Nonresponse in the Norwegian Labour Force Survey (LFS): using administrative information to describe trends

Transcription

1 Notater Documents 54/2012 Ib Thomsen and Ole Vllund Nonresponse n the Norwegan Labour Force Survey (LFS): usng admnstratve nformaton to descrbe trends

2

3 Documents 54/2012 Ib Thomsen and Ole Vllund Nonresponse n the Norwegan Labour Force Survey (LFS): usng admnstratve nformaton to descrbe trends Statstsk sentralbyrå Statstcs Norway Oslo Kongsvnger

4 Documents In ths seres, documentaton, method descrptons, model descrptons and standards are publshed. Statstcs Norway Symbols n tables Symbol When usng materal from ths publcaton, Statstcs Category not applcable. Norway shall be quoted as the source. Data not avalable.. Publshed August 2012 Data not yet avalable Not for publcaton : Nl - ISBN (prnted) Less than 0.5 of unt employed 0 ISBN (electronc) Less than 0.05 of unt employed 0.0 ISSN Provsonal or prelmnary fgure * Subject: Break n the homogenety of a vertcal seres Break n the homogenety of a horzontal seres Prnt: Statstcs Norway Decmal punctuaton mark.

5 Documents 54/2012 Nonresponse n the Norwegan Labour Force Survey (LFS) Preface Ths document s the result of a study on nonresponse n the Norwegan Labour Force Survey (LFS). The study follows a broader project on how to deal wth nonresponse, whch nvestgates several dfferent surveys. The LFS s dstnct n ts sze and the relatvely long tme seres, and therefore deal for studyng nonresponse trends n some detal. The motvaton was how to better descrbe the effects of nonresponse, as the focus on ncreasng nonresponse rates tells us lttle about the effect on offcal statstcs. Statstcs Norway, 14 August 2012 Hans Henrk Scheel Statstcs Norway 3

6 Nonresponse n the Norwegan Labour Force Survey (LFS) Documents 54/2012 Abstract Studyng data from the Norwegan Labour Force Survey sample lnked to admnstratve regster data, we focus on trends n nonresponse and ts effects. Analyzng response rates and nonresponse bas n several subgroups, we fnd relatvely stable response patterns. Although the overall response rate s decreasng, t seems to exst some structural dfferences. Factors such as age, gender, ctzenshp and employment seem to determne a response pattern that changes lttle over a relatvely long perod, when we observe clearly ncreasng nonresponse over the same perod. Usng regster-based employment status avalable for the total sample, we have estmated the nonresponse bas for employment. Defnng the response structure as the rato of response rates among employed/notemployed, we compare ths ndcator wth nonresponse rate and nonresponse bas. 4 Statstcs Norway

7 Documents 54/2012 Nonresponse n the Norwegan Labour Force Survey (LFS) Contents Preface... 3 Abstract... 4 Contents Introducton and summary Data sources Survey sample Admnstratve regsters Response rate Nonresponse Bas A smple relatonshp between response rate and nonresponse bas, for bnomnal varables References Appendx A Lst of fgures Lst of tables Statstcs Norway 5

8 Nonresponse n the Norwegan Labour Force Survey (LFS) Documents 54/ Introducton and summary The purpose of ths document s to present a straghtforward descrpton of the nonresponse development over the perod n the Norwegan Labour Force Survey (LFS). The sample s lnked to admnstratve regsters n order to compare response rates n varous subgroups. In chapter 3, response rates are presented for a number of subgroups, such as age groups, gender, ctzenshp and employment. In chapter 4, we concentrate on one varable, namely employment status n the regster. However the target varable s employment status accordng to the ILO defnton, based on the survey response. We know that the correlaton between the two varables s very hgh (Thomsen and Zhang 2001), and therefore the fndngs concernng regster status are relevant for the target varable as well. In ths study we concentrate on trends n yearly average nonresponse rates. However, as llustrated n addtonal fgures (Annex), the dfference n nonresponse rate between quarters can be relatvely large. We plan to study ths phenomenon later. We dscuss two qualty ndcators and compare ther development over tme: nonresponse rate and nonresponse bas. The man fndngs are: There are clear dfferences n response rates between subgroups, and more nterestng, the pattern remans relatvely constant over tme. Ths s an mportant fndng because, t ndcates that the representatveness of the sample of respondents changes lttle over tme compared to a steady ncrease n nonresponse rates. In spte of the clear ncrease n nonresponse rate, we fnd that the bas (of regstered employment) s more or less constant durng the studed perod. Ths s observed across several subgroups such as age, gender and natonalty. Ths shows that nonresponse rate alone s a poor ndcator of the qualty. In chapter 5 we defne response structure as a rato of response rates. A smple relatonshp between the bas of a bnomal varable and the response structure s presented. From ths relatonshp we see that the bas s a functon of the response structure and not of the nonresponse rate. 2. Data sources 2.1. Survey sample The Norwegan Labour Force Survey (LFS) s a contnuous, rotatng panel, sample survey. The sample sze s The sample frame s regstered famles of resdents between 15 and 74 years old. The sample desgn s a one-stage cluster samplng where every famly member s ncluded n the sample. Each person s ntervewed once every quarter for eght consecutve quarters. Oversamplng of some sparsely populated regons, result n a slght varaton of selecton probablty. The reference perod s one week, and the ntervew s supposed to be performed wthn ten days after the reference week. Sample unts are allocated to every week n each year, n order to cover all seasonal varaton. The ntervewers are organzed n a central unt and several local ntervewers, but the mode of data collectng s exclusvely computer-asssted telephone ntervews and uses the same questonnare. 6 Statstcs Norway

