Wave 1 and 2 Dataset

1 Demographic Information for Both Waves

2 Analyses of Both Waves

In this section, Wave 1 and 2 collected information on the perceived seriousness of heat-related risks and personal experience of heat stress, analyses follow this order respectively.

3 Seriousness of Heatwaves in the UK

A consideration to make when looking at the data is that the scale which the perceived heat risk seriousness variable has been measured on was a 1 - 4 scale asking “How serious of a problem do you think the following currently are for the UK, or do you not think they are a problem at all?”

1. Not at all serious

2. Not very serious

3. Fairly serious

4. Very serious

4 The Distribution of Housing Status Across Different Demographic Variables

4.1 Plotting the Distribution of the data

4.2 The Relationship Between Perceived Heat Risk Seriousness and Housing Category

cd_heat_model <- lm(cd_heat_mean ~ housing_category, data = cd_heat_grouped)
tab_model(cd_heat_model, pred.labels = c("Outright Owner (intercept)", "Mortgage Owner", "Social Renter", "Private Renter"), dv.labels = "Perceived Seriousness of Heat Risks")

	Perceived Seriousness of Heat Risks
Predictors	Estimates	CI	p
Outright Owner (intercept)	2.90	2.85 – 2.96	<0.001
Mortgage Owner	0.15	0.06 – 0.24	0.001
Social Renter	0.13	0.04 – 0.22	0.005
Private Renter	0.16	0.07 – 0.25	<0.001
Observations	2370
R2 / R2 adjusted	0.007 / 0.006

Warning: Removed 1 row containing non-finite outside the scale range
(`stat_ydensity()`).

The model, overall, was significant (F(3, 2522) = 5.19, p<.001 but the explained variance is very low (R² = .01). There were significant main effects of each housing category, such that mortgage owners reported higher perceived heat seriousness than outright owners (β = .12, p<.01), as well as social renters (β = .13, p<.01) and private renters (β = .16, p<.001). Importantly, the model shows that renters perceive heat risks as more serious a problem for the UK than homeowners, with private renters perceiving heat the most seriously out of all the groups.

4.2.1 Interaction Model for Perceived Heat Risk Seriousness

cd_heat_demo_model <- lm(cd_heat_mean ~ housing_binary * income_c +
                           housing_binary * age_c + education_num, 
                         data = cd_heat_grouped2)

	Perceived Seriousness of Heat Risks
Predictors	Estimates	std. Error	CI	p
Owners	2.89	0.06	2.77 – 3.01	<0.001
Renters	0.06	0.04	-0.02 – 0.14	0.123
Income	0.00	0.01	-0.02 – 0.02	0.788
Age	-0.05	0.02	-0.09 – -0.02	0.004
Education	0.02	0.01	-0.01 – 0.04	0.146
Tenure x Income	-0.01	0.02	-0.05 – 0.03	0.579
Tenure x Age	-0.01	0.03	-0.06 – 0.04	0.683
Observations	2188
R2 / R2 adjusted	0.016 / 0.013

A MLR model including income, age, and education, as well as interactions with housing tenure, indicated that age was the only significant predictor of perceived heat risk, with older individuals reporting lower levels of concern (β = −0.05, p = <.01). Housing tenure showed a minor positive association but did not significantly predict the perceived seriousness of heat risks when accounting for other demographic variables.

5 Personal Risk of Experiencing Heat Stress

The next variable within the RESiL RISK survey measures the self-reported perceived risk for experiencing heat stress. Heat stress is defined as “the net heat load that individuals experience due to the combined effects of metabolic heat production, environmental factors, and clothing requirements”. Experiencing heat stress can lead to discomfort and negatively impact work efficiency and safety, with increased risks of heat-related diseases as heat stress approaches human tolerance limits (Bishop et al., 2014). Also, a consideration to make when looking at the data is that perceived risk of experiencing heat stress has been measured on was a 1 - 3 scale asking “Do you believe you are personally at risk of experiencing heat stress during summer months in the UK?”

1. Not at all at risk

2. Possibly at risk

3. Definitely at risk

Caution should be used for analysing a scale with this few options as it restricts the amount of variability for participant responses.

