1

Is there a way to consider (say) the upper bounds for all independant variables plus intercept?

I mean there is a way to find the 95% CI for each variable(even excel data analysis includes them) but what should be changed/taken account for a joint upper bound?do the probabilities of upper bound for an indpendant variable and upper for the intercept occuring indepenantly/simultaneously?

I know the answer is found at the following link: http://www.ics.uci.edu/~jutts/st108/Regression_Inferences_in_R.doc which says: If you want simultaneous confidence intervals for both the intercept and slope, using the Bonferroni method with joint confidence level α, set the level equal to 1 – α / 2...

but I need a bit of elaboration b4 I consider my question asnwered: a numeric example & adequate analysis...starting with what is bonferroni method

Kefalas Savvas
  • 3
  • 2
savvas
  • 11
  • 1
  • Do you want a simultaneous confidence region for several coefficients? – mdewey Dec 10 '17 at 12:02
  • i think i found evidence on the subject - a sentance i extracted ffrom an r tutorial:It writes: – savvas Dec 10 '17 at 15:52
  • If you want simultaneous confidence intervals for both the intercept and slope, using the Bonferroni method with joint confidence level α, set the level equal to 1 – α / 2... can anybody elaborate on that say with a bite-size example or a process description a bit more verbose as in answering that:if i have the 95% interval for b0 and that of b1, then how confident am i if i try to use both upper or lower bounds? – savvas Dec 10 '17 at 15:59
  • (what is a bonferroni method?) – savvas Dec 10 '17 at 16:00
  • 3
    I think this question would be much clearer if you could edit it so that it just quotes the R tutorial you did not understand, and asks what that quotation means. Perhaps after that question has been answered, you will then understand enough to be able to frame a follow-up question, if necessary, about anything further you want to do? At the moment it seems you are asking about something you don't quite understand well enough to formulate into a question clear enough for people to be able to answer, and in those circumstances asking one question at a time usually helps to get unstuck. – Silverfish Dec 10 '17 at 21:38
  • Please register &/or merge your accounts (you can find information on how to do this in the **My Account** section of our [help]), then you will be able to edit & comment on your own question. – gung - Reinstate Monica Dec 11 '17 at 12:37
  • You may find this Q&A https://stats.stackexchange.com/questions/96018/joint-confidence-interval?rq=1 helpful – mdewey Dec 11 '17 at 13:37
  • To me this question is still unclear. A "95% CI for each variable" suggests you want a confidence interval for the independent variable, but other parts of the text suggest you want confidence intervals for their slopes/regression coefficients. – Silverfish Dec 11 '17 at 23:39
  • @mdewey Your link indeed goes to a closely related question: it concerns obtaining confidence intervals for *linear combinations* of coefficients. The (rather limited) answers currently there hint at how one might obtain the *joint* confidence intervals requested here. – whuber Dec 12 '17 at 14:29
  • The reference you supply is deceiving: the Bonferroni method (as described there) is an *approximation* that will work only in limited circumstances; namely, when all the coefficient estimates are approximately uncorrelated. This is the case for ordinary linear regression when the independent variable has been centered and so it frequently applies, but in general it doesn't work for multiple regression. – whuber Dec 12 '17 at 14:33

0 Answers0