Scoring patterns are probably not the answer. Social desirability and faking involve telling a lie to yourself before telling a lie to someone else.

Reducing the requirement for social desirability in the questionnaire is one possible strategy. If you are interested in, for example, exercise, you can start with excuses:

- With so many demands on my time, it's difficult to find the time to exercise

- The outdoor exercise facilities (eg parks, pedestrian paths) where I live are poor

- The indoor exercise facilities (eg gyms, pools) where I live are poor

Then ask a preparatory question :

- Thinking about the last week, how much exercise did you get compared with usual? More/about the same/Less

Then you ask 'how much exercise did you get in the last week' (avoid the word 'take' because it implies that the person who didn't exercise did so by choice).

The questions should show that you understand the external pressures on people.

And what do you do when most people say that last week they got less exercise than usual last week? Nothing. Just go with what they said they did. The question was there to discourage lying.

As you see, constructing questions around topics where people are prone to lying (I'm a fan of Dr House!) involves reducing the fear of social devaluation attached to telling the truth.

QUANTITATIUVE SCALE try to use true scale: 0 = none, 1 = low, 2 = medium and 3 = high. let the respondents select the numerical scale as the choice, not the adjective describing the number. The number should precede the description. Avoid Likert scale at all cost. Likert scale gives fake data because its lowest is 1; there is no zero.