5.1 Plotting the distribution of the data

5.2 The Relationship Between Perceived Vulnerability to Heat Stress and Housing Category

cd_heat_stress_model <- lm(heat ~ housing_category, data = cd)

	Perceived Risk of Heat Stress
Predictors	Estimates	CI	p
Outright Owner (intercept)	1.76	1.71 – 1.81	<0.001
Mortgage Owner	0.08	0.00 – 0.15	0.040
Social Renter	0.20	0.12 – 0.27	<0.001
Private Renter	0.22	0.14 – 0.29	<0.001
Observations	2371
R2 / R2 adjusted	0.019 / 0.018

The model, overall, was significant (F(3, 2475) = 15.49, p < .001), though it explains a modest amount of variance (R² = 0.02). There was a significant main effect of housing category, such that Mortgage owners reported slightly higher perceived personal heat risk than Outright owners (β = 0.077, p = .040). For the rental groups, Social renters reported higher personal heat risk (B = 0.203, p < .001) and Private renters reported the highest perceived personal heat risk (β = 0.214, p < .001). Therefore, the model suggests that renters perceive themselves as more at risk of heat stress than homeowners, with private renters reporting the highest risk.

5.2.1 Interaction Model for Perceived Vulnerability to Heat Stress

	Heat Stress Vulnerability
Predictors	Estimates	std. Error	CI	p
Owners	1.73	0.05	1.63 – 1.83	<0.001
Renters	0.15	0.03	0.09 – 0.21	<0.001
Income	0.00	0.01	-0.01 – 0.02	0.678
Age	-0.06	0.02	-0.09 – -0.03	<0.001
Education	0.02	0.01	-0.00 – 0.04	0.107
Tenure x Income	0.00	0.02	-0.03 – 0.04	0.858
Tenure x Age	-0.01	0.02	-0.05 – 0.03	0.696
Observations	2189
R2 / R2 adjusted	0.036 / 0.034

 housing_binary education_num income_c Estimate Std. Error     z Pr(>|z|)    S
              0          4.15        0  -0.0564     0.0152 -3.72   <0.001 12.3
              1          4.15        0  -0.0647     0.0153 -4.24   <0.001 15.5
   2.5 %  97.5 %
 -0.0861 -0.0267
 -0.0947 -0.0348

Term: age_c
Type: response
Comparison: dY/dX

A multiple regression model including income, age, and education, as well as interactions with housing tenure, indicated that age was the only significant predictor of perceived heat stress vulnerability, with older individuals reporting lower levels of concern (β = −0.06, p = <.001). Housing tenure showed a positive association (β = 0.15, p = <.001), with renters tending to report higher perceived vulnerability. Income and education were not significant predictors. Next, there were no interaction effects, indicating that both age and tenure were independently predicting perceived heat stress vulnerability.

6 Regional Variability in Risk Perceptions of Different Environmental Hazards

This analysis examines how housing tenure is associated with perceived vulnerability to heat stress and flooding, controlling for regional variation. These two hazards are modelled separately because public concern and perceived personal vulnerability diverge across them. While 87% of respondents report being fairly or very concerned about increased flooding (compared to 73% for heatwaves), respondents are three times more likely to report being “definitely at risk” from heat stress than flooding. This suggests that general hazard concern and perceived personal vulnerability operate differently across environmental risks.

Housing tenure is relevant to both hazards but may operate through different mechanisms. For heat stress, tenure may influence adaptive capacity (e.g., control over ventilation, retrofitting, cooling adaptations), whereas for flooding it may relate more to location, insurance access, and structural protections. Moreover, Region is being controlled for as both heat exposure and flood risk are spatially uneven across the UK. By estimating separate models for heat and flood vulnerability, the analysis assesses whether tenure status has comparable, stronger, or weaker associations across hazard types. Accounting for regional variation reduces confounding and allows a clearer interpretation of tenure effects independent of geographical hazard distribution.

Lastly, the survey asks respondents about their perceived vulnerability to heat stress and flooding, however, it is important to bear in mind that the heat stress item is relevant to health whereas the flood item asks about the risk to one’s property. Therefore, the survey is measuring perceived risk towards personal health and an asset. This distinction is important as it may active different risk perceptions. It would not be unreasonable, however, to assume that flood risks to property also pose as risks towards one’s health, though because this is not what was asked by the survey it will not be a focus of the analysis.

