Translate

Friday, August 21, 2015

Recidivism in Whatcom County Jail Bookings Records : 1/1/2011 - 7/1/2015


29,637 Bookings for Whatcom County Jail covering the 4.5 year period from 1/1/2011 to 7/1/2015. The vertical axis represents 17,256 unique names. 5,314 of those unique names had multiple bookings for the period with a range of 2  -  22 bookings per individual. Each unique name receives a random color for their bookings. See an animation of this chart here. See R code here. See notes on data/methodology at end. Click on the Charts to enlarge. 

This post discusses recidivism in Whatcom County jail. Recidivism is a primary concern for law and justice and  citizens. The bottom line on 'recidivism' is this: so many re-offend at such high rates that some criminology professors fundamentally doubt whether imprisonment has any significant effect on prohibiting crime.  Literature on this subject is abundant[1,2,3,4].  Some of us have come to the conclusion that imprisonment itself is the primary cause of recidivism; a counter intuitive analysis that has little political chance of wide scale adoption. I am looking at 'booking recidivism' or booking 'return buckets' in Whatcom County over the 4.5 year period from 1/1/2011 to 7/1/2015. (e.g. How many times a user is "booked" in that period.  For this definition of recidivism, I am counting multiple bookings or returns to jail per individual (user).   These aren't necessarily convictions. What I describe are 'return buckets' to our jail. Here are granular views at the 'return buckets' by user booking counts for these 4.5 years.  Click to enlarge chart.


Above: Chart of data table below showing of bookings per 'return bucket' (e.g. 'bookings count'/ unique individual for the 4.5 year period).

Below: Data table showing number of Users @ booking Number/per user with product of Users and Number (User * Number) = Count of  total bookings.

Far Below: Density plots for Users in red and Count in blue.

   Users Number Count 
1      1     22    22
2      0     21     0
3      0     20     0
4      0     19     0
5      1     18    18
6      3     17    51
7      3     16    48
8      3     15    45
9     12     14   168
10    16     13   208
11    14     12   168
12    28     11   308
13    40     10   400
14    69      9   621
15    89      8   712
16   141      7   987
17   216      6  1296
18   327      5  1635
19   567      4  2268
20  1172      3  3516
21  2612      2  5224
22 11942      1 11942

The tables below show that Whatcom County books the equivalent amount of 2% or more (1 out 50!) of its own population each year. (See column Unique.TotPop) In the tables below, I project totals for 2015 for all fields: charge, booked, unique. To project 2015, I doubled the the half year total. For Total Population (TotPop) I added the 2013-2014 increase to 2014 to create a 2015 estimate.

  Years Charge Booked Unique TotPop
1  2011  11333   6860   4989 203329
2  2012   9650   6523   4723 204827
3  2013   9577   6677   4890 206248
4  2014  10520   6348   4466 208351
5  2015  13304   6458   5190 210454

The table below divides Charge, Booked, Unique by Total Population:

     Years Charge.TotPop   Booked.TotPop Unique.TotPop
[1,]  2011    0.05573725    0.03373842    0.02453659
[2,]  2012    0.04711293    0.03184639    0.02305848
[3,]  2013    0.04643439    0.03237365    0.02370932
[4,]  2014    0.05049172    0.03046782    0.02143498
[5,]  2015    0.06321571    0.03068604    0.02466097

The table below is derived from the two tables above.

  Years Charge Booked Unique TotPop Years Charge.TotPop Booked.TotPop Unique.TotPop
1  2011  11333   6860   4989 203329  2011    0.05573725    0.03373842   0.02453659
2  2012   9650   6523   4723 204827  2012    0.04711293    0.03184639   0.02305848
3  2013   9577   6677   4890 206248  2013    0.04643439    0.03237365   0.02370932
4  2014  10520   6348   4466 208351  2014    0.05049172    0.03046782   0.02143498
5  2015  13304   6458   5190 210454  2015    0.06321571    0.03068604   0.02466097 


Recidivism can be expressed as levels instead of buckets. Below is a table and some charts. The first three charts give levels greater than 3, 5 or 10 as horizontal bar charts. The next three charts give bookings/ user greater than 10, 5, or 3 as points which show users by first name bottom to top on the vertical axis and bookings (time-series) left to right. This allows us to see that recidivism increased dramatically after after the year 2011 (~7200 bookings) ended. All of the users with greater than 10 bookings accumulate these after the start of 2012. Obviously, some change in technology or policy increased bookings from that point onward.  Click to enlarge the Charts:

Level   Count
15orGTR   11
13orGTR   39
10orGTR  121
5orGTR   963
3orGTR  2702
2orGTR  5314 



Notes on Data:

The Whatcom County Inmate press releases were used to compile this data. Unique identifiers were created from paste First, Middle, Last Names. In R:

paste(Booked201x$FirstMiddle,Booked201x$Last)

Unique numerical IDs weren't appended to the press releases until December 2013. Data may be subject to revision. For R code see here.

3 comments:

Dan McShane said...

Thanks Ryan. Good stuff to contemplate

Matt Thompson said...

Ryan, Thanks for the information. When looking at the total booking by Whatcom population, 2%, did you take into account that single person's will be booked multiple times? So, how many persons were booked throughout the year a single time versus more than once?

Thanks again.

Ryan M. Ferris said...

I used unique.totpop for the 2% or more figure so that should be unique (unique.(firstname+middlename+lastname) / total pop). However, I created that primary key because not all firstname+middlename+lastnmes had a unique id. firstname + middlename + lastname collisions could have happened because people can share firstname+middlename+lastname. To be honest, it would be more likely with this population that the same person was named differently for different bookings. Best I could do without consistent unique (numerical IDs).