A lot of scientists misinterpret and take the hypothesis test values under absolute terms without knowing the full implications of disregarding all P = |t| values >0.05 as if there's some magic value at which a variable becomes negligible. If you study the algebra and the details behind the way multiple linear regression works, if one just simply ignored a variable because it's P = |t| value is > 0.05, the effects that variable has upon the model doesn't just disappear. While it may not increase the coefficients by which we measure all variables, it can definitely change the values of some of the coefficients. Where does the remainder of the effects go? Into the error term. I would argue (and there is a lot of people that disagree with me), that just because a variable is less significant, doesn't make it insignificant, ESPECIALLY if it effects the way we measure the other variables in the equation. Now, if the removal of the disputed variable doesn't effect the estimates of the other coefficients in the equation, then by all means drop it as it should go into the error term. I expect that many scientists understand this concept while a majority of people who haven't studied the MLR model as much do not and as such know that they would be stigmatized for including these (allegedly) insignificant variables.