What is the significant level of 1-tailed multi regression analysis in spss and how to interpret the results of R-square,F, Sig. Beta and T statistics? what is the high, moderate, lower and normal level of significant in 1-tailed.
I am not sure, if you got the concept of significance right. The significance level is something you determine a priori, for example 5% or p=0.05. So, there is no high, moderate, low, or normal level of significance. Maybe you should consult a local statistician or a good textbook, because to fully explain the whole concept of statistical testing and how R-square, F or t values, b and beta are interwoven is beyond this question and answer section in my opinion.
In social sciences, e.g. psychology it is agreed that level of significance is 0.05. It is relatively lenient level, however social sciences is not physics. Good luck
Dear Khalid,
1) If you use a level of significance = 0,05, in one tail it corresponds to 0,05/2.
2) R square X100 measures the percentage of variation of a variable that is explained on average for the other
3) F Sdnedecor evaluates on a global basis the model and not each parameter.It is a test of the goodness of fit of the global regression equation.
4) T Student assess the additional contribution of an exogenous variable on the already existing in the model. t tests individually for each coefficient are imprecise when there is high multicollinearity.
5) Beta= is the interpretation of each regression coefficient (partial derivatives). Thus, if VI is a quantitative variable, beta means that for every a unit increase in Xi, keeping constant all remaining VI, the effect on Y is beta.
If beta as a negative signal corresponds to a decrease in Y; but if the sign of beta is positive correspondes to an increase in y.
Have a nice time
Helena
Dear Helena, very nice summary, thank you. But one thing is not clear enough: not all test can be conducted one or two tailed. For example, the test of the overall model, i.e. if R is significantly different from 0, is always one-tailed. On the other hand, the significance tests for the b values, i.e. the t-tests for each predictor, can be tested one or two-tailed.
Dear Rainer,
In terms of mutliple regression models ther are much more to say, then you are wright, but i tried to answer only directly to Khalid. ...
Have a nice day
Helena
Yes, MUCH more to say, and what you described is absolutely right in my opinion, but I was not sure if this one/two-tailed thing was clear enough, since the initial question seemed to be about R-square.
- Badges
- Science topic