

5.a What information do you need to collect the probability of selection and how does this relate to calculating design weights? (OPTIONAL)

Total number of the population and census information about their characteristics for each province. This will influence how you set up your sampling frame and to ensure that the sample is representative of the population and its parameters. By having this information, you can then compare your sample to the census information and weight accordingly – e.g. if you have only 35% females responding and you know your population has 52% females, you can weight the 35% so that it is representative of the 52% female population.

5.b What information about the population in each stratum do you need to collect/obtain (from the National Statistics Office, for example) if you want to aggregate the data and calculate national-level indicators? (OPTIONAL)

Total numbers of those interviewed (and their demographics) and total numbers vaccinated per vaccine per stratum. Aggregating is adding up.