I am a beginner in statistics so I would like some help with literature. I have a linear model $Y= b_0 + b_1 X^2$ Intercept $b_0$ is not significant (sig: 0,09 > 0,05) but $b_1$ is very significant. I cannot find any literature to support the fact that it is not a big problem for the intercept being zero and also I am not sure I am right.

I understand there's another similar question but I wonder if anyone has an official paper or book that, states the answer to my question. I haven't found something answering exactly that.

  • $\begingroup$ Related: stats.stackexchange.com/q/99267/27276 $\endgroup$ Jun 4, 2018 at 9:56
  • 2
    $\begingroup$ Did you really mean to fit a model with a quadratic but not the linear term? This is quite a specialised model. $\endgroup$
    – mdewey
    Jun 4, 2018 at 12:31
  • 1
    $\begingroup$ Whether this is "a problem" or not totally deepens on the question you are asking and what you are modelling. You won't find any source saying that it is always (never) a problem that the intercept is (non-) significant. $\endgroup$
    – Cederlöf
    Jun 4, 2018 at 12:50