Public data does not create transparency simply because it is public.
For a public accountability system to work, the information it provides must allow legislators, families, researchers, journalists, and members of the public to answer basic questions about how the system is functioning—and to independently reproduce the answers being reported.
Missouri’s child welfare dashboard represents an important step toward greater public visibility into the Children’s Division. But visibility and accountability are not the same thing.
The question is not simply whether data has been published.
The question is whether the published data can actually answer the questions the public has a right to ask.
A Simple Question Should Have a Verifiable Answer
Consider a straightforward question:
How many children in Missouri state custody are currently listed as runaway?
A legislator should be able to ask that question.
The Children’s Division should be able to answer it.
And Missouri’s public accountability dashboard should allow that answer to be independently verified.
At a recent legislative oversight hearing, the number of children in runaway status became one of the questions raised by lawmakers.
The larger issue, however, is not any disputed number given during that discussion.
It is that someone outside the agency cannot easily use the public dashboard to determine the answer for themselves.
That is a transparency problem.
What HB 1414 Was Supposed to Accomplish
Missouri’s child welfare dashboard exists because lawmakers recognized a basic accountability problem: enormous decisions are made within the child welfare system, but historically, much of the information necessary to evaluate that system has been difficult for the public and policymakers to access.
HB 1414 required greater public reporting of child welfare information.
The purpose of that kind of reporting is not merely to produce charts.
It is to make the operation of the system more visible and understandable.
That distinction matters.
A dashboard can contain thousands of data points and still leave important questions unanswered if users cannot determine what those numbers actually represent.
What the Dashboard Does Provide
The dashboard provides public access to information that previously would have required considerably more effort to locate.
Users can view measures related to areas such as:
- Children in foster care
- Placement settings
- Permanency
- Caseworker caseloads
- Parent-child visitation
- Demographic characteristics
- Entry and exit information
- Selected performance measures
That is meaningful progress.
Centralizing this information makes it easier to identify trends and gives policymakers and the public a starting point for understanding Missouri’s child welfare system.
But publication alone does not make data interpretable.
To understand a number, we also need to know how that number was created.
Placement Categories Are Not the Same as Placement Codes
One of the clearest examples is placement data.
The public dashboard presents broad placement categories.
Behind those categories, however, Missouri’s child welfare information system uses more specific placement codes.
Those two levels of information are not interchangeable.
A category might tell us that a child is included under:
Relative
Foster Care
Residential
or
Other
But an analyst trying to understand the system needs another piece of information:
Which underlying placement codes are included in each category?
Without that crosswalk, the public sees the final category without being able to reconstruct how children were assigned to it.
That becomes especially important when terminology has operational meanings inside the Children’s Division that may not match the ordinary meaning a legislator, parent, journalist, or member of the public would reasonably assume.
Terms such as relative, kinship, trial home visit, residential, and other need explicit definitions.
Otherwise, two people can look at the same chart, use the same words, and still be talking about different populations of children.
The Problem With “Other”
“Other” categories are sometimes unavoidable in large administrative datasets.
But they create a serious analytical limitation when their composition is not disclosed.
For example, Missouri’s underlying placement information includes a code for children listed as runaway.
On the public dashboard, however, runaway status is not presented as its own visible placement category. It is incorporated into the broader Other category.
That means someone looking specifically for the number of children recorded as runaway cannot simply open the dashboard and retrieve it.
Instead, they see “Other.”
But “Other” does not tell us how many of those children are runaway.
It also does not tell us, at a glance, what additional statuses have been combined with them.
That matters because these populations may represent completely different circumstances and policy concerns.
A child who is missing from placement raises very different oversight questions than a child whose placement falls into another administrative category.
Combining them may be useful for a high-level visualization.
It is not sufficient for meaningful public accountability unless the underlying composition remains accessible.
Definitions Are Part of the Data
This problem extends beyond placement categories.
A number cannot be properly interpreted without understanding its denominator, inclusion criteria, exclusions, reporting period, and underlying definition.
Consider the difference between questions such as:
How many children were in foster care during the year?
and
How many children were in foster care on the last day of the year?
Those numbers can both be accurate while describing completely different things.
The same problem arises when determining whether a metric represents:
- Children or cases
- Point-in-time counts or cumulative counts
- Placements or individual children
- Events or unique individuals
- Legal status or physical placement
- Fiscal year or calendar year
- Current status or status at a particular reporting date
These distinctions may seem technical.
They are not.
They determine what the numbers mean.
When the Logic Cannot Be Reconstructed
While reviewing Missouri’s publicly available child welfare data, I have encountered metrics where the relationship between the displayed measure, its definition, and the underlying population is difficult to reconstruct from the information provided.
That does not automatically mean the underlying data is wrong.
It means the public cannot adequately determine whether it is right.
Those are different claims—and the distinction is important.
A transparent system should not require the public to assume that an unexplained calculation is correct simply because it appears on an official government website.
