Today’s Irish Times reports that the Cabinet is examining the option of redeploying Census Enumerators back into the field once the Census is complete. Apparently the coin has dropped that the “the names on the register are very different from those of the households”. Glad to see the paper of record is only 12 months or so behind the Tribune.
“The problem facing the Government is that action will have to be taken quickly if an accurate register is to be compiled in time for the next election. The census enumerators would supply a ready-made team to deal with this.”
But buried at the bottom of the article is the comment that the CSO believes that using Census Enumerators would be “too cumbersome”. Why ever would that be?
This brings me to the first problem with this solution. Most census enumerators have been employed on short-term part-time contracts with an end-date of the date of the 21st of May or thereabouts. Many of these part-time enumerators have other jobs. Many are buggering off on holidays as soon as the grind of 3 months of late evenings driving around housing estates is done.
The anecdotal evidence I have from talking to some enumerators past and present is that they would be unlikely to do the job again even if they were offered €10,000. Come the end of the Census, there may not be a standing army of enumerators to call on.
The second problem is that a push to “get the register in order before the next election” is simply scrap and rework. What are the government proposing to do to maintain a newly cleaned electoral register?
Once it is magically cleansed and restored to order, what will stop it rapidly becoming just as bad again? What controls are to be introduced into the process (technical and/or human) to prevent the data quality from decaying? What root cause analysis will be done on the current state of the register to identify what part(s) of our democratic process have failed to keep pace with modern Ireland?
 I note that Fiona O’Malley TD (she of “I was registered twice” fame) will be tabling a motion at the PD Annual Conference to introduce a new system based on PPS numbers (similar to US Social Security Number). This is a commendable suggestion in principle – an additional data marker would make the matching of names and addresss data more accurate. The general rule is name+address+1 other. However, the capture of more information in a process that is broken without the necessary changes to processes (people processes and technology processes).
Is there a centralised electronic register within which the occurences of duplicate PPS numbers can be identified? Would the Data Protection Commissioner permit the use of PPS numbers for this purpose? What controls would be in place to ensure the accuracy of the pps numbers input (ie to prevent miskeying or transposition errors or simply incorrect PPS numbers being used)? Would all of this be ready in time?
As one of the key problems is that people are registered at multiple addresses perhaps the use of a question on a voter registration form about where the person lived before if they have been living at their current address less than 3 years would be a useful check moving forward. However in the absence of a PPS number or other additional data marker it may not be advisable to start matching people at old addresses against other regional electoral rolls.
 More data does not necessarily mean better quality information unless the processes that create, use and maintain that information are addressed to ensure that they too are capable of delivering a quality product.
What we are seeing could be described as tactics without a strategy. Deploying enumerators into the field to do scrap and rework on the data is a tactic. Without a strategy to improve, and to continously improve what W.Edwards Deming called the “means of production” – the process by which the register is complied and managed- that tactic will never achieve, on a sustainable basis, the capacity to meet or exceed the expectations of the citizens or the political classes.
If enumerators are available then they should be utilised. However they should be utilised in the context of a clear vision and strategy for the management of electoral register data.
 Some suggestions:
1) Assess the Information Architecture for the Electoral process
- Are all Local Authorities holding data in the same formats?
- Do all Local Authorities understand and call core concepts in the electoral process by the same name (ie voter or citizen – are they the same thing?
- Map the processes for Electoral data management – do they make sense?
- If the information is held locally, ask if the architecture and processes allowed for the information to be stored centrally would the risk of problems be reduced?
 2) Conduct a Root Cause Analysis on the current situation using TQM techniques
- A ‘fishbone’ diagram is a useful tool to identify what the contributing factors to the electoral register issues actually are.
- Efforts can then be focussed on actual causes rather than perceived causes and efforts can be prioritised by impacts
- ‘Speak with data’ – measure the impacts of these potential causes and prioritise.
- Institute Leadership to breakdown barriers between areas (what Pat Rabitte has called the “traditions” that prevent solutions being implemented).
 3) Clean up the historic problems, but at the same time fix defective processes to prevent recurrence.
- Scrap and Rework is a non value-adding activity.
- Cost savings, efficiencies and sustainable quality come about through tackling the ‘means of production’ to build quality in rather than inspecting defects out.
What role for the CSO in all of this longer term stuff? Once the processes are addressed to build quality in and the legacy sh!te data is cleaned up, regular measurement and control of the Electoral register will be required to ensure that the processes continue to function correctly. This is statistical process control, a well proven technique. A perfect role for statisticians.
As for the obstacles to all of this, Pat Rabitte of the Labour Party is right to point a finger at tradition and the civil service mindset of silo’d thinking. However there are certain other things such as the Data Protection Act and the Statistics Act that need to be borne in mind also.