CLS Cohort Studies. Centre for Longitudinal Studies. Data Note 2007/3 CLS. Programming Employment Histories in BCS70 Sweeps 5, 6 and

Transcription

1 CLS CLS Cohort Studies Data Note 2007/3 Centre for Longitudinal Studies Programming Employment Histories in BCS70 Sweeps 5, 6 and Kelly Ward Centre for Longitudinal Studies Bedford Group for Lifecourse and Statistical Studies Institute of Education 20 Bedford Way London WC1H 0AL Tel: Fax: cls@ioe.ac.uk Web

2 CLS Cohort Studies Data Note 2007/03 Programming Employment Histories in BCS70 Sweeps 5, 6 and Kelly Ward Centre for Longitudinal Studies, Institute of Education. 1. Introduction The following work was undertaken as part of a Gender Network project with Professors Shirley Dex and Heather Joshi, examining the occupational mobility of men and women over the lifecourse, using the retrospective employment histories from BCS70 sweeps 5, 6 and 7. This Data Note outlines the extensive work undertaken to programme the job spells collected from survey sweeps 5, 6 and 7 into an employment history data file. SPSS syntax is also provided in the Appendices. This Data Note is split into three sections as follows: 2. British Cohort Study 1970 (BCS70): A brief introduction to the BCS70 survey and details of the User Support Group are provided. 3. The second section details the work undertaken to programme SPSS to hold the employment histories in a readable and workable format. 4. The third section highlights the remainder of the work carried out on the employment history file. 2. British Cohort Study 1970 (BCS70) The 1970 British Cohort Study (BCS70) is the third birth cohort study in Britain and was designed along similar lines to the second (1958) birth cohort study, the National Child Development Study. BCS70 surveyed over 17,000 babies born in Britain in the week 5-11 April Since the survey around the time of the 1970 birth, there have been six other major data collection exercises of the BCS70 cohort members in order to monitor their health, education, social and economic circumstances. These were carried out in 1975 (age 5), 1980 (age 10), 1986 (age 16), 1996 (age 26), 1999/2000 (age 29/30) and 2004 (age 34). The information collected in the last three sweeps has covered cohort members transitions into adult life, including leaving full-time education, entering the labour market, setting up independent homes, forming partnerships and becoming parents. 1

3 The main aim of these most recent surveys was to explore the factors central to the formation and maintenance of adult identity in each of the following domains: Lifelong learning Relationships, parenting and housing Employment and income Health and health behaviour Citizenship and values User Support Group This group, supported by CLS staff, provides advice and guidance on the use of BCS data by: providing documentation about the data; collating and disseminating information on uses of the data; producing publications; producing and distributing a newsletter and working papers; providing access to non-computerised BCS data; collecting additional information; and servicing the User Group. The User Support Group can be contacted by post, 'phone, fax, or as shown below: Cohort Studies User Support Group, Centre for Longitudinal Studies, Institute of Education, 20 Bedford Way, London WC1H 0AL Tel: Fax: cohort@cls.ioe.ac.uk Internet: 2

4 2. BCS70 Employment Histories The BCS70 employment histories are based on spells of employment recorded by respondents at BCS70 sweeps 5, 6 and 7. Each of these interview sweeps provides different information that has been used to prepare a chronological employment history dataset to chart the moves that respondents make in and out of employment from when the respondent left full-time education to the latest interview at sweep 7 (2004). In the discussion below we use the terminology as follows: a. Job spell which is a job but also implies a period of employment. b. Non-employment spell is the gap between discontinuous job spells. c. Hours status is whether a job spell has been coded either working full time (code 1) or part time (code 2). Data collected at each survey sweep is collected in spells, where a spell represents one job. Respondents were asked to delineate each spell with its start date (month and year). In effect therefore, each spell s dates relate to one job and its tenure. Obviously each of these spells represents a period of employment. However, two successive spells, namely two jobs may not represent an uninterrupted period of employment. The information collected at each survey sweep does not collect the end dates (month and year) of each spell. Therefore it is assumed here that, for instance, the start date of the second spell represents the end date of the first spell, regardless of whether the second spell is a period of employment or non-employment. Where job spell dates are not continuous, this often indicates that a period of non-employment separated these two job spells. The deposited dataset that accompanies this Data Note does not contain information about spells of nonemployment. Table 1 below provides a comparative overview of the information on employment histories available from the three sweeps of the BCS70 survey, 5, 6 and 7. This table describes the information that is available and how this information has been collected and stored in the original datasets. 3

5 Table 1: Overview of employment and occupation details available. Survey Data Collected Respondent Details Collected Only collects the current spell of Main respondent Date job spell started only the employment at interview. only. start year was collected. BCS70 Sweep 5 age 26 Hours status - part time or full time. Economic activity status whether employed or nonemployed. BCS70 Sweep 6 age 30 BCS70 Sweep 7 age 34 Collects the current job spell and also collects up to 10 job spells in the employment history, covering the period from interview (1999/2000) age back to age 16 (1986) and therefore overlaps with Sweep 5, when respondents were aged 26 years. Spell one refers to the most recent job spell nearest to the Sweep 6 interview date and spell 10 refers to the most distant spell in the past. The details collected at this sweep include spells of employment and non-employment together. The start dates of the main respondent interview are arranged in the order of the most recent job to the most distant job (e.g. 94, 92, 90, 87). Collects the current job spell and also collects up to 10 job spells dating back from the interview at age 34 (2004). The details collected at this sweep include spells of employment and non-employment together. Dates range from most recent job at time of interview to the most distant past job spell. Main respondent only Main respondent only. Occupation classification based on SOC90. Date job spell started - start month and start year. Hours status - part time or full time. Economic activity status whether employed or nonemployed. Occupation classification based on SOC90. No information was collected regarding the end dates of each job spell. Date job spell started - start month and start year. Hours status - part time or full time. Activity status whether employed or non-employed. Occupation classification based on SOC90. This sweep also has the occupations coded to the SOC2000 classification. No information was collected regarding the end dates of each spell. 4

6 3. Programming SPSS The first step of programming began by merging all the spells of employment together from BCS70 Sweeps 5, 6 and 7. The sort variable used was the individual case identifier key. The process of merging longitudinal data together often means that some cases will be identified as system missing at different points in the employment history. These values will refer to those who did not complete that interview wave. It is important to ensure that all missing values (such as Not applicable (-1) and Don t Know (-8) and system missing (-5)), consistently use the same values across survey sweeps. Organising the employment histories means working with a number of job spells that contain valid month and year dates (start and end) as well as information relating to the weekly hours worked and occupation category. To begin programming SPSS, it is necessary to run a command that will set up an SPSS vector of empty variables. In this employment history we needed 23 empty variables that we could use to load the information contained in each job spell. To begin with we loaded the start month of each job spell taken from BCS70 Sweeps 5, 6 and 7, of which there were 23 spells in total. One job spell was taken from Sweep 5, eleven from Sweep 6 and eleven from Sweep 7. SPSS sets up these variables by using the following command which will compute 23 variables that SPSS will call mth(1), mth(2), mth(3)..mth(23). All variables in the mth vector are initially set to 0 (missing values need to be turned off in SPSS for all of this syntax; this is shown in more detail in Appendix Two). vector mth(23). loop #a=1 to compute mth(#a)=0. Once we have defined the number of empty variables that we need in a vector to store all the history s spell records, we must then load in the information that we require. For example, this means that we must tell SPSS which start month for each spell belongs in which empty vector variable defined in the above box. This process is continued until we reach the end of the available job spell data. The following syntax provides an example of this. Mth(1) relates to the start month of the job spell taken from Sweep 5. Mth(2) to mth(12) relate to the start month of the job spells taken from Sweep 6 and mth(13) onwards relate to Sweep 7. When loading the required information for each job spell into the vectors it was important to select only on spells that referred to a job. This is because spells of non-employment were collected alongside the job spells. Therefore when taking the start month of a job spell from Sweeps 5, 6 and 7, it was necessary to ensure these month values were only copied into the vector that related to a job spell (and if the spell was a non-employment spell, it was passed over). 5

