I have an R squared of 0.07 with a p value of .03, in a simple linear regression. Can I submit this for publication? Opinions appreciated.🙂
R squared
- Thread starter JFTSM
- Start date
- Joined
- Apr 16, 2004
- Messages
- 4,927
- Reaction score
- 6,128
Your question is nonsensical.
It's like asking: I have a manuscript that's 20 pages. Can I submit this for publication?
We need more context about what you are actually comparing and what your study entails.
It's like asking: I have a manuscript that's 20 pages. Can I submit this for publication?
We need more context about what you are actually comparing and what your study entails.
Its a simple linear regression looking for a correlation between number of days in the hospital and BMI of people that had radiation therapy in a particular hospital during a particular year. I hope that is enough? Thank for the help.
- Joined
- Feb 11, 2005
- Messages
- 598
- Reaction score
- 482
In general- anything can be submitted for publication. Whether or not it gets accetped is a harder question, and one you are not going to get from a discussion forum.
A lot goes into deciding whether or not a paper gets accepted by a given journal, including quantifiaible factors (# patients, study design- RCT > Phase I or II > retrospective, etc). Other factors include relevance to readers, novelty of question or study, appropriateness of statistics, etc.
If you want a free gauge of the quality of a potential paper- submit an abstract to a national meeting. Alternatively, simply write the paper and submit it to your journal of choice. A rejection at one journal does not mean that another journal will not accept it. A rejection should also give you feedback of what you could do to improve it.
Lastly- the data is what it is, and p values should not impact if a paper is acceptable or not. One should strive for a well-designed study and not a give p value. For example: a p value 0.05 is not magically different than a p value of 0.06- one means a 6% risk that the null hypothesis (equivalence) is true, and the other 5%.
So my suggestion- write it up and submit it, and have a good explantion why you have a poor correlation and sign. p value.
A lot goes into deciding whether or not a paper gets accepted by a given journal, including quantifiaible factors (# patients, study design- RCT > Phase I or II > retrospective, etc). Other factors include relevance to readers, novelty of question or study, appropriateness of statistics, etc.
If you want a free gauge of the quality of a potential paper- submit an abstract to a national meeting. Alternatively, simply write the paper and submit it to your journal of choice. A rejection at one journal does not mean that another journal will not accept it. A rejection should also give you feedback of what you could do to improve it.
Lastly- the data is what it is, and p values should not impact if a paper is acceptable or not. One should strive for a well-designed study and not a give p value. For example: a p value 0.05 is not magically different than a p value of 0.06- one means a 6% risk that the null hypothesis (equivalence) is true, and the other 5%.
So my suggestion- write it up and submit it, and have a good explantion why you have a poor correlation and sign. p value.
- Joined
- Jul 1, 2006
- Messages
- 150
- Reaction score
- 0
a p value of 0.06 means that given the null is true, you will accidently say it is not true 6% of the time. It has nothing to do with the % of time the null is true or not true.
try to get a mentor nearby that can help tell you determine whether something is publishable or not, and where to submit it.
try to get a mentor nearby that can help tell you determine whether something is publishable or not, and where to submit it.
- Joined
- Feb 11, 2005
- Messages
- 598
- Reaction score
- 482
You are corerct- but it is really just semantics.
null hypothesis - equivalence
p=0.05 - 5% likelihood of null hypothesis being true. If you reject null hypothesis at p=0.05, then there is a 5% likelihood (or risk) of erroneously rejecting null hypothesis (or accepting alternative hypothesis)
erroneously rejecting null hypothesis = null hypothesis true
p=0.05 = 5% 'risk' of null hypothesis being true
The original poster is a medical student- so he or she should already have a mentor, since hospitals/Universities usually do not allow human subjects research without a faculty mentor.
null hypothesis - equivalence
p=0.05 - 5% likelihood of null hypothesis being true. If you reject null hypothesis at p=0.05, then there is a 5% likelihood (or risk) of erroneously rejecting null hypothesis (or accepting alternative hypothesis)
erroneously rejecting null hypothesis = null hypothesis true
p=0.05 = 5% 'risk' of null hypothesis being true
The original poster is a medical student- so he or she should already have a mentor, since hospitals/Universities usually do not allow human subjects research without a faculty mentor.
