Document changes in the Changelog and add Application Decision Records

CHANGELOG.md

@@ -28,3 +28,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Add previous year data.
 - Add scale to framework. Calculations for scores bubble up through the framework.
 - Change the threshold for sufficiency from a static number of responses to a response rate; from a minimum threshold of 17 teacher responses per survey item in a measure to 25 percent response rate in a subcategory and a minimum of 196 student responses per survey item in a measure to 25 percent response rate in a subcategory.
+- Correct response rate for short-form schools.
+- Correct scores for short-form schools
+- Show historical data even if a response rate cannot be calculated for a year of information.
+- Cap student response rate at 100

doc/architectural_decision_records/3.md

@@ -17,3 +17,5 @@ We have changed the minumim threshold to be a 25 percent response rate at the su
 ## Consequences
 
 This story, in combination with the story for marking short-form only schools, means that small population elementary schools will be able to see the responses visualized on the dashboard where before, that data would have been filtered out.
+
+Calculating the response rate depends upon knowing the total possible number of students and teachers for a given year for a given school.  Without those numbers a response rate cannot be calculated.  Changing the response rate threshold from a static number to a response rate percentage has meant that for years where the response rate cannot be calculated, data is not shown; no graphs are visualized.  An additional story will need to be created to figure out how to handle showing information for years where the response rate cannot be calculated.

doc/architectural_decision_records/4.md

@@ -0,0 +1,19 @@
+# Decision record 4
+
+# Correct response rate and scores for response rate schools
+
+## Status
+
+Completed
+
+## Context
+
+Data for short-form schools gets incorrectly listed as insufficient data.  Short form surveys include fewer questions than the default surveys.  When calculating the response rate for a subcategory we divide the total number of survey item responses by the number of questions to get the average responses per question.  We were incorrectly dividing by a higher number of quesions than are on the short-form survey.
+
+## Decision
+
+The decision was made to take into account if a school took the short form or long form survey and display the response rate and score accordingly.  In order to keep track of which school took the short form surveys in which years, we created a new database table 'Surveys'.   We also marked which questions were on the short form survey to get the correct count of questions by which to divide the responses.
+
+## Consequences
+
+There is additional code complexity in performing the response rate and score calculations.

doc/architectural_decision_records/5.md

@@ -0,0 +1,21 @@
+# Decision record 5
+
+# Show historical data even if response rate cannot be calculated for a year of information
+
+## Status
+
+Completed
+
+## Context
+
+After story #181231394 was completed, some data stopped being visualized.  Any school/year data where the response rate could not be calculated was no longer presented to the end user.  See the Consequences section of application decision record 3.  Story #181460465 was then created to show all data even if the response rate could not be calculated.
+
+## Decision
+
+During the response rate calculation, if the total possible number of students and teachers for a school and year has not yet been populated, we just return a response rate of 100.  We could return any response rate that fulfilled the sufficiency threshold (from 25 to 100 percent).  We chose to go with 100 percent.  We needed a valid response rate in order to allow the score to be calculated.  If there is insufficient data, the score does not get calculated.  When displaying the response rate we check if the total possible students/teachers is populated in the database and set the response rate to N/A if those values don't exist.
+
+## Consequences
+
+This is a stopgap measure and will be changed as soon as we have student and teacher data for the schools and years the surveys were administered.
+
+It means there is an artificially set 100 percent response rate that does not represent the reality of a response rate that cannot be calculated.