That is correct. If you set the alpha value at 0.05, if the statistical significance of any independnet variable is higher than 0.05, then those variables do not have as much influence on the dependent variable compared to those with p
Your expression is correct. They don't have to be included in the model and Srini is very right. Depending also on the method of model selection, they can be included in the model. I have a model that a variable with p = 0.051 was included in the model. It's attached.