Download presentation
Presentation is loading. Please wait.
1
Sampling for social surveys from Labour Force Survey data – advantages and challenges
Hanna Sutela Senior reseacher, Dr. LFS Methodology Workshop, , Reykjavik
2
Finnish Quality of Work Life Survey 2013 (FQWLS)
Survey on Self-employed without Employees 2013 (SEWE) Summing up Hanna Sutela
3
Finnish Quality of Work Life Surveys
4
Finnish Quality of Work Life Surveys
1977, 1984, 1990, 1997, 2003, 2008, 2013 3 000 – respondents per data Employees aged 15 to 64/67 years, regular working hours at least 10 h/wk Response rate % Face-to-face (c. 60 min) Core questionnaire remains the same => building up time series New themes added, old stuff removed/rotated Next FQWLS in special theme digitalization of the working life Hanna Sutela
5
Finnish Quality of Work Life Surveys
Provides information on the state of work life for policymaking Provides data and analysis for research, possibility to link survey data to register data Provides reliable information on working conditions for public debate Funding of the fieldwork in co-operation with different ministries and other stakeholders (Finnish Innovation Fund Sitra, Ministry of Employment and Economy, and Ministry of Social Affairs and Health, the Finnish Work Environment Fund, the State Treasury, the Finnish Institute of Occupational Health, the Finnish Centre for Pensions, the Local Government Pensions Institution, the Centre for Occupational Safety, the Social Insurance Institution) Hanna Sutela
6
FQWLS 2013 Sampling Sampling from the II, III, IV and V rotations of the September and October LFS from among LFS interviewees aged 15 to 64 => employees wit regular working hours 10 or more => invitation to the survey => agreement of face-to-face interview FQWLS fieldwork continues until January next year Hanna Sutela
7
Problem in sampling since 2008
The fieldwork of LFS lasts only two weeks => not possible to continue to reach the non-response of LFS <=> potential FQWLS respondents lost in the LFS non-response Hanna Sutela
8
Solution in sampling 2013 Interviewers were allowed to continue to reach for the LFS non-response even after the two-weeks LFS fieldwork In case a target person was reached: =>a few LFS questions to define occupational status and working hours => in case the target persons meets the criteria for sample, invitation to the FQWLS => in case s/he agrees, a few more LFS questions = ”mini LFS” (occupation, workplace, industry, part-time employment…) Hanna Sutela
9
Lengthened contact time in 2013
LFS non-response 1,800 persons => c. 400 of them succesfully contacted after the LFS field work => c. 220 met the criteria of the sample => 40 refused => 182 participated in the FQWLS Including 54 cases who had never participated in the LFS! Hanna Sutela
10
Estimations of the size of the sample and non-response FQWLS 2013 1/2
C. 1,200 persons from the initial sample (LFS target persons aged 15 to 64 yrs) could not be contacted in LFS nor for FQWLS The age, gender, education and region of these persons known but not whether they should be included in the sample of the FQWLS (employees, regular weekly working hours at least 10) Hanna Sutela
11
Estimations of the size of the sample and non-response FQWLS 2013 2/2
The proportions by age and gender of the persons meeting the criteria on inclusion in the sample of the FQWLS calculated from the target persons of the LFS from whom interviews were obtained => possible to estimate what proportion of the persons from whom even LFS interviews were not obtained would have also belonged to the sample of the FQWLS Estimates were calculated by 5-year age groups by gender and education Sample (and non-response) of FQWLS most probably somewhat overestimated, since the non-employed are overpresented in the LFS non-response Hanna Sutela
12
Lengthened contact time – was it worth it?
