Goal setting and signposting in a rasch-based recommender system to promote household energy conservation

Recent studies have used algorithm tailoring on digital platforms to provide household energy-saving advice. Such ‘recommender systems’ have successfully used the psychometric Rasch model as an advice algorithm, matching energy-saving measures in terms of their difficulty to consumers’ ability levels. While these previous studies indicated positive user experiences, tailored advice did not lead to higher savings overall; not even when also using persuasive nudges, such as displaying social norm percentages in the system. One possible reason for these results was that the system was used exploratively, allowing users to pick energy measures as they liked without tapping into goal setting or value-based motivational frames (e.g., signposts). In this study, 202 participants used and evaluated our ‘Saving Aid’ Rasch recommender system, choosing energy-saving measures they would like to perform at home. Through a 3 × 2-between subject design, we examined whether guided goal setting and signposts (kWh/Euro/CO2) affected user experience and energy savings. Following the signpost literature, we examined the moderation of these effects by user values, such as environmental concern (New Environmental Paradigm (NEP) score). A structural equation model analysis revealed that goal setting did not affect outcome variables, while signpost framing had varying effects, although these were not in line with prior expectations. Still, the overall system remains promising, with users achieving a 316 kWh yearly savings with the chosen recommendations.