home STUDY DESIGN STUDY DESIGN
STUDY DESIGN

STUDY DESIGN

The KLIPS sample is an equal probability sample of households from 7 metropolitan cities and urban areas in 8 provinces (excluding Cheju Island) and was designed to interview 5000 households and their members (aged 15 and over).

The final number of individuals successfully interviewed during the 1st Wave (1998) was 13,321.


The KLIPS sample is an equal probability sample of households from 7 metropolitan cities and urban areas in 8 provinces (excluding Cheju Island) and was designed to interview 5000 households and their members (aged 15 and over, excluding people in military service or facility residents). The final number of individuals that we successfully interviewed during the 1st Wave (1998) was 13,321.

These 5000 households and their members make up the original panel households and household members. To sample primary sampling units and then the households therein, KLIPS used the sampling frame for the 1995 Korea (mid-term) Census. Out of the Census unit area (total = 21,675), a total of 951 sampling unit areas were finally selected, and from each sampling unit, 5-6 households were randomly selected. The households and members thus selected are designed to represent the adult population (aged 15 and over) residing in urban areas.

Follow-up Rules: Starting from the 2nd Wave (1999), individuals that come to form blood and economic ties to the original panel members are added to the original sample. For example, if a panel member gets married and forms an independent household with his/her spouse, then, his/her spouse becomes a ‘new respondent’ to the original panel and the couple is followed and interviewed thereafter. On the other hand, if one of the panel members moves into a non-member household, through getting married for instance, he/she is also followed and his/her spouse’s household members are interviewed. In this way, the size of the sample members grows and expands over the waves. This method of incorporating related households and individuals into the sample allows the expanded sample to reflect on-going demographic dynamics of the original panel members. When a panel member moves out of the original household, through divorce for instance, he/she is also tracked as long as he/she lives with his/her children.

Census PSU and PSU sampled into KLIPS

City, Province Census PSU KLIPS PSU
Households
Seoul 5,186 227 227
Busan 1,841 176 97
Daegu 1,212 155 64
Inchon 1,116 150 59
Gwangju 607 106 32
Daejun 592 101 31
Ulsan 456 70 24
Gyunggi-do 3,624 280 167
Kangwon-do 725 165 26
Choongcheongbuk-do 694 159 22
Choongcheongnam-do 865 122 25
Jeollabuk-do 967 170 40
Jeollanam-do 996 152 25
Gyeongsangbuk-do 1,372 129 54
Gyeongsangnam-do 1,422 220 58
Total 21,675 2,382 951

ORIGINAL SAMPLE AND ATTRITION RATES

Initial contact with the 5000 households originally sampled out of the Census PSU met with a success rate of about 75% (3,773 households), and the rest (25%, 1,227 households) were replaced by other randomly selected households in the same PSU.[1] The total number of individuals in the 5,000 interviewed households amounted to 17,505. As we do not interview household members 14 years old or younger, the total number of individuals personally interviewed in the base year (1998) was 13,321 (out of 13,738 persons). Among them, 4012 were paid workers, 2,415 were self-employed, and 7,315 were not employed at the time of survey.

Field work for KLIPS usually starts in May and continues up to September each year. However, in the base year, field work was delayed by a month (June to October). The majority (about 2/3) of interviews were finished by the end of August.

As mentioned above, KLIPS is based on an expanding sample. That is, we add new respondents to the original household sample, while following any original panel members who move out of the original household due to divorce, separation, etc. New respondents include non-family members who marry the family member and move into the original household, as well as the doubling up of non-panel family members who were once economically independent, etc. We call new independent households created by original panel members ‘branch households’ (Bunga Gagu). So far up to the 18th Wave, a total of 2,211 new branch households were created.

ADDITIONAL SAMPLING(2009)

To overcome the limitations posed by panel household attrition and urban concentration and improve the representativeness of the panel data, additional sampling was carried out in 2009. KLIPS used ‘two-stage cluster systematic sampling’, which is to sample firstly using unit area of the 2005 Korea Census as the sampling frame, then selecting random households within the area. In the first stage, unit areas were selected using a different method according to the administrative district. ‘Unit Area Characteristics Value(value calculated with the number of households and single-family households)’ is used to choose ‘–dong’ unit area and ‘Probabiltiy Proportionate Sampling(sampling proportionately to the household number of the unit area)’ is used for ‘-eup’ and ‘-myeon’ district. In the second stage, five households were randomly selected for each sample unit area chosen in the first stage. Total of 1,415 households was additionally sampled using this procedure.

98 SAMPLE AND CONSOLIDATED SAMPLE

KLIPS today has two panels in one dataset; '98 sample and Consolidated sample. The '98 sample contains data from the 5,000 original households starting from Wave 1 (1998). The consolidated sample contains additional households sampled in 2009. For the consolidated data, all the households that responded in 2009 (Wave 12) make up the original sample. Thus the consolidated sample consists of not only the 1,415 households additionally included in 2009, but also the existing households ('98 sample + branch households) who responded to Wave 12 survey, for a total of 5,306.

The consolidated sample is made up of 6,721 households as the 1st original panel, which includes the 3,657 out of the 5,000 original households that have been surveyed in 2009 as well as their 1,649 branch households and the additional sample of 1,415 households.