Normally you need to attach in the paper the correlation diagram, the correlation formula and the R2 factor.
You should read some papers, there are numerous papers on this questions. Normally, first u can discussed about descriptive statistics, then correlation but remember that u should not copy exact tables of SPSS out put , there are certain editing require in the table to adjust format.
You can find this page very valuable: https://statistics.laerd.com/spss-tutorials/linear-regression-using-spss-statistics.php
If the regression analysis was performed with the objective of obtaining an equation, you must present at least:
- The equation resulting from the adjustment of the model [(refer to the significance of the equation (results of F test) and the estimates of its parameters (results of t test)]
- The values of the goodness of fit analysis criteria: coefficient of determination, root of the mean square error (RMSE) or standard error of the esimative (Syx), bias and the graphs of residuals distribution (verify if these comply with the statistical assumptions of normality, homocedasticity and absence of serial autocorrelation) to verify, mainly, if there are no tendencies of overestimation or underestimation.
Regards !
You should report R square first, followed by whether your model is a significant predictor of the outcome variable using the results of ANOVA for Regression and then beta values for the predictors and significance of their contribution to the model.
Mohammad -
Linear (including multiple linear) regression?
I suggest:
1) a graphical residual analysis scatterplot
2) cross-validation; minimally a few data saved (not used for model selection or estimation of regression coefficients) to check against predictions
3) estimate variance of the prediction error - which varies by prediction under heteroscedasticity, which is natural
I suggest (1) as a minimum. It may need to go in an appendix.
See https://newonlinecourses.science.psu.edu/stat462/node/81/
"Lesson 6" is on multiple regression.
Best wishes - Jim
Dear Mohammad
Refer to the attached article on "Energy efficiency of iron–boron–silicon metallic glasses in sulfuric acid solutions".
Also, refer to the following text:
-Motulsky H and Christopoulos A 2003 Fitting Models to Biological Data Using Linear and Nonlinear Regression: a Practical Guide for Curve Fitting (San Diego, CA: GraphPad Software Inc.) (www.graphpad.com)
The above references could be useful. The 1st reference has an example of the application of non-linear regression analysis. The second reference is an excellent text on regression analysis.
Your question was rather broad. Are you talking about Linear regression, Logistic regression or Cox regression?
The type of regression will depend on the type of data and study design.
Linear regression is used to analyze numerical( continuous, discrete data). Correlation coefficient which describes the degree of association is used in interpretation.
For most qualitative studies, categorical/ordinal data are used. Logistic regression is used in analysis of predictors of outcome. Odd ratio is used for interpretation.
For follow-up studies, survival analysis is used to analyze time to event depending on study design. Cox regression is used for regression analysis and Hazard ratio is used for interpretation.
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Materials Science
- Materials