This article explores the question of what we can consider to be real in drug policy. It examines two increasingly common aspects of drug policy analysis; radical constructionist critique and successionist data science. It shows how researchers using these assumptions have produced interesting findings, but also demonstrates their theoretical incoherence, based on their shared 'flat ontology'. The radical constructionist claim that reality is produced within research methods - as seen in some qualitative studies - is shown to be unsustainably self-defeating. It is analytically 'paralyzing'. This leads to two inconsistencies in radical constructionist studies; empirical ambivalence and ersatz epistemic egalitarianism. The Humean successionist approach of econometric data science is also shown to be unsustainable, and unable to provide explanations of identified patterns in data. Four consequent, limiting characteristics of this type of drug policy research are discussed: causal inference at a distance, monofinality, limited causal imagination, and overly confident causal claims.

The article goes on to describe the critical realist approach towards 'depth ontology' and 'generative causation'. It provides examples of how this approach is deployed in critical realist reviews and discourse analysis of drug policy. It concludes by arguing that critical realism enables more deeply explanatory, methodologically eclectic and democratically inclusive analysis of drug policy development and effects.