6.1 How Does the Perceived Vulnerability to Heat Stress Vary Across the UK?

In the following model, region of the UK will be accounted for to see whether the perceived vulnerability to heat stress varies across region.

region_heat_model <- lmer(heat ~ housing_category + (1 | Region), data = cd)

	Perceived Risk of Heat Stress
Predictors	Estimates	std. Error	CI	p
Outright Owner (intercept)	1.74	0.05	1.65 – 1.83	<0.001
Mortgage Owner	0.08	0.04	0.00 – 0.15	0.040
Social Renter	0.20	0.04	0.12 – 0.27	<0.001
Private renter	0.21	0.04	0.14 – 0.29	<0.001
Random Effects
σ2	0.45
τ00 Region	0.01
ICC	0.02
N Region	6
Observations	2371
Marginal R2 / Conditional R2	0.018 / 0.037

6.2 How Does the Perceived Vulnerability to Flooding Vary Across the UK?

flood_region_model <- lmer(data = cd, flood ~ housing_category + (1 | Region))

	Perceived Risk of Flooding
Predictors	Estimates	std. Error	CI	p
Outright Owner (intercept)	1.27	0.02	1.23 – 1.32	<0.001
Mortgage Owner	0.10	0.03	0.04 – 0.16	0.001
Social Renter	0.13	0.03	0.07 – 0.19	<0.001
Private renter	0.10	0.03	0.04 – 0.17	0.001
Random Effects
σ2	0.32
τ00 Region	0.00
ICC	0.00
N Region	6
Observations	2371
Marginal R2 / Conditional R2	0.009 / 0.012

6.3 Model Summaries and Comparisons for Perceived Vulnerability to Heat Stress and Flood Risk Across Regions

6.3.1 Perceived Vulnerability to Heat Stress

A linear mixed-effects model was used to examine the association between housing tenure and heat vulnerability, with Region included as a random intercept to account for regional differences. The baseline category was outright owners. Random intercepts for Region were relatively small (SD = 0.094), indicating that while some regional variation exists, most variation in heat vulnerability occurs at the individual level.

Housing tenure was a significant predictor of heat vulnerability. Compared with outright owners, households with a mortgage had slightly higher predicted heat vulnerability (estimate = 0.078, p = 0.04), whereas social renters (estimate = 0.199, p < 0.001) and private renters (estimate = 0.214, p < 0.001) exhibited substantially higher predicted heat vulnerability. These results suggest a clear gradient in heat risk by tenure status, with renters—particularly those in the private or social rented sectors—experiencing higher vulnerability than homeowners.

Overall, the model indicates that tenure is an important determinant of heat vulnerability, even after accounting for regional differences, with the largest proportion of variation observed at the individual level rather than between Regions.

6.3.2 Perceived Vulnerability to Flooding

A linear mixed-effects model was used to examine the association between housing tenure and flood vulnerability, with Region included as a random intercept to account for potential regional differences. The baseline category was outright owners. Random intercepts for Region were very small (SD = 0.031), indicating minimal variation in flood vulnerability between Regions and suggesting that most variation occurs at the individual household level.

Housing tenure was a significant predictor of flood vulnerability. Compared with outright owners, households with a mortgage had slightly higher predicted flood vulnerability (estimate = 0.102, p = 0.001). Similarly, social renters (estimate = 0.131, p < 0.001) and private renters (estimate = 0.105, p < 0.001) exhibited higher predicted flood vulnerability. These results indicate a consistent gradient in risk, with renters, particularly those in the social rented sector, experiencing greater flood vulnerability than homeowners.

Overall, the model demonstrates that housing tenure is an important determinant of flood vulnerability, even when accounting for potential regional differences, with the majority of variability arising at the individual household level rather than across Regions.

6.3.3 Comparing the Two Models

From the two models examining vulnerability to heat stress and flooding, there are some notable differences. Firstly, heat stress shows larger housing gradients, particularly for renters, suggesting that tenure status is more strongly linked to perceived heat vulnerability than flood risk. Secondly, there are small regional effects for perceived vulnerability to heat more than flooding, implying that heat vulnerability perceptions are more geographically structured than flood risk perceptions. Overall, housing and region are more predictive of perceived heat stress than flood risk.

These results, though, may reflect stronger beliefs over the risk to one’s personal health than their property (as seen in the raw distribution of the data) and so this consideration must be made when looking at the data.

Outcome	Variance explained
Perceived heat stress vulnerability	3.7%
Perceived flooding vulnerability	1.2%

7 General Concern Over Impacts of Heatwaves

In the dataset, there were several questions relating to the risks around different environmental hazards, with many of these relating to heatwaves. These items related to an individuals concern over:

1. Disruption to public transport due to more or sustained heatwaves
2. Cities trapping in heat
3. New diseases being spread due to a warmer UK
4. Wildfires becoming more common
5. Buildings and schools not being safe during heatwaves

For the following analyses, a mean score for the above items will be taken to give a broad overview of the concern of heatwave impacts within the UK. Then, analyses looking at items 3 & 6 will follow to provide a better picture of concern over risks within a built environment.

