CODEWITHSUNDEEP
rstatistics
Andy Papadopoulos
Andy Papadopoulos

Reputation: 19

Negative Confidence Interval

This is the confidence interval of my model:

> confint(tbest,level=0.90)
                        5 %          95 %
(Intercept)        7.7300811475 14.4907225225
age                0.0163843843  0.0732836893
sex                0.7901979755  3.1819643312
betadiet           0.0001948422  0.0007758837
vituse            -2.0411193539 -0.4811121339
calories          -0.0010620015  0.0010316039
quetelet          -0.2400225686 -0.1143914971
fiber              0.1022553071  0.2886850591
smokstat           0.2034217522  4.7632612497
calories:smokstat -0.0031382219 -0.0006940886

I am interested in the interaction of calories:smokstat. It seems that I get a negative value. How do I interpret that In my results? Do I round it up to zero and conclude that I accept the null hypothesis?

Upvotes: 0

Views: 1423

Answers (1)

dvarelas
dvarelas

Reputation: 988

Interesting question. As you can see your upper bound is really close to 0, which means by selecting a slightly different alpha (by alpha I mean 1- level) the interaction between those two would be statistically insignificant.

Given that I don't know how many data points you have or how you trained your model I wouldn't trust that result.

Upvotes: 0

Related Questions