Setting Reliability Targets With Contaminated Data

The data problems do not excuse you from setting a target; they change how you set it.

1. Segment the trend rather than averaging it. Report rural interstate and urban freight separately. A single statewide reliability number would conceal a 7% decline behind a 4% gain, and the 180-day MPO process will surface that anyway.
2. Restore a usable baseline. For the 2021 vendor change, either re-process overlapping months under both methodologies to create a bridge, or exclude the discontinuity and state which months are non-comparable. Do not splice series silently.
3. Treat 2020-2021 as a structural break, not a data point. Use 2019 and the most recent 12 months as endpoints rather than a continuous trend line through the pandemic.
4. Set the target with an explicit basis - e.g., "return urban freight reliability to the 2019 level by the target year" - and state the assumptions (traffic recovery, planned projects, no new incidents).
5. Include a documented data-quality caveat in the target report.

A target is a commitment about direction and magnitude. Choosing a defensible basis and disclosing its limits is professionally sound; choosing the number the data will most likely deliver is not.

The MPO's more aggressive targets and different horizon create a real conflict that must be resolved on the record, not by hoping nobody notices.

Approach:

• Understand the difference first. The MPO may be using a different network extent, a different reliability measure, or a modeling-based projection while the state uses observed data. A 2-point gap in targets may reflect methodology rather than ambition.
• Where the difference is methodological, ask the MPO to re-run its target using the statewide measure on the MPO's network. Often the targets converge once measured the same way.
• Where the difference is genuine ambition, that is the MPO's prerogative. The state should not pressure a partner to lower a target; it should ensure both are reported and explainable.
• Commit to a joint reporting arrangement. Publish both sets of targets and both sets of results in the same document so the public sees the comparison.
• Use the 180-day window for a technical exchange, not a negotiation. Give the MPO the methodology and the data bridge before it acts.

A state that arrives at day 170 with an unexplained divergence has created a public credibility problem. A state that arrives at day 60 with a joint technical memo has not.

Three practices matter:

Report against the baseline, not just the target. Show the actual trajectory, the target line, and the annual values, including the pandemic years marked as non-comparable. A reader should be able to see whether the target was met and why.

Report the corridor-level distribution. The urban freight corridors driving the 7% decline should be named, with their own values, so the number connects to a decision. A statewide reliability index that moves 4% is a fact; three named corridors declining 7% is a program.

Guard against both sandbagging and aspiration. A target that was obviously met in advance teaches decision-makers to ignore targets; a target with no identified actions is decoration. Each target should be paired with the specific projects and strategies expected to deliver it, and the linkage reviewed each cycle.

Finally, review the measures themselves. If the probe methodology changes again, or if traffic patterns shift structurally, the right move is to document the change and re-baseline openly - not to keep reporting a series that no longer measures the same thing. The credibility of performance-based planning rests on whether the numbers are comparable year to year and whether anyone acts when they are not.