Establish Your Approach
Based on your study’s purpose, scope, and conceptual definitions, you can determine which journalism employment research approach and corresponding methodology is most appropriate. Several approaches are outlined below, with associated strengths and limitations. These include:
- Exploring Occupational and Labor Statistics
- Using Surveys of Journalists or News Organizations
- Conducting Desk Research and Manual Compilation
- Assessing Commercial or Platform-Based Data
- Exploring Content- and Byline-Based Approaches
This report also features a table comparing the strengths and weaknesses of each approach in Appendix B.
1. Exploring Occupational and Labor Statistics
A variety of federal labor datasets provide insights into the health of the journalism industry, yet they operate at varying frequencies and geographical resolutions. The primary challenge for researchers is navigating the trade-off between capturing a real-time industry temperature and conducting a deep dive into occupational specificity. The frequency of publication dictates the granularity and the ultimate purpose of these datasets; understanding these mechanics is essential for accurately measuring the news workforce.
The primary challenge for researchers is navigating the trade-off between capturing a real-time industry temperature and conducting a deep dive into occupational specificity.
High frequency datasets serve as the primary indicators of immediate economic shifts, such as mass layoffs or industry-wide contractions, but they often lack the resolution needed to distinguish specific newsroom roles. Lower frequency datasets are critical in helping us distinguish between all newspaper employees from those employed in specific journalistic roles. Table 1 summarizes the strengths and weaknesses of six federal data sets.
| DATASET | UNIT OF ANALYSIS | FREQUENCY | USE CASE | COVERAGE GAP |
|---|---|---|---|---|
| Current Employment Statistics (CES) | Establishment | Monthly | Tracking immediate newsroom headcounts. Helpful in assessing shocks/industry-wide layoffs | No occupational detail |
| Current Population Survey (CPS) | Individual | Monthly | Estimating national unemployment rates for self-identified journalists/editors. | Small/negligible sample for news |
| Quarterly Census of Employment and Wages (QCEW) | Job | Quarterly | Benchmarking total newsroom employment at different geographical levels | Misses freelance |
| Occupational Employment and Wage Statistics (OEWS) | Occupation | Annual | Wages and staffing ratios specifically for Reporters and Editors | 3-year rolling lag; misses freelance |
| American Community Survey (ACS) | Individual | Annual | Analyzing the demographics within local newsroom workforces via PUMS | High noise in small areas; limited news coverage given sample size limitations |
| Economic Census | Firm/Establ. | 5 Years | Mapping the total number of physical news establishments | Suppression in small markets |
Current Employment Statistics (CES)
www.bls.gov/ces
Published monthly, the CES is an establishment survey that tracks non-farm payroll employment, hours, and earnings. For the purposes of studying journalism employment, it provides a high-frequency count of the total workforce within specific industry classifications, such as Newspaper Publishers or Broadcasting. It is designed to reliably track monthly macroeconomic shifts across sectors rather than granular job roles. As a result, it serves as a measure of sectoral health rather than occupational density. That is, it does not distinguish between a reporter, a delivery person, or an accountant employed within the same industry, as it aggregates all payroll headcounts regardless of specific job functions.
Current Population Survey (CPS)
www.bls.gov/cps
This monthly household survey is the source of the national unemployment rate. It allows researchers to identify individuals who self-report their occupation as a “journalist” or “editor.” However, because journalists represent a small fraction of the total population, the sample size at the monthly level is often too small for reliable geographic analysis below the national or large-state level. Despite that limitation, the CPS reports “place of residence” of the workforce. Thus, if a reporter lives in and covers rural Washington, they will be counted as a resident professional within their rural county even if their place of work is Seattle.
Quarterly Census of Employment and Wages (QCEW)
www.bls.gov/cew
Derived from administrative Unemployment Insurance (UI) records, the QCEW is a near-census of U.S. jobs. Its greatest strength is its geographic resolution, providing data down to the county level. While it is the most robust tool for measuring local industry health, it remains limited by its focus on industry classification (NAICS). Thus, it can tell us if a local newspaper is shedding jobs, but not which types of roles are being eliminated.
However, most freelancers and independent contractors will be invisible in the QCEW. Since they are not “employees” in the legal sense, a newsroom does not report them on their quarterly UI filings. Furthermore, this is a “place of work” report. In the age of remote work, a journalist living in rural Washington might work for a paper headquartered in Seattle. The QCEW attributes that job to Seattle, masking the existence of local journalistic labor in a rural county.
