0
$\begingroup$

This is my exam preparation, which will be held completely in R. Teacher said that similar tasks will be there with very limited amount of time given for solving it. Here I need to recover linear regression values (i.e., where are a-i missing values) having only this R output. I don't have the input data so I can't just copy the command and reproduce the result.

Given coefficient is R output. I don't have the input data.

My question is how can I make it using R? Are there any specific commands? Or could it be done only manually? If it is possible only by manual calculation than please explain how to do it more effectively.

Please, describe it in detail.

Thank you so much for your help and deep explanation.

Improve this question
$\endgroup$
7
  • $\begingroup$ Please add [self-study] tag and check stats.stackexchange.com/tags/self-study/info $\endgroup$
    – Tim
    Commented Dec 11, 2015 at 21:09
  • $\begingroup$ Here's a hint t = Estimate/Std. Error You're welcome $\endgroup$
    – SASsy
    Commented Dec 11, 2015 at 21:10
  • $\begingroup$ Welcome to CV! Software questions are generally off-topic here—you may have more luck on StackOverflow. $\endgroup$
    – Sean Easter
    Commented Dec 11, 2015 at 21:34
  • $\begingroup$ @SeanEaster this is not a software question and it is on-topic, it asks about strictly statistics stuff. I do not understand the two votes for moving it to SO. What is the programming issue in this question that makes it suitable for SO? This is strictly about understanding regression, Wald test, F test etc. $\endgroup$
    – Tim
    Commented Dec 12, 2015 at 22:19
  • 1
    $\begingroup$ @Tim "How can I make it using R" sounds exactly like a software question. If that question I quoted is not asking for R commands it needs to be reworded so it doesn't read like that's what it's asking for. $\endgroup$
    – Glen_b
    Commented Dec 13, 2015 at 5:23

1 Answer 1

Reset to default
3
$\begingroup$

Since this seems to be a self-study question let me give you some hints. First, do you know what the columns in your output mean? Estimate is $\beta$ parameter, Std. Error is standard error and t value is $t$-value (or standardized parameter). Hypothesis tests are conducted using Wald test, i.e. $\beta/\mathrm{SE}(\beta)$. Knowing all this you can easily compute the values (a)-(g). As you can notice from column name Pr(>|t|) it is a probability from $t$-distribution, so knowing the previous values you can read the values from statistical tables for $t$-distribution or compute using appropriate R function. As about (i), you know the adjusted $R^2$, so you have just to revert the adjustment. By the way, it would be probably good idea to check in your handbook information about degrees of freedom in multiple regression and ANOVA (generally for the $F$ test) since this information may be helpful. See also this thread that also describes the lm output.

Improve this answer
$\endgroup$
6
  • 1
    $\begingroup$ @RLearnsMath no need to thank. If you find some answer helpful in here you can upvote it (the upward arrow) or accept it (the "v" button) -- you can upvote multiple answers and accept only a single one. Unfortunately, you can upvote only after gaining some reputation. See stats.stackexchange.com/tour and welcome on our site! $\endgroup$
    – Tim
    Commented Dec 11, 2015 at 21:24
  • $\begingroup$ Yes but how will I figure out the amount of observation in the data set? By manually computing each variable? $\endgroup$
    – Ruslan Seletskiy
    Commented Dec 11, 2015 at 22:32
  • $\begingroup$ @RLearnsMath have a closer look at Tim's last sentence. $\endgroup$
    – Stefan
    Commented Dec 12, 2015 at 2:15
  • $\begingroup$ Yes but I can do nothing with ANOVA just because I don't have sample size of the given data. $\endgroup$
    – Ruslan Seletskiy
    Commented Dec 12, 2015 at 11:11
  • $\begingroup$ Knowing how degrees of freedom are calculated you can easily compute it from DF. $\endgroup$
    – Tim
    Commented Dec 12, 2015 at 14:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.