How Transparent, Proactive, and Explainable Data Makes for Better Informed Decision-Making in Government

Government data is integral to government programs' success and carries higher consequences than industry data. While industry data issues can result in a million dollars in damage and lost jobs, government data mistakes can lead to a trillion dollars lost and even ruined lives. Public sector programs must choose purpose-built technology for programmatic efficiency to avoid negative societal impacts. While more commercial efforts like social media, advertising, and e-commerce have wiggle room to pivot and improve, government tools don’t have that forgiveness. Without robust solutions, it’s the difference between a constituent receiving unemployment benefits or a child not receiving critical education resources.

Some methodologies can help government entities deliver the programs they promise. These fundamental principles can transform government programs from error-prone and painful to reliable and delightful. They encompass transparency, proactivity, and explainability. Let’s begin with an example of how transparency can halt and help government programs.

Transparent Claim Management Supports Child Enrichment in Ohio

For government program beneficiaries, transparency means knowing where they stand and what steps they must take to receive benefits. This is particularly important when money is on the line.  An example of a lack of transparency was clearly shown during the spike in unemployment from COVID-19 in California. The Employment Development Department (EDD) was stretched to its limit. With the EDD’s attempt to crack down on fraud, thousands of Californians were without necessary benefits.

Beneficiaries who were a part of this program were compounded by the fear of running out of money and the uncertainty of what would happen next. “I feel like I’m a hostage,” said a California office manager, “I had no money, and I kept saying: ‘How long is this going to take?’” 

This sentiment is a fundamental sign of a program lacking transparency: ambiguous steps without communication leave constituents confused and afraid. The problem is even worse for people with unusual living situations or those who fill out paperwork incorrectly, as they often slip through the cracks.  In the words of an LA-based lawyer, “The way I see it, the EDD is punishing regular civilians who are just filing for benefits and making honest mistakes.”

Transparent programs present constituents with all their program data and provide well-defined pathways for fixing data problems, whether due to a form filled out incorrectly or changing life circumstances. When waiting for program administrator action—especially if there are long delays—expected wait times are available, and the purpose of the pending admin action is provided.

When government programs model a transparent approach, such as Ohio's Afterschool Child Enrichment program—which provides $1,000 per child annually to parents for extracurricular activities—they set a positive example. Eligibility requirements are well defined, and parents are informed if and why they don’t qualify. Parents are presented with their data online and have easy digital access to program admins to fix data issues. For claims, families are informed why each claim is or is not reimbursable through the web or a mobile app. This transparency contributed to the ACE program becoming the largest supplemental ESA in the country, with 75,000+ enrolled students and $125+ million allocated.

How Proactive Data Measures Help Unemployment Insurance

As illustrated, large-scale government programs require high accuracy in constituent data and effective communication and service delivery for success. Without these guiding principles for government data, programs may face significant challenges in program implementation. Historically, the sheer volume of data made it impractical to double-check everything, leading to overlooked inaccuracies and compromised interactions. 

However, proactive data management is feasible with advanced tools and AI technologies. Organizations can significantly enhance their data quality by meticulously combing constituent data and addressing discrepancies. This proactive approach ensures that every interaction is based on precise information, mitigating the negative impacts of compromised data on individuals. Embracing these modern solutions transforms data management from complicated tasks into a strategic advantage, increasing trust and efficiency in every constituent relationship.

The unemployment insurance system is a prime example of a government program that could embrace modern solutions for a proactive approach. No one could’ve foreseen the global pandemic—but when it came to the Unemployment Insurance system and allocating funds to Americans nationwide to offset the financial burdens faced by layoffs and stay-at-home orders, the program quickly realized many challenges: program design and variation in how states were administering, which caused disparity in benefit distribution. Additional challenges included “providing customer service providing customer service, delivering timely benefits, implementing new programs, and using and modernizing legacy IT systems.”

The Unemployment Insurance system aimed to aid Americans but faced delays, exacerbating financial hardship. In the future, government initiatives could use AI technologies to swiftly notify eligible Americans so that there is no question about the timing of the delivery of unemployment benefits. 

Embracing and Conquering Program Complexity with Explainability

Federal and state government programs inevitably become complex as services scale to millions of people and diverse regions. Grappling with this complexity is the hallmark of a well-run government program. Government programs might strive to avoid complexity by enforcing rigid rules, often excluding many constituents. Or they can embrace complexity by providing flexible programs that meet people in the messy real world. Embracing complexity has historically led to ruin, but explainability offers a solution.

A heartbreaking example of complexity crushing a government program is the rebuilding of the VA Hospital EHR (Electronic Health Record) program. The VA pioneered EHRs, revolutionizing patient care through technology, but the rebuilding has been plagued with problems. Even though the new software will cost taxpayers 10 years and $16 billion, it is so complex that only 44% of department staff passed an initial proficiency test.

Representative Cathy McMorris Rodgers, a Republican from Spokane, highlighted the severity of these issues: “I have one report of a VA doctor ordering two medications for a veteran, who then received 15 incorrect medications. I also have multiple reports of delayed prescriptions, which, in one case, caused a veteran to suffer withdrawal. These impacts are dangerous and unacceptable.”

One recent VA EHR rebuild challenge: 250,000 VA patients may have received the wrong medication. These clerical issues hurt real veterans: “A patient with post-traumatic stress disorder and traumatic brain injury wasn't given medication they needed to treat adrenal insufficiency because the residential rehabilitation program they were in didn't see a prescription for the medication.” One of the most frustrating issues with this problem is the uncertainty - out of those 250,000 veterans, program administrators can’t determine who received the correct medication and who did not. The new VA Hospital EHR program lacks explainability.

Explainable data systems provide a complete audit trail of both manual and automated data management tasks. Let’s take an example of a veteran’s medical records: when a new diagnosis is made by a doctor, that record is added to the veteran’s profile, and all metadata about that event (which doctor input the record when it occurred and diagnostic equipment used, etc). If the diagnosis is later augmented with additional test results added via API by a testing partner, the same metadata is captured again. This metadata forms a lineage graph that can be traversed through time to understand how data was derived - providing a full explanation of that data throughout its lifetime. Hospital staff or patients can swiftly leverage this metadata to know how their data came to be in its current state. In our medication example, caregivers and their patients could understand which automated system had made a mistake and rectify it without ambiguity.

Looking Forward with Stronger Government Data

In the future, government systems should have data that is: 

• Transparent so that constituents can understand what the government knows about them and what they need to do next to access benefits
• Proactive so constituents can be promptly notified of data issues requiring their attention.
• Explainable so constituents can track each piece of data back to its source and understand why decisions were made.

With these principles implemented, programs run by government agencies will be easier to work with, more efficient, and free of data errors. If governments fail to support their constituents with better technology, we’ll see more lives ruined as people fall through government programs' cracks and public institutions' distrust deepens.

At Merit, we’ve built transparent, proactive, and explainable tools by developing and adopting cutting-edge AI, provenance, monitoring, cleaning, and visualization technologies. We use these tools to support programs for many government agencies across workforce development, education, and emergency response. Ready to learn more about how this technology can help your constituents?