7.1 Plotting the Distribution of the Data

7.2 General Concern Over the Impacts of Heatwaves by Housing Category

hw_risk_concern_model <- lm(data = cd_hi_grouped, cd_hi_mean ~ housing_category)

	Concern Over Heatwave Impacts
Predictors	Estimates	CI	p
Outright Owner (intercept)	3.21	3.17 – 3.26	<0.001
Mortgage Owner	0.16	0.10 – 0.23	<0.001
Social Renter	0.06	-0.01 – 0.12	0.079
Private renter	0.16	0.09 – 0.22	<0.001
Observations	2479
R2 / R2 adjusted	0.014 / 0.012

7.2.1 Model Summary for Concern for Heatwave Risks by Housing Category

The overall model was significant (F(3, 2475) = 11.18, p < .001), accounting for a small amount of variance (R² = 0.01). Main effects of housing were observed for Mortgage owners (β = 0.16, p < .001) and Private Renters (β = 0.15, p < .001). This suggests that concern about heatwave effects is elevated among mortgage holders and private renters, but not among social renters, compared to outright owners.

7.2.2 Interaction Model for Concern over Heatwave Risks

hw_risk_concern_demo_model <- lm(cd_hi_mean ~ housing_binary * income_c + housing_binary * age_c + education_num, data = cd_hi_grouped)

	Concern Over Heatwave Impacts
Predictors	Estimates	CI	p
Owners	3.25	3.16 – 3.33	<0.001
Renters	0.02	-0.03 – 0.08	0.466
Income	0.01	-0.00 – 0.03	0.096
Age	-0.04	-0.06 – -0.01	0.005
Education	0.01	-0.01 – 0.03	0.407
Tenure x Income	-0.02	-0.05 – 0.01	0.143
Tenure x Age	-0.00	-0.04 – 0.03	0.819
Observations	2292
R2 / R2 adjusted	0.012 / 0.010

7.3 Model Summary for General Concern for Heatwave Risks and Housing Category Whilst Controlling for Demographics

The effect of tenure on the general concern for heatwave risks was diminished when accounting for demographic variables. Risk perceptions among renter groups was not significant ((β = 0.02, p >.05), however, age was shown to have a main-effect on risk perceptions ((β = -0.04, p < .01), although this effect was minor. Next, there were no interactions observed between tenure and age.

7.4 Concern Over Heatwave Risks Within Built Environments

7.5 Concern over HRBE by Housing Category

hw_built_env_concern_model <- lm(data = cd_hi_built_env_grouped, cd_hi_built_env_mean ~ housing_category)

	Concern Over Heatwave Impacts Within Built Environments
Predictors	Estimates	CI	p
Outright Owner (intercept)	3.11	3.06 – 3.15	<0.001
Mortgage Owner	0.20	0.13 – 0.27	<0.001
Social Renter	0.10	0.03 – 0.17	0.004
Private renter	0.20	0.13 – 0.27	<0.001
Observations	2479
R2 / R2 adjusted	0.018 / 0.017

7.5.1 Model Summary for Concern Over HRBE by Housing Category

The overall model was significant (F(3, 2475) = 16.23, p < .001), accounting for a small amount of variance (R² = 0.02). Main effects of housing were observed for all housing groups, Mortgage owners (β = 0.23, p < .001) and Private Renters (β = 0.22, p < .001) and Social Renters (β = 0.13, p < .001). Interestingly, concern for HRBE was strongest in mortgage owners, not renters, although this difference was marginal.

7.5.2 Interaction Model for Concern Over HRBE

hw_built_env_demo_concern_model <- lm(data = cd_hi_built_env_grouped, cd_hi_built_env_mean ~ housing_binary * income_c +
                                        housing_binary * age_c +
                                        education_num)

	Concern Over HRBE
Predictors	Estimates	CI	p
Owners	3.13	3.04 – 3.23	<0.001
Renters	0.04	-0.02 – 0.10	0.149
Income	0.01	-0.00 – 0.03	0.121
Age	-0.05	-0.07 – -0.02	0.002
Education	0.01	-0.01 – 0.03	0.171
Tenure x Income	-0.02	-0.06 – 0.01	0.131
Tenure x Age	-0.01	-0.05 – 0.03	0.718
Observations	2292
R2 / R2 adjusted	0.018 / 0.015

7.6 Model Summary for Concern over HRBE and Housing Category whilst Controlling for Demographics

The effect of tenure on the concern over heatwave risks within built environments was absent when accounting for demographic variables. Risk perceptions among renter groups was not significant ((β = 0.04, p >.05), however, age was shown to have a main-effect on risk perceptions ((β = -0.05, p < .01), although this effect was minor. Next, there were no interactions observed between tenure and age.