9 Documents 54/2012 Nonresponse n the Norwegan Labour Force Survey (LFS) The varable of nterest here s employment, defned n the sample as havng worked at least one hour durng the reference week. Ths defnton s n lne wth recommendatons from the ILO (Internatonal Labour Organzaton). Our goal was to study as long as possble tme seres. The Norwegan LFS started n 1972, but avalable data collected before 1996 do not contan relevant nformaton about nonresponse at the mcro level. For the present study, we have selected data startng wth 1 st Quarter 1996 up to and ncludng 4 th Quarter The resultng study data conssts of over 1.5 mllon records for people; ncludng over 1.35 mllon ntervews from nearly people Admnstratve regsters Regster data used n ths study are collected by governmental agences for admnstratve purposes. The man source s the Norwegan Labour and Welfare Servce s employee regster. It s mandatory for employers to report employee jobs to the regster, wth few exceptons. Most data are sent to the regster drectly from the employers own IT-systems. Thus, the regster data are ndependent of the survey responses, and qute comprehensve and accurate as well. In addton, the regster data are somewhat revsed at Statstcs Norway, for nstance mergng records that pertan to the same job, and adjustng status by checkng aganst other regsters. Another mportant source of admnstratve data s the Central Populaton Regster, run by the Tax Agency. Demographc attrbutes, such as age, gender, ctzenshp and place of resdence, are collected only from regster data. The survey sample data s lnked to the regster data at the mcro level, usng the personal dentfcaton and approxmate reference tme. It can be descrbed as lnkng at the person level and not at job level. We remark that due to dfferences n defntons, data collectng routnes and tme lag, there s some dvergence at mcro- and macro level between survey employment and regster-based employment. For a dscusson about these matters, se for nstance Vllund (2010). 3. Response rate We defne nonresponse rate as n t R,t t = 1 = where n t, t response status for ndvdual at tme t n sample of sze n t. r R s a bnary varable for the Fgure 3 1 shows response rates over tme, for each quarter and the yearly mean. Response rates show a declnng trend over the years ( ). In addton to ths trend, there s a varaton between the years due to random and nonrandom factors (for nstance ntervew staff reorganzatons, countermeasures amed at ncreasng response, etc.). A seasonal pattern s observed n the perod , when the response rate s sgnfcantly lower n the second quarter. We beleve ths can be explaned by the allocaton of ad-hoc modules, whch result n more burdensome ntervews. From 1996 to 2005 these extra questons where allocated to the 2 nd quarter, now they are dstrbuted over the whole year. Statstcs Norway 7

10 Nonresponse n the Norwegan Labour Force Survey (LFS) Documents 54/2012 Fgure 3 1. Response rate by quarter. LFS The next fgures llustrate varaton n response rates between demographc groups. In ths study, we had a few varables avalable such as ctzenshp, age and gender. The response rate s generally hgher for people over 40 years old, for women, and for natonals. People under 40 years old have lower response rate, wth less gender dfference than total. Nonnatonals have even lower response rate, and greater dfference between men and women. Ths pattern seems rather constant over the whole perod. Fgure 3 2. Response rate, by age and gender. LFS Statstcs Norway

