Summative Project Topics

Choose one topic for your group project. Each topic has a clear research question, accessible public data, and a genuine policy dimension. Your group will design an investigation, analyse data, and each write an individual 4-page policy report.

Before you commit: Check with your instructor that no other group has chosen the same topic. First come, first served.

Topic 1: Does wind farm output vary by UK region?

Question: Do onshore wind farms in Scotland produce more electricity per unit of capacity than those in England and Wales?

Data: DESNZ Regional Renewable Statistics — renewable electricity by local authority (Excel, annual). GOV.UK: Regional Renewable Statistics

Variables: Installed capacity (MW), actual generation (GWh), load factor (%), region, technology type, year. Local-authority level, 2014–present.

Suggested comparison: Compare load factors across UK nations or English regions for onshore wind. Is the difference statistically significant? How large is it?

Analytical approach: ANOVA (4+ regions) or t-test (2-group comparison, e.g. Scotland vs England). Effect sizes.

Confounders to consider: Turbine age and model, altitude and terrain, grid curtailment, year-to-year weather variation.

Difficulty: ⬟⬟⬡ Moderate — data is well-structured; main challenge is controlling for confounders.

Topic 2: Is domestic electricity consumption falling?

Question: Has household electricity use per customer declined over the past 20 years, and does the rate of decline differ between UK regions?

Data: DESNZ Sub-national Electricity Consumption Statistics — regional and local authority level (Excel, annual 2005–2024). GOV.UK: Regional and local authority electricity consumption

Variables: Total domestic consumption (GWh), number of meters, mean consumption per meter (kWh), region, local authority, year.

Suggested comparison: Compare the rate of decline in domestic consumption per meter across regions. Is it faster in some areas? Has the decline stalled?

Analytical approach: Regression (consumption vs year, by region). Compare slopes. Alternatively, t-test/ANOVA on recent vs historical consumption.

Confounders to consider: Temperature/heating degree days, energy prices, housing stock (new builds vs old), population change, economic deprivation, uptake of heat pumps and EVs (which increase electricity use).

Difficulty: ⬟⬟⬡ Moderate — rich dataset; challenge is choosing a focused question from many possible comparisons.

Topic 3: Solar panel output — does latitude matter in the UK?

Question: Within the UK, how much does solar PV generation per MW of installed capacity vary with latitude?

Data: DESNZ Regional Renewable Statistics (generation and capacity by local authority) combined with local authority latitude. GOV.UK: Regional Renewable Statistics

Variables: Solar PV installed capacity (MW), actual generation (GWh), calculated load factor (%), local authority, latitude.

Suggested comparison: Scatter plot of solar load factor vs latitude. Fit a regression. How much of the variation does latitude explain?

Analytical approach: Linear regression. R² and slope interpretation. Prediction intervals.

Confounders to consider: Panel orientation and tilt, local cloud cover, shading (urban vs rural), proportion of roof-mounted vs ground-mounted, data quality (small installations may not report accurately).

Difficulty: ⬟⬡⬡ Easier — clean relationship expected; good for regression practice. Extension: add a second predictor (e.g. urban/rural classification).

Topic 4: Has nuclear output kept pace with closures?

Question: As older UK nuclear stations have closed, has total nuclear generation declined proportionally, or have remaining stations increased output?

Data: DUKES Chapter 5 — electricity generation by fuel type (annual, 2000–2024). Individual station data from EDF Energy annual reports or DUKES Table 5.11 (nuclear stations). GOV.UK: DUKES Electricity

Variables: Nuclear generation (TWh), number of operating stations, total capacity (GW), load factor (%), year.

Suggested comparison: Track capacity and generation over time. Calculate load factor trends. Has output per GW of capacity changed?

Analytical approach: Regression (generation vs year, generation vs capacity). Compare actual output to what you’d predict from capacity alone.

Confounders to consider: Planned maintenance outages, unplanned shutdowns (graphite cracking), Hinkley Point C delays, electricity demand changes.

Difficulty: ⬟⬟⬡ Moderate — requires careful separation of capacity reduction from output-per-station trends.

Topic 5: Do countries that invest more in renewables have lower carbon intensity?

Question: Across countries, is higher renewable electricity capacity associated with lower CO₂ emissions per kWh of electricity generated?

Data: IRENA Renewable Capacity Statistics (country-level, 2000– 2024, free download) + IEA or Ember electricity carbon intensity data. IRENA Data & Statistics Ember Global Electricity Data

Variables: Renewable capacity (MW) by technology and country, electricity generation (GWh), carbon intensity (gCO₂/kWh), GDP, year.

Suggested comparison: Cross-country scatter plot: renewable capacity per capita vs carbon intensity. Is the relationship significant? Does it hold after controlling for GDP?

Analytical approach: Regression with potential multiple predictors. Effect sizes and confidence intervals.

Confounders to consider: Nuclear power (low carbon but not renewable), fossil fuel exports (e.g. Norway has high renewables but also oil), country size and geography, GDP per capita, historical energy mix.

Difficulty: ⬟⬟⬟ Harder — international dataset, multiple confounders, risk of ecological fallacy. Ambitious but rewarding.

Topic 6: The gas price shock — did energy consumption change?

Question: Did UK domestic gas consumption fall significantly after the 2022 energy price spike, and has it recovered?

Data: DESNZ Sub-national Gas Consumption Statistics — regional and local authority level (Excel, annual 2005–2024). GOV.UK: Sub-national gas consumption data

Also: quarterly energy prices from DESNZ (for price variable).