7.7 Are There Regional Effects for HRBE within a London Population?

In this analysis, it is important to consider how heatwave risks for built environments interact with housing tenure. However, without looking at a specific population within a built environment, such as London, responses from rural populations may obscure the model outputs. Therefore, as London was the only city-specific regional variable recorded, it will be used to test whether the relationship between tenure status and HRBE is stronger within a London population.

hw_built_env_concern_london <- lm(cd_hi_built_env_mean ~ housing_category,
  data = subset(cd_hi_built_env_grouped, Region == "London"))

	Concern Over Heatwave Impacts Within Built Environments (London sample)
Predictors	Estimates	CI	p
Outright Owner (intercept)	3.18	3.05 – 3.32	<0.001
Mortgage Owner	0.14	-0.07 – 0.35	0.202
Social Renter	0.14	-0.05 – 0.32	0.142
Private renter	0.24	0.04 – 0.43	0.019
Observations	293
R2 / R2 adjusted	0.019 / 0.009

7.7.1 Summary of Regional Effects for HRBE within a London Population

The overall model did not reach statistical significance (F(3, 289) = 2.157, p = 0.093). Despite there being main effects of housing, such that private renters had stronger concerns over heatwave risks within built environments more so than outright owners, it would be unlikely that these effects are robust given the small amount of variance accounted for by the model (R² = 0.01).

7.8 Interaction Model for Regional Effects for HRBE

hrbe_london_demo_model <- lm(cd_hi_built_env_mean ~ housing_binary * income_c + housing_binary * age_c + education_num,
  data = subset(cd_hi_built_env_grouped, Region == "London"))

	Concern Over Heatwave Impacts
Predictors	Estimates	CI	p
Owners	3.20	2.91 – 3.49	<0.001
Renters	0.15	-0.01 – 0.31	0.072
Income	0.03	-0.01 – 0.07	0.120
Age	-0.05	-0.13 – 0.02	0.151
Education	-0.00	-0.06 – 0.05	0.908
Tenure x Income	-0.04	-0.11 – 0.02	0.187
Tenure x Age	-0.01	-0.12 – 0.09	0.775
Observations	285
R2 / R2 adjusted	0.046 / 0.025

7.8.1 Summary of Interaction Model

When using a subset population of London only, the MLR model showed that there were no main effects of tenure or any other demographic variables. Tenure was marginally non-significant (β = 0.15, p = .07), suggesting that the effect of tenure could be obscured by the differences in private renters (β = 0.24, p = .02) and social renters (β = 0.14, p = .14) as seen in the previous model.

8 Wave 1 Only Analysis

One of the measured variables in the first wave of the survey related to the level of concern participants had over more frequent climate-change related events by 2050. This variable was not used in the second wave, however, it remains a variable of interest.

The scale used to measure people’s concern over more frequent climate-change related events was:

1. Not at all concerned
2. Not very concerned
3. Fairly concerned
4. Very concerned

8.1 Plotting the Distribution of the Data

8.2 The Relationship Between Concern Over Increased Heatwave Frequency and Housing Category

hc_model <- lm(data = cd_hc_heat_grouped, cd_hc_heat_mean ~ housing_category)

	Concern Over Increased Heatwave Frequency
Predictors	Estimates	CI	p
Outright Owner (intercept)	2.85	2.77 – 2.94	<0.001
Mortgage Owner	0.23	0.10 – 0.36	0.001
Social Renter	0.15	0.02 – 0.28	0.025
Private renter	0.25	0.12 – 0.38	<0.001
Observations	1378
R2 / R2 adjusted	0.013 / 0.011

The model overall was significant F(3, 1374) = 8.39, p < .001, with housing category showing to have a small effect (R² = 0.02). Compared to outright owners, all groups showed significantly more concern for an increase in heatwave frequency (Mortgage, (β = 0.245, p < .001) and Social renters (β = 0.186, p = .003)), with Private Renters showing the greatest concern (β = 0.250, p < .001).