7 Note: When working with vectors SPSS often requires that parts of the variable name are in parentheses, for example Mth(1), however once you have defined the vector as outlined in the example above SPSS will load the information shown below without the need for parentheses. compute mth1=6. if (b26act>=3) mth1=-1. if (b26act<0) mth1=-1. compute mth2=startm10. if (act10=3) mth2=-1. compute mth3=startmo9. if (act9=3) mth3=-1. compute mth4=startmo8. if (act8=3) mth4=-1. compute mth5=startmo7. if (act7=3) mth5=-1. compute mth6=startmo6. if (act6=3) mth6=-1. compute mth7=startmo5. if (act5=3) mth7=-1. compute mth8=startmo4. if (act4=3) mth8=-1. compute mth9=startmo3. if (act3=3) mth9=-1. compute mth10=startmo2. if (act2=3) mth10=-1. compute mth11=startmo. if (act1=3) mth11=-1. compute mth12=cstartmo. if (cact=3) mth12=-1. compute mth13=b7stmo10. if (b7act10=3) mth13=-1. compute mth14=b7stmo09. if (b7act9=3) mth14=-1. compute mth15=b7stmo08. if (b7act8=3) mth15=-1. compute mth16=b7stmo07. if (b7act7=3) mth16=-1. compute mth17=b7stmo06. if (b7act6=3) mth17=-1. compute mth18=b7stmo05. if (b7act5=3) mth18=-1. compute mth19=b7stmo04. if (b7act4=3) mth19=-1. compute mth20=b7stmo03. if (b7act3=3) mth20=-1. compute mth21=b7stmo02. if (b7act2=3) mth21=-1. compute mth22=b7stmo01. if (b7act1=3) mth22=-1. compute mth23=b7cstamo. if (b7cact=3) mth23=-1. execute. variable label mth1 "bcs 26 - start month job one". variable label mth2 "bcs 30 - start month job ten". variable label mth3 "bcs 30 - start month job nine". variable label mth4 "bcs 30 - start month job eight". variable label mth5 "bcs 30 - start month job seven". variable label mth6 "bcs 30 - start month job six". variable label mth7 "bcs 30 - start month job five". variable label mth8 "bcs 30 - start month job four". variable label mth9 "bcs 30 - start month job three". variable label mth10 "bcs 30 - start month job two". variable label mth11 "bcs 30 - start month job one". variable label mth12 "bcs 30 - start month job current". variable label mth13 "bcs 34 - start month job ten". variable label mth14 "bcs 34 - start month job nine". variable label mth15 "bcs 34 - start month job eight". variable label mth16 "bcs 34 - start month job seven". variable label mth17 "bcs 34 - start month job six". variable label mth18 "bcs 34 - start month job five". variable label mth19 "bcs 34 - start month job four". variable label mth20 "bcs 34 - start month job three". variable label mth21 "bcs 34 - start month job two". variable label mth22 "bcs 34 - start month job one". variable label mth23 "bcs 34 - start month job current". 6

8 value labels mth1 mth2 mth3 mth4 mth5 mth6 mth7 mth8 mth9 mth10 mth11 mth12 mth13 mth14 mth15 mth16 mth17 mth18 mth19 mth20 mth21 mth22 mth23-8 'don t know' -1 'not applicable'. Execute. Once this syntax has been run, the empty vector variables that were originally set to zero should now contain the start month of all the job spells taken from Sweeps 5, 6 and 7. Once this information is loaded into the empty vector variables, the data will not be in any particular chronological order, as information obtained at each sweep can potentially overlap with the previous sweep. It is also possible that respondents did not participate at all three sweeps, which will also mean that the vector data at this point will not be chronological and will have missing spells. Therefore the vector should look like this: Table 1: Start month data Details obtained from Sweep 5 Sweep 6 Sweep 7 How the original sweep data appears: Variable names of the start month variables in each job spell (as loaded into the startm10 startm09 startm08 b7stmo10 b7stmo09 vector in the example above) syntax Month start date values that should be loaded into the empty vector variables The start month of this job spell was not collected. We have set all start months for these job spells to (Not Applicable) -1 (Not Applicable) How the vector data will appear: 4 6 Vector name Mth(1) Mth(3) Mth(4) Mth(5) Mth(16) Mth(17) Values found in these vectors should match the variable were the (Not Applicable) -1 (Not Applicable) 4 6 information was obtained, as represented here Setting up a vector of each job spell s start month represents only the initial stages of programming the employment histories. Once the start month details had been programmed, the other information about each spell also needs to be stored in its own 23-space vector, namely the spell s start year, weekly hours, end month and year dates (please see note below) and the occupation code. This requires a total of 6 vectors of 23 variables in each vector. Throughout each stage, it was necessary to recode all system 7

9 missing values into a common missing category which is set to -5. Appendix Two provides full details of the syntax used at this stage of the process. Job spell end dates The end dates of each job spell were not collected in the BCS70 employment histories in survey sweeps 5, 6 and 7. To identify the end dates of the job spells we decided to use the start date of the following spell, either the following (in time) job or non-employment spell. Empty vectors were declared to load in the end month and year dates for each job spell, and the dates used to load the information were the start dates of a job spell. If a start date represented a spell of non-employment it was still used to identify an end date of a previous job spell. Therefore, if job spell (1) was followed by a spell of non-employment, the start date of the non-employment spell would be used to represent the end date for job spell (1). If job spell (1) was followed by job spell (2), the start date of the job spell (2) would be used to represent the end date of job spell (1). 4. Restructuring the data file Once all the required variables are programmed into vectors it is important to restructure the files. The first restructuring undertaken was to remove the missing values from the histories (as shown in Table 1, these missing values can crop up in the middle of the employment history). Restructuring therefore meant re-ordering the dataset so that all valid values for each job spell followed in a chronological order and all missing not applicable values were found at the end of the employment history file. To ensure that the correct variables were removed I focused on the start year of each job spell and selected on those with valid values greater than zero. If a missing cell was removed using its start year then it was necessary to remove the corresponding information relating to this particular spell and position from the other vectors (start month, end month, end year etc). To facilitate this, storage vectors were set up to receive the correct valid responses and these responses were then re-ordered in each vector accordingly. Storage vectors will temporarily hold the information in the SPSS memory but do not set up any further variables. The next step was to reorder the job spells chronologically from the most distant job spell to the most recent job spell. Therefore if job(3) started in 1989 and job(10) started in 1987, job(10) would be moved within the vector to come before job(3) and the variable names were reordered to continue in the format of job(1), job(2),. job(23). Again, any reordering that takes place must also carry through to reorder the corresponding details for that spell s position in all of the vectors. So for example the occupation category for job(10) would also need to be moved to come before the occupation category for job(3). 8

