Thank you for this video. Is there an option for doing this test in SPSS not just with a simple non-linear regression model but also a multiple linear regression model? Thank you for your help.
@@jensk.perret6794 Thank you for your reply. However, I meant to ask if there is an option to do a curve fitting test with a multiple linear regression model rather than just a simple linear regression model.
In this case you might want to check out the actual non-linear approach of SPSS. This allows you to use multiple variables in a model of your design. Alternatively, for polynomial, exponential or logarithmic terms you could generate new transformed variables and introduce them into a linear regression approach.
Hi, thanks for the nice video. Just confirming something: should we first check that the equation is significant, then checking the R-square and F values? I have some data and the Sig. F change is 0.053 for linear and 0.034 for logarithmic. The linear being close to significance, I am unsure which one to choose. Any advice would be greatly appreciated. Thank you.
I read your comment in the way that the numbers that you report are the significance levels?! In this case it would mean that the logarithmic model fits your data better. In particular if you assume the standard threshold for significance of 5%, in the first case the model would be unfit while in the second case you would have a model with a significant explanatory power. Thus, if the second model makes sense in practical terms I would go with it.
Great video and explanation. Thanks!
Great video. Helped a lot. Thanks. Please keep making this type of videos.
Thank you for this video.
Is there an option for doing this test in SPSS not just with a simple non-linear regression model but also a multiple linear regression model?
Thank you for your help.
There is in the form of the command Non-linear. Here you can find a tutorial on it:
ua-cam.com/video/epVuS06jqJM/v-deo.html
@@jensk.perret6794 Thank you for your reply. However, I meant to ask if there is an option to do a curve fitting test with a multiple linear regression model rather than just a simple linear regression model.
thanks. clear and helpful.
Hey. Could you please let me know to to run more complex models and do curve fitting and r^2 estimation etc. on them?
In this case you might want to check out the actual non-linear approach of SPSS. This allows you to use multiple variables in a model of your design. Alternatively, for polynomial, exponential or logarithmic terms you could generate new transformed variables and introduce them into a linear regression approach.
Hi, thanks for the nice video.
Just confirming something: should we first check that the equation is significant, then checking the R-square and F values?
I have some data and the Sig. F change is 0.053 for linear and 0.034 for logarithmic. The linear being close to significance, I am unsure which one to choose.
Any advice would be greatly appreciated. Thank you.
I read your comment in the way that the numbers that you report are the significance levels?! In this case it would mean that the logarithmic model fits your data better. In particular if you assume the standard threshold for significance of 5%, in the first case the model would be unfit while in the second case you would have a model with a significant explanatory power. Thus, if the second model makes sense in practical terms I would go with it.
@@jensk.perret6794 Thank you very much!