A Survey on Household Incomes and Expenditures was introduced in Georgia in the second half of 1996 and has been carried out each year since then (on a quarterly basis). The survey was revised significantly in 2002. National Statistics Office of Georgia following to a large extent unchanged poverty measurement methodology established in 1997, has regularly calculated and published poverty measurement results.
The Household Integrated Survey (HIS) in Georgia has been conducted regularly from 1996 and has served to assess the level of consumption-based poverty since then. The HIS represents quarterly panel data. The survey covers 13,404 households over the year. Each month 1/12 of the sample is refreshed (about 228 households are changed in 25 census units).
Kind of Data
Sample survey data [ssd]
Unit of Analysis
The scope of the Household Integrated Survey includes:
- living conditions, housing, durables;
- household composition;
- household expenditures;
- agricultural activity;
- public and private transfers;
- employment and income.
The survey covers 13,404 households over the year. Each month 1/12 of the sample is refreshed (about 228 households are changed in 25 census units).
The survey covers all household members, excluding persons fully supported by the state, for example persons staying in homes for the elderly and the disabled, children in public care institutions, prisoners, etc.
Producers and sponsors
National Statistics Office of Georgia
National Statistics Office of Georgia
The survey consists in quarterly interviewing households in Tbilisi and nine regions of Georgia:
3. Shida Kartli, including Mtskheta-Mtianeti1;
4. Kvemo Kartli;
10. Imereti, including Racha-Lechkhumi and Kvemo Svaneti.
The 1989 Population Census served as a sampling frame of the household survey untill 2002. A new census was conducted in early 2002, which enabled development of new sampling design. Transiton to the new sampling design began in April 2002.
In the new design of the survey, stratification of each reagion was mainly carried out by settlement type and settlement altitude. Three types of settled areas according to the structure of employment and incomes were identified: (a) large cities (with population over 50,000); (b) small cities (with population under 50,000); and (c) villages. By altitude, the settlements can be divided into two groups: (1) highland settled areas; and (2) lowland settled areas.
The households are selected according to the same principle as in the old design, but using information from 2002 Population Census.
1. In each stratum, census units are selected with probability proportional to the number of households in census units.
2. The selected census units are actualized. Namely, the household list in a selected census unit is given to an interviewer who determines the changes that have occurred since the last population census (by visiting all households in the unit) and records the existing situation (i.e., whether the address indicated in the list exists or not, whether the dwelling on the indicated address is being used for living, who lives in the indicated address; he/she also enumerates all new dwellings built in the census unit territory from the last census).
3. Households are selected from actualized list according to systematic random sampling method.
4. After the selected households are interviewed four times (once in quarter), they are replaced by households which are next in the list.
5. After all households in census unit participate in the survey, this census unit is replaced by another census unit selected randomly from the same stratum.
When the data of the sample survey are expanded to the total population, each household included in the sample is given a statistical weighting which characterizes the total number of households represented by the selected household. The average value of basic weighting is 600 i.e. Each selected household is a representation of about 600 households.
The following factors are used for constructing a basic weighting:
Probability of selecting each big city, small town and village council.
Probability of selecting each constituency in big cities/small towns and the zone in village councils.
Probability of selecting each household within the constituency and the zone included in the sample.
The inverse value of the product of the three probabilities is the basic weighting of a household:
WB_k = -------------------- ,
p1_k * p2_k *p3_k
k is the number of a household.
The total of all basic weightings gives the estimate of the number of all household of the country and is constant during the whole year:
sum_(i= 1 to n ) WB_k = N ,
n is the size of sample;
n is the estimate of the total population.
The earlier described method of constructing a household sample did not take into account that some households will not be surveyed for one or the other reason such as impossibility to get the household at home despite numerous attempts, illness, categorical refusal to provide information or for other reasons. If the estimate comprises only the households that agreed to be surveyed, the results of the survey will be understated.
There are different methods of making adjustments for missing answers. One of them is the adjustment of basic weighting i.E. Statistical re-weighting which is based on overstating the weighting of respondents having certain characteristics which are peculiar for those who refused from participating in the survey.
Based on checking the data the variables are selected which will be then used for statistical re-weighting such as :
Place of residence (big city, small town, village).
Type of accommodation (individual apartment, individual house, communal flat, part of a house, hostel).
Size of the household (single individuals and households consisting of 2, 3, 4 and more people).
ADJUSTMENT OF BASIC WEIGHTING FOR NON-RESIDENTIAL APARTMENTS - During the baseline interview there might be a situation when the selected apartment is non-residential (it can be used as a shop, office premises and etc.). As the addresses of such apartments were accounted for when forming a sample (i.E. They influenced the probability of selecting other addresses), it is needed to adjust basic weightings to eliminate distortions. To enhance the accuracy of calculations, all selected households are divided into groups depending on regional location. These groups are called weighting units. Basic weightings are thus adjusted in accordance with the formula:
sum_(j= 1 to L_k )WB_k_j + sum_(j= 1 to M_k )WB_k_j
WV_k, = --------------------------------------------------------------------------------- ,
sum_(j= 1 to M_k )WB_k_j
k = 1,7 is the number of the region;
WV_k_i is the weighting of the “i” household of the “k” region adjusted for non-residential apartments;
WB_k_i is the basic weighting of the “i” household of the “k” region;
L_k is the number of non-residential apartments in the “k” region;
M_k is the number of the rest of the households in the “k” region.