10 To complete the reordering process efficiently, it was necessary to identify the cases where successive spells started in the same year. To find out which spell was earliest, the start and end months of each spell needed to be compared to order the job spells in the correct chronological sequence. For example, if the start year of job spell (1) and job spell (2) were both 1999 and the start months equalled August and March respectively, then the file was reordered again so that the month and year of job spell (2) became job spell (1) and job spell (1) became job spell (2). Due to the complexity of this, I again made use of temporary storage vectors. At the seam points between BCS70 interviews (where the retrospective employment history from Sweep 7 (or Sweep 6) reached back to Sweep 6 s (or Sweep5 s) interview date, it was possible for duplicate details of job spells to be recorded. Where duplicate cases were identified, we kept the data which was closest to the time of its collection. For example, if a respondent identified a job in Sweep 5 as starting on 9/1980 and this job spell was also identified in Sweep 6, then the details given in Sweep 6 would be dropped from the employment history and the Sweep 5 information about this job spell would be kept. It is also important to identify the cases that had said they were still employed when the interview took place these cases are censored at interview. For these cases the end date values were set to -6. Step-by-step details are outlined in Appendix Three. The final job spell for any individual depends on whether they participated in all three interview sweeps, or only two or only one of them. Where the final job spell appears in the 23 space vector record will depend on how many jobs were reported in total across the survey sweeps. A value of -6 in the end dates of a respondent s employment history appears only once for respondents who had entered their most recent job across the three interviews, but the tenure of this job was censored at interview. Derived Variables I have included one additional derived variable in the employment history dataset deposited at the Data Archive. This variable identifies the following: a) Censored 0/1 variable which identifies the cases censored at interview (code 1) by picking up all the -6 values. 9

11 APPENDIX One: Programming SPSS Once the variables at each cross-section are cleaned we start by merging the datasets together. The syntax for the path to these data files (e.g. F:\Employment History\...) will need to be amended to point at the directory in which you have stored them on your own PC. GET FILE='F:\Employment History\BCS26 - employment data.sav'. SORT CASES BY key (A). SAVE OUTFILE='F:\Employment History\BCS26 - employment data.sav' /COMPRESSED. GET FILE='F:\Employment History\BCS30 - employment data.sav'. SORT CASES BY key (A). SAVE OUTFILE='F:\Employment History\BCS30 - employment data.sav' /COMPRESSED. GET FILE='F:\Employment History\BCS34 - employment data.sav'. SORT CASES BY key (A). SAVE OUTFILE='F:\\Employment History\BCS34 - employment data.sav' /COMPRESSED. **NOW TO MERGE THE THREE SURVEY SWEEPS TOGETHER USING A COMMON IDENTIFIER** GET FILE='F:\Employment History\BCS26 - employment data.sav'. MATCH FILES /FILE=* /FILE='F:\Employment History\BCS30 - employment data.sav' /BY key. EXECUTE. SAVE OUTFILE='F:\Employment History\BCS Employment History.sav' /COMPRESSED. GET FILE='F:\Employment History\BCS Employment History.sav'. SORT CASES BY key (A). MATCH FILES /FILE=* /FILE='F:\Employment History\BCS34 - employment data.sav' /BY key. EXECUTE. SAVE OUTFILE='F:\Employment History\BCS Employment History.sav' /COMPRESSED. First need to recode the activity status of each job in the BCS70 survey sweeps because these variables include information of non-employment spells. *recode economic activity for age BCS 5 (age 26) ** recode b26empstat2 (sysmis=-5) (else=copy) into b26act. variable label b26act "BCS 26 Economic Activity". value labels b26act 1 'works ft' 2 'works pt' 3 'not in work' 4 'at home ft' 5 'other' -1 'not applicable' -5 'system missing'. 10

12 execute. **recode economic activity variables for BCS 6 (age 30) ** recode activi10 activit9 activit8 activit7 activit6 activit5 activit4 activit3 activit2 activity econact (1,3=1) (2,4=2) (5 thru hi=3) (-5, sysmis=-5) (else=copy) into act10 act9 act8 act7 act6 act5 act4 act3 act2 act1 cact. variable labels act10 "BCS 30 Job status activity 10". variable labels act9 "BCS 30 Job status activity 9". variable labels act8 "BCS 30 Job status activity 8". variable labels act7 "BCS 30 Job status activity 7". variable labels act6 "BCS 30 Job status activity 6". variable labels act5 "BCS 30 Job status activity 5". variable labels act4 "BCS 30 Job status activity 4". variable labels act3 "BCS 30 Job status activity 3". variable labels act2 "BCS 30 Job status activity 2". variable labels act1 "BCS 30 Job status activity 1". variable labels cact "BCS 30 Job status activity at interview". add value labels act10 act9 act8 act7 act6 act5 act4 act3 act2 act1 cact -8 'don t know' -5 'system missing' 1 'full time' 2 'part time' 3 'non-employed'. execute. **recode economic activity variables for BCS 7 (age 34)** recode bd7oth10 bd7oth09 bd7oth08 bd7oth07 bd7oth06 bd7oth05 bd7oth04 bd7oth03 bd7oth02 bd7oth01 bd7ecact (1,3=1) (2,4=2) (5 thru hi=3) (sysmis=-5) (else=copy) into b7act10 b7act9 b7act8 b7act7 b7act6 b7act5 b7act4 b7act3 b7act2 b7act1 b7cact. variable labels b7act10 "BCS 34 Job status activity 10". variable labels b7act9 "BCS 34 Job status activity 9". variable labels b7act8 "BCS 34 Job status activity 8". variable labels b7act7 "BCS 34 Job status activity 7". variable labels b7act6 "BCS 34 Job status activity 6". variable labels b7act5 "BCS 34 Job status activity 5". variable labels b7act4 "BCS 34 Job status activity 4". variable labels b7act3 "BCS 34 Job status activity 3". variable labels b7act2 "BCS 34 Job status activity 2". variable labels b7act1 "BCS 34 Job status activity 1". variable labels b7cact "BCS 34 Job status activity at interview". add value labels b7act10 b7act9 b7act8 b7act7 b7act6 b7act5 b7act4 b7act3 b7act2 b7act1 b7cact -9 'refusal' -8 'don t know' -5 'system missing' -1 'not applicable' 1 'full time' 2 'part time' 3 'non-employed'. execute. 11

