combine two equivalent points

Author: drbenvincent

docs/source/notebooks/graded_intervention_time_series_single_channel_ols.ipynb

@@ -61,9 +61,8 @@
     "Standard regression assumes **independent errors** — that the unexplained variation at time $t$ is unrelated to time $t-1$. When this assumption fails (as it almost always does in time series):\n",
     "\n",
     "1. ✅ **Coefficient estimates remain unbiased:** still correct on average\n",
-    "2. ❌ **Standard errors are WRONG:** typically too small, leading to overconfident inference\n",
+    "2. ❌ **Standard errors are WRONG:** typically too small, making confidence intervals too narrow\n",
     "3. ❌ **Hypothesis tests are invalid:** false positives, misleading p-values\n",
-    "4. ❌ **Confidence intervals are too narrow:** underestimate true uncertainty\n",
     "\n",
     "This means you might conclude an intervention \"works\" when it actually doesn't, or claim high precision when you're actually quite uncertain!\n",
     "\n",