11 Documents 54/2012 Nonresponse n the Norwegan Labour Force Survey (LFS) Fgure 3 3. Response rate, by ctzenshp and age. LFS Fgure 3 4 Response rate, by ctzenshp and gender. LFS We now turn to varaton n response rate between employed and not-employed people. The next three fgures llustrate varaton n response rates between employed and not-employed people by age, gender and ctzenshp. We observe a clear dfference n response rate between employed and not-employed, across the demographc groups. Statstcs Norway 9

12 Nonresponse n the Norwegan Labour Force Survey (LFS) Documents 54/2012 Fgure 3 5. Response rate, by employment and age. LFS Fgure 3 6. Response rate, by employment and gender. LFS Statstcs Norway

13 Documents 54/2012 Nonresponse n the Norwegan Labour Force Survey (LFS) Fgure 3 7. Response rate, by employment and ctzenshp. LFS Nonresponse Bas We defne nonresponse bas as where pˆ p = n t (X,tR,t ) = 1 n t X s a bnary varable for the regster employment status for ndvdual, t n sample n at tme t, and R as defned n the response rate formula, pˆ the employment rate n the response sample, and p the employment rate n the total sample. The prevous secton demonstrated the varaton n response rate by employment. From earler, we know that regster-based employment status s correlated wth survey-based employment status and survey response status. Therefore, regsterbased employment s a key varable for analyzng nonresponse bas. We defne nonresponse bas as employment rate n the response sample mnus employment rate n the total sample. Thus, a postve fgure means that the response sample over-represents employment. In order to observe the relatonshp between the bas and the rate over tme, we compare the trends for nonresponse rate and nonresponse bas over the perod Fgure 4 1 shows the results for ths perod. Whle the nonresponse rate shows a growng trend, the nonresponse bas shows no clear trend. Takng averages over 4-year perods, we actually fnd smaller bas n the perod than n the prevous three perods,.85 percentage pont compared to = 1 R,t n t = 1 n X t,t Statstcs Norway 11

14 Nonresponse n the Norwegan Labour Force Survey (LFS) Documents 54/2012 From ths, we draw two prelmnary conclusons: Nonresponse bas does not show any clear trend durng ths perod, but possbly a slow declne. The nonresponse rate s ncreasng and shows no obvous relatonshp wth the bas. Thus t seems to be of lttle use n assessng the mpact of nonresponse. Fgure 4 1. Nonresponse rate and -bas. LFS Non-natonals and non-response Snce earler fndngs (e.g. Vllund 2008) have shown low response rates for mmgrants, we wsh to look closer at ths group. Although stll a relatvely small group, the mmgrant populaton has grown from 5 percent to 13 percent from 1996 to In comparson, the proporton of nonnatonals n the sample has grown from around 3.5 percent to nearly 8 percent.fgure 4 2 Compares the nonresponse for natonals and nonnatonals (thn lnes). We observe a hgher nonresponse rate and bas for nonnatonals seems to be rather constant over tme. 12 Statstcs Norway

15 Documents 54/2012 Nonresponse n the Norwegan Labour Force Survey (LFS) Fgure 4 2. Comparng nonresponse rate and bas, by ctzenshp. LFS A smple relatonshp between response rate and nonresponse bas, for bnomnal varables Let X and R denote the bnomnal varable and response status as defned n the prevous chapter. Furthermore, we defne the followng probabltes: r1 = P(R = 1 X = 1) r0 = P(R = 1 X = 0) p = P(X = 1) pˆ = P(X = 1 R = 1) Consder now the estmated odds of beng employed: pˆ = (1 pˆ) n = 1 n = 1 X R (1 X )R Takng the expected value we fnd the followng approxmaton: E pˆ E (1 pˆ) E n = 1 n = 1 X R (1 X )R = n = 1 n = 1 E E ( E(X R ) X ) 0 ( E((1 X )R ) X ) p r = 1 p r 1 Statstcs Norway 13

16 Nonresponse n the Norwegan Labour Force Survey (LFS) Documents 54/2012 From ths we see that the bas of the observed odds, and therefore of the observed r1 proporton, s a functon of the rato but not of the response rate. In other r0 words, response rates can vary consderably wthout affectng the nonresponse bas, f ths rato remans constant. Fgure 5 1 llustrates that ths n fact s the case n the data we have studed. We plan to compare several sample surveys regardng ths. Fgure 5 1. Comparng nonresponse rate, bas and response structure. LFS Statstcs Norway

