The 2005/6 Household Income and Expenditure Survey is the second nationwide survey of households undertaken by Solomon Islands Statistics Office (SISO) since 1992.
The primary objectives of the HIES includes:
• Re-basing of the weights of the current basket of goods and services in the Consumer Price Index (CPI). The survey also aimed to provide data on the behavior of household consumption expenditure patterns that will help form the weights that would reflect the relative importance that consumers attach to commodities and services;
• Obtaining relevant data for purposes of updating the series of national accounts aggregates particularly the Gross Domestic Product.
The secondary objectives of the HIES were to:
• Obtain data on housing and general demographic characteristics of households;
• Obtain data on poverty measures, income and income inequality measures;
• Obtain relevant data for the Millennium Development Goals (MDG), particularly health and education; and
• Obtain other relevant data where necessary
The field data collecting exercise was undertaken from October 2005 to March 2006 and that seasonality effects on expenditure was not fully considered.
Kind of Data
Sample survey data [ssd]
National. The HIES operation covered both the Urban and Rural areas focusing on Honiara, Other Urban Areas and the Rural Areas of the ten (9) provinces, and aimed to produce estimates at the country national and provincial levels only.
Unit of Analysis
The survey targeted private households whilst collective households in hospital, hotels, prison and educational institutions were excluded. A household is considered in the scope for the survey if the household have resided in the Solomon Islands for the last 12 months or more, or if not, they intend to live in Solomon Islands for the next 12 months.
Producers and sponsors
Authoring entity/Primary investigators
Solomon Islands Statistics Office (SISO)
Ministry of Finance, Government of Solomon Islands
The survey was based on a two-stage sampling strategy using probability proportional to size (PPS) selection and random selection. The strategy for selection of each area type is slightly different depending also on enumerator workload schedule and the need to accommodate estimates at the National and Provincial level as well as Urban and Rural splits.
The Survey was designed to collect data for national and provincial level estimates and covered both urban and rural areas. The survey covered Honiara, provincial centers and rural areas within these provinces.
The sampling scheme used was a stratified two stage design with the Enumeration Areas (EA) as the Primary Sampling Unit (PSU) and the households within the sample areas as the secondary sampling unit (SSU). In the first stage the EAs were selected with probability proportional to their population size based on the 1999 population census. In the second stage households were selected using systematic sampling with a random start. The next stage was allocating the sample to each provinces proportional to the square-root of the population. This should mean that estimates of each province would roughly have the same level of accuracy. The sample was then split for each province between the provincial centers (considered to be urban) and the remaining rural population. Given the need for urban and rural estimates the sample was split between the two areas proportional to the square-root of the population based on the 1999 census. The last stage in the process involved modifying the final counts to accommodate the workloads for interviewers during the fieldwork. The interviewers were expected in the field for six months and could accommodate 10 households per month (60 household in total). It was desirable to have the total workloads for each province divisible by 60 to give each interviewer an even sized workload and have the sample spread out evenly across each month.
Since Honiara (capital of Solomon Islands) consists of a mix of areas which covers high income, middle income and low income areas, it was advisable that the EAs be grouped based on the class best suited to their situation. Thus for Honiara the EA list was sorted by the income group category for selection. The number of EAs to select from Honiara is simply the desirable sample size (480 households) divided by the number of households to be selected for each EA. It was decided that 10 households should be selected from each selected EA. Therefore the number of EAs that were selected was equivalent to (480 / 10) = 48 EAs.
A sample of 4,320 households was planned for the country and about 3,822 households (88.5%) responded favorably satisfying the survey requirements.
Despite efforts made by the enumerators and follow up attempts by the supervisors in most of the cases, there was non-response encountered during the survey.
The reasons for non response by the household were due mainly to the following:
• The household was out of scope of the survey
• Dwelling was vacant or not being lived in
• The household could not be contacted after a number of attempts
• Household excluded for other reasons like death in the family, refusals, customary reasons etc
A weight indicates how many people in the population a person represents. The weight assigned to each EA and or household was based on the sampling technique used as mentioned in the survey design. For the HIES, each household carried two weights:
• A Strata weight – representing the weight of the household within the geography or locality strata, that is within Honiara, the provincial centers and the rural areas;
• A National weight – representing the weight of the household within the whole of the Solomon Islands.
The EAs as the primary sample unit were initially stratified according to respective localities in rural and urban areas and a total 18 strata was derived. The strata are as follows;
The sample selection scheme used the standard method of selecting the primary sampling units (EAs) probability proportional to size (the size being the latest household count for each EA), and then selecting a fixed number of households within each selected EA. So the sample weight fora household would be:
Weight = 1 / P(selection)
P = probability
P(selection) = P(selected stage 1) * P(selected stage 2)
P(selected stage 1) = (# hholds in EA on frame)/(skip for that province)
P(selected stage 2) = (# hholds responded in EA)/(actual # hholds in EA)
Dates of Data Collection (YYYY/MM/DD)
Mode of data collection
Type of Research Instrument
The HIES is a relatively complex survey and the instruments to collect data was implemented through the following questionnaires and associated sections:
• Household Control Form – household composition and particulars;
• Household Expenditure Form – housing amenities, facilities and major household, expenditure on tenure, fixed capital, land, property etc;
• Personal Income Form – Income pattern of household members and other income earning activities;
• Household Dairy – Daily expenditure by type of goods and services • An additional health module was included – health facility utilization, immunization, motherhood, mortality, breast feeding & family planning, Malaria and miscellaneous
The Statistics Programme at the Secretariat of the Pacific Community (SPC) provided the assistance in data processing. A HIES data entry program was setup in CSPro version 2.6 and data entry started soon after the first workload was registered in the Statistics Office in November 2005 until May 2006. Logic procedures for data editing are prepared in Microsoft Access and data editing for all questionnaires were done in CSPro, except for the Diary where the editing is done in Microsoft Excel. Data management queries are done in Microsoft Access and the production of tables was done in Microsoft Excel. This report was prepared in Microsoft Word. Data verification of 5 per cent is done to check the accuracy of data input, though data edit checks are carried out for completeness, consistency and accuracy including the outliers. Anomalies of data were amended appropriately.
Estimates of Sampling Error
No formal measures of sample errors have been calculated for the survey results.
Non sampling errors cannot be readily measured. These included:
o A response difficulty caused by misunderstanding of what was required from the survey and survey instruments by both households and interviewers.
o The questionnaires were in English, which is at least a second language for interviewers and respondents.
o The fact that some expenditure are seasonal and would not have been picked up in the survey period.
o The exclusion of remote areas and institutions from the sampling frame.
Solomon Islands National Statistics Office (SINSO)