Spanish households: Car insurance with deductibles

After analyzing the five imputed datasets there are no remarkable differences between them and it is decided to use the first imputed dataset (EFF-derived DATOSIMP1 and databol1 files). Together there are 6,313 surveys representing 18,821,645 households according to the weights of the facine3 variable in the databol1 file. First, we select how many families have any vehicle, and we obtain that of the 6,313 surveys, a total of 5,074 surveys correspond to households that have any vehicle (in the population as a whole, they represent 14,042,117 households). Secondly, those households with negative or null wealth are discarded, since in order to apply the formulations it is necessary for this variable to have a positive value. Overall, 413 surveys representing 1,398,145 households are discarded for the entire survey. Finally, we calculate how many households have any vehicle and have positive wealth, obtaining a total of 4,774 surveys, representing 13,203,356 households. Next we generate a vector containing the total number of vehicles in the total population, for this we multiply the number of vehicles that each household in the survey has by the number of households that each one represents in the total population (weight). We obtain a total of 20,571,785 vehicles, of which we must randomly choose 28.41%, i.e. 5,844,444 vehicles, which are those that will be franchised. In total we selected 5,851,932 vehicles, corresponding to vehicles from 984 different surveys (households). As in some households we have selected more than one vehicle we check how many surveys we have ended up selecting, which are 873 surveys.