Statistical Data Sources

The generation of energy statistics draws on two main kinds of sources: statistical data sources, collected exclusively for statistical purposes from censuses and/or sample surveys, and administrative data sources, created originally for purposes other than statistical production [IRES, Ch. VII, para. 7.36, PDF p. 106, 2018].

Census vs. sample survey

The typical statistical data sources for compiling energy statistics are surveys of the units in the population of interest, conducted either by enumerating all units in the population (a census) or a scientifically selected subset of representative units (a sample survey) [IRES, Ch. VII, para. 7.37, PDF p. 106, 2018].

Censuses are generally time-consuming, resource-intensive and costly, and impose a high overall response burden, so they are unlikely to be used very often. Depending on the population of interest, available resources and national circumstances, however, a complete census of units in the energy industry may be appropriate — for example, when a country does not maintain an up-to-date business register, when there are few energy producers (in which case they should be included in a “take all” stratum of appropriate surveys), or when there is significant user interest in detailed energy data [IRES, Ch. VII, para. 7.38, PDF p. 106, 2018].

Sample surveys collect information from a portion of the total population (the sample) to draw inferences about the whole population, and are almost always less costly than censuses. Energy statistics can use different survey types depending on the sampling units: (i) enterprise surveys, (ii) household surveys, and (iii) mixed household-enterprise surveys.

Recommendation (7.39): it is recommended that countries make efforts to establish a programme of sample surveys that satisfies the needs of energy statistics in an integrated way — i.e., as part of an overall national sample survey programme of enterprises and households — to avoid duplication of work and minimize response burden [IRES, Ch. VII, para. 7.39, PDF p. 106, 2018] — recommendations tracker row VII/7.39.

Survey design

Before carrying out a survey, a proper survey design is fundamental. Design proceeds through a sequence of steps [IRES, Ch. VII, paras 7.40–7.45, PDF pp. 107–108, 2018]:

1. Identify information needs and goals. Establish the project’s specific goals, with special emphasis on priorities, feasibility, budget and geographic breakdown, drawing on experience from similar projects in other statistical domains, relevant international recommendations (e.g., those published in the International Recommendations for Industrial Statistics 2008, IRIS) and applicable national laws and regulations. This phase needs input from energy specialists as well as specialists in sample design, interviewing techniques and analysis procedures, requiring cooperation among the national statistical office, relevant ministries and academic institutions [IRES, Ch. VII, para. 7.40, PDF p. 107, 2018].

2. Establish periodicity. Energy surveys should be designed to ensure regular conduct, so periodicity should be established from the outset. The survey design should be optimized around the intended use and inferences from results, avoiding non-essential information, and structured to guarantee the greatest analytical benefit and consistency over time given the survey’s cost.

   Recommendation (7.41): it is recommended that the periodicity of energy surveys be established from the very beginning, with survey design optimized for the intended use of results and non-essential information avoided as far as possible [IRES, Ch. VII, para. 7.41, PDF p. 107, 2018] — recommendations tracker row VII/7.41.

3. Select data items. Once the survey’s topic(s) are determined, the next stage is to select the data items to collect, using the reference list in Chapter VI as a reference and ensuring the selection follows an appropriate classification and precise definition of each data-item concept [IRES, Ch. VII, para. 7.42, PDF p. 107, 2018].

4. Select the target population/sample. The number of units to interview must be decided to ensure representativeness, balancing time availability, budget constraints and the necessary degree of precision. The sampling technique depends on the population(s) being sampled and on information available from other regular survey programmes and business registers, which may give a better picture and context for the project [IRES, Ch. VII, para. 7.43, PDF p. 107, 2018].

5. Design questionnaires. Decide the interviewer’s profile, the interviewing method best suited to the survey’s purpose (personal interviews, telephone, mail, computer-direct, email, Internet, and others), the temporal scope of data items, and how items and related concepts will be presented and asked. Determine question type and sequence, favouring clear, direct, straightforward language, and select proper measurement units for the respondent — small units such as kilowatt-hour or cubic metre suit consumers or gasoline stations, but not energy supply industries [IRES, Ch. VII, para. 7.44, PDF p. 107, 2018].

6. Prepare instructions; pilot and train. Concise, clear instructions should be prepared for potential respondents, and the survey design adapted to the specific context, geographical scope, informant, interviewer and planned procedures. Questionnaires must be tested in a context similar to the one in which they will be applied before final adjustments are made. Interviewers need careful training in techniques for measuring different fuels; for biomass in particular, availability of measurement instruments (e.g., scales for fuelwood and charcoal) for physically measuring fuels actually consumed is extremely important and should be ensured where possible [IRES, Ch. VII, paras 7.45, PDF pp. 107–108, 2018].