13 VECTORS *LOADS THE START MONTH VALUES INTO SPSS* ************************************************************************************************** *load in start month of each job spell making sure it isn't a spell of non-employment* ************************************************************************************************** recode startm10 startmo9 startmo8 startmo7 startmo6 startmo5 startmo4 startmo3 startmo2 startmo cstartmo b7stmo10 b7stmo09 b7stmo08 b7stmo07 b7stmo06 b7stmo05 b7stmo04 b7stmo03 b7stmo02 b7stmo01 b7cstamo (-5=-1) (sysmis=-5) (else=copy). missing values startm10 startmo9 startmo8 startmo7 startmo6 startmo5 startmo4 startmo3 startmo2 startmo cstartmo b7stmo10 b7stmo09 b7stmo08 b7stmo07 b7stmo06 b7stmo05 b7stmo04 b7stmo03 b7stmo02 b7stmo01 b7cstamo b26act act10 act9 act8 act7 act6 act5 act4 act3 act2 act1 cact b7act10 b7act9 b7act8 b7act7 b7act6 b7act5 b7act4 b7act3 b7act2 b7act1 b7cact (). vector mth(23). loop #a=1 to compute mth(#a)=0. compute mth1=6. if (b26act>=3) mth1=-1. if (b26act<0) mth1=-1. compute mth2=startm10. if (act10=3) mth2=-1. compute mth3=startmo9. if (act9=3) mth3=-1. compute mth4=startmo8. if (act8=3) mth4=-1. compute mth5=startmo7. if (act7=3) mth5=-1. compute mth6=startmo6. if (act6=3) mth6=-1. compute mth7=startmo5. if (act5=3) mth7=-1. compute mth8=startmo4. if (act4=3) mth8=-1. compute mth9=startmo3. if (act3=3) mth9=-1. compute mth10=startmo2. if (act2=3) mth10=-1. compute mth11=startmo. if (act1=3) mth11=-1. compute mth12=cstartmo. if (cact=3) mth12=-1. compute mth13=b7stmo10. if (b7act10=3) mth13=-1. compute mth14=b7stmo09. if (b7act9=3) mth14=-1. 12

14 compute mth15=b7stmo08. if (b7act8=3) mth15=-1. compute mth16=b7stmo07. if (b7act7=3) mth16=-1. compute mth17=b7stmo06. if (b7act6=3) mth17=-1. compute mth18=b7stmo05. if (b7act5=3) mth18=-1. compute mth19=b7stmo04. if (b7act4=3) mth19=-1. compute mth20=b7stmo03. if (b7act3=3) mth20=-1. compute mth21=b7stmo02. if (b7act2=3) mth21=-1. compute mth22=b7stmo01. if (b7act1=3) mth22=-1. compute mth23=b7cstamo. if (b7cact=3) mth23=-1. execute. variable label mth1 "bcs 26 - start month job one". variable label mth2 "bcs 30 - start month job ten". variable label mth3 "bcs 30 - start month job nine". variable label mth4 "bcs 30 - start month job eight". variable label mth5 "bcs 30 - start month job seven". variable label mth6 "bcs 30 - start month job six". variable label mth7 "bcs 30 - start month job five". variable label mth8 "bcs 30 - start month job four". variable label mth9 "bcs 30 - start month job three". variable label mth10 "bcs 30 - start month job two". variable label mth11 "bcs 30 - start month job one". variable label mth12 "bcs 30 - start month job current". variable label mth13 "bcs 34 - start month job ten". variable label mth14 "bcs 34 - start month job nine". variable label mth15 "bcs 34 - start month job eight". variable label mth16 "bcs 34 - start month job seven". variable label mth17 "bcs 34 - start month job six". variable label mth18 "bcs 34 - start month job five". variable label mth19 "bcs 34 - start month job four". variable label mth20 "bcs 34 - start month job three". variable label mth21 "bcs 34 - start month job two". variable label mth22 "bcs 34 - start month job one". variable label mth23 "bcs 34 - start month job current". value labels mth1 mth2 mth3 mth4 mth5 mth6 mth7 mth8 mth9 mth10 mth11 mth12 mth13 mth14 mth15 mth16 mth17 mth18 mth19 mth20 mth21 mth22 mth23 0 'doing same job' -8 'don t know' -1 'not applicable'. missing values mth1 mth2 mth3 mth4 mth5 mth6 mth7 mth8 mth9 mth10 mth11 mth12 mth13 mth14 mth15 mth16 mth17 mth18 mth19 mth20 mth21 mth22 mth23 startm10 startmo9 startmo8 startmo7 startmo6 startmo5 startmo4 startmo3 startmo2 startmo cstartmo b7stmo10 b7stmo09 b7stmo08 b7stmo07 b7stmo06 b7stmo05 b7stmo04 b7stmo03 b7stmo02 b7stmo01 b7cstamo (-9 thru -1). recode mth1 mth2 mth3 mth4 mth5 mth6 mth7 mth8 mth9 mth10 mth11 mth12 mth13 mth14 mth15 mth16 mth17 mth18 mth19 mth20 mth21 mth22 mth23 (sysmis=-5) (else=copy). add value labels mth1 mth2 mth3 mth4 mth5 mth6 mth7 mth8 mth9 mth10 mth11 mth12 mth13 mth14 mth15 mth16 mth17 mth18 mth19 mth20 mth21 mth22 mth23-5 'system missing'. 13

15 missing values mth1 mth2 mth3 mth4 mth5 mth6 mth7 mth8 mth9 mth10 mth11 mth12 mth13 mth14 mth15 mth16 mth17 mth18 mth19 mth20 mth21 mth22 mth23 (-9 thru -1). exe. **************************************************************************************************************** *LOAD IN START YEARS FOR EACH JOB SPELL, IGNORING NON-EMPLOYMENT SPELLS* **************************************************************************************************************** *recode format of bcs 26 current job spell*** missing values b (). compute b26styr=b if (b960270>0) b26styr=1900+b if (b960270=-7 b960270=-3) b26styr=-8. if (b960270=-5) b26styr=-5. variable labels b26styr "BCS 26 Start year of current job". value labels b26styr -8 'don t know' -5 'system missing'. execute. missing values b26styr (-9 thru -1). freq b26styr. recode b26styr starty10 startyr9 startyr8 startyr7 startyr6 startyr5 startyr4 startyr3 startyr2 startyr cstartyr b7styr10 b7styr09 b7styr08 b7styr07 b7styr06 b7styr05 b7styr04 b7styr03 b7styr02 b7styr01 b7cstayr (- 5=-1) (sysmis=-5) (9998=-8) (9999=-9) (else=copy). missing values b26styr starty10 startyr9 startyr8 startyr7 startyr6 startyr5 startyr4 startyr3 startyr2 startyr cstartyr b7styr10 b7styr09 b7styr08 b7styr07 b7styr06 b7styr05 b7styr04 b7styr03 b7styr02 b7styr01 b7cstayr (). vector jobs(23). loop #e=1 to compute jobs(#e)=0. compute jobs1=b26styr. if (b26act>=3) jobs1=-1. if (b26act<0) jobs1=-1. compute jobs2=starty10. if (act10=3) jobs2=-1. compute jobs3=startyr9. if (act9=3) jobs3=-1. compute jobs4=startyr8. if (act8=3) jobs4=-1. compute jobs5=startyr7. if (act7=3) jobs5=-1. compute jobs6=startyr6. if (act6=3) jobs6=-1. compute jobs7=startyr5. if (act5=3) jobs7=-1. compute jobs8=startyr4. if (act4=3) jobs8=-1. compute jobs9=startyr3. if (act3=3) jobs9=-1. compute jobs10=startyr2. 14