17 Documents 54/2012 Nonresponse n the Norwegan Labour Force Survey (LFS) Tables Table 1. Response rates. LFS yearly average Total... 91,7 91,5 89,9 90,3 89,9 87,7 90,0 89,6 90,4 89,2 87,4 87,5 86,4 87,0 85,1 84,4 Gender Men... 90,8 90,6 88,9 89,2 89,0 86,7 89,0 88,8 89,9 88,5 86,7 87,1 85,6 86,0 84,1 83,8 Women... 92,6 92,3 91,0 91,4 90,9 88,8 91,0 90,5 91,0 90,0 88,1 88,0 87,2 88,1 86,2 85,0 Age år... 92,3 90,7 89,7 90,5 89,7 87,7 90,4 89,0 90,1 89,4 88,1 88,1 86,7 88,6 86,2 86, år... 90,6 91,1 89,1 88,6 87,6 84,6 87,6 87,3 87,5 85,2 83,6 84,1 80,8 80,7 78,7 77, år... 92,0 91,6 90,0 90,9 90,6 88,3 90,2 89,8 91,2 89,8 87,7 88,2 87,3 87,5 84,7 84, år... 91,9 91,3 89,7 90,5 90,9 89,5 91,5 91,6 92,2 91,8 89,8 89,7 89,9 90,4 89,7 89, år... 93,0 94,2 93,8 94,3 95,1 94,2 94,8 95,2 95,6 95,9 92,7 90,4 92,3 94,2 94,0 93,6 Resdence Other resdences 92,6 92,4 91,1 91,3 91,1 89,3 91,2 90,6 91,4 90,3 88,7 88,8 87,6 88,3 86,3 85,7 Large 92,0 91,8 88,6 90,5 90,1 87,2 90,9 90,0 90,8 89,7 86,8 88,1 87,0 87,1 85,1 84,8 Muncpaltes Oslo (CAPITAL).. 84,9 84,2 83,5 82,6 81,2 76,4 79,5 81,9 83,3 81,2 79,1 78,3 76,9 78,1 77,2 75,2 Employment Not employed... 89,1 88,8 87,3 87,8 87,7 85,6 87,6 87,3 88,1 86,9 85,0 84,7 83,9 85,2 83,2 82,4 Employed... 93,3 93,0 91,3 91,6 91,1 88,9 91,2 90,9 91,7 90,5 88,8 89,2 87,7 88,0 86,2 85,5 Table 2. Nonresponse bas. LFS yearly average Total... 0,00 0,00 0,00 0,00 0,00 0,00 0,00 0,00 0,00 0,00 0,00 0,00 0,00 0,00 0,00 0,00 Gender Men... -0,49-0,47-0,57-0,61-0,53-0,59-0,57-0,49-0,32-0,41-0,39-0,28-0,47-0,61-0,60-0,37 Women... 0,49 0,47 0,57 0,61 0,53 0,59 0,57 0,49 0,32 0,41 0,39 0,28 0,47 0,61 0,60 0,37 Age ' ,10-0,14-0,03 0,04-0,04-0,01 0,06-0,10-0,06 0,04 0,13 0,10 0,06 0,29 0,22 0,32 ' ,37-0,14-0,32-0,62-0,80-1,09-0,81-0,80-0,97-1,29-1,21-1,05-1,72-1,90-2,00-2,30 ' ,10 0,05 0,03 0,19 0,21 0,20 0,07 0,05 0,23 0,19 0,11 0,23 0,32 0,14-0,16-0,04 ' ,04-0,03-0,03 0,04 0,16 0,34 0,30 0,41 0,38 0,54 0,53 0,47 0,81 0,79 1,07 1,09 ' ,13 0,26 0,36 0,36 0,47 0,57 0,38 0,44 0,42 0,53 0,44 0,25 0,54 0,68 0,88 0,93 Resdence Other resdences.. 0,73 0,76 0,96 0,82 0,94 1,38 1,04 0,83 0,76 0,86 1,11 1,03 1,06 1,10 1,02 1,11 Large 0,06 0,06-0,23 0,04 0,03-0,08 0,15 0,06 0,06 0,09-0,10 0,10 0,11 0,01 0,00 0,09 muncpaltes... Oslo (captal)... -0,79-0,82-0,73-0,86-0,97-1,29-1,19-0,89-0,82-0,95-1,01-1,14-1,17-1,11-1,03-1,21 Employment Not employed... -1,10-1,04-1,01-0,93-0,86-0,85-0,91-0,94-0,91-0,93-1,01-1,18-0,98-0,73-0,84-0,86 Employed... 1,10 1,04 1,01 0,93 0,86 0,85 0,91 0,94 0,91 0,93 1,01 1,18 0,98 0,73 0,84 0,86 P1-P Table 3. Response rate, by age and gender. LFS yearly average Total Men Women Total Other age groups years olds Total Other age groups years olds Total Other age groups years olds ,6 92,0 90,2 90,7 91,3 88,6 92,5 92,8 91, ,4 91,6 90,8 90,5 90,9 89,5 92,3 92,3 92, ,9 90,3 88,6 88,9 89,5 86,9 90,9 91,1 90, ,2 90,9 88,0 89,1 90,0 86,1 91,4 91,8 89, ,8 90,6 87,2 88,9 89,9 85,7 90,8 91,4 88, ,6 88,8 83,6 86,6 87,9 82,3 88,6 89,7 85, ,9 90,8 87,1 88,9 89,9 85,8 91,0 91,7 88, ,6 90,5 86,4 88,7 89,6 85,6 90,5 91,3 87, ,4 91,4 86,7 89,8 90,8 86,2 91,0 92,0 87, ,2 90,5 84,1 88,4 89,8 83,0 89,9 91,1 85, ,4 88,5 82,4 86,7 87,8 81,9 88,0 89,3 82, ,5 88,5 83,1 87,0 88,0 82,7 88,0 89,0 83, ,3 88,0 79,1 85,5 87,2 78,1 87,2 88,7 80, ,0 88,7 79,0 85,9 87,7 77,9 88,1 89,8 80, ,1 86,9 76,8 84,1 86,0 75,7 86,1 87,9 78, ,3 86,4 75,4 83,7 85,7 75,2 85,0 87,1 75,7 Statstcs Norway 15

