We are pleased that our study has been discussed among a group of practising psychiatrists and appreciate Dr Yates' interest.

We disagree, however, that the clinical applicability of the study findings are limited. As van Os highlighted in the accompanying editorial to our paper, ^{Reference van Os1} psychotic experiences elicited through the Psychosis Screening Questionnaire (PSQ) are probably indicative of a dimensional phenotype which is frequently present in common mental disorders, as well as present in psychotic disorders. Although a proportion of people screening positive on the PSQ may later convert to clinical psychosis (cited in van Os ^{Reference van Os1} and Das-Munshi et al ^{Reference Das-Munshi, Becares, Boydell, Dewey, Morgan and Stansfeld2} ), for common mental disorders the presence of these symptoms usually suggests a poorer prognosis. ^{Reference van Os1} The clinical applicability of our findings is that there may be important risk factors in the environment which are associated with an increased risk of mental disorders in ethnic minority groups. From both a public health perspective and a clinical perspective, the findings suggest ways in which one may intervene to modify these social risk factors. As van Os suggests, ^{Reference van Os1} the advantage of using the PSQ as an outcome measure also meant that the findings of the study cut across traditional diagnostic boundaries. This may be why we found similar associations in a related analysis on common mental disorders. ^{Reference Das-Munshi, Becares, Dewey, Stansfeld and Prince3}

We wholeheartedly agree that associations of psychotic experiences with ethnic density should be seen within a multifactorial model where individual-level factors are understood from within a contextual framework. The inference of causality in a complex condition such as psychosis (or any other mental disorder) should be multifactorial, and should also assess possible interactions between causes, ^{Reference Rothman and Greenland4} as we believe our study has done.

The assertion that ‘the general trend was supportive’ of density associations for all of the ethnic minority groups was a reference to Fig. 1 in the paper, which suggests that only the White British group were not party to protective density effects. Focusing on an arbitrary significance cut-off of P = 0.05 masks an understanding of the study, which we feel must also be informed by looking at effect sizes as well as associated confidence intervals. ^{Reference Gardner and Altman5}

An advantage of our analysis was the ability to examine contextual and individual-level experiences by ethnic group. We did assess and present data relating to the ‘combined’ ethnic minority sample as well (Figs 2–4). ^{Reference Das-Munshi, Becares, Boydell, Dewey, Morgan and Stansfeld2} Although we assert that research should refrain from ‘lumping’ minority groups together, some experiences (especially those relating to discrimination, social support and adversity) may have salience and cut across cultural or ethnic differences, and so we felt justified in presenting this for the full sample.

Finally, Dr Yates' suggestion for some of the ‘cross-over effects’ shown in the interaction models are fully justified as this is a cross-sectional study and so temporality cannot be assumed. We are pleased that the study lends speculation as well as an understanding to this area, and hope that future research may help address some of the issues highlighted by our paper.