how to boost the coefficient difference between tow groups, OLS

[Amal Alfawzan]

On my phd I have an argument states that gender is a predicted variable for the relationship between innovation and religion. Men is more likely to have higher effect than women. The data showed me a significant difference between genders, however, the difference is very small regarding the coefficient only 0.05. Although its significant by performing t-test, my supervisor is not happy with this result. My questions are:
1- what is the acceptable coefficient differences between tow groups
2- are there any methods that can boost the coefficient differences between the 2 genders? I am using STATA 2015

thanks

[Carlo Lazzaro]

Amal:
unfortunately, the answer to both your questions is: no.
Kind regards,
Carlo
(Stata 16.0 SE)

[Amal Alfawzan]

thank Dr Carlo ,

i cant understand your answers for this question
what is the acceptable coefficient differences between tow groups?

[Carlo Lazzaro]

Amal:
sorry for my being unclear.
I meant that there's no White/black magic to set a threshold that splits acceptable vs non-acceptable coefficients.
The same holds for possible magic wands that move up (or down) our coefficiients according to our desires/instructions.
I'm also surprised that your supervisor is not happy with your results: data are what they are, and we have simply to live with them.

[Amal Alfawzan]

I meant what is the minimum number of coefficient differences between tow groups can be acceptable ?

[Amal Alfawzan]

Thanks
mu supervisor said that ,there is no differences between gender with this results Because the coefficient difference here is not academically and convincingly, the difference must be greater than 0.05

[Carlo Lazzaro]

Amal:
unfortunately, my answer cannot change.

[Phil Bromiley]

Since you do not appear to be understanding Carlo's very appropriate answer, let me take another try to explain the problems.

First, there is a difference between statistical significance (being confident the parameters differ) and substantive significance (the difference in predicted value being meaningful in context, often that it makes a meaningful difference in the predicted value).

Second, the statistical results are whatever the statistical results are. It is inappropriate to attempt to manipulate what you're doing to get a larger difference to make anyone, including an advisor, happy. There is nothing that guarantees a given parameter will have a given magnitude until you estimate the parameter.

If this is a simple linear regression, a coefficient of .05 on a dummy variable for gender would mean the predicted value is .05 larger for one gender than the other. Whether this is substantively significant depends entirely on your context. For example, if I measure something in dollars .05 will have a different implication that if I measure something in pennies. And in either case, we have to consider the context – a difference in predicted value of $1 million would mean a lot if we're talking about individual income but would mean little if we are talking about income for massive corporations. Sometimes, people compare the difference in predicted value to the standard deviation in the dependent variable.

[Amal Alfawzan]

Dear Dhil

Thank you so much For your answer and time. It is very clear now. I performed 2-tailed t-tedt and it is significant . Does this support Does this support the research hypothesis which is men are more innovative than women?