18 Nonresponse n the Norwegan Labour Force Survey (LFS) Documents 54/2012 Table 4. Response rate, by ctzenshp and gender. LFS yearly average Total Men Women Total Nonnatonal Natonal Total Nonnatonal Natonal Total Nonnatonal Natonal ,6 76,3 92,2 90,7 72,8 91,4 92,5 80,1 93, ,4 75,4 92,0 90,5 72,0 91,2 92,3 79,0 92, ,9 75,8 90,4 88,9 71,7 89,5 90,9 80,2 91, ,2 75,2 90,8 89,1 70,8 89,8 91,4 80,3 91, ,8 76,5 90,3 88,9 75,0 89,4 90,8 78,1 91, ,6 71,9 88,2 86,6 69,1 87,3 88,6 74,7 89, ,9 77,2 90,4 88,9 73,5 89,6 91,0 81,0 91, ,6 77,7 90,1 88,7 76,1 89,3 90,5 79,4 91, ,4 78,0 90,9 89,8 77,1 90,4 91,0 78,9 91, ,2 74,8 89,9 88,4 74,0 89,1 89,9 75,5 90, ,4 71,3 88,1 86,7 68,2 87,5 88,0 73,9 88, ,5 73,1 88,2 87,0 69,8 87,9 88,0 76,3 88, ,3 69,6 87,3 85,5 66,4 86,7 87,2 73,3 87, ,0 69,1 88,1 85,9 65,1 87,3 88,1 73,7 88, ,1 67,2 86,3 84,1 62,7 85,6 86,1 72,4 87, ,3 67,0 85,7 83,7 63,2 85,4 85,0 71,2 86,0 Table 5. Response rate, by ctzenshp and resdence LFS yearly average Total Nonnatonal Natonal Total Other Oslo Total Other Oslo Total Other Oslo resdences resdences resdences ,6 92,5 84,4 76,3 82,5 64,8 92,2 92,7 87, ,4 92,3 84,0 75,4 81,9 62,4 92,0 92,5 86, ,9 90,6 83,3 75,8 81,3 65,0 90,4 90,9 85, ,2 91,1 82,2 75,2 80,6 63,6 90,8 91,4 84, ,8 90,9 80,8 76,5 82,2 63,7 90,3 91,1 82, ,6 88,9 75,8 71,9 78,3 55,9 88,2 89,3 78, ,9 91,1 79,4 77,2 81,3 62,6 90,4 91,5 81, ,6 90,5 81,9 77,7 81,7 64,0 90,1 90,8 83, ,4 91,2 83,2 78,0 81,4 67,1 90,9 91,6 85, ,2 90,1 81,1 74,8 79,3 60,3 89,9 90,6 83, ,4 88,3 79,0 71,3 74,7 60,3 88,1 88,9 81, ,5 88,6 78,2 73,1 76,2 61,3 88,2 89,2 80, ,3 87,5 76,9 69,6 73,0 56,2 87,3 88,2 79, ,0 88,1 78,0 69,1 71,3 61,0 88,1 89,0 80, ,1 86,1 77,2 67,2 69,9 58,0 86,3 87,0 80, ,3 85,5 75,1 67,0 69,4 57,3 85,7 86,6 77,8 Table 6. Response rate, by age and resdence LFS yearly average Total Total Other resdences Oslo Other age groups years olds Total Other age groups years olds Total Other age groups years olds ,6 92,0 90,2 92,5 92,8 91,2 84,4 84,8 83, ,4 91,6 90,8 92,3 92,4 91,8 84,0 83,8 84, ,9 90,3 88,6 90,6 90,9 89,6 83,3 84,1 81, ,2 90,9 88,0 91,1 91,7 89,1 82,2 83,1 80, ,8 90,6 87,2 90,9 91,5 88,5 80,8 81,7 78, ,6 88,8 83,6 88,9 90,0 85,2 75,8 76,8 73, ,9 90,8 87,1 91,1 91,8 88,6 79,4 80,2 77, ,6 90,5 86,4 90,5 91,3 87,4 81,9 82,4 80, ,4 91,4 86,7 91,2 92,1 87,6 83,2 84,1 81, ,2 90,5 84,1 90,1 91,4 84,9 81,1 81,8 79, ,4 88,5 82,4 88,3 89,5 83,4 79,0 79,8 76, ,5 88,5 83,1 88,6 89,5 84,5 78,2 79,2 75, ,3 88,0 79,1 87,5 89,0 80,1 76,9 78,1 73, ,0 88,7 79,0 88,1 89,6 80,3 78,0 80,2 72, ,1 86,9 76,8 86,1 87,7 78,0 77,2 79,4 71, ,3 86,4 75,4 85,5 87,4 76,5 75,1 77,1 70,2 16 Statstcs Norway

