diff --git a/HandlingIdentificationEtc.md b/HandlingIdentificationEtc.md
new file mode 100644
index 0000000..db7abfe
--- /dev/null
+++ b/HandlingIdentificationEtc.md
@@ -0,0 +1,25 @@
+Key points
+
+This is an attempt at measuring the effect of extending the enrollment period.
+The main issue is that the interaction between enrollment levels, enrollment status, and timing is confounded due to endogeneity.
+This can be addressed
+
+The other concerns are:
+- endogeneity between market and population.
+    I this isn't a caual issue because it is contained between the two, can be treated as a single RV and controlled for together.
+- ommitted variable bias. Did I forget or miss anything?
+    - The DAG is based on the details outlined based on FDA rules. I NEED TO LOOK THOSE UP AGAIN. The Assumptions that allow this to work are:
+        1. timeliness/accuracy in reporting open and close
+        2. updating certain details (open/close recruitment) is helpful because this is part of your marketing. (Concerns about measurement error)
+        3.
+- Where did the DAG come from?
+
+In spite of the endogeneity issue, I chose to continue modelling as if it were causal, because:
+1. If we assume an intervetion that is handles the joint timing/enrollment  status together, then it is causally identified (but hard to interpret)
+    - Walking away from identification is an issue in that you lose the use of this analysis
+    - Interpretation is as follows: changing enrollment status but breaking out of the standard timing of these things. Need a better way to say that.
+2.
+
+
+This is the only attempt I've found that tries to address this in a causal way, everything else is just descriptive.
+It also differs in being the first econ literature on measuring the impact of an operational concern.