search

data-liberation-project / epa-rmp-coordination

A repository to track efforts to publish the EPA RMP database and associated documentation.
1 stars 0 forks source link

To analyze: How consistently do facilities submit updates after accidents? #3

Open jsvine opened 1 year ago

jsvine commented 1 year ago

This question, drawn from the documentation draft seems to be empirically analyzable — at least to some degree. An initial test might do this:

  1. For each accident reported through the Accident History portion of the RMPs (tblS6AccidentHistory) ...
  2. ... with an accident date after the facility's first-ever RMP submission ...
  3. ... how long elapsed between the accident date and the next RMP submission?

For the purposes of discussion, let's call this metric days_until_submitted. What is the distribution of days_until_submitted overall, by accident year (are facilities submitting more/less timely?), by facility industry, et cetera? What have been the biggest delays? What can we learn about RMP data quality from those?

As context, the RMP submission guide suggests facilities are required to submit an RMP update after a reportable accident (“e.g., changes in emergency contact information, facility address, or change in accident history”). The relevant statute, however, does not seem to say so explicitly. (Although perhaps it is implied by “Within six months of a change that requires a revised PHA or hazard review.”)

jsvine commented 1 year ago

An observation: In the tblS1Facilities table, there is a RMPSubmissionReasonCode field, with corresponding lookup table tlkpSubmissionReasonCodes. Reason code C05 is described as "New accident history information". Unfortunately, it appears that this code was used in a couple hundred submissions received 2004–2009, but none before or after:

year count
2004 16
2005 57
2006 47
2007 60
2008 52
2009 32

Will add a note to the documentation that these submission-reason codes aren't necessarily used consistently over time.

jsvine commented 1 year ago

Update: Per table in https://github.com/data-liberation-project/epa-rmp-coordination/issues/7, the disappearance of RMPSubmissionReasonCodes is not specific to this particular reason — it seems to extend to all C0_ codes. So, a follow-up question: Is there a corresponding R0_ code that fully captures C05 after that code stopped being used? I'm concerned that there might not be.

jsvine commented 1 year ago

Have now asked the EPA about the RMPSubmissionReasonCodes, and whether there's a corresponding R0_ code we could use.