Sequential sampling is a non-probabilistic sampling technique, initially developed as a tool for product quality control. The sample size, *n,* is not fixed in advanced, nor is the timeframe of data collection. The process begins, first, with the sampling of a single observation or a group of observations. These are then tested to see whether or not the null hypothesis can be rejected. If the null is not rejected, then another observation or group of observations is sampled and the test is run again. In this way the test continues until the researcher is confident in his or her results.

This technique can reduce sampling costs by reducing the number of observations needed. If a whole batch of light bulbs is defective, sequential sampling can allow us to learn this much more quickly and inexpensively than simple random sampling. However, it is *not* a random sample and has other issues with making statistical inference.

### Advice

#### Advice for CHOOSING this option (tips and traps)

- Consult an expert, preferably an applied statistician or methodologist, to understand why this option is appropriate and how it should be conducted in order to obtain statistically valid results.
- Sequential modeling is best used to test quality control – food reserves, water purity, and industrial products, for example. It does not require sampling at the same time point. Nor does it require large samples at any particular time. Most of the standard social, economic, political, and health related questions, on the other hand, do require large sample sizes over the same timeframe. If interested in any of the standard social or health related questions – income disparity, household savings, health inequalities, political processes, racial/ethnic or urban/rural differences, then sequential modeling is probably not the correct sampling option.

#### Advice for USING this option (tips and traps)

- Be sure to understand the limitations of the technique. Sequential modeling is not a probabilistic sampling option. It can lead to valid statistical conclusions but the means in which these are obtained is separate from probabilistic sampling techniques.

### Sources

Bakeman, R & Gottman, J.M. (1997) *Observing Interaction: An Introduction to Sequential Analysis* Cambridge: Cambridge University Press.