General Information
Project description
SEAI is conducting a monthly behavioural tracking survey to track everyday energy-related behaviours (home energy use and travel) in Ireland. The survey is being run to support ongoing government communications around the current energy crisis. We will run the survey over a given week (7-10 days) each month on a nationally representative adult sample. It is expected that the survey will run for 13 iterations initially.
In order to capture an accurate snapshot of current energy-related behaviours, we have developed a survey using the “Day Reconstruction Method” (DRM) in which respondents are asked to reconstruct the previous day using a structured online questionnaire. The primary energy consuming activities we will initially collect information on relate to home energy use (heating, hot water, cooking and electrical appliance usage) and transport use, although further iterations of the survey may include additional behaviours. In addition, we will collect data on the socioeconomic characteristics of respondents, their dwelling and household characteristics, psychological/motivational factors, perception of and confidence in government response, awareness of an ongoing government information campaign, as well as indicators of energy poverty (approximate monthly energy costs, ability to pay and energy-related deprivation).
Data will be collected as a repeated cross-section, however participants will be invited to participate every third wave potentially enabling a longitudinal analysis.
Analysis Plan
Pre-analysis plan: Is there a pre-analysis plan associated with this registration?
Hypothesis
This study is exploratory in nature rather than setting out to test specific hypotheses.
The key research questions are:
Energy behaviours
Q1: What energy consuming activities are people in Ireland undertaking most frequently and when do they occur?
Q2: How do activities vary in relation to the socioeconomic characteristics of people and where they live?
Q3: How do activities vary over time and in response to external factors (such as changing weather, energy prices or information)?
Q4: Who are the excess energy users/energy wasters (socioeconomic, demographic, location)? What factors influence their energy consumption (motivational, psychological, lifestyle patters, household composition)?
Q5: Who are peak energy users (socioeconomic, demographic, location)? What factors influence their energy consumption (motivational, psychological, lifestyle patters, household composition)?
Impact of government information campaigns
Q6: How do energy behaviours targeted by the government information campaign vary over time? Are there any indications of the campaign having an impact on behaviour?
Q7: How do people perceive their own efforts to conserve energy compared with that of others and how do they feel about the government’s handling of the crisis? How does this relate to actual behaviour?
Energy costs, poverty and deprivation
Q8: How do the composition of fuel costs (oil, gas, solid, electricity, motor) vary across different people and over time?
Q9: What is the relationship between energy costs, ability to pay and energy related deprivation and how does this relationship change over time and across different heating seasons?
Longitudinal analysis
Q10: Where possible, the above questions will be explored on a longitudinal basis
How hypothesis will be tested
Research questions will be investigated using descriptive and graphical statistical analysis, correlation analysis, factor analysis and multivariate regression.
Dependent variables
Behaviours
Travel:
• Previous week: modes of transport most frequently used
• Previous day: no. of journeys, mode of transport, distance, duration
Home heating:
• Previous week: no. of days home heated
• Previous day: frequency and intensity, time of day, heating of unoccupied rooms, heating when no one at home, thermostat setting, use of secondary heating
Water heating:
• Previous day: frequency and intensity, time of day, shower duration
Cooking & electrical appliances
• Previous week: no. of days certain appliances used
• Previous day: appliances used, time of day, frequency and intensity
Variable creation
Excess energy users/energy wasters: composites variable categorising excess users/energy wasters will be created by combining responses across the following domains:
• Travel: respondents who travelled by private car or taxi for short distances (less than 2-5km) when a public transport option was available.
• Heating: (i) respondents who heated unoccupied rooms (unless at a lower temperature) or (ii) who indicate the heating was on during periods the home is empty (iii) who have their thermostat set to 21 or higher (iv) turn on heating for extended periods (e.g. more than 12 hours)
• Water: (i) respondents who showered for 10 mins or longer or (ii) who filled the bath more than half full (iii) heat water for extended periods (e.g. more than 12 hours)
• Cooking: respondents who use high intensity appliances for extended periods but produce a low number of portions
• Appliances
o Washing machine, dishwasher, dryer: (i) respondents who wash at high temps (e.g. 50 °C and above for washing machine) or (ii) run with a less than half full load, (iii) do not use eco settings
o Respondents who use the washing machine/tumble dryers on 3 or more days per week
o Respondents who do not unplug computers/laptops after use
Peak users: a composite variable categorising peak users will be created by combining:
• respondents who use electric heating, hot-water and cooking between 4-7pm each day
• respondents who use large electrical appliances/ a number of smaller appliances between 4-7pm each day
Energy poverty: a variable will be created to quantify the proportion of household disposable income spent on energy costs (electricity, heating, other fuels and/or transport). This variable will be used to identify energy poor (those spending 10% or more on energy) and the energy burden more broadly.
Energy deprivation: a composite measure will be created combining participants who unable to pay bills, have had to go without heating through lack of money and have had to cut down on other essentials through lack of money.
Please note: The above descriptions are indicative of methods we will use to create composite variables. Actual variables will depend to some degree on the distribution of responses in data collected. Variables will most likely be ordinal
Analyses
Q1-Q3, Q8 will be assessed using descriptive statistical analysis. Graphical presentations of key energy and travel activities and costs for each survey cross-section and over time.
Outcomes will be presented for different segments of the population. Correlation and multivariate regression analysis will be undertaken to examine relationship between energy behaviours, energy poverty and deprivation and socio-economic, demographic and locational variables.
Q4,5 will be assessed by creating composite indicates of high-energy consumers using additive methods or, if appropriate, more complex aggregations such as principal component/factor analysis
Q6 will involve linking external datasets such as (i) daily weather data from Met Eireann (ii) energy price data from sources such as SEAI, the Commission for Utility Regulation and Bonkers.ie (iii) information on content and timing of Irish Govt Communications on the energy crisis. The impact of external factors will be assessed using methods such as multivariate analysis and event studies.
Q7 will use descriptive analysis to compare survey responses to questions on their own behaviour vs that of others. Descriptive and regression modelling will be undertaken to assess the relationship between predicted and actual behaviour
Q9 will be assessed by examining the relationship between expenditure-based energy poverty and energy-related deprivation. Using descriptive statistics and correlation analysis we will assess to what extent these metrics overlap, and whether people not considered energy poor as defined by expenditure and curtailing heating or consumption of other essentials. We will assess how these factors vary by socio-economic, demographic and locational variables, and time of year.
Q10 will use all of the above methods. Additionally, it may be possible to estimate models using individual-specific effects (fixed effects) to control for time-invariant, respondent-specific factors.
Sample Size. How many observations will be collected or what will determine sample size?
1000 per month, nationally representative on age, gender, location and social group
Data Exclusion
We have included an attention check in the questionnaire. Those who fail the check are given a second chance to answer correctly before being excluded from the experiment (not included in sample of 1,000).
We will conduct sensitivity analyses by omitting those who failed the attention check the first time before responding correctly.
We will also check other responses for internal consistency and conduct sensitivity analysis by removing those who are not consistent.
We will also conduct sensitivity analysis omitting those who complete the survey in a very short amount of time (under 5 minutes).
Treatment of Missing Data
N/A
Who is behind the project?
Project status:
Pre-registration
Methods
What is the project about?
Date published:
6 January 2023