Why Longitudinal O*NET Analysis Is Hard • onet2r

O*NET is one of the most useful public sources for occupation-level skills, tasks, abilities, work activities, and work contexts. It is also easy to misuse as a time series. The most important fact is simple: O*NET was not built as a longitudinal panel. The Web Services API serves the current release, while version-over-version analysis has to be reconstructed from the downloadable release archives (National Center for O*NET Development, n.d.b).

That reconstruction is why onet2r includes archive, bridge, and reconciliation helpers. They do not make the historical data automatically causal or comparable. They make the assumptions visible.

The 3 Problems Users Must Handle

1. The API Is Current-Release First

Most API workflows ask “what does this occupation look like now?” Longitudinal work asks “what changed between release A and release B?” Those are different questions. For panels, use archive releases rather than the current-release API.

panel <- onet_panel(
  "Abilities",
  versions = c("30.2", "30.3"),
  scale = "IM",
  archives = c(`30.2` = archive_302, `30.3` = archive_303),
  release_dates = c(`30.2` = "2026-02-01", `30.3` = "2026-05-01")
)

panel |>
  count(release_version, release_date, soc_vintage, domain) |>
  onet_kable()

release_version	release_date	soc_vintage	domain	n
30.2	2026-02-01	2019	Abilities	7
30.3	2026-05-01	2019	Abilities	7

The table above is produced by onet_panel(). In a live workflow, the same call can download archive ZIP files by version. In a package vignette, local archive paths keep the output reproducible.

2. Occupation Codes Change across Taxonomies

O*NET-SOC vintages do not line up one-to-one forever. Occupations split, merge, appear, and disappear. Hosseinioun et al. (2025) handle this by composing adjacent official O*NET-SOC crosswalks, which is the same strategy used by onet_crosswalk_bridge().

The synthetic fixture below spans the 2010-to-2019 O*NET-SOC seam. One 2010 software occupation maps to 2 2019 occupations, so the bridge is uncertain for within-occupation comparison.

cross_panel <- onet_panel(
  "Abilities",
  versions = c("24.3", "25.1"),
  scale = "IM",
  archives = c(`24.3` = archive_243, `25.1` = archive_251),
  release_dates = c(`24.3` = "2020-08-01", `25.1` = "2020-11-01")
)

bridge_2010_2019 <- tibble::tibble(
  from_vintage = "2010",
  to_vintage = "2019",
  from_onet_soc_code = c("15-1132.00", "15-1132.00", "29-1141.00"),
  to_onet_soc_code = c("15-1252.00", "15-1253.00", "29-1141.00"),
  map_type = c("split", "split", "one_to_one"),
  crosswalk_weight = c(0.5, 0.5, 1)
)

cross_changes <- onet_panel_reconcile(cross_panel, bridge_2010_2019)

cross_changes |>
  select(
    from_onet_soc_code,
    to_onet_soc_code,
    element_name,
    change_type,
    crosswalk_uncertain,
    transition_data,
    safely_comparable
  ) |>
  onet_kable()

from_onet_soc_code	to_onet_soc_code	element_name	change_type	crosswalk_uncertain	transition_data	safely_comparable
15-1132.00	15-1252.00	Oral Comprehension	transition_data	TRUE	TRUE	FALSE
15-1132.00	15-1253.00	Oral Comprehension	transition_data	TRUE	TRUE	FALSE
29-1141.00	29-1141.00	Oral Comprehension	real_update	FALSE	FALSE	TRUE
15-1132.00	15-1252.00	Problem Sensitivity	dropped	TRUE	FALSE	FALSE
15-1132.00	15-1253.00	Problem Sensitivity	dropped	TRUE	FALSE	FALSE

3. A Release-to-Release “Change” May Not Be a True Content Change

Many O*NET archive tables include a source date and domain source. These fields matter because a value can appear unchanged simply because the occupation was not re-surveyed, or a value can change without a source-date change because of a recode or recalculation. Handel (2016) is a useful entry point for the broader measurement and comparability cautions.

changes <- onet_panel_reconcile(
  panel,
  bridge = onet_crosswalk_bridge("2019", "2019")
)

changes |>
  distinct(value_changed, date_changed, change_type, safely_comparable) |>
  arrange(change_type) |>
  onet_kable()

value_changed	date_changed	change_type	safely_comparable
FALSE	FALSE	stale_carryforward	TRUE
TRUE	TRUE	real_update	TRUE
TRUE	TRUE	real_update	FALSE
FALSE	TRUE	resampled_stable	TRUE
TRUE	FALSE	recode_or_recalc_flag	FALSE

comparison_counts <- cross_changes |>
  mutate(comparability = if_else(safely_comparable, "Safe", "Not safe")) |>
  count(comparability, name = "rows")

ggplot2::ggplot(comparison_counts, ggplot2::aes(
  x = comparability,
  y = rows,
  fill = comparability
)) +
  ggplot2::geom_col(width = 0.6, show.legend = FALSE) +
  ggplot2::coord_flip() +
  ggplot2::scale_fill_manual(
    values = c("Safe" = onet2r_colors[["teal"]], "Not safe" = onet2r_colors[["rose"]])
  ) +
  ggplot2::labs(
    title = "How Many Rows Are Safe to Compare?",
    subtitle = "The count comes from the cross-vintage example, not a hand-built table.",
    x = NULL,
    y = "Rows"
  ) +
  onet2r_theme()

Horizontal bar chart showing safely comparable and not safely comparable rows in the cross-vintage example.