The following figure shows the number of employees in newspaper publishing in the first quarter of 2024. You will notice that figures from QCEW are only slightly understated compared with the Economic Census (discussed later) which produces a similar count. This makes the Quarterly Census a very valuable source of high frequency data.
Occupational Employment and Wage Statistics (OEWS)
www.bls.gov/oes
This annual survey of employers is the primary tool for measuring specific roles, such as News Analysts, Reporters, and Journalists. It provides critical data on employment levels and wages at the national, state, and Metropolitan Statistical Area (MSA) levels. To generate its annual estimates, the OEWS pools data from six semi-annual panels collected over a three-year cycle. However, it faces significant geographic suppression. This means that in smaller or less competitive markets, the BLS may suppress data to protect the confidentiality of the few local news employers in that region.
The OEWS is aimed at providing a precise overview of employment and wage estimates for specific occupations. To do so, the BLS combines six panels of data collected over a three year period. Thus, it is not entirely suitable to study shocks to the economy, but excellent for structural analysis. Since it is also an establishment survey, coverage for self employed or freelancing workers is limited.
Furthermore, occupation codes follow the individual worker, rather than their employer. So, if a large corporation (let’s say, an insurance company) hires an individual for the purposes of brand journalism, they will still be counted as journalists based on their day to day tasks.
The figure below depicts the estimated number of individuals employed as journalists across states according to the May 2024 estimates of the OEWS.
American Community Survey (ACS)
www.census.gov/programs-surveys/acs.html
As an ongoing household survey, the ACS provides the most detailed socioeconomic look at the journalism workforce, including age, race, and education. Using Public Use Microdata Sample (PUMS) files, researchers can analyze journalists at the PUMA level (areas of ~100,000 people). While it offers high geographic flexibility, the data is subject to high Margins of Error (MOE) for niche occupations, making it noisy for researchers looking at specific rural areas. In some cases, depending on the goals of your study, you may need to default to the five year ACS to get a large enough N for a local enough analysis.
Economic Census
www.census.gov/programs-surveys/economic-census.html
Conducted every five years, this is the most comprehensive measure of American business. It provides the ground truth for the number of physical news establishments and total revenue. It is the most accurate tool for identifying long-term structural shifts.
Conducted in years ending in ‘2’ and ‘7’, the goal of the economic census is to provide comprehensive, industry-specific statistics on the number of establishments, revenue, and payroll to benchmark GDP and other economic indicators.
It offers the highest potential for geographic granularity, often drilling down to the county and place level. However, its utility is restricted by disclosure rules. If an industry like Newspaper Publishers is not sufficiently “publishable” in a specific region, the data is suppressed to protect firm confidentiality. In practice, this results in data being available for only about 364 out of 900+ CBSAs and 499 out of 3,100+ counties, creating significant blind spots when studying rural news ecosystems.
Furthermore, the primary Geographic Area Statistics only cover employer businesses (those with paid employees). This means that these tables potentially systematically understate journalistic activity in regions where production is heavily reliant on freelancers or independent contractors.
The figure below shows the number of people employed in the newspaper sector by state, according to the Economic Census of 2022.
Ultimately, federal labor statistics should be viewed as a floor rather than the ceiling for measuring the journalism workforce. Given that these datasets are tied to administrative tax records or Unemployment Insurance filings, they primarily capture the minimum count of formal, payroll-based staff jobs. They effectively ignore the fluid non-employer margin of the industry, often comprising independent contractors, freelancers, and solo-entrepreneurs who increasingly define the local news ecosystem.
In sum, although federal databases have several strengths, including a national scope, consistency over time, and legitimacy for trend analysis, they also have significant limitations.
In sum, although federal databases have several strengths, including a national scope, consistency over time, and legitimacy for trend analysis, they also have significant limitations. These large datasets include occupational categories that align poorly with newsroom roles, they have a limited ability to distinguish local work from non-local work and inadequately capture part-time, freelance, intern, or volunteer labor. Additionally, they suppress local journalism employment data and have weak geographic specificity, making national labor datasets one part of the equation when measuring journalism employment, but requiring additional measures to fully capture the entire picture.
