Readme File for UK 1995

README FileUK 1995 Survey Data

(World 5.5 – Release 3; World 5.8 & World 6.0 Release -1)

21 September 2009

Contents:

Technical description of the survey

Information on the original files

Time-use variables

Variable-specific information

TECHNICAL DESCRIPTION OF THE SURVEY

INFORMATION ON THE ORIGINAL FILES

TOTAL2.sav is theoriginal data file which contains the original demographic, episode and aggregate data. Episode information presented horizontally using ninety six 15 minute timeslots. A number of the original variables, including the weights, are not labelled in the data file, and no information on these variables is available in the documentation. We converted this file into SAS.

Time-Use Variables

Activity Variable Documentation for UK 1995

Assumptions:

The original activity code 10 Breaks meant breaks at main economic activity, thus is coded as AV1

There is only one code 7 Travel. If a travel time appeared before paid work or education, or a traveltime appeared after paid work or education, this time is calculated as AV5 (travel to/from work or education) and the rest is calculated as AV17 (leisure travel)

Code 29Nothing meant "doing nothing"

Variable-Specific Information

AGEKID2, ALONE, CARER, CHILD, CITIZEN, COMPUTER, DISAB, EMPINCLM, EMPSP, HEALTH, ICT, OAD, OCCUP, PARNTID1, PARNTID2, PARTID, RUSHED, SEC, SECTOR, SPPART, URBAN, WORKHRS

These variables cannot be constructed

CDAY, DAY, MONTH

The documentation indicates that the survey took place in May, yet the original variable month has 41 6’s (June). Nevertheless, some of these cases where month=6 have a calendar day of 31 (and while there are 31 days in May there are only 30 days in June). 3 diaries did not have a value for month. As the missing month and month set to June are most likely errors in the data, the value of MONTH for all cases is set to 5 (MAY).

The original variables date(calendar day) and desigday(day-of-the-week) are used accordingly as CDAY and DAY the diary was completed. There are 137 cases that the values of date and desigday do not correspond. In those cases, if date has a value, then DAY is calculated based on the 1995 calendar, else CDAY is left blank and leave DAY as is.

EDUCA

The original variable lefted(age completed full time education) is used as EDUCA with no conversion. leftedvalues and labels are as follows:

HHTYPE, HHLDSIZE, SINGPAR

There are still a small number of cases of which these three variables do not correspond to one and other. Logical adjustments have been made to correct the discrepancies of the obvious cases but some are left behind without adjustments.

By studying the distribution of the original household variables, a number of cases contain values that do not corresponded with one or the others. The table below lists these cases and their fixes. The fixes to q12, q16, q17, or q18are made based on some logical assumptions and are highlighted in red. Some of the cases that aredifficult to make any assumptions uponare left alone and they are indicated by a question mark (?) in the Assumption column.

INCORIG

The original variable q73 (gross income) is used as INCORIG with no conversion. q73 values and labels are as follows:

NCHILD, AGEKIDX, FAMSTAT

The original variables q13/q14/q15 are the number of children aged 0-4/5-10/11-15 accordingly. The number of co-resident children aged 16-18 are not captured in the survey. Therefore:

1) NCHILD includes the number of children up to aged 15, plus 1 if the diarist if aged between 16 and 18

2) AGEKIDX=2/3/4 are not precise

3) FAMSTAT=0,2,3 are based on information about children aged up to 15

PERSID & ID

These are taken from serialno. Examination of the serialno variable revealed that:

1)Two of the serialno numbers (16088249505 and 5016069505) were used twice for different cases; to fix these duplicates, 1 is added to these numbers and that produced serialno 16088249506 and 5016069506 for the duplicates.

2)27 cases have blank serialno and with 1440 minutes of sleep (code 1) but have almost full profile information; these cases are kept, are assigned a new serialno starting with a number 22000000000 + the row number, and are assigned as BADCASE=31.

PROPWT

wta is the original weight variable, but there is one good diary case (serialno=14071159505) with no original weight (wta=0). The Median wta (0.527909426) is calculated of a similar profile (same age, sex, marital status, and education) group and is assigned to this special case. After the SAS programme was written, propwt had one final adjustment,

propwt=propwt*(2005/1994.3)

1994.3 is the sum of weights prior to this final adjustment.

RELREFP

The original data did not provide the categories whether the relationship of the diarist to the household is a sibling or an in-law, thus only RELREFP=1/2/3/4/9/10 are possible.

RETIRED

99% of the cases with a missing value of q119 (last week activity) correspond to respondents who are employed or self-employed. In these cases, if their age is 65+, then set RETIRED=1, else set RETIRED = 0. Regardless, if theage is 70+ and not in paid work, then they are coded as RETIRED=1.

STUDENT

The original variable q116 (age left full education) has a value 97 to label ‘still in full time education’. This isused to compute STUDENT.