ADJUSTMENT FOR NON-ANSWERS DURING THE BASELINE INTERVIEW - To make an adjustment for non-answers in the baseline interview, the weighting units are formed based on the following characteristics:
the region, 7 values (the Brest region, ..., the Mogilev region); type of settlement, 3 values (big city, small town, village); type of accommodation, 2 values (individual accommodation (an individual apartment in big cities and small towns and an individual house in villages) or other type of accommodation (communal flat, part of a house, hostel); size of a household, 4 values (1 person, 2 persons, 3 persons, 4 and more persons). After the units are formed, the distribution of households is analyzed in a way that one unit should contain at least 20 households. The new weightings are then calculated based on the modified units. For the “i” household included in the “k” unit, the weighting WBL_ki adjusted for non-answers in the baseline interview can be
sum_(j= 1 to M_k )WV_kj + sum_(j= 1 to N_k )WV_kj
WBL_ki = ----------------------------------------------------------------------- * WV_ki ,
sum_(j= 1 to N_k )WV_kj
WV_k_i is the weighting of the “i” household included in the “k” weighting unit adjusted for non-residential apartments;
M_k is the number of non-answers per a baseline interview in the “k” unit;
N_k is the number of answers per a baseline interview in the “k” unit.
In 1996 after an adjustment for non-answers in the baseline interview the average weighting increased to 692.
ADJUSTMENTS IN QUARTERLY INTERVIEWS - The units for an adjustment for non-answers in quarterly interviews are formed based on the same characteristics which are used for the baseline interview. For the “i” household included in the “k” unit formed for the quarter, the weighting WQ_alpha_k_i adjusted for non-answers in the quarterly interview is calculated in as follows:
sum_(j= 1 to M_k )WQ_a - 1_kj + sum_(j= 1 to N_k )WQ_a - 1_kj
WQ_a_k_i = ---------------------------------------------------------------------------------------------- * WQ_a - 1_ki ,
sum_(j= 1 to N_k )WQ_a-1_kj ,
a= 1,2,3,4 is the number of the quarter;
WQ_a-1_k_i is the weighting adjusted for non-answers in the quarterly interview for the quarter a-1 for the “j” household included in the “k” unit formed for quarter a and WQ_0_ki ;
M_k is the number of non-answers in the quarterly interview in the “k” unit for quarter a;
N_k is the number of answers in the quarterly interview in the “k” unit for quarter a.
After the baseline interview the average weighting has slightly increased due to high percentage of answers after the first interview. The average weightings are equal to 708, 716, 723 and 726 accordingly for 4 quarters in 1995. It should be noted that the households who miss one interview are withdrawn from the survey. Thus the households who answered the questions of the last interview had also answered the questions of all previous interviews. Therefore, the weighting calculated for the last interview can be used for the households who answered the questions during the whole year. The annual report uses the data on the households who answered the questions during the whole year re-weighted by means of using the last quarter weighting.
Dates of Data Collection
Data Collection Mode
National Statistics Office of Georgia
The questionnaire consists of eight sections:
- Shinda01: general information about living conditions, housing, durables, etc. This section remained unchanged since the household survey was introduced in 1996.
- Shinda02: household composition. This section also remained unchanged since the survey inception.
- Shinda03: diary expenditure form. This section includes all diary expenditures during one week and it is filled out four times during the households' period of survey.
- Shinda04: quarterly expenditures and agricultural activity form. This section covers quarterly expenditures on durables, energy supplies, health care, education, and other services. The questionnaire also collects information about harvest and processing of agricultural products produced by the household, sale and income from selling these products. The questionnaire is filled out four times, simultaneously with diary expenditures form. This section also features "reminder questions", which help households remember their expenditures.
- Shinda05: Information about public and private transfers, as well as on changes in household financial and demographic conditions is collected in the section. The substance of the questions was not changed; however their phrasing was adjusted to make them more understandable for respondents.
- Shinda05-1: includes information on employment and incomes from employment of adult household members.
- Shinda07: refusal form. This section covers information on non-response or non-eligibility. This form helps correct the weights before data processing.
- Shinda09: monitoring of poverty in Georgia.
"Shinda" is a Georgian abbreviation for "observation of households."
National Statistics Office of Georgia
The use of the datasets must be acknowledged using a citation which would include:
- the identification of the Primary Investigator (including country name)
- the full title of the survey and its acronym (when available), and the year(s) of implementation
- the survey reference number
- the source and date of download (for datasets disseminated online)
National Statistics Office of Georgia. Georgia Household Integrated Survey (HIS) 2006, Ref. GEO_2006_HIS_v01_M. Dataset downloaded from [URL] on [date].
Disclaimer and copyrights
The user of the data acknowledges that the original collector of the data, the authorized distributor of the data, and the relevant funding agency bear no responsibility for use of the data or for interpretations or inferences based upon such uses.
DDI Document ID
The World Bank
Documentation of the study
Development Data Group
The World Bank
Review of metadata
Date of Metadata Production
DDI Document version
Version 02 (April 2016)
The additional data "Shinda 5" was added to the exicting dataset.