This is a modeling question. You don't "make the overall model ... look significant". You apply a model that makes sense on theoretical grounds. Fishing around for significance is known as p-hacking and is very bad practice. It would generally lead to rejecting the article.
On your second question, if the overall F statistic is not significant, you should generally not consider significance of the individual independent variable tests. There are a number of reasons for possibly producing that phenomenon, although the reverse situation is more common.
You can see a discussion of this in more detail here
--
Original Message:
Sent: 8/22/2023 10:00:00 AM
From: Alice Fujiwara
Subject: RE: The reason of different regression results between "enter" and "stepwise" methods
Thank you very much!
Do you know any ways to make the overall model (given by "enter" method) look significant?
OR
The results obtained through the 'enter' method can be reported directly in our paper. But if the overall model is not significant, does it still make sense for a particular independent variable to have a significant prediction on the dependent variable?
------------------------------
Alice Fujiwara
------------------------------
Original Message:
Sent: Tue August 22, 2023 09:26 AM
From: Jon Peck
Subject: The reason of different regression results between "enter" and "stepwise" methods
Well, it must be the case that the stepwise method selected different variables from the enter method. Stepwise would select variables based on their effectiveness in the regression. If there are missing values, that could also make the samples different and the results noncomparable.
However, you should be aware that the significance levels of the coefficients as well as the predictive power (R**2) of the regression are exaggerated by using the stepwise method.
------------------------------
Jon Peck
Original Message:
Sent: Tue August 22, 2023 07:46 AM
From: Alice Fujiwara
Subject: The reason of different regression results between "enter" and "stepwise" methods
When conducting linear regression analysis in SPSS, there is a significant difference in my regression results obtained using the "enter" and "stepwise" methods. All other settings of the two regression analyses are the same, except for the methods. There are five regressors and 18 cases.
With the "enter" method, the model is not significant (R2=.45, adjusted R2=0.15, p>0.2). But the predictor (regressor) of interest shows a significant effect on the dependent variable (p<0.05).
On the other hand, with the "stepwise" method, the model is significant (R2=.40, adjusted R2=0.37, p<0.01), and the predictor of interest also has a significant effect (p<0.01).
I would like to know why there is such a substantial difference between the two methods. Additionally, if I must use the "enter" method for regression, are there any ways to make the overall model significant?
------------------------------
Alice Fujiwara
------------------------------