chrismclarke
commented
6 years ago From Alex:
Estimation: the aim of the sample survey is to estimate one or more of the population characteristics (i.e. the average population income, the proportion of female-headed households, …). When a sample is selected, the relevant value (called statistic) is computed using the available sample data (following the examples above, the sample household income average or the sample proportion of female-headed households). We call these values estimates of the population characteristics. However, if the sample was different from the original one, the estimates would be different. Both estimates are aiming at the same unknown population value, and (provided that the samples are equally representative and large) equally valid. If we could select infinite samples, we would obtain infinite estimates and we could, therefore, have an idea of how different (or how variable) these estimates can be. However, we usually select just one sample. That is the reason why we need, along with the estimate, a measure of precision of such estimate, i.e. how different could the estimate be should we select a different sample. When providing an estimate, a measure of precision such as the standard error or the margin of error should be provided.
Now just need to decide how/where this fits in best. Will leave open for now.
Prepare an explanation about why estimation is a key part of the objective. Include: selection of key indicator, need for margin of error