Variables: Domestic gas consumption per meter (kWh), number of meters, gas price index, region, local authority, year.

Suggested comparison: Before-and-after comparison (pre-2022 vs post-2022) of consumption per meter. Did consumption fall? Has it bounced back? Does the effect differ by region or deprivation level?

Analytical approach: t-test (before vs after), or regression with a price term. ANOVA if comparing multiple regions.

Confounders to consider: Winter temperatures (heating degree days), boiler efficiency improvements, heat pump adoption, housing insulation upgrades, population changes.

Difficulty: ⬟⬟⬡ Moderate — topical question, clear before/after design. Risk of confounding with temperature variation.

Topic 7: How fast is the UK installing heat pumps?

Question: Is the rate of heat pump installation accelerating, and does it vary by region or housing type?

Data: Ofgem MCS installation data (Microgeneration Certification Scheme — public quarterly reports), plus DESNZ Boiler Upgrade Scheme statistics. MCS Data Dashboard GOV.UK: Boiler Upgrade Scheme statistics

Variables: Number of installations by month/quarter, technology type (ASHP, GSHP), region or local authority, property type.

Suggested comparison: Is the installation rate accelerating (test for non-linearity in the trend)? Do rural areas adopt faster than urban? Do certain regions lead?

Analytical approach: Regression (installations vs time), ANOVA (regions). Test for acceleration (quadratic term or compare slopes across periods).

Confounders to consider: Government subsidy levels and changes, housing stock (detached houses more suitable), gas grid connectivity, income/deprivation.

Difficulty: ⬟⬟⬡ Moderate — data access is straightforward; good policy angle. May need some cleaning.

Topic 8: Offshore wind — are newer farms more productive?

Question: Do recently commissioned offshore wind farms have higher capacity factors than older ones?

Data: DESNZ Renewable Energy Planning Database (REPD) — quarterly extract with individual project details (capacity, technology, status, commissioning date). Plus DUKES/Energy Trends for generation data. GOV.UK: Renewable Energy Planning Database

Variables: Farm name, capacity (MW), commissioning year, location, turbine model (if available), generation (GWh), calculated load factor.

Suggested comparison: Regression of load factor on commissioning year. Is there a significant positive trend? How much has efficiency improved per decade?

Analytical approach: Regression. Possibly ANOVA if grouping farms by era (pre-2015, 2015–2020, post-2020).

Confounders to consider: Location (distance from shore, water depth, wind resource), turbine size (larger turbines are more efficient), grid curtailment, maintenance and availability.

Difficulty: ⬟⬟⬟ Harder — requires linking two datasets and cleaning individual farm records. Rewarding for students interested in engineering.

Topic 9: Does energy-from-waste reduce landfill?

Question: In local authorities with energy-from-waste (EfW) facilities, has landfill use declined faster than in authorities without EfW?

Data: Defra Local Authority Collected Waste Statistics (annual, by local authority — tonnes sent to landfill, incineration, recycling). GOV.UK: Local authority collected waste statistics

Plus DESNZ regional renewable statistics (for EfW generation data).

Variables: Waste tonnage by destination (landfill, incineration/ EfW, recycling), local authority, year, presence/absence of EfW facility.

Suggested comparison: Compare landfill rates in local authorities with and without EfW facilities. t-test or ANOVA. Control for region and deprivation.

Analytical approach: t-test (with vs without EfW), or regression with EfW as a binary predictor alongside confounders.

Confounders to consider: Recycling rates (some LAs recycle more, not burn more), population density, waste composition, distance to nearest EfW facility (waste may be exported to neighbouring LAs).

Difficulty: ⬟⬟⬡ Moderate — nice natural-experiment design. Students must define “with EfW” carefully (facility in the LA, or waste sent to one?).

Topic 10: Electric vehicles — is charging infrastructure keeping up?

Question: Is the number of public EV charge points growing in proportion to the number of registered EVs, or is the ratio deteriorating?

Data: DfT/OZEV EV charging device statistics (quarterly, by local authority) + DfT vehicle licensing statistics (registered plug-in vehicles by LA). GOV.UK: EV charging device statistics GOV.UK: Vehicle licensing statistics

Variables: Number of public charge points (total, rapid, slow), number of registered plug-in vehicles, local authority, region, quarter/year.

Suggested comparison: Track the charge-point-per-vehicle ratio over time. Is it improving or worsening? Does it vary by region (urban vs rural)? Is there a relationship between charge point density and EV uptake?

Analytical approach: Regression (charge points vs vehicles, over time). ANOVA across regions. Two-variable regression.

Confounders to consider: Home charging (most EVs charge at home, so public points matter more in urban areas without driveways), devolved policy differences, commercial fleets vs private.

Difficulty: ⬟⬟⬡ Moderate — highly topical. Data is quarterly and well-structured. Good policy angle.

Choosing your topic

When picking a topic, consider:

  1. Interest. You’ll spend 6 weeks on this. Choose something your group genuinely wants to understand.
  2. Data accessibility. Before committing, download the data and open it. Can you read it into R? Does it have the variables you need?
  3. Statistical approach. Your project should use at least one formal test (t-test, ANOVA, or regression). Check that your topic maps to one of these.
  4. Confounders. The most interesting topics have genuine confounders that you’ll need to discuss. This is where the analytical depth comes from.
  5. Group strengths. Some topics need more data cleaning; others need more statistical thinking. Play to your group’s strengths.

Your instructor can help you refine a question within your chosen topic. The question in the title is a starting point, not a straitjacket.