8.2.1 Interaction Model for Concern Over Increase in Heatwave Frequency

hc_demo_model <- lm(cd_hc_heat_mean ~ housing_binary * income_c + housing_binary * age_c + education_num, data = cd_hc_heat_grouped)

	Concern Over Heatwave Impacts
Predictors	Estimates	CI	p
Owners	2.77	2.60 – 2.94	<0.001
Renters	0.03	-0.08 – 0.13	0.645
Income	0.01	-0.02 – 0.04	0.572
Age	-0.10	-0.15 – -0.05	<0.001
Education	0.05	0.01 – 0.08	0.010
Tenure x Income	-0.03	-0.09 – 0.03	0.281
Tenure x Age	-0.02	-0.09 – 0.05	0.548
Observations	1301
R2 / R2 adjusted	0.044 / 0.039

8.3 Summary of Concern Over an Increase in Heatwave Frequency

Tenure was no longer an independent predictor for the concern over heatwave impacts as previously observed in the model prior. When including demographics, Age (β = -0.10, p < .001) and Education (β = 0.01, p = .01) emerged as significant predictors of heatwave frequency concerns.

9 Comparative Interpretations for Risk Perceptions

9.0.1 Unadjusted Models

	Heat Risk Seriousness				Heat Stress Risk				Heat Stress Risk (Region)				Flood Risk				Heatwave Concern				Built Environment Heat Concern				Built Environment Concern (London)				Heatwave Frequency Concern
Predictors	Estimates	std. Error	CI	p	Estimates	std. Error	CI	p	Estimates	std. Error	CI	p	Estimates	std. Error	CI	p	Estimates	std. Error	CI	p	Estimates	std. Error	CI	p	Estimates	std. Error	CI	p	Estimates	std. Error	CI	p
(Intercept: Outright Owner)	2.90	0.03	2.85 – 2.96	<0.001	1.76	0.02	1.71 – 1.81	<0.001	1.74	0.05	1.65 – 1.83	<0.001	1.27	0.02	1.23 – 1.32	<0.001	3.21	0.02	3.17 – 3.26	<0.001	3.11	0.02	3.06 – 3.15	<0.001	3.18	0.07	3.05 – 3.32	<0.001	2.85	0.04	2.77 – 2.94	<0.001
Mortgage Owner	0.15	0.05	0.06 – 0.24	0.001	0.08	0.04	0.00 – 0.15	0.040	0.08	0.04	0.00 – 0.15	0.040	0.10	0.03	0.04 – 0.16	0.001	0.16	0.03	0.10 – 0.23	<0.001	0.20	0.04	0.13 – 0.27	<0.001	0.14	0.11	-0.07 – 0.35	0.202	0.23	0.07	0.10 – 0.36	0.001
Social Renter	0.13	0.05	0.04 – 0.22	0.005	0.20	0.04	0.12 – 0.27	<0.001	0.20	0.04	0.12 – 0.27	<0.001	0.13	0.03	0.07 – 0.19	<0.001	0.06	0.03	-0.01 – 0.12	0.079	0.10	0.04	0.03 – 0.17	0.004	0.14	0.09	-0.05 – 0.32	0.142	0.15	0.07	0.02 – 0.28	0.025
Private Renter	0.16	0.05	0.07 – 0.25	<0.001	0.22	0.04	0.14 – 0.29	<0.001	0.21	0.04	0.14 – 0.29	<0.001	0.10	0.03	0.04 – 0.17	0.001	0.16	0.03	0.09 – 0.22	<0.001	0.20	0.04	0.13 – 0.27	<0.001	0.24	0.10	0.04 – 0.43	0.019	0.25	0.07	0.12 – 0.38	<0.001
Random Effects
σ2									0.45				0.32
τ00									0.01 Region				0.00 Region
ICC									0.02				0.00
N									6 Region				6 Region
Observations	2370				2371				2371				2371				2479				2479				293				1378
R2 / R2 adjusted	0.007 / 0.006				0.019 / 0.018				0.018 / 0.037				0.009 / 0.012				0.014 / 0.012				0.018 / 0.017				0.019 / 0.009				0.013 / 0.011

