Sample bias, weights and efficiency of weights in a continuous web voluntary survey

Using micro data from a continuous voluntary web survey, the Wage Indicator, the paper analyses the type of bias that such a sampling method produces and discusses a methodology to weight the data in order to correct such bias and make it possible to run analyses to obtain results and conclusions applicable to the whole population. In order to evaluate the efficiency of the weighting methodology to solve the potential sample bias of web surveys, the results are confronted with those obtained from an alternative standard labour survey dealing with the same issues. Since the Wage Indicator is a survey oriented to labour market issues, we considered that a labour market case study was most appropriate for the evaluation of the results. The method of evaluation followed is to calculate mean salaries, inequality indexes and salary regressions before and after implementing the weights using the Wage Indicator Survey data for Spain. The results are compared with those reached using the Structure of Earnings Survey, a wage survey run by the Spanish Statistical Institute.