Developing a survey questionnaire for Structural Equation Modelling (SEM) can be critical. Often times I wonder if I should ask questions in separate sections with all the items of the same construct or keep all items from all the constructs together, without any separation.

Clustering items could, in a way, bias the respondents.

Putting all items together could lead to straight-lining.

How to counteract these problems while developing survey questionnaire?