The Student Room Group
Uni students - Complete our survey >>
Uni students - Complete our survey >>
Uni students - Complete our survey >>

Regression Testing

Morning all,

I don't study Maths/Statistics, however for a project I'm working on I need to determine the relationship between variables. I've undertaken Correlation analysis on SPSS to confirm there is a relationship between the variables to begin with.

I've moved onto Linear regression analysis on SPSS in order to establish the strength of the relationship using the Beta Value (trying to keep it as simple as possible).

What range can Beta fall into? Is it -1 to 1?

At what value does it become significant? If I have a Beta score of .217 (p<0.01), does that state that it's highly significant?

If you'd be able to keep any responses as simple as possible that'd be great as I don't have a great understanding of statistics and only recently started using SPSS.

Any help would be greatly appreciated!
Original post by Dean_Ed

I've moved onto Linear regression analysis on SPSS in order to establish the strength of the relationship using the Beta Value (trying to keep it as simple as possible).

What range can Beta fall into? Is it -1 to 1?


The beta values can be anything. In linear regression you are modelling your outcome variable as a linear combination of your explanatory variables plus some normal random noise. So we typically write

yi=β0+β1xi1+β2xi2+⋯+βnxin+ϵi\displaystyle y_i = \beta_0 + \beta_1 x^1_i + \beta_2 x^2_i + \cdots + \beta_n x^n_i + \epsilon_i

where the yiy_i are your outcome variables, the xikx^k_i are your explanatory variables and the βk\beta_k are the regression coefficients estimated by your software and ϵi\epsilon_i represent the normal random noise.

At what value does it become significant? If I have a Beta score of .217 (p<0.01), does that state that it's highly significant?


The value of beta represents the strength of association between the explanatory variable and the outcome. A beta of 0.217 says that increasing the corresponding explanatory variable value by one increases the outcome variable by 0.217. The p-value tells you whether this association is statistically significant. p<0.01 is usually taken as being reasonably strong evidence of an association.

Quick Reply

Related discussions

Latest

Trending

Trending

Articles for you