19 Documents 54/2012 Nonresponse n the Norwegan Labour Force Survey (LFS) References Thomsen, I. and Zhang, L.-C.: The Effects of Usng Admnstratve Regsters n Economc Short Term Statstcs: The Norwegan Labour Force Survey as a Case Study Journal of Offcal Statstcs, Vol. 17, No. 2, 2001, pp Vllund, O.: Evaluatng employment classfcaton. A qualty study lnkng survey data and regster data Statstcs Norway Documents 2010/09. Vllund, O.: Immgrant partcpaton n the Norwegan Labour Force Survey Statstcs Norway Documents 2008/7 Statstcs Norway 17

20 Nonresponse n the Norwegan Labour Force Survey (LFS) Documents 54/2012 Appendx A Ths annex contans some techncal nformaton and more detaled results made durng the study. Ths materal s ncluded for reference and background nformaton for future studes. Documentaton of the sample data Fgure 1compares the sample sze of drect- and proxy response, as well as nonresponse and nelgble unts. Proxy response means that a spouse or parent s ntervewed on behalf of the study unt. Drect means self-response,.e. the study unts answer themselves. Passve unts are old-age pensoners and nsttutonalzed people that are ntentonally not ntervewed, thus not classfed as nonresponse. The small group Out conssts of people who are dead or emgrated between samplng and ntervew week. Fgure 1. Composton of total sample. LFS The purpose of ths study was not to analyze the reasons or underlyng causes of nonresponse. However, durng feldwork, the nonrespondents are classfed accordng to reason for nonresponse. Ths classfcaton s of course manly based on the ntervewers percepton of the stuaton, snce we can t ask the potental respondent about the real reason. Some groups are relatvely specfc, such as language problems and refusals, these groups are often relatvely small. Many of the major groups tell us lttle about the actual causes, such as non-contact and unspecfed reasons. Fgure 2 shows the composton of the nonresponse, wth respect to selected types. The man group Other conssts manly of non-contact and other plausble reasons for not beng ntervewed. Outrght refusal consttutes a relatvely small part of nonresponse, and wthout a clear trend. Explctly dentfed language problems are growng somewhat n later years, n lne wth an ncreasng mmgrant populaton. We suspect that the proportons of both refusals and language problems are underestmated. For nstance when people do not answer the telephone, t s hard to know the real reason. 18 Statstcs Norway

