Survey ID Number
LVA_1996_HBS_v01_M
Title
Household Budget Survey 1996
Sampling Procedure
Sampling Frame
The target population of the HBS consists of all households in Latvia. Persons living in institutional households (elderly people boarding house, disabled children boarding house, student hostels, hotels, barracks, hospitals, sanatoriums, penal institutions, etc.) and homeless people are excluded from the current survey.
During the preparation stage of the survey it was decided to use the population register formed in the early 90s. At the beginning of 1995, it included more than 99% of the Latvian population.
In the pre-pilot survey in 1995 we found out that due to the different mistakes in the population register, several households were not found in the given addresses. Almost 10% of the entries of the register were outdated. Some respondents were dead. Other deficiencies were found with the addresses, physically non-existing addresses were also ascertained as well as other mistakes with the register. One should admit that the usage of the population register to form the sample of HBS makes the data processing procedures more difficult. However, in towns the population register was the only available register which allowed to create the HBS sample.
It could mean that the principal survey should consider non-response, the amount of which could seriously impair the representativeness of the survey.
Alternatives for the sampling frame were, therefore, sought out. One of the alternatives turned out to be the complete list of households of the local governments -- the civil parish. Of 32 parishes included in the first phase of HBS, 11 parishes already had a full list of households due to a previous agricultural survey. Therefore, it was necessary to prepare the identical lists only for the remaining 21 parishes.
Sample Size and Sample Rotation
The total annual sample size of the new HBS is equal to approximately 7992 households (666 per month with a complete monthly renewal till December 1996). Half of the monthly samples of the 1996 will form a panel in the sample of the corresponding months during the coming 3 years.
Samples for the HBS have been selected separately for the following domains:
• Riga and 6 large cities;
• middle and small towns, and
• rural areas.
The monthly sample size is distributed across the major strata in the following way: 222 households in Riga; 116 in the six other large cities; 90 in middle-sized towns (over 7,000 inhabitants); 33 in small towns; 205 households in rural areas. In Riga, the large cities and the middle and small towns, the population register was used to create the sample. In the rural areas, however, the list of parishes and list of households in the parish was used. It is necessary to rememeber that the sampling procedures are different in different territories, therefore the probability for an individual household to be included in the survey should be calculated differently in different territories. The allocation of the sample (distribution of the sampling community) between towns and countryside as well as among 5 large regions of Latvia (Kurzeme, Zemgale, Vidzeme, Latgale and Riga region) is described inTable 1 on page 15 of the user manual.
Sample Design (Rural and Urban Areas)
Rural Areas
In order to obtain the sample in rural areas, a two-stage stratified random sampling of households is applied. The administrations of the parishes provided the data. There were 59 parishes, where the number of households was less than 300. 54 of them were added to the neighboring parish. Therefore, 47 primary sampling units (PSU) consist of 2 or 3 parishes. All other 419 parishes form separate PSU. The total number of PSU in rural areas equals 466. These PSU are distributed among 5 strata (geographical regions of Latvia) according to their location. This is shown in table 2. At the first stage, the PSU were selected within each stratum with probabilities proportional to the number of households of PSU. The total number of selected PSU is 29 and 3 of them consist of 2 parishes each. For all of the 32 parishes selected , a complete list of households was prepared. These lists of households are used to obtain the sample of households at the second stage.
Urban Areas
The Population Register is used to obtain the sample of households in urban areas. The 6 administrative districts of the capital Riga and each of the 6 other large towns of Latvia (Daugavpils, Jelgava, Jsrmala, Liepâja, Rèzekne, Ventspils) form 12 separate strata. Simple random sampling of persons (of age 15 years and more) is applied within each stratum. The sample size per strata is proportional to the population size. All other towns of Latvia are distributed into 10 strata according to their size (more than 7000 inhabitants or less than or equal to 7000 inhabitants) and their location (5 geographical regions). Two stage stratified random sampling is used to obtain the sample. The PSU are the towns of strata and the selection of PSU is conditional to the probabilities proportional to the total number of inhabitants in PSU. At the second stage simple random sampling of persons (of age 15 years and more) is applied within each selected PSU. In urban areas the households of the sampled persons are surveyed. The total number of PSU per strata, and the number of PSU included in the HBS in each strata are shown in tables 3 and 4.
Non Response and Replacenment Procedure
In a multi-functional survey as the HBS, it is exceptionally important to receive responses as complete as possible. Households that refuse to participate in the survey or respond to the questions of the survey, or households that are not found at the given address, may have an impact on the precision of the acquired results that should not be neglected.
In order to limit the effects of non-response or refusal on the survey results, a sequential sampling approach is applied. A refusing or non-responding household is replaced by another from a reserve list and surveyed. According to the survey procedure, households for replacement are taken in strict order. Considering that the reserve list has been made on a random selection basis, the households from the reserve list have the same probability of being selected.
For this reason the supervisor of the field-work has two sampling lists at his disposal:
i) basic list: Sampling list with household addresses to be surveyed during a month's time. Every month, a new list is given to both supervisor and interviewer. The total number of households equals 666;
ii) reserve list: This list is given to supervisors only. The reserve list contains households in random order. This list is used upon the decision of the supervisor in the case that a sampled household does not participate in the survey. Then, the supervisor assigns a new household, keeping strictly to the sequence of the list. The reserve list is used until all households included are being surveyed. Afterwards the supervisor receives the next reserve list.
According to the regulations of the survey, the term for replacement of the household is limited. The latest possible date is the 5th day of the survey month. In case the interviewer fails to replace a household from the basic list by the one from the reserve list and to start the interview, the household shall be excluded from the survey.
The interviewer states the reasons if a household included in the survey (from the basic list or reserve list) is not surveyed in the respective month, filling in a REPORT ON REASONS OF NON-RESPONSE . On the basis of these reports the analysis on non-response is made.
Since the beginning of the survey the number of households replaced by the reserve list reached 30% from the total number of households included in the sampling list.