poststratifikáció

Poststratification is a statistical adjustment technique used in survey research to improve the representativeness of a sample. It is applied after data collection has been completed, hence the prefix "post". The primary goal of poststratification is to correct for potential biases that may exist in the sample due to non-response or under- or over-sampling of certain demographic groups.

The process involves dividing the sample into subgroups (strata) based on known population characteristics, such as age, gender, race, or education level. These strata are defined using auxiliary information from a reliable external source, typically census data or other population benchmarks. The sample proportions within each stratum are then compared to the known population proportions.

If the sample proportions do not match the population proportions, weights are applied to the sample data. These weights are calculated to adjust the sample so that its distribution across the defined strata more closely reflects the distribution in the target population. For example, if a particular age group is underrepresented in the sample, its observations will be given a higher weight to compensate. Conversely, overrepresented groups will receive lower weights.

The outcome of poststratification is a set of adjusted survey estimates that are considered more accurate and generalizable to the population of interest than unadjusted estimates. This technique is particularly useful when the sampling frame is imperfect or when response rates vary significantly across different population segments, leading to a non-representative sample. It helps to mitigate the impact of sampling errors and non-response bias.

---