[SMARTKeymaster]

In a context where households are not well organized it is recommended to use simple random sampling. If the cluster is large it can be broken down first by implementing segmentation (PPS). SMART does not recommend to use the modified EPI method.

[SMARTKeymaster]

For cluster surveys t=2.045 is used in the sample size calculation because the number of clusters is typically 30 clusters or less. In Simple Random Surveys (non-cluster) z=1.96 is used because the degrees of freedom is high (over 100).

[SMARTKeymaster]

Using the numbers on a bank note to randomly select the first household should only be used as a last resort. It is recommended to use a Random Number Table (RNT) to maintain randomness as well as to ensure that all SMART surveys are using the same procedures. Kindly send an email to slogue@actioncontrelafaim.ca and I will send the RNT procedures.

[SMARTKeymaster]

Conducting a SMART survey will ensure a statistically representative sample. Stunting can be used as the main indicator when determining the sample size calculation. Since the prevalence of stunting is usually quite high (will depended on your sampling frame context), the level of precision can be reduced. This will result in a lower sample size for your survey.

[SMARTKeymaster]

Just to reiterate what has already been stated; in your example if the data is disaggregated into sub-districts the confidence intervals will be wide because the sample size will be quite low. As a result, the findings will likely not be able to be used for programmatic decision making processes. During the planning stage of a survey(s) if the sampling universe contains two distinct livelihoods such as pastoralist and agriculturalist (and it is believed there will be a significance difference in prevalence) it would be best to conduct two separate SMART surveys.

[SMARTKeymaster]

Thank you for your response Blessing. I would also like to reiterate that in your example you will select ‘Random’ sampling in the ‘Planning’ Tab and the plausibility check will automatically be adjusted accordingly. Kindly note that the Poisson distribution will appear in the Plausibility Check summary table (page 1) but will not be calculated as it is used for cluster sampling. Also, the Design Effect for your survey will be 1.

[SMARTKeymaster]

There is no rule of thumb for estimating non-response rate. As you mentioned, non-response is based on several factors such as context and previous survey reports.
A 5% estimate for non-response in a camp setting will be sufficient if:
a) Sensitization occurs ahead of time and the residents are aware when the data collection will take place.
b) The SMART survey does not contain any additional variable(s) that some individuals may consider invasive (such as drawing blood for anaemia) and may refuse to participate in the survey.

[SMARTKeymaster]

Unfortunately, ENA for SMART software is currently only available on the Windows Operating System.

[SMARTKeymaster]

You are correct that the Poisson distribution applies to cluster sampling and examines level of heterogeneity or the presence of pockets of malnutrition. The Poisson distribution does NOT apply to an exhaustive survey or simple/systematic random surveys because there are not any clusters to compare levels of heterogeneity. Please disregard the Poisson distribution analysis in the plausibility report if a cluster survey design was not used.

[SMARTKeymaster]

Thank you Blessing! The correct link is http://www.thl.fi/publications/monica/bp/bpqa.htm . Alternatively, the study can be found by performing a google search: WHO MONICA study of blood pressure

[Author]

Posts