Reshaping energy policy based on social and human dimensions: an analysis of human-building interactions among societies in transition in GCC countries

The study addresses how human and social dimensions shape energy use in the building sector in rapidly developing, demographically diverse GCC countries, using Qatar as a representative case. Buildings account for about 40% of annual global GHG emissions, with occupants’ behaviors introducing significant stochasticity and performance gaps between predicted and actual consumption. The research question centers on which demographic, socioeconomic, behavioral, and building-related factors most strongly influence human-building interactions and energy consumption patterns, and how these insights can inform targeted, human-centered energy policies (e.g., pricing, awareness, incentives, DR programs) that balance energy efficiency and occupant comfort/productivity. The study is motivated by the GCC context of high cooling loads, subsidized energy, and populations comprised largely of expatriates and migrant workers, where understanding human factors is essential for effective policy design and sustainability outcomes.

The paper reviews evidence that occupant behavior significantly affects building energy use and comfort, with wide variability driven by demographics (age, gender, culture), psychology, and building characteristics. Past work highlights that techno-economic approaches alone are insufficient to achieve energy targets or comfort; human-building interactions (HBIs) are a critical yet under-addressed factor. Barriers to energy efficiency (e.g., bounded rationality, imperfect information, split incentives) underscore the need for policy interventions that integrate human behavior. Studies link demographics to thermal comfort, energy audit uptake, and energy burdens; socioeconomic status to energy expenses, audit participation, and efficiency; and attitudes/awareness to energy-saving behaviors and DR participation. Evidence from EU and China indicates policy effectiveness improves when shifting toward outcome management and behavioral insights. Research also shows building age and features modulate behavior (e.g., setpoints), and actual consumption often far exceeds predictions. Behavioral drivers (norms, attitudes, responsibility, trust) and program design (incentives, convenience) influence DR participation and energy-saving actions.

- Setting and sample: Doha, Qatar; survey of 2200 participants representing a highly diverse population (about 90% expatriates and migrant workers). - Data collected: demographics (age, gender, ethnicity); socioeconomics (household income, expenses, marital, employment, ownership/rental); residential building attributes (construction year, type, floor area, AC technology, modern/old style); workplace attributes (construction year, type/style, perceived technological advancement); behavioral and psychological factors (climate change consequence awareness, motivations for home energy efficiency, perceived responsibility drivers, habits); indoor comfort perceptions and preferences (thermal and lighting at home and work); human-building interactions (thermostat override, window opening); financial drivers (monthly electricity bills, whether bills are included/subsidized; price sensitivity; DR willingness and preferred incentives). - Derived variables: Energy Use Intensity (EUI) proxy calculated as average monthly electricity bill divided by home floor area (Qatar’s residential rates are not dynamic, enabling this comparison). - Clustering: k-means on EUI to classify consumption patterns into low and high consumers. Due to missing data, 1021 complete cases were used; clustering yielded 829 low consumers (category 1; centroid EUI ≈ 2.89 in bill-per-m² units) and 192 high consumers (category 2). - Feature importance analysis: Two random forest classifiers (500 trees; bootstrap aggregating; random patches; Gini impurity) with 10-fold cross-validation. 1) Model A inputs: age range, gender, ethnicity group, household income, monthly expenses; target: consumption class (high/low). Average accuracy 93.5% (range 90.0–97.1%). Feature importance assessed via Gini, permutation importance, and SHAP. 2) Model B inputs: awareness of consequences, motives for home energy efficiency (financial/social/environmental/none), perceived responsibility-improvement factor (education/media/incentives/coercive); target: consumption class. Average accuracy 92.6% (range 90.0–95.1%). Feature importance assessed via the same three techniques. - Analysis of distributions and associations: Kernel density and violin plots for comfort perceptions; cross-tabulations and visualizations for demographic/socioeconomic and building attribute associations with comfort, behaviors, and financial drivers; examination of DR participation preferences by housing and ownership type. - Simulations context: EnergyPlus simulations for Doha indicate 46–61% of annual electricity use in buildings is devoted to cooling, framing the importance of HBIs in cooling demand.

