Q&A
In case personal information, advertising information or defamatory contents are included, your posting may be deleted.
Household survey sampling methodology
Answer
- WriterTommy Lee
- Date2026-04-02
- Views3820
- files
Hello, I am writing to ask about the sample design of the Household Income and Expenditure Survey (가계동향조사) across the period 2006–2025. Specifically, I would like to clarify the following: How are sample households ed for a given year? I understand the survey used a 36-month rotation sample through 2016, and shifted to a different rotation structure (6-6-6) from 2019 onward. Could you confirm how each design works in practice? Within a given year, should we expect the same household to appear across multiple quarters? For instance, in the pre-2019 design, it appears that most households are observed in all four quarters of a given year. Is this also the case under the post-2019 design? Across years, is there overlap in sampled households? That is, can the same household appear in consecutive years' surveys under the same household ID? I am asking because I am working with the microdata for 2006, 2015, and 2025, and want to ensure I correctly understand the panel structure (or lack thereof) when constructing annual household-level averages. Thank you for your time.
Answer
Ministry of Data and Statistics/2026-04-07
Dear Tommy Lee, Thank you for visiting the KOSIS website. Regarding your inquiry, the Short-Term Household Income and Expenditure Statistics Division provided information as follows; The followings are commonly applied to all sampling designs. 1. When a specific household is selected as a sample even in a month of a quarter, data on this household are included in a given quarter. 2. In addition to regular sample rotations, a household may be included to or excluded from a sample when the household migrates owing to its own reasons. 3. The two sampling methods always have households that are newly included to a sample and excluded from a sample every month. In a 36-month rotation sample, each sample household is covered for 36 months. ① When a household that is newly included to a sample in January 2013, it will be covered until December 2015. In other words, data on this same household are included for all four quarters for the 2013-2015 period. ② When a household that is newly included to a sample in February 2013, it will be covered until January 2016, this same household is covered for all four quarters of three years. In other words, data on this same household are included from all four quarter of 2023 to the first quarter of 2016. Therefore, the same household can be included in the sample for 36 months consecutively. According to the time point of the entry to the survey, the household in the ① example can remain in the sample for three years (12 quarters consecutively). The household in the ② example can remain in the sample for four years (13 quarters consecutively. In a 6-6-6 rotation sampling, a household can be included for 6 months, excluded for the next 6 months, and then included for the next 6 months. ① When a household that is newly included to a sample in January 2026, it will be covered for the January-June period of 2026 and the January-June period of 2027. In other words, data on this same household are included for the first and second quarters of 2026 and for the first and second quarters of 2027. ② When a household that is newly included to a sample in February 2026, it will be covered for the February-July period of 2026 and the February-July period of 2027. In other words, data on this same household are included for the first, second and third quarters of 2026 and for the first, second and third quarters of 2027. Therefore, the same household (like the ② example) can be included maximally for three quarters of the year. Besides, the same household can be included to the sample for two years consecutively. However, the same household cann’t be included to the sample for three years consecutively. Finally, micro data can’t identify the same household with household serial number or ID. Warm regards, Ministry of Data and Statistics