{"doi":"10.1111/add.15595","title":"Commentary on Robinson <i>et al</i> . (2021): Evaluating theories of change for public health policies using computer model discovery methods","abstract":"Recent developments in computer modelling—known as model discovery—could help to confirm the mechanisms underpinning Robinson and colleagues' important early findings for the effectiveness of minimum unit pricing, and to test the complete theory of change underpinning this crucial evaluation. Robinson and colleagues provide important evidence for understanding the short-term impact of minimum unit pricing (MUP) in Scotland [1]. Their finding of a 3.5% reduction in off-trade alcohol sales since the introduction of the policy on 1 May 2018 aligns with previous research reporting a reduction in alcohol purchasing by households in the first 8 months post-implementation [2]. Evaluators increasingly seek evidence on the mechanisms underlying policy effects. Mechanism-focused evidence is important because it helps identify wider secondary effects (positive or negative), helps interpret claims of causality, supports assessments of longer-term effectiveness and can inform the transfer of policies to different contexts [3]. It is reassuring that the present study is part of a much broader MUP evaluation programme examining a variety of mechanisms [4]. The programme is informed by a theory of change that identifies a central pathway anticipated to determine policy effectiveness and additional pathways, which could interact with the policy [4]. The central pathway posits that retailer compliance with MUP legislation increases retail prices of some alcohol products; these price increases lead to reduced purchasing and consumption of alcohol products, which reduces the risk of alcohol-related harm to individuals and society. This pathway aligns with computer model-based appraisals that informed the MUP decision [5, 6]. The additional pathways consider market responses to MUP and impacts on the alcohol industry, changes in social norms and attitudes, changes in spending on food and essentials among heavier drinkers and increased demand for specialist alcohol treatment services. The theory of change explicitly acknowledges that MUP impact may also be shaped by external factors, such as changes in disposable income. One possibility absent from the programme is the use of formal computer modelling to test and confirm the rich set of mechanisms in the MUP theory of change using an approach known as model discovery. Developed in the physical sciences, model discovery combines computer modelling with machine learning to search through the possible explanations for a phenomenon to identify which explanation can best represent the observed data [7]. Model discovery goes beyond best practice approaches for calibrating models to data [8], by identifying not just model parameters, but also model structures (i.e. the formal equations and their corresponding computer code). State of the art model discovery techniques are now moving to investigate social phenomena that involve contested explanations and reflexive human behaviour, including alcohol use [9-11]. Our own approach draws on the work of the American sociologist, James Coleman, who developed a typology of explanatory mechanisms based on individuals and social structures such as norms and institutions [12]. In terms of the MUP evaluation, Coleman's typology suggests we can model how social norms interact with MUP using three mechanism types: situational mechanisms explain how social structures affect individuals (i.e. how social norms are transmitted between and perceived by individuals); action mechanisms explain how perceived norms influence individuals’ drinking behaviours; and transformational mechanisms explain how drinking behaviours collectively change norms as social structures. These mechanisms can most naturally be expressed as the ‘building blocks’ of an agent-based model (ABM)—a computational approach to modelling interacting agents and social structures that is gaining prominence in social psychology and the addictions field [13, 14]. Model discovery has already identified competing e","journal":"Addiction","year":2021,"id":221140,"datarank":0.0,"base_score":0.0,"endowment":0.0,"self_citation_contribution":0.0,"citation_network_contribution":0.0,"self_endowment_contribution":0.0,"citer_contribution":0.0,"corpus_percentile":null,"corpus_rank":null,"citation_count":1,"citer_count":0,"citers_with_citation_signal":0,"citers_with_endowment":0,"datacite_reuse_total":0,"is_dataset":false,"is_dataset_confidence":0.9492,"is_data_producer":false,"deposit_databanks":null,"is_oa":true,"file_count":0,"downloads":0,"has_version_chain":false,"published_date":"2021-01-01","fair_score":null,"fair_percentile":null,"algorithm_id":"datarank_citation_only_1hop_v6","ranking_scope":"data_only","authors":[{"id":392911,"name":"Charlotte Buckley","orcid":"0000-0002-8430-0347","position":1,"is_corresponding":false},{"id":308318,"name":"Alan Brennan","orcid":"0000-0002-1025-312X","position":2,"is_corresponding":false},{"id":240671,"name":"John Holmes","orcid":"0000-0001-9283-2151","position":3,"is_corresponding":false},{"id":392915,"name":"Robin C. Purshouse","orcid":"0000-0001-5880-1925","position":0,"is_corresponding":true}],"reference_count":19,"raw_metadata":null,"created_at":"2026-07-18T23:53:50.838581Z","pmid":"34184346","pmcid":null,"fwci":null,"citation_percentile":null,"influential_citations":0,"oa_status":null,"license":null,"views":0,"total_file_size_bytes":0,"version_count":0,"fair_f":null,"fair_a":null,"fair_i":null,"fair_r":null,"fair_zscore":null,"fair_rationale":null,"fair_model":null,"fair_agent_version":null,"fair_fulltext_source":null,"fair_has_llm":null,"fair_computed_at":null,"clinical_trials":[],"software_tools":[],"db_accessions":[],"linked_datasets":[],"topics":[]}