2. Using Surveys of Journalists or News Organizations
While journalists have been directly surveyed for more than five decades,1 these surveys have traditionally focused more on journalistic roles, work conditions, and professional values rather than direct staffing estimates. Surveys such as the American Society of News Editors (ASNE) Newsroom Employment Census (later known as the Newsroom Employment Diversity Survey) operated as a benchmark mechanism for tracking the demographic makeup of newsroom staff and evolved to include digital-only outlets, newsroom leadership diversity, and gender. In later years, the survey was plagued by limited participation and low response rates, leading to a temporary pause in data collection2 before relaunching in 2025 under new ownership as the American Press Institute’s Media Inclusion and Impact Survey. Updates from this survey are anticipated to arrive by late 2026.
One ongoing national survey of nonprofit organizations is conducted by the Institute for Nonprofit News (INN). INN has been surveying its members since 2018 for information related to staff size, revenue, and audiences, among other metrics.3 INN’s index of business and editorial statistics allows its members to benchmark their progress and develop strategies for further development. Additionally, Project Oasis, a global research project focused on independent digital media and led by SembraMedia with the support of LION Publishers, the Google News Initiative, and others, has surveyed outlets across the globe, including outlets’ staffing numbers.
The local TV industry has also been surveyed over the years, including measures such as staff size, layoffs, and salaries, but these surveys have focused only on local broadcast organizations, limiting the overall journalism employment picture across mediums, and the unit of analysis has been at the newsroom level rather than the individual journalist level. Additionally, the sample frame and definition of a journalist can vary, making direct comparison across decades and survey waves difficult.
Scholars have begun conducting surveys of local news organizations in their respective states to try to ascertain staffing numbers across news organizations, including radio, TV, online-only, and newspaper. In Washington, scholars disseminated a survey to 353 news organizations who met the inclusion criteria for their database and specifically asked staffing questions, including how many full-time editorial staff, how many part-time editorial staff, and how many overall staff existed in each news organization. The survey was disseminated specifically to management level employees like editors, publishers, and owners who were in a position to provide up-to-date data on this metric. In Minnesota, scholars disseminated a survey to a subset of news organizations in the state and asked similar questions, but structured their survey so newspaper owners could answer the survey for up to five newspapers because of the high number of small local chains in the area. As such, scholars were able to differentiate how much time staff members worked part-time at each publication.
The strengths of state-level surveys include the ability to tailor definitions and adapt them to the local context.
The strengths of state-level surveys include the ability to tailor definitions and adapt them to the local context. Additionally, the state level surveys mentioned above occurred after the scholars had already mapped the local news ecosystem in their states and had a database of verified news organizations to send the survey to, allowing for a more robust and accurate response. Additionally, survey questions and survey logic can be tailored to capture part-time, freelance, intern, and volunteer labor; effort which is essential for keeping news organizations afloat and yet is masked or opaque in broader data collection efforts, including at the national level with BLS data. Additionally, state-level surveys can provide “ground truth” to other measurement approaches, thereby serving as an important triangulation tool.
Even with these strengths, state-level surveys have limitations. Like most surveys used in social science research, state-level surveys on journalism employment can suffer from low or uneven response rates among participants. Journalists may not respond to survey requests because of time constraints, competing priorities, or suspicions that such requests are phishing attempts. Additionally, surveys that ask questions related to annual operating budgets and staffing numbers may be sensitive for many newsrooms. These topics may bring up painful emotions related to layoff decisions and colleague departures and may feel like a criticism to managers who have had to oversee shrinking newsrooms.
Another challenge with survey research is that researchers may define their concepts differently (e.g., who is a journalist), use different units of measurement (newsroom vs. journalist), and/or have different aims or goals in their surveys, making replication and/or longitudinal studies difficult. Conducting surveys can be resource-intensive with high labor costs for researchers. Depending on the strengths and skillsets of the researchers involved, researchers may need to out-source certain components of the survey process, such as survey design, dissemination, follow-up, or analysis. Surveys also typically require an institutional review board (IRB) process by a university body, which can lengthen the overall time for a project.
3. Conducting Desk Research and Manual Compilation
In the absence of a comprehensive database, researchers might elect to conduct desk research to take stock of the number of journalists in a particular region. This approach generally involves bringing together multiple data sources, such as assembling a list of news outlets in the region and estimating staffing based on outlet websites, social media pages, public filings, union records, and/or association lists. For example, a researcher might start with a comprehensive list of news outlets in a particular region — newspapers, digital news sites, radio or TV stations, etc. — and then record the number of journalists employed by each outlet, based on publicly-available information on their website, LinkedIn page, or other sources.
