A curvilinear theory suggests that the relationship between religiosity and death anxiety is curvilinear. That is, both high and low scorers on religiousness report lower levels of death anxiety. Only those who are moderately religious report higher levels of death anxiety. There are three rival theories. What is the best way to atest the theory?
Hi,
The most common solution is to perform regression or ANOVA algorithms that include the testing the culvilinear relationship (in SPSS, for example, commands “Regression” : “Curve Estimation” ; in R: variants of lm or glm procedure dealing with curvilinear relations).
Some sites addressing the issue:
SPSS
http://www.ucdenver.edu/academics/colleges/nursing/Documents/PDF/NonlinearExampleHowTo.pdf
https://www.google.si/?gws_rd=cr,ssl&ei=Bak5VtCuEsaesAHf94awBw#q=testing+curvilinear+relationship+spss
http://essedunet.nsd.uib.no/cms/topics/regression/3/2.html
http://home.moravian.edu/users/soc/medwh02/0_DATA/ch8/ch8in.pdf
http://faculty.cas.usf.edu/mbrannick/regression/curvilinear.html
R
http://ww2.coastal.edu/kingw/statistics/R-tutorials/simplenonlinear.html
http://rcompanion.org/rcompanion/e_03.html
Regards,
Janek
Hi,
The most common solution is to perform regression or ANOVA algorithms that include the testing the culvilinear relationship (in SPSS, for example, commands “Regression” : “Curve Estimation” ; in R: variants of lm or glm procedure dealing with curvilinear relations).
Some sites addressing the issue:
SPSS
http://www.ucdenver.edu/academics/colleges/nursing/Documents/PDF/NonlinearExampleHowTo.pdf
https://www.google.si/?gws_rd=cr,ssl&ei=Bak5VtCuEsaesAHf94awBw#q=testing+curvilinear+relationship+spss
http://essedunet.nsd.uib.no/cms/topics/regression/3/2.html
http://home.moravian.edu/users/soc/medwh02/0_DATA/ch8/ch8in.pdf
http://faculty.cas.usf.edu/mbrannick/regression/curvilinear.html
R
http://ww2.coastal.edu/kingw/statistics/R-tutorials/simplenonlinear.html
http://rcompanion.org/rcompanion/e_03.html
Regards,
Janek
Hi Naser,
Check out these helpful YouTube videos (6 parts) that walk you through curvilinear regression in SPSS, which I know you've used in the past: https://www.youtube.com/watch?v=HtzE-HmAcwk
If you can't see that link let me know and I'll get them to you another way.
Hope all is well, take care,
Jason
I think the simple answer is to square the X variable and enter it in a hierarchical regression model in a second step after the linear (unsquared X) effect is entered on step one. As follows:
Step 1: X --> Y
Step 2: X Squared --> Y
If there is significant variability accounted for in Y by X squared in the second step, then there is a curvilinear effect. Keep in mind X squared will test just the quadratic effect. That is, U shaped or inverted U shaped relationships.
Dr. Toussaint,
Thanks for your answer. Could you please send me some texts or any papers that used this method? I'd appreciate it.
I just added a pdf that is a pretty comprehensive example. The example is not mine, but one that I like to use.
Its also important to include the linear term together with the curvilinear term. Also, check for outlyers. It is all well explained in Aiken & West (1991).
Religiosity tends to have curvilinear relationship with a lot of other variables. Always depends on you measure it. Anyhow, I would be careful with regression and avoid simple linear regression, of course. Try to estimate the curve first. Or if you are sure is a clean curvilinear then use a logit model.
Dr. Tröster,
Thanks. I dont think i'm familiar with Aiken & West (1991). Could you please give me the full bibliographical citation, so I can find it?
Aiken, L. S., and S. G. West (1991). Multiple regression: Testing and interpreting
interactions. Newbury Park: Sage Publications.
https://www.minitab.com/uploadedFiles/Content/News/Published_Articles/interactions_in_multiple_regression.pdf
- Badges
- Science topic
- Similar topics
- Psychology
- Social Psychology