Timeline for Simple linear regression output interpretation
Current License: CC BY-SA 3.0
13 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Mar 13, 2012 at 22:36 | history | edited | Macro | CC BY-SA 3.0 |
Fixed typesetting
|
| Jul 20, 2011 at 20:34 | history | edited | jedfrancis | CC BY-SA 3.0 |
more changes to 3rd paragraph
|
| Jul 20, 2011 at 20:24 | comment | added | jedfrancis | @whuber I clarified the 3rd paragraph. Although one can fit a linear model to data with a non-linear relationship you would be right to suggest caution in its use and interpretation. | |
| Jul 20, 2011 at 19:48 | history | edited | jedfrancis | CC BY-SA 3.0 |
clarified 3rd paragraph
|
| Jul 20, 2011 at 17:10 | comment | added | whuber♦ | @jed Thanks. Whatever you did, it seems to have worked, because the comments and votes stuck with your answer. By the way, the questions of linearity and normality of residuals are really separate ones; lack of normality does not necessarily imply a linear model is wrong (but it does suggest a more cautious approach). | |
| Jul 20, 2011 at 16:30 | comment | added | jedfrancis | @whuber thanks, it wouldnt let me edit it, so i deleted it and re-posted the edited version: I meant to say in a linear model, slope with a value close to +1 is indicative of a strong positive relationship between the variables IF the variables are standardized. | |
| Jul 20, 2011 at 16:12 | history | edited | jedfrancis | CC BY-SA 3.0 |
deleted 54 characters in body
|
| Jul 20, 2011 at 13:08 | comment | added | whuber♦ | @jed It would be good if you were to correct your reply. | |
| Jul 20, 2011 at 1:56 | comment | added | jedfrancis | @wolf.rauch gotcha | |
| Jul 19, 2011 at 23:54 | comment | added | wolf.rauch | @whuber is saying that the value of the slope does not tell you anything about the strength of the association unless variables are standardized. See shabbychefs answer. | |
| Jul 19, 2011 at 21:29 | comment | added | whuber♦ | Welcome to our site, @jed, and thanks for your reply! Please note that the slope itself says almost nothing about the correlation, apart from its sign, because correlation does not depend on the units in which X and Y are measured but the slope does. | |
| Jul 19, 2011 at 19:51 | comment | added | Mog | Thanks @jed. Yes, I'd checked the normality of the residuals, and all was well. Your suggestion that the data is spread widely around that regression line is exactly right - the data points looks like a cloud around the line of regression plotted by the software. | |
| Jul 19, 2011 at 19:32 | history | answered | jedfrancis | CC BY-SA 3.0 |