Clarifications following 538 (#1)

* Update data_dict.md

* Update README.md

* Update README.md

* fix typos

Co-authored-by: Peter Ganong <ganong123@users.noreply.github.com>

Author: peterobertson

README.md

@@ -61,7 +61,13 @@ Benchmarks source: Unemployment Insurance Benefit Accuracy Measurement data
 
 The calculator computes the benefits of a worker without dependendents as of January 2020. It also can be used to simulate alternative policies, such as Federal Pandemic Unemployment Compensation.
 
-Many states have complex UI benefit rules. We follow the rules described in ["Significant Provisions of State Unemployment Laws"](https://oui.doleta.gov/unemploy/content/sigpros/2020-2029/January2020.pdf). This document, and therefore our calculator, do not capture every aspect of a state's UI rules. A dictionary of the features used by the calculator can be found in [`data_dict.md`](data_dict.md). 
+Many states have complex UI benefit rules. We follow the rules described in ["Significant Provisions of State Unemployment Laws"](https://oui.doleta.gov/unemploy/content/sigpros/2020-2029/January2020.pdf). If a state has multiple ways of satisfying eligibility or multiple ways of calculating benefits, we include only the first listed way. This document, and therefore our calculator, do not capture every aspect of a state's UI rules. A dictionary of the features used by the calculator can be found in [`data_dict.md`](data_dict.md). 
+
+
+### Acknowledgements
+If you find a problem, please open a github issue or even better propose a fix using a pull request. 
+
+Thank you to Maya Sweedler for helpful feedback.
 
 ### License
 This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

data_dict.md

@@ -15,9 +15,9 @@ benefit amount.
 
 | Scenario                  | Weekly Benefit Amount |
 | ------------------------- | --------------------- |
-| `wba_formula < min`       | `min`                 |
-| `min < wba_formula < max` | `wba_formula`         |
-| `wba_formula` \> `max`    | `max`                 |
+| `wba_formula` < `min`       | `min`                 |
+| `min` < `wba_formula` < `max` | `wba_formula`         |
+| `wba_formula` > `max`    | `max`                 |
 
 The `wage_concept` records how the `base_wage` should be calculated and
 has five possible
@@ -70,3 +70,15 @@ the majority of values in `state_eligibility.csv` are
 | `wba_2hqw`          | minimum earnings in two highest quarters as a multiple of the weekly benefit amount            |
 | `abs_2hqw`          | minimum dollar amount in two highest quarters                                                  |
 | `hqw_2hqw`          | minimum in two highest quarters as multiple of high quarter wages                              |
+
+## State Specific Notes
+
+Colorado, Iowa, Minnesota, Ohio and Wyoming all include statewide averages as part of their calculations. We use averages [from the BLS](https://www.bls.gov/oes/current/oessrcst.htm).
+
+Colorado, Minnesota and New York have benefits schedules which change if the worker's income is in excess of a particular amount. We include the `inc_thresh` variable explained above to account for this. 
+
+New Hampshire benefits are set to 1% of annual wages. This is the lower bound of the range of 1% - 1.1% in the Significant Provisions document. We could not readily find information on how the range between 1% and 1.1% is determined.
+
+North Dakota has a `wage_concept` which is not common to any other states and is coded as `ND` and described in the `wage_concept` table above.
+
+Alaska, Pennsylvania and Utah all have replacment rates which vary with income as a result of a fixed payment (or deduction) from the benefit. This is coded in the `intercept` column.