The perfect model structure would see theory (the AO1 knowledge) being separate from evaluation, but should also show them linked to each other effectively.
For instance, when talking about the biological explanations of depression, you may discuss genetic factors. You'd establish that if depression was passed genetically, in adoption studies, there would be a higher incidence of depression in the biological relatives of a depressed child. Once this has been established, to tie in your evaluation (AO2) point, it's best to use a strong linkback phrase such as 'this is shown by..' or 'highlighted in..', and use a relevelant case study that has been studied in class. A good example here would be the Wedner et al case, where it was found that there was a higher incidence of depression (8x higher) of depression, than in a non-depressed control group.
In summary, although the theory and evaluation are kept separate in mind, they must be linked and together once you are writing the essay.