Directories such as the News Media Yellow Book and Bacon’s Newspaper Directory have historically provided a useful starting place for such studies because they previously included contact information for reporters and editors working at newspapers. However, these directories are limited because they only contain information for journalism employees in newspapers, which constitute a shrinking segment of the media ecosystem. Neither the News Media Yellow Book or Bacon’s Newspaper Directory are being published in print, although older versions can be accessed for historical work. Additionally, both databases have transitioned to online subscription databases that can be prohibitively expensive. More recently, journalism support organizations like INN and LION have made their membership databases freely available, providing a generative starting place for studies involving desk research and manual compilation.
While they may be highly accurate and painstakingly precise, these studies are hard to scale, sustain, and update, which therefore limits their utility.
At its best, this approach to measuring journalism employment can be highly accurate when carefully executed, making it a particularly useful approach in situations when precision is required (e.g., policymaking). It is also highly adaptable to the local context and larger media ecosystem in a particular region, making it useful for place-based philanthropies. Additionally, desk research can be combined with other methods — including the aforementioned national labor statistics or survey data — to resolve any gaps and provide additional precision, when required. Unlike survey methods, though, desk research does not require researchers to recruit news organizations to participate, because it is largely based on publicly-available information, thus reducing the burden on time- and resource-strapped news organizations.
One downside of desk research and manual compilation is the fact it is extremely labor- and resource-intensive. Depending on the scope of the analysis, such studies will generally require multiple researchers and/or research assistants to achieve the level of precision and detail necessary to meaningfully assess local journalism employment. Additionally, they are highly dependent on the researcher’s judgment, making them difficult for other researchers to replicate (and thus measure change over time). Therefore, while they may be highly accurate and painstakingly precise, these studies are hard to scale, sustain, and update, which therefore limits their utility.
4. Assessing Commercial or Platform-Based Data
Commercial platforms such as Coresignal, Revelio, and Muck Rack offer alternative methods for measuring the journalism workforce by aggregating professional identities of professionals’ presence across digital media and platforms. Unlike administrative records, these sources identify individuals based on their self-reported roles and public-facing work. Publicly-available and/or purchased data from these sources and others can be used as a method to measure and assess local journalism employment.
Coresignal and Revelio
www.coresignal.com and www.reveliolabs.com
Data providers like Coresignal and Revelio aggregate structured labor market data from professional networking sites like LinkedIn and Glassdoor. They provide large-scale datasets focused on employment and organizational history.
Sources like these can potentially offer coverage across time on firm-level headcounts, employee seniority, and career trajectories. They may be effective for mapping personnel movement as well, tracking when individuals leave journalism for other sectors, such as public relations or corporate communications.
Muck Rack
www.muckrack.com
Muck Rack is a media database that identifies journalists by scraping news websites for bylines and monitoring social media activity. It maintains a directory of professionals that is updated through a combination of automated web crawls and manual verification.
Individual profiles include current outlet affiliations, historical byline archives, specific topical beats and social media engagement metrics. It identifies contributors based on active production, allowing the inclusion of freelancers, independent contractors, and remote reporters who may not appear in a specific newsroom’s local payroll data but are actively producing content for that outlet.
The Local Journalist Index (LJI) by Rebuild Local News is an example of effective and innovative usage of such platform data. Relying on extensive data extraction from the Muck Rack platform, LJI provides estimates of how many journalists remain at various geographic levels, including the county. Further, they provide an improvised statistic, the Local Journalist Equivalent, which estimates how many journalists cover counties even if they lack their own local media.
As the comparison between the raw counts of journalists from LJI and BLS data show (Fig. 4), platforms like Muck Rack provide more extensive coverage across states and counties because they are not constrained by federal disclosure-avoidance rules. Furthermore, because these platforms identify journalists through byline activity rather than payroll records, they capture independent and freelance labor that falls outside the BLS definition of a “wage and salary” employee. At the same time, platforms like Muck Rack may provide an overly optimistic view of the market by including content creators or bloggers who are not journalistic in nature and by not regularly removing journalists who have been laid off or have left the industry. As a result, platform data may be more reflective of the content economy rather than newsroom health.
To best facilitate a fair assessment of these various datasets as well as the journalistic labor market, one way to think about this is to consider federal figures as the floor of the analysis, and the platform data as the ceiling. Bounding expectations in this way can result in more reasonable estimates than either data alone can provide. Triangulating between these two bounds allows for a measurement of the market that accounts for both, the state of legacy newsrooms and the rise of decentralized content production.