Lengthened contact time for FQWLS brought 181 ”bonus” respondents in the data Response rate of FQWLS 2.5 p.p. higher than otherwise Estimated response rate increased (68.7 %) from the 2008 survey (67.6 %) Inreased costs Interviewers opinions were mixed - but more on negative side Complications in the data compilation Hanna Sutela
13
Survey on Self-employed without employees 2013 (SEWE)
14
Survey on Self-employed without employees 2013
On-demand survey for urgent policy needs, preparations ad hoc Sample of 2,000 persons drawn from the Labour Force Survey from among those self-identificating as sole entrepreneurs, own-account workers, freelancers or grant recipients and having no paid employees (excl. agricultural sector) Mixed mode (telephone 30 min / web survey) Pre-research: 11 qualitative interviews Conducted by Statistics Finland, co-funded by the Ministry of Employment and Economy, and Ministry of Social Affairs and Health Hanna Sutela
15
Sample design SEWE Initial plan was to sample btw January and August 2013 from all rotations One person = once in the sample => not enough persons in the sample by end of the summer, so the sampling continued in September and October One rotation = about target persons 23 rotations Hanna Sutela
16
Inclusion in the sample defined during the LFS interview => inquiry about willingness to participate (later on) in the Survey on Self-employed without employees ”Finally I tell you that Statistics Finland will conduct a survey for sole entrepreneurs, own-account workers, freelancers and grant recipients in the next autumn. Since you belong to this group, we would like you to participate. Can we contact you once the survey will start?” In the survey, we want to gather information on the working conditions, work situation and issues concerning social security of self-employed without employees. You can participate in a telephone inteview or in web-survey. Could we have your -address in case you want to answer in the internet? (Finally, addresses could not be used) Hanna Sutela
17
Mixed-mode design SEWE 1/4
Sub-sample SEWE0= Those sampled btw January-July were sent in August a letter with an invitation to the web survey, including an user name and password (no matter whether they had initially agreed or not to participate) The recipients were informed that in case they would not have an opportunity to participate in the CAWI, an interviewer would call them later WEB-survey was open 19.8 – The sending of the letters was centralised to the Data Collection Unit (not interviewers themselves) Hanna Sutela
18
Mixed-mode design SEWE 2/4
Sub-sample SEWE1= Those sampled in August + those from January-July sample, who had not answered in CAWI => CATI Sub-sample SEWE2= The ones sampled in September => directly CATI Sub-sample SEWE3= The ones sampled in October => directly CATI CATI fieldwork initially , but continued until Hanna Sutela
19
Mixed-mode design SEWE 3/4
The fieldwork of FQWLS 2013 going on at the same time ( ), but with different sample criteria (employees, regular weekly working hours at least 10 vs. self-employed without employees) => In theory, no risk of overlapping Hanna Sutela
20
Mixed-mode design SEWE 4/4
In practice: the target persons of FQWLS and SEWE were partially overlapping! A same person may have been sampled to SEWE in January as a freelancer, but to FQWLS in September as an employee Confusion in the field: two telephone calls for the same person from different interviewers re different surveys In case all interviewers would have had their ”own” targets and sent the letters themselves, overlapping would not have been a problem ”Combo-employment” => should have been taken into account not only in the questionnaire design, but in the sample design as well!! Hanna Sutela
21
SEWE 2013 data Sample 2,010 (excl. non-response of LFS)
Response rate 79 % (excl. non-response of LFS) 1,583 interviews (representing well the structure in LFS) Co-operation during the fieldwork with different entrepreneur organisations, unions and NGOs 22 % responded by CAWI, 78 % by CATI Hanna Sutela
22
Was it worth to send an invitation letter also for those who initially refused to participate?
Initially, 78% of sampled persons (Jan-July) agreed to participate 26% of which participated in CAWI, 60% CATI, 14% non-response 22 % refused to take part, when initially (Jan-July) asked 9 % of which answered in CAWI, 43 % CATI, 48 % non-response => C. 200 persons initially refused, but participated after all – c. 200 initially agreed, but did not participate Hanna Sutela
23
To sum up: advantages Ready-made pre-sample
Unique opportunity to draw a up-to-date sample of a specific sub-population ”on-line” via LFS questions Opportunity to integrate the information collected in the LFS to the additional survey without lengthening the survey questionnaire “Social pressure” of target persons to participate in the additional survey: they already once gave their agreement and the interviewer is now familiar Hanna Sutela
24
To sum up: challenges 1/2 The exact size of the sample impossible to forecast The effect of LFS non-response: The exact size of the final sample impossible to be calculated even afterwards (LFS non-response) – only estimations LFS non-response accumulates with the non-response of the additional survey Unnecessary non-response fot the additional survey in case the time for the fieldwork of LFS is very short The exact size of the non-response not known – only estimations Hanna Sutela
25
To sum up: challenges 2/2 One has to be cautious as how far to go in trying to persuade LFS target persons to participate in the additional survey => impact on LFS non-response later on? Topic of the additional survey preferably in accordance with the theme of LFS Hanna Sutela
26
Sutela, Hanna & Lehto, Anna-Maija: Työolojen muutokset , SVT, Tilastokeskus Pärnänen, Anna & Sutela, Hanna (2018) Self-employed without employees in Finland 2013, Statistics Finland Hanna Sutela
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.