A.V. Garcia Luengo

University of Almería (SPAIN)
The objective of this communication is to prepare software applications that are of a great help in efficiently conveying information, with a corresponding saving of time. Doing this, we encourage the introduction into the classroom of materials, resources and new technologies to help with education and the learning of statistical methods. The use of statistical packages is essential for the resolution of problems with a great number of calculations. This is especially true for higher education and for students of the University of Almeria (Spain). The following example was set out.

In most socio-economic studies, several variables are considered simultaneously and in sample surveys. It is traditional to make use of information on auxiliary variables to increase the precision of estimators. It is well known, especially in mail surveys, that information is usually not obtained from all the sample units, even after call backs. Because of this, the objective of this communication is to address this situation by preparing an estimate of the population mean on the second occasion in mail surveys, when several auxiliary variables are correlated positively and negatively with the analyzed variable.

The estimate of the population mean in mail surveys has been adjusted for unit non-response on both occasions. Using the theory of successive sampling, we develop the optimum estimator by combining two mean estimates: a double sampling multivariate ratio-product estimator for the matched part of the sampling, and a simple estimator for the mean of the unmatched part. This assumes that total response from the sample units are available on both occasions. An empirical illustration demonstrates the performance of the proposed combined estimator.

All the results have been carried out using the "SPSS" statistical package, the Mathematica program and LaTeX language. The outputs are shown and interpreted to pupils with Power-Point.