Hey Guys!
On the table, since it is linear regression and not multiple regression, will we only have step 1? If not, where does step 2 come from? Under model are we supposed to have a 1 and a 2? Also, for the write up, are we supposed to make up a hypothesis? And should we refer to the table, even though in step one we have not made it to the table yet? Also, is there a certain order that the write out has to be in in reporting B, p, R, etc.? On the table are we not supposed to report the weight in pounds, and in the example handout out why does the education variable appear twice? I am not sure that I computed the correlation part right. Are we to do it like the bivariate example or should we do it differently since there are more than 2 variables? I know this is a long question but I have a hunch that there are other people out there wondering about the same things? Thanks so much for taking the time to answer!
Subscribe to:
Post Comments (Atom)
2 comments:
By steps, do you mean "blocks" or "models"? Yes, it is a linear regression, not multiple regression, so you are not adding another block of variables like you would in multiple regression. You should only have Model 1 listed in the output. I think you are referring to step 2 because you are looking at the multiple regression handout. A linear regression does not have a step 2 like the multiple regression example in the handout.
For the write up, you do not need a hypothesis. You are writing up the results. However, it is helpful to start with telling the reader what has been examined. Perhaps something like this: "The correlation between ___ and ___ was examined using a linear regression..." Then go on to describe the results. To answer question one, you should not need to refer to the table - just to the linear regression you computed.
In terms of the write-up, yes, there is a certain way/order to report them. I think you must have gone over this in class. Make sure to include both descriptive and inferential statistics. If you have the APA manual, the "Statistical Presentation" section of the Results part (p.21-24 in the 5th edition) would be helpful in answering your questions.
The table should be made from the correlations between ALL variables (which includes weight in pounds). The education variable you are referring to is in the Multiple Regression handout - disregard that handout for this hw because it is on Linear Regression not multiple regression.
For the correlation part, you are to make a correlation matrix which includes all the variables. I don't have the data set in front of me, but I would think you could use bivariate correlations to get all of the answers you need.
Please let me know if there is anything I missed or anything that isn't clear.
Kelly's comments are accurate, but let me just amplify a couple of the things she said.
First, the contrast for correlation is bivariate versus partial or distances, neither of which we have talked about (and one of which we WON'T talk about this quarter). Even though you have multiple variables in that correlation matrix, each individual correlation is a bivariate correlation. That is, even though weight and age are also in the table, the correlation you report at the intersection of the column of height in inches and height in feet is a bivariate correlation, including only the variables of height in inches and height in feet. Only those two variables contribute to that correlation, thus making it a bivariate correlation. There are many bivariate correlations reported in the table, but they are all bivariate.
Second, imbedding the hypothesis in the write up does serve to organize both you and the reader, so having the hypothesis in there is a good idea. However, it is possible to write it without it--although I would avoid using two analysis names in the description. Perhaps "The association between ____ and ___ was examined using a linear regression," to avoid using correlation in the sentence.
Post a Comment