ADP Research · Sep 10, 2025
Tim Decker explains in ADP Research how it adjusts bias in employment estimates
Markdown · JSON · Original source
Tim Decker’s article for ADP Research was published September 10, 2025. It argues that bias is part of data, but estimates can remain useful when bias is measured and addressed. ADP’s payroll data come from businesses that choose to outsource payroll and human resources, which may not represent employment across an industry or geography. In manufacturing, ADP says employment growth among its clients exceeded overall growth reported by the Quarterly Census of Employment and Wages (QCEW), which it describes as more comprehensive but less timely. Decker explains that a simple adjustment does not fit the data well, then describes using a linear model divided into yearly sections. ADP says the approach produces a tight fit with QCEW data for manufacturing, while noting that the fit depends on the period used. The article concludes that bias is a natural consequence of limited resources and measurement, and that responsible methods can help estimate and remove it when a largely bias-free reference series is available. How well the adjustment performs against employment evidence beyond the QCEW remains unknown.
Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.
Original source
ADP Research · Article
Source details
Original source date: Sep 10, 2025 (2025-09-10).
Date basis: Date stated on the publisher’s page.
Added to Escape Meridian: 2026-11-12T05:08:26.471Z. First observed: 2026-10-30T02:55:52.107Z. Retrieved: 2026-10-30T09:50:08.685Z.
Referenced by
Recorded citations and public file attachments. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
Text as retrieved on October 30, 2026; originally published September 10, 2025. The article presents ADP Research’s account of its adjustment and manufacturing comparison; it does not establish performance against employment evidence beyond the QCEW. The article says that the linear fit depends on the start and end times of the series, so results from one period do not establish how the adjustment performs across other periods.
Open question
How well does the adjustment perform when compared with employment evidence beyond the QCEW?
Source-stated
ADP Research argues that bias is present in data, but estimates can remain useful when bias is estimated and addressed.
Bias is everywhere. Yet, in many cases, data can still provide valuable insights. So how to avoid throwing out the data with the bias? What’s needed is a systematic way to estimate and remove that bias.
Source-stated
ADP Research says its payroll data come from businesses that chose to outsource payroll and human resources, and may not represent employment across an industry or geography.
ADP data comes from businesses that have chosen to outsource payroll and human resources activities. As such, these employers might not perfectly represent overall employment for their industry or geography.
Source-stated
ADP Research says that, in its manufacturing comparison, employment growth among ADP clients exceeded overall market growth reported by the QCEW.
While the shapes of the QCEW and ADP measures are remarkably similar, the employment growth experienced by ADP clients far exceeds the overall market growth reported by the QCEW.
Source-stated
ADP Research says it uses a piecewise linear bias adjustment for the National Employment Report, with each section one year long.
For the National Employment Report, we use a piecewise linear bias adjustment, where each section is one year long.
Source-stated
ADP Research says its manufacturing adjustment yields a tight fit with QCEW data, while the article also notes that the linear fit depends on the time period used.
When it comes to manufacturing employment, our piecewise, weekly bias correction yields a tight fit between our employment report and QCEW data.
Source-stated
ADP Research concludes that responsible methods can estimate and remove bias when a largely bias-free reference series is available.
Appropriate and responsible methods exist to estimate and remove these biases. The linear model bias correction outlined here is one such responsible method, at least for cases where a largely bias-free reference series, such as the federal government’s QCEW, is available.
Source and contributor context
ADP Research · Publisher
The publisher is identified here. No individual author has been inferred.
Source media
Media retrieval status: unchecked.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
TIME
TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Editorial provenance
Version 1. Publication ID: 82d56bff-0680-32ff-c3d1-859ca4bacaa0.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-12T05:04:34.529Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model.
Observation: d6a38a67c0b1acb41cfeaf9c78c9e45396efa330eddec9e9b4d513583a4d3c4f. Source text hash: 0b1058f3a93d959a468ec0cc352250af1d86e8303ff99f347563f147da92a291.
Revision and source-check history
Version 1 · 2026-11-12T05:08:26.471Z
Tim Decker explains in ADP Research how it adjusts bias in employment estimates
Tim Decker’s article for ADP Research was published September 10, 2025. It argues that bias is part of data, but estimates can remain useful when bias is measured and addressed. ADP’s payroll data come from businesses that choose to outsource payroll and human resources, which may not represent employment across an industry or geography. In manufacturing, ADP says employment growth among its clients exceeded overall growth reported by the Quarterly Census of Employment and Wages (QCEW), which it describes as more comprehensive but less timely. Decker explains that a simple adjustment does not fit the data well, then describes using a linear model divided into yearly sections. ADP says the approach produces a tight fit with QCEW data for manufacturing, while noting that the fit depends on the period used. The article concludes that bias is a natural consequence of limited resources and measurement, and that responsible methods can help estimate and remove it when a largely bias-free reference series is available. How well the adjustment performs against employment evidence beyond the QCEW remains unknown.
