Chair: Agustín Cañada
Room: S2 Wawel
Time: 11:30 - 13:00
Date: 28 June
|Title: <<< The "EUROPA 2020" LFS indicators on regional level in Poland. How to improve the indicators precision without the increase of the LFS sample? >>>
In case when there is a need to evaluate some phenomenon by selected characteristics or by territorial division, the often encountered problem is finding an appropriate data source which delivers reliable results. Two of the five “Europe 2020” headline targets are monitored with EU Labour Force Survey (LFS) indicators. Moreover, several other indicators covering different EU policy domains are based on the EU-LFS results. Every Member State has to disseminate comparable and reliable data at regional level for regional policies monitoring needs. How to do this when the data are based on a sample survey, like the LFS, and response rate has been decreasing over time? The objective of the research work carried out in Polish Central Statistical Office (CSO) was division of main LFS indicators (not only “Europe 2020”) into three groups:
1) ready for publication at regional level;
2) maybe possible - the precision of which may be expected to be improved to the acceptable level after the introduction of the significant methodological changes into the survey;
3) not possible at all for dissemination due to a low precision.
For indicators from second groupseveral simulation tests were done.
The two most important effects were achieved within the framework of the research work, which significantly have developed possibilities of the public statistics as regards the publication of the LFS regional data:
1) Specification of the base of indicators (from the list being the object of the research work) together with the codes of precision/quality embracing the historical data;
2) Elaboration of the variants of changes in the survey methodology, allowing to extend significantly possibilities for meeting the information requirements included in the scope of the research work.
|Title: <<< Quality Improvements using indicators, business process improvement and change management >>>
Business Process Improvement in LFS. The CSO has moved recently from the Quarterly National Household Survey (QNHS) to the Labour Force Survey (LFS). Significant changes have been made during this transition including the use of CATi for later waves, a new IT System to monitor and measure metadata and paradata and a separation of the LFS from the General Household Survey (GHS). The GHS is now the instrument to collect the once off surveys like ICT, Crime and Victimisation and other European and National Surveys that are done on an ad-hoc basis. This was a large project for CSO and has had many challenges over the last number of years. The LFS will go live on Q1 2018 and because of the challenges it has become part of the work of the Quality Assurance to review the process from questionnaire design to final publication. This paper will detail how we reviewed the processes of the many teams involved and how the quality division in conjunction with the relevant teams changed some of the older QNHS practices to fit the new LFS data lifecycle. The paper will detail how we assimilated the data, how the relevant teams involved produced some of the solutions themselves and how when we put all the existing processes together the teams had immediate suggestions for improvement. Quality indicators and risks were identified at a number of stages and this were built into the process control. Change management was key to this process improvement and the Quality team managed the change very well. The improvements were detailed and celebrated and this certainly gave momentum to continue the improvement. This paper will detail the various stages of change and subsequent improvement.
|Title: <<< Analysis on nonresponse bias for the Swedish Labour Force Surveys (LFS) >>>
All statistics have some degree of uncertainty that have an effect on the accuracy. The focus in this paper will be on object nonresponse and its effect on the nonresponse bias in the Swedish Labour Force Surveys (LFS). An analysis has been conducted by approximating LFS variables with variables from different registers. The variables being analyzed are; employed, unemployed, not in the labour force, employees, students, group after income and not in employment, education or training (NEET). Two different estimates are produced for each register variable by using the General Regression Estimator currently in use in the Swedish LFS, one by using the respondents and the other by using the sample. These estimates are used in order to estimate bias, relative bias and confidence intervals for the biases. All estimates are produced for the total population aged 16-74 and for relevant subgroups. Results for age group 16-74 in December 2015, age group 16-24 in year 2014 for NEET, show that the relative bias is significantly different from zero for almost all of the analyzed register variables, the exception is unemployed. Looking at the relative bias for employed, unemployed and not in the labour force over time, year 2011-2015, one can see that it has been relatively stable over the time period. Therefore, there is no clear indication of an increasing nonresponse bias as the nonresponse increases. In order to analyze the nonresponse in estimates of change, estimates of change comparing corresponding months from one year to the next have been computed for employed and unemployed for year 2013-2015. For unemployed, the estimates based on the respondents are similar to the estimates based on the sample. The same pattern is in general also seen for employed with exception for subgroups after education.
|Title: <<< Coherence between surveys and based-register data in labour market statistics >>>
The study of the labour market has always had a special interest in society, both economically and socially. In Spain, there are different sources of information that allow us to approach and measure this reality. Statistics based on surveys are available, such as the Labour Force Survey (LFS) - carried out by the National Statistics Institute of Spain (INE) -; and others using administrative data, such as those based on the social security affiliation register and the unemployed persons registered in public employment services, provided by the Ministry of Employment and Social Security. Therefore it is interesting to study the coherence among these sources of information as an aspect of improving the quality of statistics. Under the framework of the High Council on Statistics the working group of Short-Term Labour Market Statistics draws up periodically a report with the aim of analyzing and comparing the data provided by the LFS and those provided by the Ministry of Employment, using for that purpose, a harmonized methodology. In addition, a series of short-term analyses are also carried out to study the coherence between them from the point of view of the final published data. This project allows, on the one hand, to compare, study and review methodological differences, and on the other to analyze the information that users receive from different sources of information.
|Title: <<< On the exciting path to add a mode – redesigning the Austrian Microcensus Housing Survey into a multi-mode questionnaire >>>
The Microcensus Housing Survey is currently being redesigned to meet the growing demand of a multi-mode capable questionnaire. Besides computer assisted personal interviewing and computer assisted telephone interviewing, respondents will have the option to fill out the questions online. Although three different interviewing methods will be offered, the survey will be based on only one common questionnaire which needs to be highly user-friendly for all modes. Planning and redesigning the new survey instrument, specific challenges have to be taken into account: (1) the Housing Survey is jointly conducted with the Labour Force Survey within the framework of the Microcensus, (2) the output is linked to other statistical products, e.g. the Consumer Price Index, (3) the data is collected continuously throughout the year, (4) the selected households are interviewed quarterly and (5) the households have to answer questions on housing and housing costs, whereby precise information is indispensable. To implement a new mode means to develop new questions and to change the data structure. At different stages of the project, specific quality assurance measures are applied to ensure an optimal transformation from the old to the new survey instrument and to minimize breaks in time series. First of all, respondent debriefings as well as interviewer debriefings were carried out. Different data sets on housing were analysed. Considering all the analysis, a draft questionnaire was elaborated and tested by using qualitative methods. This questionnaire is now being redesigned. In mid-2018 a quantitative test of the final multi-mode questionnaire is planned.