Thank you for your very clear and informative videos. I have found your videos and explanations to be more helpful to me than most others I have come across for both Curvilinear Regression and other methods.
Thank you so much for this very informative channel. The videos are very well done, and I really appreciate all the confidence and straight forward teaching style. Keep up the good work.
I believe so. Often in practice, usually both the linear and quadratic effect are statistically significant. However, if only the quadratic effect is statistically significant, then you would go with the interpretation that there is only a quadratic effect. Bear in mind that the linear and quadratic terms are often substantially correlated which does affect the size of the standard errors associated with the beta weights. It's conceivable that you will see changes from sample to sample.
Hi! thank you very much for this video, its the only one i found to this topic and very informativ. i`m going to use a quadratic regression for a research- could you suggest some literature (articles, books) about how to calculate a quadratic regression via SPSS?
Thank you for your videos! I am wondering if you can enter multiple independent variables when using this method, particularly to include group and interaction variables. For example: IQ, sex (male or female), and IQ*sex interaction
Thank you for the video. My question is: do you not need to convert IQ into Z-scores (or center around the zero) before squaring? I'm imagining a curve, and it seems to me that squaring without centering around zero would only explain the variance in one side of the curve. It makes sense that your results would still be significant in your example, as the left side of your curve is more defined. Thank you in advance for any response.
Great video. If my residuals Vs fitted plot shows a distinct pattern in a nonlinear function for example a weibull function does it still mean that the model is improper?
How do you interpret the unstandardized coefficients for the quadratic portion. This seems to be skipped over in the video and any documents I have read.
Jacques Lauperr It's very difficult to interpret the quadratic coefficients meaningfully. If it is negative in direction, then it would imply an inverted U association. If it's positive in direction, then a U function would be implied. Other than that, your best plotting the effect in a scatter plot with the corresponding line of best fit.
Thank you for your very clear and informative videos. I have found your videos and explanations to be more helpful to me than most others I have come across for both Curvilinear Regression and other methods.
Thank you so much for this very informative channel. The videos are very well done, and I really appreciate all the confidence and straight forward teaching style. Keep up the good work.
I believe so. Often in practice, usually both the linear and quadratic effect are statistically significant. However, if only the quadratic effect is statistically significant, then you would go with the interpretation that there is only a quadratic effect. Bear in mind that the linear and quadratic terms are often substantially correlated which does affect the size of the standard errors associated with the beta weights. It's conceivable that you will see changes from sample to sample.
Stephanie: Yes, you can use curvilinear regression in the context of multiple regression and test for interactions.
Hi! thank you very much for this video, its the only one i found to this topic and very informativ. i`m going to use a quadratic regression for a research- could you suggest some literature (articles, books) about how to calculate a quadratic regression via SPSS?
Thank you for your videos! I am wondering if you can enter multiple independent variables when using this method, particularly to include group and interaction variables. For example: IQ, sex (male or female), and IQ*sex interaction
It's totally fine to add a cubic function to this type of analysis.
Thank you for the video. My question is: do you not need to convert IQ into Z-scores (or center around the zero) before squaring? I'm imagining a curve, and it seems to me that squaring without centering around zero would only explain the variance in one side of the curve. It makes sense that your results would still be significant in your example, as the left side of your curve is more defined. Thank you in advance for any response.
Does curvilinear regression model is the cubic regression model?
Great video. If my residuals Vs fitted plot shows a distinct pattern in a nonlinear function for example a weibull function does it still mean that the model is improper?
can you explain more about the meaning of F value or show me where can i find out more about this ?
Is it acceptable to do three blocks of this analysis: linear, quadratic, and then cubic? Or is it restricted to just two?
How do you interpret the unstandardized coefficients for the quadratic portion. This seems to be skipped over in the video and any documents I have read.
Jacques Lauperr It's very difficult to interpret the quadratic coefficients meaningfully. If it is negative in direction, then it would imply an inverted U association. If it's positive in direction, then a U function would be implied. Other than that, your best plotting the effect in a scatter plot with the corresponding line of best fit.
Can you share your data used in your analysis?
Thank you! :)