14. Sampling techniques

Summary

Real-world data collection rarely allows a simple random sample to be carried out. For example, issues such as the operational costs of covering a large territory are common in any social statistics study. This is why different sampling techniques have emerged, along with estimators associated with them, so that the approximation to the true population parameters is as good as possible (which is the goal of any estimator).

That said, there is a very important initial consideration when sampling begins: the sample size. If it is too small, representativeness may not be adequate (with everything that implies, as noted earlier); if it is too large, it is essentially an unnecessary expense of resources. Determining an adequate sample size is therefore the starting point of a good data analysis, which is then tied to the technique used for sampling.

In that regard, 2 resources stand out: the use of strata and the use of clusters.

The former (strata) are meant for studying regions or groupings of interest to the research, obtaining statistics not only at the overall level but within each one. It is a division that can be made on the population (the whole set of units of analysis) before or after sampling (post-stratification), on the condition that the complete set of strata is studied, since each stratum is likely a partial representation of the population. A random sample is drawn from each stratum and then studied together with the rest. An example of strata could be the neighborhoods of a city.

The latter (clusters) are groupings of units of analysis that tend to occur naturally together, with the particularity that each is a representation of the population in itself, and that they also make the study / sampling easier. Clusters are not sampled internally; instead, all the units within them are surveyed. The sampling unit is therefore the cluster, and it is the clusters that are selected, usually systematically. An example of clusters could be the passengers on a flight.

Further reading

  • Métodos de muestreo. Lind, Marchal and Wathen, Estadística aplicada a los negocios y la economía (in Spanish), 15th ed., pp. 266–271.
  • Elección del tamaño adecuado de una muestra. Lind, Marchal and Wathen, Estadística aplicada a los negocios y la economía (in Spanish), 15th ed., pp. 316–321.
  • Estimating Population Mean and Total under SRS. STAT 506: Sampling Theory and Methods, Penn State, lesson 1.
  • Confidence Intervals and Sample Size. STAT 506: Sampling Theory and Methods, Penn State, lesson 2.
  • Stratified Sampling. STAT 506: Sampling Theory and Methods, Penn State, lesson 6.
  • Cluster and Systematic Sampling. STAT 506: Sampling Theory and Methods, Penn State, lessons 7 and 8.
  • Multi-Stage Designs. STAT 506: Sampling Theory and Methods, Penn State, lesson 9.