16 if (act2=3) jobs10=-1. compute jobs11=startyr. if (act1=3) jobs11=-1. compute jobs12=cstartyr. if (cact=3) jobs12=-1. compute jobs13=b7styr10. if (b7act10=3) jobs13=-1. compute jobs14=b7styr09. if (b7act9=3) jobs14=-1. compute jobs15=b7styr08. if (b7act8=3) jobs15=-1. compute jobs16=b7styr07. if (b7act7=3) jobs16=-1. compute jobs17=b7styr06. if (b7act6=3) jobs17=-1. compute jobs18=b7styr05. if (b7act5=3) jobs18=-1. compute jobs19=b7styr04. if (b7act4=3) jobs19=-1. compute jobs20=b7styr03. if (b7act3=3) jobs20=-1. compute jobs21=b7styr02. if (b7act2=3) jobs21=-1. compute jobs22=b7styr01. if (b7act1=3) jobs22=-1. compute jobs23=b7cstayr. if (b7cact=3) jobs23=-1. execute. variable label jobs1 "bcs 26 - start year job one". variable label jobs2 "bcs 30 - start year job ten". variable label jobs3 "bcs 30 - start year job nine". variable label jobs4 "bcs 30 - start year job eight". variable label jobs5 "bcs 30 - start year job seven". variable label jobs6 "bcs 30 - start year job six". variable label jobs7 "bcs 30 - start year job five". variable label jobs8 "bcs 30 - start year job four". variable label jobs9 "bcs 30 - start year job three". variable label jobs10 "bcs 30 - start year job two". variable label jobs11 "bcs 30 - start year job one". variable label jobs12 "bcs 30 - start year job current at interview". variable label jobs13 "bcs 34 - start year job ten". variable label jobs14 "bcs 34 - start year job nine". variable label jobs15 "bcs 34 - start year job eight". variable label jobs16 "bcs 34 - start year job seven". variable label jobs17 "bcs 34 - start year job six". variable label jobs18 "bcs 34 - start year job five". variable label jobs19 "bcs 34 - start year job four". variable label jobs20 "bcs 34 - start year job three". variable label jobs21 "bcs 34 - start year job two". variable label jobs22 "bcs 34 - start year job one". variable label jobs23 "bcs 34 - start year job current at interview". 15

17 value labels jobs1 jobs2 jobs3 jobs4 jobs5 jobs6 jobs7 jobs8 jobs9 jobs10 jobs11 jobs12 jobs13 jobs14 jobs15 jobs16 jobs17 jobs18 jobs19 jobs20 jobs21 jobs22 jobs23 0 'doing same job' -8 'don t know' -2 'not answered' -1 'not applicable'. missing values jobs1 jobs2 jobs3 jobs4 jobs5 jobs6 jobs7 jobs8 jobs9 jobs10 jobs11 jobs12 jobs13 jobs14 jobs15 jobs16 jobs17 jobs18 jobs19 jobs20 jobs21 jobs22 jobs23 b26styr starty10 startyr9 startyr8 startyr7 startyr6 startyr5 startyr4 startyr3 startyr2 startyr cstartyr b7styr10 b7styr09 b7styr08 b7styr07 b7styr06 b7styr05 b7styr04 b7styr03 b7styr02 b7styr01 b7cstayr (-9 thru -1). recode jobs1 jobs2 jobs3 jobs4 jobs5 jobs6 jobs7 jobs8 jobs9 jobs10 jobs11 jobs12 jobs13 jobs14 jobs15 jobs16 jobs17 jobs18 jobs19 jobs20 jobs21 jobs22 jobs23 (sysmis=-5) (else=copy). add value labels jobs1 jobs2 jobs3 jobs4 jobs5 jobs6 jobs7 jobs8 jobs9 jobs10 jobs11 jobs12 jobs13 jobs14 jobs15 jobs16 jobs17 jobs18 jobs19 jobs20 jobs21 jobs22 jobs23-5 'system missing'. execute. **LOADS THE JOB STATUS - part time or full time** vector stat(23). loop #e=1 to compute stat(#e)=0. compute stat1=b26act. if (b26act>=3) stat1=-1. if (b26act<0) stat1=-1. compute stat2=act10. if (act10=3) stat2=-1. compute stat3=act9. if (act9=3) stat3=-1. compute stat4=act8. if (act8=3) stat4=-1. compute stat5=act7. if (act7=3) stat5=-1. compute stat6=act6. if (act6=3) stat6=-1. compute stat7=act5. if (act5=3) stat7=-1. compute stat8=act4. if (act4=3) stat8=-1. compute stat9=act3. if (act3=3) stat9=-1. compute stat10=act2. if (act2=3) stat10=-1. compute stat11=act1. if (act1=3) stat11=-1. compute stat12=cact. if (cact=3) stat12=-1. compute stat13=b7act10. if (b7act10=3) stat13=-1. compute stat14=b7act9. if (b7act9=3) stat14=-1. 16

18 compute stat15=b7act8. if (b7act8=3) stat15=-1. compute stat16=b7act7. if (b7act7=3) stat16=-1. compute stat17=b7act6. if (b7act6=3) stat17=-1. compute stat18=b7act5. if (b7act5=3) stat18=-1. compute stat19=b7act4. if (b7act4=3) stat19=-1. compute stat20=b7act3. if (b7act3=3) stat20=-1. compute stat21=b7act2. if (b7act2=3) stat21=-1. compute stat22=b7act1. if (b7act1=3) stat22=-1. compute stat23=b7cact. if (b7cact=3) stat23=-1. execute. variable label stat1 "bcs 26 - job status one". variable label stat2 "bcs 30 - job status ten". variable label stat3 "bcs 30- job status nine". variable label stat4 "bcs 30 - job status eight". variable label stat5 "bcs 30 - job status seven". variable label stat6 "bcs 30 - job status six". variable label stat7 "bcs 30 - job status five". variable label stat8 "bcs 30 - job status four". variable label stat9 "bcs 30 - job status three". variable label stat10 "bcs 30 - job status two". variable label stat11 "bcs 30 - job status one". variable label stat12 "bcs 30 - job status current at interview". variable label stat13 "bcs 34 - job status ten". variable label stat14 "bcs 34 - job status nine". variable label stat15 "bcs 34 - job status eight". variable label stat16 "bcs 34 - job status seven". variable label stat17 "bcs 34 - job status six". variable label stat18 "bcs 34 - job status five". variable label stat19 "bcs 34 - job status four". variable label stat20 "bcs 34 - job status three". variable label stat21 "bcs 34 - job status two". variable label stat22 "bcs 34 - job status one". variable label stat23 "bcs 34 - job status current at interview". value labels stat1 stat2 stat3 stat4 stat5 stat6 stat7 stat8 stat9 stat10 stat11 stat12 stat13 stat14 stat15 stat16 stat17 stat18 stat19 stat20 stat21 stat22 stat23-8 'don t know' -1 'not applicable' 1 'Full time' 2 'Part time'. missing values stat1 stat2 stat3 stat4 stat5 stat6 stat7 stat8 stat9 stat10 stat11 stat12 stat13 stat14 stat15 stat16 stat17 stat18 stat19 stat20 stat21 stat22 stat23 (-9 thru -1). recode stat1 stat2 stat3 stat4 stat5 stat6 stat7 stat8 stat9 stat10 stat11 stat12 stat13 stat14 stat15 stat16 stat17 stat18 stat19 stat20 stat21 stat22 stat23 (sysmis=-5) (else=copy). add value labels stat1 stat2 stat3 stat4 stat5 stat6 stat7 stat8 stat9 stat10 stat11 stat12 stat13 stat14 stat15 stat16 stat17 stat18 stat19 stat20 stat21 stat22 stat23-5 'system missing'. 17