9.0.2 Adjusted Models

	Heat Risk Seriousness				Heat Stress Vulnerability				Heatwave Concern				Built Environment Heat Concern				Heatwave Concern (London)				Heatwave Frequency Concern
Predictors	Estimates	std. Error	CI	p	Estimates	std. Error	CI	p	Estimates	std. Error	CI	p	Estimates	std. Error	CI	p	Estimates	std. Error	CI	p	Estimates	std. Error	CI	p
(Intercept: Mean levels)	2.89	0.06	2.77 – 3.01	<0.001	1.73	0.05	1.63 – 1.83	<0.001	3.25	0.04	3.16 – 3.33	<0.001	3.13	0.05	3.04 – 3.23	<0.001	3.20	0.15	2.91 – 3.49	<0.001	2.77	0.09	2.60 – 2.94	<0.001
Housing (Renter vs Owner)	0.06	0.04	-0.02 – 0.14	0.123	0.15	0.03	0.09 – 0.21	<0.001	0.02	0.03	-0.03 – 0.08	0.466	0.04	0.03	-0.02 – 0.10	0.149	0.15	0.08	-0.01 – 0.31	0.072	0.03	0.06	-0.08 – 0.13	0.645
Income (centred)	0.00	0.01	-0.02 – 0.02	0.788	0.00	0.01	-0.01 – 0.02	0.678	0.01	0.01	-0.00 – 0.03	0.096	0.01	0.01	-0.00 – 0.03	0.121	0.03	0.02	-0.01 – 0.07	0.120	0.01	0.02	-0.02 – 0.04	0.572
Age (centred)	-0.05	0.02	-0.09 – -0.02	0.004	-0.06	0.02	-0.09 – -0.03	<0.001	-0.04	0.01	-0.06 – -0.01	0.005	-0.05	0.01	-0.07 – -0.02	0.002	-0.05	0.04	-0.13 – 0.02	0.151	-0.10	0.03	-0.15 – -0.05	<0.001
Education	0.02	0.01	-0.01 – 0.04	0.146	0.02	0.01	-0.00 – 0.04	0.107	0.01	0.01	-0.01 – 0.03	0.407	0.01	0.01	-0.01 – 0.03	0.171	-0.00	0.03	-0.06 – 0.05	0.908	0.05	0.02	0.01 – 0.08	0.010
Tenure × Income	-0.01	0.02	-0.05 – 0.03	0.579	0.00	0.02	-0.03 – 0.04	0.858	-0.02	0.01	-0.05 – 0.01	0.143	-0.02	0.02	-0.06 – 0.01	0.131	-0.04	0.03	-0.11 – 0.02	0.187	-0.03	0.03	-0.09 – 0.03	0.281
Tenure × Age	-0.01	0.03	-0.06 – 0.04	0.683	-0.01	0.02	-0.05 – 0.03	0.696	-0.00	0.02	-0.04 – 0.03	0.819	-0.01	0.02	-0.05 – 0.03	0.718	-0.01	0.05	-0.12 – 0.09	0.775	-0.02	0.04	-0.09 – 0.05	0.548
Observations	2188				2189				2292				2292				285				1301
R2 / R2 adjusted	0.016 / 0.013				0.036 / 0.034				0.012 / 0.010				0.018 / 0.015				0.046 / 0.025				0.044 / 0.039

Housing tenure shows its clearest and most substantial association with perceived personal vulnerability to heat stress. After accounting for age, income and education, there generally was no effect of tenure, other than for the perceived vulnerability to heat stress variable. For the demographic variables, age appeared as a strong predictor of lower risk perceptions across several risk types (apart from concerns about heatwaves in a London sample), and education showed a minor positive association for concerns over an increase in heatwave frequency. For the unadjusted models, tenure shows strong effects likely because of the confounds of age. For example, in the sample characteristics, proportionately, renters were younger and homeowners were older. The attenuation of tenure after adjustment suggests that differences in age across tenure groups account for a substantial portion of the observed unadjusted associations.

Importantly, the effect of tenure remained for personal vulnerability to heat stress. One possible interpretation of this is that tenure differences are more salient when individuals evaluate immediate or personally experienced risks, whereas abstract or future-orientated risks show weaker effects of tenure.

10 Heat Adaptation

10.1 Support for Policies in Relation to Heat Adaptation

Within the survey, there were items which asked participants whether they would support certain adaptation policies. There was one item relating to heat adaptation, which asked about introducing tighter regulations on buildings to be able to deal with hotter and drier weather (e.g., insulation and air-conditioning).