While these sources provide a granular view of the workforce, they introduce specific biases and technical challenges that require careful consideration. For instance, these platforms rely on a digital footprint. As a result, they can over-represent journalists at national outlets or digital-native startups in urban centers. Rural reporters or those at legacy publications with limited web presence may be undercounted or omitted.
While these sources provide a granular view of the workforce, they introduce specific biases and technical challenges that require careful consideration.
Selection bias is more pertinent for data like CoreSignal or Revelio, which, while reliable when it comes to macro trends, still depend on individuals creating a LinkedIn, Glassdoor or Indeed profile. Those who do not create or update their profiles, will continue to remain invisible or miscounted.
Occupational categorization on these platforms often relies on user-selected tags. This creates a significant risk of occupational conflation, where content marketers, brand managers, and PR specialists are grouped with newsroom professionals due to overlapping keyword descriptions like “writer” or “content creator.”
The accuracy of the data depends on the frequency of platform scraping as well as user updates. Profiles often remain active long after an individual has left a specific role or the industry entirely, leading to potential overcounts of active labor in the field. In the same vein, utilizing platform data to aid over time analyses requires significant technical and computational resources.
5. Exploring Content- and Byline-Based Approaches
Another way to assess journalism employment is to use publication output as a proxy for journalism labor. Specifically, researchers can measure content volume, track bylines, or content similarity, to assess newsroom employment capacity and capabilities.
Researchers who measure content volume assume that when staffing cuts occur, a news organization’s original content volume will likely drop as well. Through empirical research in U.S. and French newsrooms, scholars have shown how content volume is credible as an explicit theory of newsroom production because it reveals changes in reporting effort within an outlet or market over time. That said, it still serves as a proxy measure for exact journalism employment numbers because it does not pinpoint whether decreased output is a result of fewer staff, platform shifts in the news organization, or changes in newsroom workflow.
Scholars have shown how content volume is credible as an explicit theory of newsroom production because it reveals changes in reporting effort within an outlet or market over time.
Another practice involves counting the number of journalists based on the public bylines available on a news organization’s website to estimate journalism labor. Scholars can code articles by employment basis (e.g., specialized staff, general staff, freelancers, etc.) to better identify the type of employment available within a news organization. A limitation to this approach is that authorship and byline conventions differ depending on the newsroom’s norms, with some news organizations choosing to not have bylines on their articles, creating external validity and replication challenges.
Recently, scholars developed a scalable and semi-automated approach to assess digital news content according to journalism quality standards, capturing both the volume and quality of content, including bylines of journalists. This method, when combined with other approaches, may be a successful strategy for measuring journalism employment.
Lastly, researchers can assess content similarity within a news organization to infer labor capacity. The underlying logic of this approach suggests that fewer original stories reflects fewer employed journalists and greater reliance on external copy, such as wire stories or even press releases.
The strengths of these content- and byline-based approaches include a lower burden on newsrooms because researchers can typically bypass newsroom involvement and instead scrape news organizations’ websites or manually download news stories to conduct content analysis. The limitations of these approaches include the fact that they are proxy measures for direct staff counts of journalists at news organizations, and if using a web crawler, typically require a certain level of technical expertise to develop and deploy it.
In sum, these five approaches to measuring journalism employment (occupational and labor statistics, surveys, desk research/manual compilation, commercial or platform data, and content and byline approaches) can be combined to compensate for gaps in any single approach (e.g., combining manual desk research to account for any gaps in a newsroom survey).
- John W. C. Johnstone, Edward J. Slawski, and William W. Bowman, The News People: A Sociological Portrait of American Journalists and Their Work (Urbana: University of Illinois Press, 1976); David H. Weaver, Lars Willnat, and G. Cleveland Wilhoit, “The American Journalist in the Digital Age: Another Look at U.S. News People,” Journalism & Mass Communication Quarterly 96, no. 1 (2019): 101–30, https://doi.org/10.1177/1077699018778242.
↩︎ - American Press Institute, “Survey History,” accessed May 4, 2026, https://americanpressinstitute.org/api-media-inclusion-impact-survey/survey-history/ ↩︎
- Institute for Nonprofit News, “The 2025 INN Index,” accessed May 4, 2026, https://inn.org/research/inn-index/2025-index/about-the-index/
↩︎
