Thank you.

The beta value in the regression isn't really a useful effect size metric (it is hard to interpret with collinearity and other factors). The unstandardised coefficient (usually B in SPSS).

This is interpreted as the change in Y associated with a 1 unit change in X when all other variables in the regression are fixed (i.e., don't change in value).

So if I regress height (X1) in cm and age (X2) in year on salary (Y) in dollars then a B coefficient of 433 for height means that each +1 cm change in height is associated with a +£433 increase in salary.

Effect size

Thom S. Baguley's jolly example about the regression of height and salaries shows the misunderstanding of a relation. Small women earn less money than than tall men. The addition of the overlapping distribution of height might result in a linear construction.

My recommendation for training statistics is: Don't read any books about statistics. Don't prepare Your statistic by a cooking machine like SPSS. Calculate some simple connections and then look for each difficulty, each question in a book or ask some statistical bakers by trade. So You will learn inductively. Build Your own cooking machines by macros. By this way You will discover that there are sometimes no solutions but approximations.

 Thank you Sir Thom S Baguley, for your response.

Can you please elaborate what effect size or beta value is to consider as statistically significant? can you please tell me how to do it linear regression in SPSS selecting certain variables as predictor? and also what is the significance of preforming this test? As I could not get any satisfied or authentic answers for these questions. 

Thank you so much Miss Sofia Cividini