sepe: Sample Environment Protection Expenditure Survey.
In modi: Multivariate Outlier Detection and Imputation for Incomplete Survey Data

sepe	R Documentation

Sample Environment Protection Expenditure Survey.

Description

The sepe data set is a sample of the pilot survey in 1993 of the Swiss Federal Statistical Office on environment protection expenditures of Swiss private economy in the previous accounting year. The units are enterprises, the monetary variables are in thousand Swiss Francs (CHF). From the original sample a random subsample was chosen of which certain enterprises were excluded for confidentiality reasons. In addition, noise has been added to certain variables, and certain categories have been collapsed. The data set has missing values. The data set has first been prepared for the EU FP5 project EUREDIT and later been data protected for educational purposes.

Usage

sepe

Format

A data frame with 675 rows and 23 variables:

idnr: identifier (anonymous)
exp: categorical variable where 1 = 'non-zero total expenditure' and 2 = 'zero total expenditure, and 3 = 'no answer'
totinvwp: total investment for water protection
totinvwm: total investment for waste management
totinvap: total investment for air protection
totinvnp: total investment for noise protection
totinvot: total investment for other environmental protection
totinvto: overall total investment in all environmental protection areas
totexpwp: total current expenditure in environmental protection area water protection
totexpwm: total current expenditure in environmental protection area waste management
totexpap: total current expenditure in environmental protection area air protection
totexpnp: total current expenditure in environmental protection area noise protection
totexpot: total current expenditure in other environmental protection
totexpto: overall total current expenditure in all environmental protection
subtot: total subsidies for environmental protection received
rectot: total receipts from environmental protection
employ: number of employees
sizeclass: size class (according to number of employees)
stratum: stratum number of sample design
activity: code of economic activity (aggregated)
popsize: number of enterprises in the population-stratum
popempl: number of employees in population activity group
weight: sampling weight (for extrapolation to the population)

Details

The sample design is stratified random sampling with different sampling rates. Use package survey or sampling to obtain correct point and variance estimates. In addition a ratio estimator may be built using the variable popemple which gives the total employment per activity.

There are two balance rules: the subtotals of the investment variables should sum to totinvto and the expenditure subtotals should sum to totexpto.

The missing values stem from the survey itself. In the actual survey the missing values were declared as 'guessed' rather than copied from records.

The sampling weight weight is adjusted for non-response in the stratum, i.e. weight=popsize/sampsize.

References

Swiss Federal Statistical Office (1996), Umweltausgaben und -investitionen in der Schweiz 1992/1993, Ergebnisse einer Pilotstudie.

Charlton, J. (ed.), Towards Effective Statistical Editing and Imputation Strategies - Findings of the Euredit project, unpublished manuscript available from Eurostat and https://www.cs.york.ac.uk/euredit/euredit-main.html.