Hi everyone
Can someone help with the following observation while applying simple linear regression in SPSS and STATA
Dependent variable: Ferritin levels (ng/ml)
Independent: Age of the respondent (years)
STATA results: Correlation coefficient = 1.7 (see the image), How can this be possible?
SPSS results: Correlation coefficient = 0.17 (different from STATA) and a p value of 0.01 for such a weak coefficient.
DATA set is SAME in both STATA and SPSS
Can someone explain the coefficient of more than 1 in STATA and a different result in SPSS
In your first image, you highlight the (raw score) regression coefficient for MA1. In your second image, you show the Pearson correlation between the same two variables. They would have the same value only if you standardized the X and Y values (i.e., converted them to ZX and ZY) before estimating the regression model.
If you want to estimate the same regression model in SPSS that you estimated in Stata, use the REGRESSION command in SPSS. In the GUI, click on Analyze > Regression > Linear. Or if you want to compute Pearson r with Stata, use either the -correlate- command or the -pwcorr- command.
Dear Zeeshan Kibria
All statistical programs give you the same results when used correctly.
You are incorrectly comparing two tables for different outputs
The table you have from Stata output contains the regression coefficients and the regression constant = 1.7
(no correlation coefficient)
In the second image, you have the Pearson output, which contains the correlation coefficient.
Note the value of R squared
From STAT output = 0.029 which is identical to the Pearson coefficient squared 0.17
Therefore, if your work on the two programs is correct, the outputs will be identical
In the case of simple correlation and regression,
I advise you to rely on SPSS. Zeeshan Kibria
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Materials Science
- Materials
- Inorganic Materials
- Metals
- Alloys
- Steel