Those rows are not a hand-built truth table. They are the observed classifications from the example archive panels. The patterns are the minimum checks users should make before interpreting any cross-release difference.

How This Differs from the Standard Task Approach

The canonical task-framework literature often avoids O*NET’s version-over-version variation. Instead, it holds task scores fixed and gets time variation from changing occupational employment shares (Autor, Levy, and Murnane 2003; Autor and Dorn 2013; Autor 2013). That approach is standard because it is cleaner for many labor-market questions. onet2r enables a complementary route: directly comparing archived O*NET content after marking taxonomy and survey-timing problems.

Direct within-occupation change work is closer to what the longitudinal module does. Consoli et al. (2023) combine DOT and O*NET to study within-occupation task change over a long period. Chen et al. (2026) report a narrower O*NET version comparison as one stylized fact inside a paper whose main data source is online vacancies.

Official Resources to Check

The O*NET Database Releases Archive is the source for release ZIP files (National Center for O*NET Development, n.d.b).
The O*NET Data Dictionary documents table structure, date fields, source fields, and historical content changes (National Center for O*NET Development, n.d.a).
The O*NET-SOC taxonomy files provide the official adjacent crosswalks used by the bridge layer (National Center for O*NET Development, n.d.c).
BLS provides a methodological overview for mapping Employment Projections and O*NET data, including split, merge, and imputation issues that motivate future employment-weighted bridge options (U.S. Bureau of Labor Statistics 2021).

Adjacent Tools to Verify before Citing

Several tools map O*NET to other classification systems rather than across O*NET’s own versions. They may be useful, but they are intentionally not in inst/REFERENCES.bib until verified to citation standard:

Institute for Structural Research occupational task crosswalk files.
occupationcross, reported as an R package for occupational reclassification.
The Dais NOC-to-O*NET crosswalk for Canada.
European Commission ESCO-O*NET crosswalks.

References

Autor, David H. 2013. “The ‘Task Approach’ to Labor Markets: An Overview.” Journal for Labour Market Research 46 (3): 185–99. https://doi.org/10.1007/s12651-013-0128-z.

Autor, David H., and David Dorn. 2013. “The Growth of Low-Skill Service Jobs and the Polarization of the US Labor Market.” American Economic Review 103 (5): 1553–97. https://doi.org/10.1257/aer.103.5.1553.

Autor, David H., Frank Levy, and Richard J. Murnane. 2003. “The Skill Content of Recent Technological Change: An Empirical Exploration.” The Quarterly Journal of Economics 118 (4): 1279–1333. https://doi.org/10.1162/003355303322552801.

Chen, Hangyu, Yongming Sun, and Yiming Yuan. 2026. “Artificial Intelligence and Skills: Evidence from Contrastive Learning in Online Job Vacancies.” arXiv:2601.03558; SSRN 5924344. https://doi.org/10.2139/ssrn.5924344.

Consoli, Davide, Giovanni Marin, Francesco Rentocchini, and Francesco Vona. 2023. “Routinization, Within-Occupation Task Changes and Long-Run Employment Dynamics.” Research Policy 52 (1): 104658. https://doi.org/10.1016/j.respol.2022.104658.

Handel, Michael J. 2016. “The O*NET Content Model: Strengths and Limitations.” Journal for Labour Market Research 49 (2): 157–76. https://doi.org/10.1007/s12651-016-0199-8.

Hosseinioun, Moh, Frank Neffke, Letian Zhang, and Hyejin Youn. 2025. “Skill Dependencies Uncover Nested Human Capital.” Nature Human Behaviour 9 (4): 673–87. https://doi.org/10.1038/s41562-024-02093-2.

National Center for O*NET Development. n.d.a. “O*NET Data Dictionary.” O*NET Resource Center. https://www.onetcenter.org/dictionary/30.3/excel/.

———. n.d.b. “O*NET Database Releases Archive.” O*NET Resource Center. https://www.onetcenter.org/db_releases.html.

———. n.d.c. “O*NET-SOC Taxonomy and Crosswalk Files.” O*NET Resource Center. https://www.onetcenter.org/taxonomy.html.

U.S. Bureau of Labor Statistics. 2021. “Mapping Employment Projections and O*NET Data: A Methodological Overview.” Monthly Labor Review.