Fix incorrect p-value calculation

[raphaelsaavedra]

In handle_p_values, the function wants to compute two-sided Wald p-values, but the current formula is incorrect. It currently computes a computes a probability under N(beta_hat, std_err) evaluated at Z = beta_hat / std_err.

It should compute a probability under the null distribution of the estimator, not the estimated distribution, i.e., under the null hypothesis H_0: beta = 0, the statistic is Z = beta_hat / std_err and Z ~ N(0, 1). What we're doing is Z ~ N(beta_hat, std_err).

@guilhermebodin please double check if this sounds right...

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 91.29%. Comparing base (0b0198d) to head (0bd708b).

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #342      +/-   ##
==========================================
- Coverage   91.30%   91.29%   -0.01%     
==========================================
  Files          36       36              
  Lines        3713     3711       -2     
==========================================
- Hits         3390     3388       -2     
  Misses        323      323              

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[guilhermebodin]

Your fix is correct, by definition Z ~ N(0, 1). This should be the distribution where we test the p-values.