PROPWT is a 16 digit numeric variable.
PROPWT reports the proposed weight used when population and day combined weights need to be rescaled.
The main MTUS weight, PROPWT, is calculated as follows:
- Begin from the original survey weight (deflated if the original weight was inflated). If there is no original weight, a population weight is constructed by dividing the percentage of the population expected to be in each age/sex (and in some cases employment status) group by the percentage of cases actually in that age/sex/employment status category in the sample data (expected/achieved).
- Create a good-diary inflation factor by dividing the total number of diaries collected in the survey by the number of good diaries (total diaries/good diaries).
- Create an interim weight that starts with the same value of the weight in step 1. Second, set the value of this interim weight to 0 for all low-quality diaries (NOWGHT=0). Third, multiply the interim weight by the good diary inflation factor created in step 2 for all cases (it will stay 0 for the bad diaries).
- Compute ASEWT - a sum of the weights for each age/sex/employment status group (across all days of the week).
- Compute two further sums: ASEDayWT -- a sum of weights for each age/sex/employment status) group separately for each day of the week, and also: ASEDayCASE - the sum of the number of cases in each age/sex/employment status) group for each day of the week. (This step differs from step 4 as the sums are calculated separately for each day of the week).
- An expected sum of weights is calculated (ESW) for each day of the week: ESW = ASEWT / 7 (divided by 7 as there are 7 days of the week).
- PROPWT = (ESW / ASEDayWT) / (ASEDayCASE / ASEDayWT)
- PROPWT should have a mean of 1, and should sum to the total number of diary cases. When the weight has been applied, the weighted frequency of the diaries by the days of the week should be evenly distributed (14.3% for each day). Similarly, each age/sex group should have an even distribution of diaries across each day of the week.
Users who are interested in the original weight may refer to OCOMBWT.
Comparability — Index
This variable is comparable across years and countries when available.
Comparability — France [top]
In 1985, since the original weight is not available, the proposed weight, propwt, is created by using the population taken from the National Institute of Statistics and Economic Studies in France.
Comparability — South Africa [top]
In 2000, the computation of PROPWT differs slightly from other surveys; the original population weight first had to be divided by 10,000 (according to the South African data manual).
Comparability — United Kingdom [top]
In 1983, we calculate the weight as the expected percentage (from 1985 census) divided by the actual percentage by gender and age group.
Since some of the records have no age, in the calculation of PROPWT, we used the sum of weigh instead of sum of good case for adjustment.
In 1987, where an adhoc weight is constructed, the MTUS balances the distribution of the sample survey in relation to the age and sex distribution of the national sample. As people aged 62 and older are not representative of people aged 62 and older as a whole, these diarists are 0-weighted using the PROPWT. Users seeking to include all good quality diaries should use the OCOMBWT, but also should be aware that the older diarists in this survey represent only older people who live with people aged 20-61, not the whole population aged 62 and older.
- All persons.
- Austria: 1992
- Bulgaria: 2001
- Canada: 2010
- Finland: 1979, 2009
- France: 1985, 1998, 2009
- Hungary: 1999, 2009
- Israel: 1991
- Italy: 2002
- Netherlands: 1975, 1980, 1985, 1990, 1995, 2000, 2005
- South Africa: 2000
- Spain: 2002, 2009
- United Kingdom: 1974, 1983, 1987, 1995, 2000, 2005, 2014
- United States: 1965, 1975, 1985, 1993, 1995, 1998, 2003-2012