Thanks @user9876 , having barely scraped a C at my Maths O level, the statistical methods go way above my head. I can understand ranges, but I confess I don't understand your terms. Would it be possible to explain mean difference method and linear scales? Apologies for my ignorance.
A standard way to look at results for trials is to take the average improvement for a group. So look at the average score at the start and the average score at the end for each group. Then look at the difference. There needs to be a significance test between the results for the different groups to test whether the difference is likely to have been just due to random effects or whether it is likely to be due to the intervention.
My issue with the scales comes from looking at the questions. Say you have a person who scores 0 on the SF36 scale at the start and improved by 1 question then they are likely to have started by saying they were limited a lot by "bathing or dressing themselves" and changed that to limited a little. But someone who scores around the mid point of the scale may change from finding it easier "to walk a block" or finding "it easier to walk up a flight of stairs". Someone near the top of the scale may find it easier to do "Vigourous activities such as running, lifting heavy objects or participating in strenuous sport"
So the first question is does change in a single question such as changing from being limited a lot in bathing and dressing to being limited a little represent the same change in physical ability as say being limited a little in walking a block to not being limited at all. Or being limited a little in vigorous exercise to not being limited at all. If these are not equal changes in physical function this means that a change in the score for one person is not the same as a change in the score for a different person and they are not compatible.
The second question is are some people likely to improve on two questions with the same amount of improvement on physical function so is someone likely to say find is easier to walk up a flight of stairs at the same time as finding it easier to walk a block. In this case improvements are likely to be exaggerated around the middle of the scale but it may also depend on the answers given. Again this makes it hard to compare scores.
If the changes are big then this won't matter but with smaller changes the start point (in terms of questions answered) may have an unknown impact on the level of change.