The methodology should allow the calculation to be reproduced.
Transparency Requires Reproducibility
This is the standard I believe Missouri should adopt:
If a public metric is intended to provide accountability, an informed outside user should be able to reproduce it from the definitions and source information provided.
That does not mean publishing confidential case information.
It does not mean releasing children’s names, family information, case narratives, addresses, or personally identifiable information.
It means publishing enough aggregate methodology to understand how the reported numbers were produced.
For each metric, the public should be able to determine:
What is being counted?
What is excluded?
What time period does the number represent?
Is the unit a child, case, placement, episode, event, or something else?
Which administrative codes are included?
What is the denominator?
How frequently is the measure updated?
What source system produced it?
Were historical values recalculated when definitions or systems changed?
If those questions cannot be answered, independent verification becomes extraordinarily difficult.
And without independent verification, a dashboard functions primarily as a publication tool rather than an accountability tool.
The FACES Problem
There is another layer to this issue.
Much of Missouri’s child welfare reporting ultimately depends on information recorded in FACES, the state’s case management system.
That means public reporting inherits many of the limitations of the underlying administrative system.
If a status is inconsistently entered, difficult to extract, reclassified over time, or grouped differently between reports, those decisions can affect what ultimately appears in public-facing data.
This makes documentation especially important.
The public does not need access to individual FACES records.
But it does need to understand how information moves from:
case record → administrative code → reporting logic → public metric
Without that chain, the public sees the end product without being able to examine how it was produced.
A Dashboard Should Help Legislators Ask Better Questions
The recent oversight hearing demonstrated why this matters.
Legislators should not have to rely exclusively on an agency’s verbal response when asking basic quantitative questions about the system they are responsible for overseeing.
Ideally, a committee member should be able to ask:
How many children are currently listed as runaway?
And then open the public reporting system and verify the answer.
From there, lawmakers could ask the more important questions:
How long have those children been missing?
How many have been missing longer than 30 days?
How many entered runaway status from residential care?
How many have experienced repeated runaway episodes?
What percentage have been located?
How quickly are missing children typically recovered?
Are particular placement types associated with higher runaway rates?
Those are oversight questions.
Good public data should make them easier to ask—not harder.
What Would Make the Dashboard More Useful?
Missouri does not necessarily need to rebuild the dashboard from scratch.
Several relatively straightforward additions would dramatically improve its analytical value.
1. Publish a Data Dictionary
Every metric should have a plain-language definition describing exactly what is being measured.
2. Publish the Placement-Code Crosswalk
The state should identify which administrative placement codes are included within each public-facing placement category.
If codes change, the historical crosswalk should remain available.
3. Make “Other” Expandable
If multiple statuses must be combined for the primary visualization, users should be able to expand “Other” and see its component categories.
4. Identify Numerators and Denominators
Percentage measures should clearly state both.
5. Distinguish Children From Events
When a child can appear more than once in a measure, the dashboard should explicitly say so.
6. Identify Reporting Periods
Point-in-time, monthly, annual, calendar-year, and fiscal-year measures should be unmistakably labeled.
7. Publish Methodology Notes
If data is excluded, suppressed, undergoing validation, or affected by a system change, the dashboard should explain what happened and what impact it has on interpretation.
8. Provide Downloadable Underlying Aggregate Data
Charts are useful for quickly understanding trends.
Downloadable tables are necessary for analysis.
Providing CSV or spreadsheet exports would allow researchers, journalists, legislators, advocates, and members of the public to independently evaluate the information.
9. Maintain Version History
When definitions, calculations, or historical figures change, those changes should be documented.
Public data should not silently become different data.
This Is Not About Catching Someone in a Wrong Number
There is an important distinction between identifying weaknesses in a reporting system and accusing the people operating that system of dishonesty.
Child welfare data is complicated.
Administrative systems are complicated.
Definitions change. Codes change. Data systems change. Reporting requirements change. Human beings enter information differently.
That is precisely why transparent methodology matters.
A strong accountability system should not depend on trusting that every number is correct.
It should make the numbers checkable.
When discrepancies appear, reproducible data allows policymakers and the agency to determine whether the cause is a reporting-period difference, a definition change, a coding issue, a calculation error, or something else entirely.
That protects the public.
It also protects the agency.
The Question Behind the Question
The most important lesson from Missouri’s child welfare dashboard may not be any individual discrepancy.
It is the difficulty of answering a much more fundamental question:
Can an informed member of the public independently determine what these numbers mean and reproduce the state’s reported results?
If the answer is no, the work of transparency is not finished.
Missouri has already taken an important step by making more child welfare information publicly available.
The next step is making that information genuinely auditable.
Because public accountability requires more than publishing numbers.
It requires giving people enough information to understand where those numbers came from, what they represent, and whether the conclusions drawn from them can be independently verified.
Data becomes accountability when the public can ask a question—and the system gives them the tools to verify the answer.
Leave a Reply