SimSam is a program to simulate a survey. If you carry out a survey, you select a sample from a population. Then you use the sample data to estimate characteristics of the population. SimSam is a program to simulate samples from finite populations. The program assumes the objective of the sample selection to be estimation of the population mean of a variable.
SimSam simulates samples from a populations. By repeating the selection of the sample a large number of times, and computing an estimate for the population mean for each sample, the distribution of the estimate can be portrayed on the screen. By taking a look at this distribution, it becomes clear how well an estimation procedure is able to estimate the population mean.
Two types of sample designs can be simulated:
Four different estimators can be computed:
- Simple random samples without replacement, and with equal probabilities.
- Samples with replacement, and with probabilities proportional to the values of some auxiliary variable.
To study the effects of nonresponse on an estimator, it is possible to generate nonresponse in the samples.
- The simple sample mean.
- The ratio estimator.
- The regression estimator
- The post-stratification estimator
The result of a simulation is displayed on the screen in the form of a histogram of the distribution of the estimates. This histogram can be saved in a bitmap file for use in other applications.
You can also analyse the distribtion of the estimator by drawing confidence intervals for each sample:
You can ask the program to produce a numerical summary with theoretical and simulated means and standard errors: