 # Linear Regression

Hello!

One thing that strikes me a bit is whether it is possible to (if regression coefficient, b= -0.35) interpret the association between IV and DV as follows: as age increases by one unit, the time spend on following politics decreases by 0.35 units. And therefore, we can conclude that the older people are, they tend to be less interested in politics-or is this sentence relevant only when interpreting correlation coeffient?

Thanks

Welcome to the community!

You can surely make such an interpretation, as long as `b` is the regression coefficient of `y` on `x`, where `x` denotes `age` and `y` denotes the time spent on following politics.

(I don't know what `IV` and `DV` mean, and hence I'm using generic `x` and `y`. I'm sure you'll be able to relate it.)

On the other hand, you can't make such an interpretation based on the correlation coefficient `r`. The sign of `r` will tell you the direction of the linear relationship, and its magnitude will inform you about the magnitude (strong, moderate, week, etc.) of that relationship.

Thank you! By IV and DV, I meant independent and dependent variable.

If your question's been answered (even by you!), would you mind choosing a solution? It helps other people see which questions still need help, or find solutions if they have similar problems. Here’s how to do it:

Hello, I have another small question!

I have three predictor variables, namely:
age(b=0.000049; t=16.58)
following politics news(b=0.027; t=4.53)
left-right self placemeny(b= -0.063;t= -5.19)

All predictor variables are statistically significant as p is less then 0.01.

However, I am not sure whether I can say that there is a notable association between these predictor variables and my outcome variable. Altough p values are less then 0.01, I would say that there is no notable association between predictor variables and dependent variable as b-values are very low.

Is my way of reasoning correct?

Thank you

I think that for talking about magnitud of association you need to check for r and r^2 and for magnitude of influence then you can check for B coefficients.

sorry, I forgot to mention but r-squared is 31.25% and adjusted r-squared is 31.18%. Therefore, I still think that there is no notable association. There is some association but predictor variables dont explain a large amount of variance in the outcome variable.

@andresrcs what do you think?

Well, I think that could be a valid interpretation, but to be sure, you should check for for all of the linear models assumptions and also check for outliers and leverage points.

The values of the regression coefficients are very small, so large values of the t statistic seems a little odd (but not impossible, of course) to me. Especially, the values of `b` and `t` for age is very surprising.

@Yarnabrina
I have checked it again and got the same numbers

If you're satisfied that all the assumptions for OLS is satisfied, and there are no unusual observations, you can proceed with your interpretation.

However, I'm very surprised that you're able to distinguish a value as small as `0.000049` from `0`, and from the value of the test-statistic, p-value will be extremely small, almost `0`.