19 execute. ******************************************************************************************************************** *DEFINE END MONTHS FOR JOBS - THESE WERE NOT COLLECTED, THEREFORE AN END JOB WILL BE SET TO THE DATE DEFINED IN THE FOLLOWING SPELL - THIS DATE WILL BE WHERE RESPONDENTS HAD A NEW JOB OR WHERE RESPONDENTS INDICATED THAT THEY HAD A SPELL OF NON-EMPLOYMENT* ******************************************************************************************************************** missing values startmo9 startmo8 startmo7 startmo6 startmo5 startmo4 startmo3 startmo2 startyr cstartmo b7stmo10 b7stmo09 b7stmo08 b7stmo07 b7stmo06 b7stmo05 b7stmo04 b7stmo03 b7stmo02 b7stmo01 b7cstamo (). vector mend(23). loop #e=1 to compute mend(#e)=0. compute mend1=startm10. compute mend2=startmo9. compute mend3=startmo8. compute mend4=startmo7. compute mend5=startmo6. compute mend6=startmo5. compute mend7=startmo4. compute mend8=startmo3. compute mend9=startmo2. compute mend10=startmo. compute mend11=cstartmo. compute mend12=b7stmo10. compute mend13=b7stmo09. compute mend14=b7stmo08. compute mend15=b7stmo07. compute mend16=b7stmo06. compute mend17=b7stmo05. compute mend18=b7stmo04. compute mend19=b7stmo03. compute mend20=b7stmo02. compute mend21=b7stmo01. compute mend22=b7cstamo. compute mend23=-1. execute. variable label mend1 "bcs 26 - end month job one". variable label mend2 "bcs 30 - end month job ten". variable label mend3 "bcs 30 - end month job nine". variable label mend4 "bcs 30 - end month job eight". variable label mend5 "bcs 30 - end month job seven". variable label mend6 "bcs 30 - end month job six". variable label mend7 "bcs 30- end month job five". variable label mend8 "bcs 30 - end month job four". variable label mend9 "bcs 30 - end month job three". variable label mend10 "bcs 30 - end month job two". variable label mend11 "bcs 30 - end month job one". variable label mend12 "bcs 30 - end month job current at interview". variable label mend13 "bcs 34 - end month job ten". variable label mend14 "bcs 34 - end month job nine". variable label mend15 "bcs 34 - end month job eight". variable label mend16 "bcs 34 - end month job seven". variable label mend17 "bcs 34 - end month job six". variable label mend18 "bcs 34 - end month job five". variable label mend19 "bcs 34 - end month job four". variable label mend20 "bcs 34 - end month job three". variable label mend21 "bcs 34 - end month job two". variable label mend22 "bcs 34 - end month job one". 18

20 variable label mend23 "bcs 34 - end month job current at interview". value labels mend1 mend2 mend3 mend4 mend5 mend6 mend7 mend8 mend9 mend10 mend11 mend12 mend13 mend14 mend15 mend16 mend17 mend18 mend19 mend20 mend21 mend22 mend23 0 'doing same job' -8 'don t know' -5 'system missing' -2 'not answered' -1 'not applicable'. missing values mend1 mend2 mend3 mend4 mend5 mend6 mend7 mend8 mend9 mend10 mend11 mend12 mend13 mend14 mend15 mend16 mend17 mend18 mend19 mend20 mend21 mend22 mend23 startmo9 startmo8 startmo7 startmo6 startmo5 startmo4 startmo3 startmo2 startyr cstartmo b7stmo10 b7stmo09 b7stmo08 b7stmo07 b7stmo06 b7stmo05 b7stmo04 b7stmo03 b7stmo02 b7stmo01 b7cstamo (-9 thru -1). recode mend1 mend2 mend3 mend4 mend5 mend6 mend7 mend8 mend9 mend10 mend11 mend12 mend13 mend14 mend15 mend16 mend17 mend18 mend19 mend20 mend21 mend22 mend23 (sysmis=-5) (else=copy). add value labels mend1 mend2 mend3 mend4 mend5 mend6 mend7 mend8 mend9 mend10 mend11 mend12 mend13 mend14 mend15 mend16 mend17 mend18 mend19 mend20 mend21 mend22 mend23-5 'system missing'. execute. **LOADS THE END YEAR VALUES** missing values startyr9 startyr8 startyr7 startyr6 startyr5 startyr4 startyr3 startyr2 startyr cstartyr b7styr10 b7styr09 b7styr08 b7styr07 b7styr06 b7styr05 b7styr04 b7styr03 b7styr02 b7styr01 b7cstayr (). vector endy(23). loop #e=1 to compute endy(#e)=0. compute endy1=starty10. compute endy2=startyr9. compute endy3=startyr8. compute endy4=startyr7. compute endy5=startyr6. compute endy6=startyr5. compute endy7=startyr4. compute endy8=startyr3. compute endy9=startyr2. compute endy10=startyr. compute endy11=cstartyr. compute endy12=b7styr10. compute endy13=b7styr09. compute endy14=b7styr08. compute endy15=b7styr07. compute endy16=b7styr06. compute endy17=b7styr05. compute endy18=b7styr04. compute endy19=b7styr03. compute endy20=b7styr02. compute endy21=b7styr01. compute endy22=b7cstayr. compute endy23=-1. execute. variable label endy1 "bcs 26 - end year job one". variable label endy2 "bcs 30 - end year job ten". variable label endy3 "bcs 30 - end year job nine". variable label endy4 "bcs 30 - end year job eight". variable label endy5 "bcs 30 - end year job seven". 19