21 Documents 54/2012 Nonresponse n the Norwegan Labour Force Survey (LFS) Fgure 2. Composton of nonresponse. LFS Quarterly results The fgures presented prevous chapters are yearly average results, wth the ntenton of lookng for structures and trends. In fact, results from the quarterly data show much more varaton. The followng fgures llustrate ths. Fgure 3. Response rate, by age and gender. LFS Statstcs Norway 19

22 Nonresponse n the Norwegan Labour Force Survey (LFS) Documents 54/2012 Fgure 4. Response rate, by ctzenshp and age. LFS Fgure 5. Response rate, by ctzenshp and gender. LFS Statstcs Norway

23 Documents 54/2012 Nonresponse n the Norwegan Labour Force Survey (LFS) Fgure 6. Response rate, by employment and age. LFS Fgure 7. Response rate, by employment and gender. LFS Statstcs Norway 21

24 Nonresponse n the Norwegan Labour Force Survey (LFS) Documents 54/2012 Fgure 8. Response rate, by employment and ctzenshp. LFS Place of resdence The two next fgures llustrate dfference n response rate by place of resdence. We have not found a clear assocaton between response rate and urbanzaton n general, but there are dfferences between the captal and other places of resdence. These dfferences cut across demographc varables. Fgure 9. Response rate, by place of resdence and age. LFS Statstcs Norway

25 Documents 54/2012 Nonresponse n the Norwegan Labour Force Survey (LFS) Fgure 10. Response rate, by place of resdence and ctzenshp. LFS Fgure 11 compares nonresponse rate and nonresponse bas trends over the years Whle the nonresponse rate shows a growng trend, the nonresponse bas shows a small but possbly decreasng trend, but wth relatvely large fluctuatons over tme. Fgure 11. Nonresponse rate vs nonresponse bas. LFS Statstcs Norway 23

26 Nonresponse n the Norwegan Labour Force Survey (LFS) Documents 54/2012 Lst of fgures Fgure 3 1. Response rate by quarter. LFS Fgure 3 2. Response rate, by age and gender. LFS Fgure 3 3. Response rate, by ctzenshp and age. LFS Fgure 3 4 Response rate, by ctzenshp and gender. LFS Fgure 3 5. Response rate, by employment and age. LFS Fgure 3 6. Response rate, by employment and gender. LFS Fgure 3 7. Response rate, by employment and ctzenshp. LFS Fgure 4 1. Nonresponse rate and -bas. LFS Fgure 4 2. Comparng nonresponse rate and bas, by ctzenshp. LFS Fgure 5 1. Comparng nonresponse rate, bas and response structure. LFS Fgure 1. Composton of total sample. LFS Fgure 2. Composton of nonresponse. LFS Fgure 3. Response rate, by age and gender. LFS Fgure 4. Response rate, by ctzenshp and age. LFS Fgure5. Response rate, by ctzenshp and gender. LFS Fgure 6. Response rate, by employment and age. LFS Fgure 7. Response rate, by employment and gender. LFS Fgure 8. Response rate, by employment and ctzenshp. LFS Fgure 9. Response rate, by place of resdence and age. LFS Fgure 10. Response rate, by place of resdence and ctzenshp. LFS Fgure 11. Nonresponse rate vs nonresponse bas. LFS Lst of tables Table 1. Response rates. LFS yearly average Table 2. Nonresponse bas. LFS yearly average Table 3. Response rate, by age and gender. LFS yearly average Table 4. Response rate, by ctzenshp and gender. LFS yearly average Table 5. Response rate, by ctzenshp and resdence LFS yearly average Table 6. Response rate, by age and resdence LFS yearly average Statstcs Norway

27

28 B NO-2225 Returadresse: Statstsk sentralbyrå Kongsvnger Statstcs Norway 54/2012 Statstcs Norway Oslo: PO Box 8131 Dept NO-0033 Oslo Telephone: Telefax: Kongsvnger: NO-2225 Kongsvnger Telephone: Telefax: E-mal: ssb@ssb.no Internet: ISBN Prnted verson ISBN Electronc verson ISSN Nonresponse n the Norwegan Labour Force Survey (LFS) Desgn: Sr Boqust