Original source
ADP Research · Article
Source details
Original source date: Sep 10, 2025 (2025-09-10).
Date basis: Date stated on the publisher’s page.
Added to Escape Meridian: 2026-11-12T05:08:26.471Z. First observed: 2026-10-30T02:55:52.107Z. Retrieved: 2026-10-30T09:50:08.685Z.
Limits
Text as retrieved on October 30, 2026; originally published September 10, 2025. The article presents ADP Research’s account of its adjustment and manufacturing comparison; it does not establish performance against employment evidence beyond the QCEW. The article says that the linear fit depends on the start and end times of the series, so results from one period do not establish how the adjustment performs across other periods.
Open question
How well does the adjustment perform when compared with employment evidence beyond the QCEW?
Source-stated
ADP Research argues that bias is present in data, but estimates can remain useful when bias is estimated and addressed.
Bias is everywhere. Yet, in many cases, data can still provide valuable insights. So how to avoid throwing out the data with the bias? What’s needed is a systematic way to estimate and remove that bias.
Source-stated
ADP Research says its payroll data come from businesses that chose to outsource payroll and human resources, and may not represent employment across an industry or geography.
ADP data comes from businesses that have chosen to outsource payroll and human resources activities. As such, these employers might not perfectly represent overall employment for their industry or geography.
Source-stated
ADP Research says that, in its manufacturing comparison, employment growth among ADP clients exceeded overall market growth reported by the QCEW.
While the shapes of the QCEW and ADP measures are remarkably similar, the employment growth experienced by ADP clients far exceeds the overall market growth reported by the QCEW.
Source-stated
ADP Research says it uses a piecewise linear bias adjustment for the National Employment Report, with each section one year long.
For the National Employment Report, we use a piecewise linear bias adjustment, where each section is one year long.
Source-stated
ADP Research says its manufacturing adjustment yields a tight fit with QCEW data, while the article also notes that the linear fit depends on the time period used.
When it comes to manufacturing employment, our piecewise, weekly bias correction yields a tight fit between our employment report and QCEW data.
Source-stated
ADP Research concludes that responsible methods can estimate and remove bias when a largely bias-free reference series is available.
Appropriate and responsible methods exist to estimate and remove these biases. The linear model bias correction outlined here is one such responsible method, at least for cases where a largely bias-free reference series, such as the federal government’s QCEW, is available.
Source and contributor context
ADP Research · Publisher
The publisher is identified here. No individual author has been inferred.
Source media
Media retrieval status: unchecked.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
TIME
TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Editorial provenance
Version 1. Publication ID: 82d56bff-0680-32ff-c3d1-859ca4bacaa0.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-12T05:04:34.529Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model.
Observation: d6a38a67c0b1acb41cfeaf9c78c9e45396efa330eddec9e9b4d513583a4d3c4f. Source text hash: 0b1058f3a93d959a468ec0cc352250af1d86e8303ff99f347563f147da92a291.
uncertain · 2026-11-12T05:04:34.529Z · anthropic/claude-opus-5.5
Publication: 82d56bff-0680-32ff-c3d1-859ca4bacaa0
attribution — pass: The reading credits the article to Tim Decker, identified in the source as lead architect of the ADP National Employment Report, published by ADP Research. Its claims are framed as ADP Research's statements rather than as established facts. The text is a single explanatory article, not a listing of other pieces.
Tim Decker is the lead architect of the ADP National Employment Report.
summary — uncertain: The summary says ADP reports a tight fit 'while noting that the fit depends on the time period used.' This attaches the period-dependence to the final yearly piecewise approach. In the source, that dependence is a problem of a single linear fit, and the yearly sections are introduced to address it; only the last section's parameters still drift. Smallest repair: say a single linear fit depended on the chosen period, so ADP split it into yearly sections whose parameters stay fixed except for the latest year. Otherwise the summary is understandable to a general reader: it explains QCEW and why ADP's sample may be biased.
N-1 of them will have fixed start and end points; the bias parameter estimates will drift over time only for the last of them.
date — pass: The reading gives September 10, 2025, which matches originalPublished at day precision. Nothing is presented with hindsight or as happening after that date. The evidence is a publisher page, not an archive copy, so no archive disclosure is required. The retrieval date appears only in the limitations.
support — pass: Each claim's excerpt is an exact source passage that supports the claim. The question and limitation about performance beyond the QCEW are framed as unknowns, not as assertions. The limitation that one period's results do not establish performance in other periods follows from the source's note that the fit depends on start and end times. No unsupported benefits or costs are implied.
The linear fit between these two series will depend on the start and end times of the series.
media — pass: No media was selected, and no verified candidates were supplied. The absence of media is appropriate.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.
TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Explore TIMETIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Explore TIMETIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Explore TIMETIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Explore TIME