21 variable label endy6 "bcs 30 - end year job six". variable label endy7 "bcs 30 - end year job five". variable label endy8 "bcs 30 - end year job four". variable label endy9 "bcs 30 - end year job three". variable label endy10 "bcs 30 - end year job two". variable label endy11 "bcs 30 - end year job one". variable label endy12 "bcs 34 - end year job current at interview". variable label endy13 "bcs 34 - end year job ten". variable label endy14 "bcs 34 - end year job nine". variable label endy15 "bcs 34 - end year job eight". variable label endy16 "bcs 34 - end year job seven". variable label endy17 "bcs 34 - end year job six". variable label endy18 "bcs 34 - end year job five". variable label endy19 "bcs 34 - end year job four". variable label endy20 "bcs 34 - end year job three". variable label endy21 "bcs 34 - end year job two". variable label endy22 "bcs 34 - end year job one". variable label endy23 "bcs 34 - end year job current at intervew". value labels endy1 endy2 endy3 endy4 endy5 endy6 endy7 endy8 endy9 endy10 endy11 endy12 endy13 endy14 endy15 endy16 endy17 endy18 endy19 endy20 endy21 endy22 endy23 0 'doing same job' -8 'don t know' -5 'system missing' -2 'not answered' -1 'not applicable'. missing values endy1 endy2 endy3 endy4 endy5 endy6 endy7 endy8 endy9 endy10 endy11 endy12 endy13 endy14 endy15 endy16 endy17 endy18 endy19 endy20 endy21 endy22 endy23 startyr9 startyr8 startyr7 startyr6 startyr5 startyr4 startyr3 startyr2 startyr cstartyr b7styr10 b7styr09 b7styr08 b7styr07 b7styr06 b7styr05 b7styr04 b7styr03 b7styr02 b7styr01 b7cstayr (-9 thru -1). recode endy1 endy2 endy3 endy4 endy5 endy6 endy7 endy8 endy9 endy10 endy11 endy12 endy13 endy14 endy15 endy16 endy17 endy18 endy19 endy20 endy21 endy22 endy23 (sysmis=-5) (else=copy). add value labels endy1 endy2 endy3 endy4 endy5 endy6 endy7 endy8 endy9 endy10 endy11 endy12 endy13 endy14 endy15 endy16 endy17 endy18 endy19 endy20 endy21 endy22 endy23-5 'system missing'. execute. **LOADS THE OCCUPATION CATEGORIES ALL USE SOC90 CLASSIFICATIONS** recode b26soc (1 thru 99=-8) (-99=-9) (else=copy). missing values b26soc soc soc2 soc3 soc4 soc5 soc6 soc7 soc8 soc9 soc10 soc11 b7soc910 b7soc909 b7soc908 b7soc907 b7soc906 b7soc905 b7soc904 b7soc903 b7soc902 b7soc91 b7soc90 (). vector occ(23). loop #e=1 to compute occ(#e)=0. compute occ1=b26soc. if (b26act>=3) occ1=-1. if (b26act<0) occ1=-1. compute occ2=soc11. if (act10=3) occ2=-1. compute occ3=soc10. if (act9=3) occ3=-1. compute occ4=soc9. if (act8=3) occ4=-1. compute occ5=soc8. if (act7=3) occ5=-1. 20

22 compute occ6=soc7. if (act6=3) occ6=-1. compute occ7=soc6. if (act5=3) occ7=-1. compute occ8=soc5. if (act4=3) occ8=-1. compute occ9=soc4. if (act3=3) occ9=-1. compute occ10=soc3. if (act2=3) occ10=-1. compute occ11=soc2. if (act1=3) occ11=-1. compute occ12=soc. if (cact=3) occ12=-1. compute occ13=b7soc910. if (b7act10=3) occ13=-1. compute occ14=b7soc909. if (b7act9=3) occ14=-1. compute occ15=b7soc908. if (b7act8=3) occ15=-1. compute occ16=b7soc907. if (b7act7=3) occ16=-1. compute occ17=b7soc906. if (b7act6=3) occ17=-1. compute occ18=b7soc905. if (b7act5=3) occ18=-1. compute occ19=b7soc904. if (b7act4=3) occ19=-1. compute occ20=b7soc903. if (b7act3=3) occ20=-1. compute occ21=b7soc902. if (b7act2=3) occ21=-1. compute occ22=b7soc91. if (b7act1=3) occ22=-1. compute occ23=b7soc90. if (b7cact=3) occ23=-1. execute. variable label occ1 "bcs 26 - occupation one". variable label occ2 "bcs 30 - occupation ten". variable label occ3 "bcs 30 - occupation nine". variable label occ4 "bcs 30 - occupation eight". variable label occ5 "bcs 30 - occupation seven". variable label occ6 "bcs 30 - occupation six". variable label occ7 "bcs 30 - occupation five". variable label occ8 "bcs 30 - occupation four". variable label occ9 "bcs 30 - occupation three". variable label occ10 "bcs 30 - occupation two". 21