Enterprise surveys

Enterprise surveys are surveys in which the sampling units are enterprises (or statistical units belonging to them, such as establishments or kind-of-activity units) as the reporting and observation units. They require a sampling frame of enterprises, and depending on the frame’s source can be classified as list-based (the initial sample is drawn from a pre-existing list of enterprises or households) or area-based (the initial sampling units are geographical areas; after one or more selection stages, enterprises or households within the selected areas are listed and sampled). List-based surveys are generally preferred, since enumerating enterprises within an area can be difficult, and area-based sampling is poorly suited to large or medium-sized enterprises operating across several areas — it is hard to collect data from just the parts of an enterprise lying within the selected areas. A stratified sampling technique should be used whenever appropriate and feasible to improve accuracy [IRES, Ch. VII, para. 7.46, PDF p. 108, 2018].

Use of a statistical business register as the sampling frame is discussed in full on its own page; in principle, the sampling frame should contain all units in the survey’s target population without duplication or omission [IRES, Ch. VII, para. 7.47, PDF p. 108, 2018].

Ad hoc energy statistics surveys

Specially designed energy statistics surveys are extremely useful for compensating for gaps in the mechanisms and instruments described above. An example is an energy consumption survey designed specifically to measure quantities of energy products consumed. The sampling unit is likely to be the household, and possibly small-scale rural industry sites below the normal threshold for sample enquiries; data generally cover weights (or volumes, where realistic conversion to weight is possible) of different fuels consumed for different purposes. Where fuel usage shows a seasonal pattern, interviews must be spread over the entire year to be representative of all seasons, and results analysed by household size to obtain per capita consumption ranges [IRES, Ch. VII, para. 7.49, PDF p. 108, 2018].

Designing and implementing such surveys can be demanding in financial and human resources, and often requires multidisciplinary expertise to identify the appropriate sample design, interviewing techniques and analysis procedures. Ad hoc surveys are very useful for assessing energy consumption activities, monitoring the impacts of energy programmes, tracking the potential for energy efficiency improvements, and assessing the feasibility of future programmes [IRES, Ch. VII, para. 7.50, PDF p. 109, 2018].

Household surveys and mixed household-enterprise surveys

Household surveys are surveys in which the sampling units are households. In mixed household-enterprise surveys, a sample of households is selected and each household is asked whether any member owns and operates an unincorporated enterprise (also called an informal-sector enterprise in developing countries); the resulting list of enterprises is then used as the basis for selecting enterprises from which data are finally collected. Mixed household-enterprise surveys are useful for covering unincorporated (household) enterprises, which are numerous and hard to register directly [IRES, Ch. VII, para. 7.51, PDF p. 109, 2018; see also IRIS 2008, paras 6.19–6.24, footnote 61, for the advantages and disadvantages of mixed household-enterprise surveys].

Although household surveys are not designed specifically for energy data compilation, they can give a broad overview of residential energy consumption by end-use and, potentially, of household energy production. Given the complexity of household energy-consumption characteristics, consumption estimates should be derived from such surveys using their own metadata. Useful information includes the number and average size of households, appliance penetration and ownership, appliance attributes and usage parameters, fuels used for cooking and heating/air conditioning, electricity sources (national grid, solar electricity, auto-production, etc.) and types of bulbs used for illumination; the age and efficiency of the household appliance stock can also be determined via administrative registers or appliance-sales surveys [IRES, Ch. VII, para. 7.52, PDF p. 109, 2018].

The frequency of household surveys matters, since behaviour in this sector often varies with prices, technologies and fuel availability, and new household appliances entering the market create new consumption habits that should be taken into account [IRES, Ch. VII, para. 7.53, PDF p. 109, 2018]. These surveys should be representative not only at the national level, but also for rural and urban areas and by region, to permit proper analysis of the data [IRES, Ch. VII, para. 7.54, PDF p. 109, 2018].

Related

• Statistical Business Register — the sampling frame recommended for list-based enterprise surveys
• Administrative Data Sources — the second main category of data source
• Data Compilation Methods — how collected survey data is validated, imputed and grossed up
• Energy Data Reporter Groups — the reporter groups these surveys collect from
• International Recommendations for Industrial Statistics — IRIS 2008, cited for survey-design practice and mixed household-enterprise survey trade-offs
• Energy Balance — the reconciliation of data gathered from these statistical sources (and from administrative sources) feeds directly into the energy balance’s data-reconciliation recommendations (IRES Ch. VIII, Section E)