- Consumption classification and predictors: - k-means separated 1021 complete cases into 829 low and 192 high consumers based on monthly bill-per-m² EUI. - Random forest feature importance (Model A) ranked predictors: household expenses (strongest), age, ethnicity, gender, income. Monthly household expenses better predicted consumption class than income, likely encoding building/appliance characteristics and use patterns. Model accuracy averaged 93.5%. - Random forest feature importance (Model B) ranked: perceived responsibility factor (strongest), energy efficiency motives, awareness. Awareness alone did not necessarily determine consumption class. Model accuracy averaged 92.6%. - Financial drivers and DR: - 52.9% reported bills included/exempt (waived or included in rent); average monthly bills were right-skewed. - 60% said higher electricity bills would lead them to reduce consumption; price sensitivity stronger among males, and Arab, Asian, and Indian groups; less pronounced in higher-income groups. - If bills are waived/included, 32.7% expected to use more, 48.8% no change, 18.5% unsure; likelihood of increased use higher among Arab and Indian groups and lower-income respondents; more females reported no change. - DR willingness: 46.4% opposed DR. Among those open to DR, 34.6–34.7% favored partial credit (percentage of monthly bill), 16.2–16.3% favored additional fixed credit, and 49.1% would not accept offers. Indicative attractive incentives: ~20% of monthly bill as partial credit or ~500 QR additional credit. - DR participation more likely among apartment residents, renters, and occupants of older-style buildings; homeowners and villa occupants were largely uninterested. - Comfort, behavior, and workplace: - At work, 45.3% preferred temperatures below 22°C; 11.3% preferred above 24°C; many preferred moderate lighting. Female respondents tended to favor colder temps compared with males’ moderate preference. Arabs and North Americans favored colder, Asians favored hotter temps. - Ability to adjust setpoints/windows varied; where possible, roughly half reported overriding thermostats or opening windows; those reporting thermal discomfort were more likely to engage in such overrides. - Thermal and lighting comfort at work were reported as important to performance by a majority (around two-thirds); yet perceived actual comfort lagged perceived importance. - Open office layouts were associated with greater reported discomfort; personal offices with higher comfort. - Residential comfort and building attributes: - Home comfort is important for quality of life; however, not all reported being comfortable at home. - Newer constructions (post-1990s), modern homes, villas and Arabic-style homes showed higher reported comfort than flats/part-of-unit. Central ducted and wall-mounted AC provided higher comfort than window AC and ceiling fans. - Attitudes and motivations: - About 44% were not sure or not concerned about climate change; concern was higher among females and highest-income respondents, and lower among Arab and Indian groups. - Main motives for home energy efficiency: financial (about 40.85%), with 24.1% social benefits and 29.1% environmental benefits; financial motives more important among males and lower-income groups; Qataris more often cited social/environmental motives. - Perceived drivers to improve energy responsibility: 40.8% favored coercive actions/law enforcement; others cited media, financial incentives, and education. Females reported higher self-perceived responsibility at work (about 65% positive overall; 29% not responsible). - Consumer class associations: - High consumption patterns were more likely among older age groups; North American and Arab ethnicity groups appeared more often in high-consumption classes. - Respondents motivated primarily by financial factors were more represented among high consumers. - Contextual energy end-uses: - EnergyPlus simulations for Doha show cooling comprises about 46–61% of annual building electricity, highlighting the centrality of occupant behavior in managing cooling loads.

Findings demonstrate that integrating human dimensions into energy policy can materially improve targeting and effectiveness in GCC contexts. Household expenses, age, and ethnicity are strong predictors of consumption behavior, enabling tailored interventions (e.g., targeted pricing, segmented awareness campaigns, incentives focused on specific demographics). Awareness alone did not classify consumption class, suggesting that combining motivations and perceived responsibility with demographics and expenses provides better leverage points for policy. The analysis shows how subsidies and waived bills can weaken price signals and risk overconsumption in some segments, indicating the need to reconsider subsidy structures and rental marketing practices involving utilities. Comfort perceptions and workplace attributes correlate with behaviors (e.g., thermostat overrides), indicating that better comfort management can reduce adverse interactions and save energy. Residential building attributes (modernity, AC type) link to higher comfort and likely efficiency, supporting targeted retrofit programs for low-comfort/low-efficiency segments. DR program design should factor in user convenience, equity, and tailored incentives; apartment residents and renters are more receptive, while homeowners and villa occupants are more resistant. Overall, addressing social norms (e.g., law enforcement, media), personalized incentives, and community-based education, alongside techno-economic measures, can reduce uncertainty in HBIs, close performance gaps, and align energy efficiency with occupant well-being and productivity.

The study contributes an empirically grounded framework for reshaping energy policy around human and social dimensions in GCC-like contexts. Using a large survey (n=2200) in Doha, Qatar, and machine learning analyses, it identifies key demographic, socioeconomic, behavioral, and building factors that drive consumption and comfort-related behaviors. Clustering and feature importance reveal that household expenses, age, and ethnicity, along with responsibility/motivation factors, effectively differentiate high- and low-consumption households. Practical implications include: targeted electricity pricing and subsidy reforms; community-tailored awareness and education; incentive models for building efficiency and DR that reflect user motivations and convenience; and guidelines that balance comfort and energy savings in residential and workplace settings. Policymakers should also address regulatory barriers to flexible pricing and DR, incentivize private-sector efficiency investments, and ensure DR program quality (privacy, reliability, security). Future research should analyze updated datasets including the COVID-19 period, decompose end-use consumption, and conduct finer-grained behavioral and technical evaluations to further refine human-centered policy instruments.

- Generalizability: Findings are based on Qatar (Doha) and may not fully generalize across all GCC contexts despite similarities. - Self-reported survey data: Subject to recall and social desirability biases; building entry and direct measurements were limited. - EUI proxy: Consumption class derived from monthly bill per floor area; actual metered end-use data and dynamic tariffs were not used; residential rates are assumed non-dynamic. - Missing data: Clustering used 1021 respondents with complete data, reducing sample for some analyses. - Cross-sectional design: Limits causal inference; observed associations may be confounded by unmeasured variables (e.g., precise building performance parameters, appliance efficiency). - Data availability: Datasets are not publicly available due to collaboration agreements and respondent anonymity, limiting external replication.