Burden reduction in Prodcom Experiences of Statistics Denmark Ingeborg Vind, inv@dst.dk
Outline Prodcom in Statistics Denmark Response burden – different types Measures to reduce response burden Prefill in survey form Online data validation Use of admin. data and other sources Idep.web Use of CN instead of Prodcom codes
Prodcom in Statistics Denmark Prodcom - quarterly survey in Denmark for quarterly National Accounts All enterprises in manufacturing (B+C) with 10+ employees are surveyed (~3,000) Survey is mandatory Electronic survey form Enterprises report all sales according to CN codes and additional categories for services and other turnover
Two kinds of response burden Measured response burden Time spent to fill out survey (x salary of person doing the work, x number of enterprises, x number of surveys pr year) Quantitative measure Comparable across surveys and countries Experienced response burden Irritation, sense of uselessness/redundancy/irrelevance, etc. Qualitative assessment Often just as significant to the respondents as the measured burden We work on reductions of both types
Pre-fill of survey form Survey form is pre-filled with: codes, values and quantities from previous report For new enterprises in survey, prefill is with codes typical for their NACE code Easy for respondents to remember how to fill Stability: same code used for same products Perhaps sometimes ‘too easy’, i.e. a tendency to always use the same codes, even if production changes
Online validation – at ‘Virk.dk’ Prompts when the form is filled Missing fields (e.g. quantity) Unlikely values (compared to last report) Reduces burden by reducing re-concact by phone to verify data Respondents expect electronic forms to be ‘smart’ But not always easy to communicate through simple one-way messages
Example, online validation What does the respondent do? Fill in actual quantity? Nothing? Fill in fictive number (e.g. ‘1’)?
Use of admin. data and other survey data Tools for easy comparison of Prodcom data with VAT data and IPI survey data during data validation Less re-contacting through cross-check; if same development is found in VAT or IPI data, we will not re-contact to verify Future initiatives easier comparison with Intrastat export data analysis of possibilities to reduce sample for quarterly survey, by using VAT data for smaller enterprises
Idep.web – easy reporting of many codes Idep.web: already in use for Intrastat 1/3 of Prodcom respondents already use Idep.web for their Intrastat reporting (more among larger companies with many products) Online reporting tool Upload reporting file (spreadsheet) generated by accounting system in enterprise Build-in validation of codes Faster than the online form (Virk.dk) when enterprises have many codes Adapted for Prodcom use possible because we use CN-codes To be launched for data collection for 3rd quarter 2014
Use of Combined Nomenclature (CN) Danish ‘Prodcom’ survey is older than Prodcom We have ‘always’ used CN-based codes, for data collection as well as publication Prior to 2007, national sub-divisions of some CN-codes Discontinued as a burden reduction initiative ~80 pct. of Prodcom respondents are also exporters reporting to Intrastat/Extrastat Response burden is reduced as enterprises only need to know one set of codes Disadvantages services are only covered in very broad categories z-codes must be estimated