23 variable label occ11 "bcs 30 - occupation one". variable label occ12 "bcs 30 - occupation current at interview". variable label occ13 "bcs 34 - occupation ten". variable label occ14 "bcs 34 - occupation nine". variable label occ15 "bcs 34 - occupation eight". variable label occ16 "bcs 34 - occupation seven". variable label occ17 "bcs 34 - occupation six". variable label occ18 "bcs 34 - occupation five". variable label occ19 "bcs 34 - occupation four". variable label occ20 "bcs 34 - occupation three". variable label occ21 "bcs 34 - occupation two". variable label occ22 "bcs 34 - occupation one". variable label occ23 "bcs 34 - occupation current at interview". value labels occ1 occ2 occ3 occ4 occ5 occ6 occ7 occ8 occ9 occ10 occ11 occ12 occ13 occ14 occ15 occ16 occ17 occ18 occ19 occ20 occ21 occ22 occ23-8 'don t know' -1 'not applicable'. missing values occ1 occ2 occ3 occ4 occ5 occ6 occ7 occ8 occ9 occ10 occ11 occ12 occ13 occ14 occ15 occ16 occ17 occ18 occ19 occ20 occ21 occ22 occ23 b26soc soc soc2 soc3 soc4 soc5 soc6 soc7 soc8 soc9 soc10 soc11 b7soc910 b7soc909 b7soc908 b7soc907 b7soc906 b7soc905 b7soc904 b7soc903 b7soc902 b7soc91 b7soc90 (-9 thru -1). recode occ1 occ2 occ3 occ4 occ5 occ6 occ7 occ8 occ9 occ10 occ11 occ12 occ13 occ14 occ15 occ16 occ17 occ18 occ19 occ20 occ21 occ22 occ23 (sysmis=-5) (else=copy). add value labels occ1 occ2 occ3 occ4 occ5 occ6 occ7 occ8 occ9 occ10 occ11 occ12 occ13 occ14 occ15 occ16 occ17 occ18 occ19 occ20 occ21 occ22 occ23-8 'don t know' -5 'system missing'. execute. ****************************************************************************************************************** Need to recode all the system missing values now. Cases that are system missing because only entered one or two of the surveys are set to -33. Not applicable, refusals and Don t Know (only approx 100 cases had don t know values) values are set to -22. ********************************************************************************************************************* recode mth1 mth2 mth3 mth4 mth5 mth6 mth7 mth8 mth9 mth10 mth11 mth12 mth13 mth14 mth15 mth16 mth17 mth18 mth19 mth20 mth21 mth22 mth23 jobs1 jobs2 jobs3 jobs4 jobs5 jobs6 jobs7 jobs8 jobs9 jobs10 jobs11 jobs12 jobs13 jobs14 jobs15 jobs16 jobs17 jobs18 jobs19 jobs20 jobs21 jobs22 jobs23 stat1 stat2 stat3 stat4 stat5 stat6 stat7 stat8 stat9 stat10 stat11 stat12 stat13 stat14 stat15 stat16 stat17 stat18 stat19 stat20 stat21 stat22 stat23 mend1 mend2 mend3 mend4 mend5 mend6 mend7 mend8 mend9 mend10 mend11 mend12 mend13 mend14 mend15 mend16 mend17 mend18 mend19 mend20 mend21 mend22 mend23 endy1 endy2 endy3 endy4 endy5 endy6 endy7 endy8 endy9 endy10 endy11 endy12 endy13 endy14 endy15 endy16 endy17 endy18 endy19 endy20 endy21 endy22 endy23 occ1 occ2 occ3 occ4 occ5 occ6 occ7 occ8 occ9 occ10 occ11 occ12 occ13 occ14 occ15 occ16 occ17 occ18 occ19 occ20 occ21 occ22 occ23 (-5=-33) (-99, -9, -1=-22) (9998, -8=-8). add value labels mth1 mth2 mth3 mth4 mth5 mth6 mth7 mth8 mth9 mth10 mth11 mth12 mth13 mth14 mth15 mth16 mth17 mth18 mth19 mth20 mth21 mth22 mth23 jobs1 jobs2 jobs3 jobs4 jobs5 jobs6 jobs7 jobs8 jobs9 jobs10 jobs11 jobs12 jobs13 jobs14 jobs15 jobs16 jobs17 jobs18 jobs19 jobs20 jobs21 jobs22 jobs23 stat1 stat2 stat3 stat4 stat5 stat6 stat7 stat8 stat9 stat10 stat11 stat12 stat13 stat14 stat15 stat16 stat17 stat18 stat19 stat20 stat21 stat22 stat23 mend1 mend2 mend3 mend4 mend5 mend6 mend7 mend8 mend9 mend10 mend11 mend12 mend13 mend14 mend15 mend16 mend17 mend18 mend19 mend20 mend21 mend22 mend23 endy1 endy2 endy3 endy4 endy5 endy6 endy7 endy8 endy9 endy10 endy11 endy12 endy13 endy14 endy15 endy16 endy17 endy18 endy19 endy20 endy21 endy22 endy23 occ1 occ2 occ3 occ4 occ5 occ6 occ7 occ8 occ9 occ10 occ11 occ12 occ13 occ14 occ15 occ16 occ17 occ18 occ19 occ20 occ21 occ22 occ23-8 'don t know - value can be recoded on dates' -22 'na and refusals' -33 'system missing'. execute. ******************************************* *SAVE FILE* ****************************************** 22

24 SAVE OUTFILE='F:\Work Kelly\Occupational mobility work\bcs DATA\Employment History\BCS Employment History.sav' /COMPRESSED. 23

25 APPENDIX THREE: Restructuring the data file ******************************************************************************************************** Need to turn the missing values off so that the vectors do not pick up the -33 and -22 values these are now turned to -1. ******************************************************************************************************** **NOW REMOVE MISSING FROM START MONTH AND YEAR VARIABLES** recode mth1 mth2 mth3 mth4 mth5 mth6 mth7 mth8 mth9 mth10 mth11 mth12 mth13 mth14 mth15 mth16 mth17 mth18 mth19 mth20 mth21 mth22 mth23 jobs1 jobs2 jobs3 jobs4 jobs5 jobs6 jobs7 jobs8 jobs9 jobs10 jobs11 jobs12 jobs13 jobs14 jobs15 jobs16 jobs17 jobs18 jobs19 jobs20 jobs21 jobs22 jobs23 stat1 stat2 stat3 stat4 stat5 stat6 stat7 stat8 stat9 stat10 stat11 stat12 stat13 stat14 stat15 stat16 stat17 stat18 stat19 stat20 stat21 stat22 stat23 mend1 mend2 mend3 mend4 mend5 mend6 mend7 mend8 mend9 mend10 mend11 mend12 mend13 mend14 mend15 mend16 mend17 mend18 mend19 mend20 mend21 mend22 mend23 endy1 endy2 endy3 endy4 endy5 endy6 endy7 endy8 endy9 endy10 endy11 endy12 endy13 endy14 endy15 endy16 endy17 endy18 endy19 endy20 endy21 endy22 endy23 occ1 occ2 occ3 occ4 occ5 occ6 occ7 occ8 occ9 occ10 occ11 occ12 occ13 occ14 occ15 occ16 occ17 occ18 occ19 occ20 occ21 occ22 occ23 (-8, -33, -22=-1) (else=copy). exe. missing values mth1 mth2 mth3 mth4 mth5 mth6 mth7 mth8 mth9 mth10 mth11 mth12 mth13 mth14 mth15 mth16 mth17 mth18 mth19 mth20 mth21 mth22 mth23 jobs1 jobs2 jobs3 jobs4 jobs5 jobs6 jobs7 jobs8 jobs9 jobs10 jobs11 jobs12 jobs13 jobs14 jobs15 jobs16 jobs17 jobs18 jobs19 jobs20 jobs21 jobs22 jobs23 stat1 stat2 stat3 stat4 stat5 stat6 stat7 stat8 stat9 stat10 stat11 stat12 stat13 stat14 stat15 stat16 stat17 stat18 stat19 stat20 stat21 stat22 stat23 mend1 mend2 mend3 mend4 mend5 mend6 mend7 mend8 mend9 mend10 mend11 mend12 mend13 mend14 mend15 mend16 mend17 mend18 mend19 mend20 mend21 mend22 mend23 endy1 endy2 endy3 endy4 endy5 endy6 endy7 endy8 endy9 endy10 endy11 endy12 endy13 endy14 endy15 endy16 endy17 endy18 endy19 endy20 endy21 endy22 endy23 occ1 occ2 occ3 occ4 occ5 occ6 occ7 occ8 occ9 occ10 occ11 occ12 occ13 occ14 occ15 occ16 occ17 occ18 occ19 occ20 occ21 occ22 occ23 (). ********************************************************************************************************************* These first sets of vectors use the information loaded from Appendix one and remove any missing cells from each job spell to order all spells one after the other. ********************************************************************************************************************** vector stomth(23). vector stojob(23). vector stostat(23). vector stomend(23). vector stoendy(23). vector stoocc(23). vector jobsmt=mth1 to mth23. vector jobsyr=jobs1 to jobs23. vector jobstat=stat1 to stat23. vector jobmend=mend1 to mend23. vector jobendy=endy1 to endy23. vector jobocc=occ1 to occ23. loop #B=1 to compute stomth(#b)=-1. + compute stojob(#b)=-1. + compute stostat(#b)=-1. + compute stomend(#b)=-1. + compute stoendy(#b)=-1. + compute stoocc(#b)=-1. compute #J=1. loop #I=1 to 23. do if (jobsyr(#i)>0). + compute stojob(#j)=jobsyr(#i). + compute stomth(#j)=jobsmt(#i). + compute stostat(#j)=jobstat(#i). 24