Dear Aarash, 

Your 7-point Likert-scale data can be treated as a continuous scale. Of course your data are different in accordance with  the Gauss curve (M, +/- 3 SD). Your software will take care of your outliers depending on what methods you are using. 

Greetings,

The data in likert scale items cant be considered outliers.

If your respondents are coming from 2 different points of view, then you may want to compare between them as 2 groups. you can use independent sample ttest. if no significant differences, then the answers are homogenous.

of course you need to get the total score to be used in ttest or other inferential statistics.

The usual goal with LIkert-scored items is to combine them into a single scale, then assess the adequacy of that scale using Cronbach's alpha and exploratory factor analysis. It your scale performs as expected, then you would examine it for outliers -- rather than doing that on a one-by-one basis with the items that make up that scale.

And then your Cronbach's alpha should be at least 0.70

after you have had to delete some items in the entity that measure planned behaviour e.g. in the traffic to speed or not to speed

 Dear Béatrice ,

Thanks for the reply. Since I have considered different variables between various culture as  you and Dr. David mentioned for Coronbach's alpha, For all variables except Control belief (alpha =0.695) , alpha is greater than 0.8.,

Dr. Muayyad Ahmad 

Thanks for the reply, I also think it will be logical since people have different attitudes and I have to consider other parameters to avoid reducing variances in my data set

Dear Aarash, 

Your alphas are very good (alpha = .695 = .70)

have you considered examining your scale using Rasch? 

is the 7point likert providing ordered thresholds for each question?

http://www.rasch.org/rmt/rmt63i.htm is worth perusing for a discussion of strata and person separation index values (equiv to chronback alpha).