Hi,
There is an article that I want to know which statistical method has been used, regression or Pearson correlation.
However, they don't say which one. They show the correlation coefficient and standard error.
Based on these two parameters, can I know if they use regression or Pearson correlation?
Could you email the authors and ask for clarification? It would normally be seen as good practice to clarify what exactly is reported in the results section (e.g., regression or correlation and what type), and that information should have been included in the article.
Not sure I understand your question. If there is a single predictor and by regression you mean linear OLS regression with the raw data (e.g., not ranked data), then the r is the same. Can you provide more details>
As Daniel Wright said, the r value from a typical regression is identical to a correlation coefficient. The fact that they gave a standard error suggests that this indeed a Pearson's r value.
The presence of a correlation coefficient and standard error in an article does not necessarily indicate whether regression or Pearson correlation was used in the analysis. Both regression and Pearson correlation can produce a correlation coefficient, and the standard error can be used to assess the precision of the estimate of the correlation coefficient.
To determine whether regression or Pearson correlation was used, you should look for additional information in the article, such as the research question being addressed, the type of variables being analyzed, and the specific methodology used to calculate the correlation coefficient.
If the article is focused on the relationship between two continuous variables, and the research question involves testing the strength and direction of the linear association between those variables, then it is likely that Pearson correlation was used.
On the other hand, if the article is focused on the relationship between a dependent variable and one or more independent variables, and the research question involves testing the relationship between those variables after controlling for other variables, then it is likely that regression was used.
If the authors do not provide enough information to determine the method used, it may be helpful to contact them for clarification.
Once again, regression and correlation are identical -- they are algebraically the same thing. The presence of a standard error indicates that it is more likely to be a Pearson's rather than a Spearman's correlation.
Yes, you are correct. Regression and correlation are similar concepts and are often used interchangeably. The presence of a standard error indicates that a Pearson's correlation is more likely to have been used, as the standard error is a measure of the variability of the sample estimate of the population correlation coefficient, and is only applicable for Pearson's correlation. However, without more information about the data and the analysis performed, it may still be possible that a Spearman's rank correlation was used, especially if the data is ordinal or non-parametric. It's always best to check the methods and results sections of the article for more information.
Mayur Wanjari
Hey,
Thank you for the explanations. It is more clear now.
In the article, they published their values and I utilized the CORREL function in Excel, which suppose to be Pearson correlation coefficient- thus, I got the same r as they published. Does it mean they indeed used the Pearson correlation?
Their variables are quantitative. However, they have only 9 examples- does it mean they can not show normal distribution?
In my work, I have 21 examples, the variables are quantitative, and since I have less than 30 examples, I used linear regression analysis to obtain R-squared (Excel-->ToolPak->regression). My question is: In my thesis, can I say two variables have a good correlation if R-squared is bigger than 0.99? Or should I only use the term relationship?
yes. Actually the correlation you find in packages was first discovered by the French mineralogist Bravais, so it should be called Bravais-Pearson correlation coefficient.
Dear Bar Friedman , I agree with You, the CORREL function in Excel is a Pearson correlation.
Correlation coefficient does not necessarily mean Pearson Correlation coefficient because we also have spearman rank correlation coefficient. On the other hand, the inclusion of standard error points to the fact that it could be Pearson's correlation.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Control Theory
- Control Systems