10.1.1 Plotting the Distribution of the Data

10.1.2 The Relationship Between Support for Heat Adaptation Policy and Housing Category

adap_model <- lm(policy1_8 ~ housing_category, data = cd)

	Heat Adaptation Policy
Predictors	Estimates	CI	p
Outright Owner (intercept)	4.04	3.98 – 4.10	<0.001
Mortgage Owner	0.05	-0.05 – 0.15	0.316
Social Renter	-0.08	-0.18 – 0.01	0.092
Private renter	0.03	-0.07 – 0.12	0.564
Observations	2370
R2 / R2 adjusted	0.003 / 0.002

10.2 Intention to Engage with Heat Adaptation Measures

The survey included three heat adaptation items, each pertaining to different aspects of adaptation. There was one ‘soft’ measure, which asked whether people would read about avoiding heat stress, and two ‘hard’ measures which pertained to increasing shade in their gardens and installing AC. In order to fully understand the intentions between renters and homeowners, the data will first be aggregated to include an overall intention score for all measures, then the measures will be analysed separately. The reason behind this is that some intentions may vary across measures and combining them may obscure some of the nuances behind people’s intentions.

10.2.1 Plotting the Distribution of the Data

adap_behav_model <- lm(data = cd_ha_grouped, cd_ha_mean ~ housing_category)

	Intentions Across all Adaptation Measures
Predictors	Estimates	CI	p
Outright Owner (intercept)	3.12	3.06 – 3.18	<0.001
Mortgage Owner	0.24	0.14 – 0.33	<0.001
Social Renter	0.14	0.04 – 0.24	0.005
Private renter	0.13	0.04 – 0.23	0.007
Observations	2371
R2 / R2 adjusted	0.010 / 0.009

10.3 Analysing Heat Adaptation Measures Separately

	Increase Shade			Install AC			Increase Heat Literacy
Predictors	Estimates	CI	p	Estimates	CI	p	Estimates	CI	p
Outright Owner (intercept)	3.36	3.27 – 3.44	<0.001	2.38	2.30 – 2.47	<0.001	3.62	3.54 – 3.69	<0.001
Mortgage Owner	0.25	0.11 – 0.38	<0.001	0.33	0.19 – 0.46	<0.001	0.13	0.01 – 0.25	0.033
Social Renter	0.04	-0.09 – 0.17	0.531	0.33	0.19 – 0.46	<0.001	0.04	-0.08 – 0.17	0.470
Private renter	0.01	-0.12 – 0.14	0.849	0.20	0.06 – 0.33	0.004	0.19	0.07 – 0.30	0.002
Observations	2370			2369			2369
R2 / R2 adjusted	0.007 / 0.005			0.013 / 0.012			0.005 / 0.003

Warning: `geom_errobarh()` was deprecated in ggplot2 4.0.0.
ℹ Please use the `orientation` argument of `geom_errorbar()` instead.

10.4 Housing Category as a Moderator for the Relationship Between Heat Risk Perceptions and Adaptation Intentions

10.4.1 Is There a Relationship Between Heat Risk Perceptions and Adaptation Intentions?

rp_ai_model <- lm(cd_ha_mean ~ heat, data = cd_ha_grouped)

	Adaptation Intentions
Predictors	Estimates	CI	p
Intercept	2.37	2.27 – 2.47	<0.001
Personal Risk to Heat Stress	0.46	0.41 – 0.51	<0.001
Observations	2371
R2 / R2 adjusted	0.126 / 0.126

10.4.2 Testing Whether Housing Category is a Moderator

	Adaptation Intentions
Predictors	Estimates	CI	p
Outright Owner (Intercept)	3.17	3.11 – 3.23	<0.001
Personal Risk to Heat Stress	0.48	0.39 – 0.56	<0.001
Mortgage Owner	0.20	0.10 – 0.29	<0.001
Social Renter	0.04	-0.05 – 0.13	0.402
Private renter	0.03	-0.06 – 0.12	0.532
HS*Mortgage Owner	-0.11	-0.24 – 0.03	0.126
HS*Social Renter	0.05	-0.09 – 0.18	0.496
HS*Private Renter	-0.02	-0.15 – 0.11	0.788
Observations	2371
R2 / R2 adjusted	0.135 / 0.132

10.4.3 Summary of Housing Category as a Moderator for the Relationship Between Heat Risk Perceptions and Adaptation Intentions

The absence of moderation suggests that tenure does not influence how people translate heat risk into adaptation intentions. In other words, perceived risk appears to motivate action in broadly similar ways across tenure groups. What differs is the baseline level of intention, which is only observed in mortgage owners (different from outright owners) and not any other tenure group. This implies that tenure does not clearly predict adaptive capacity.