How to get hired as a policy analyst in 2026-27

Several different jobs share the title, and the employer decides which one you applied for

The phrase policy analyst covers at least five jobs that have almost nothing in common day to day. They pay differently, they are hired differently, they reward different skills, and the resume that wins one is often ignored by another. Before you write a word of application material, work out which one you are aiming at, because that single decision determines whether your time is better spent learning Stata or learning a legislature's committee calendar.

In a government agency, a policy analyst is usually implementing. You are the person who turns an enacted statute into something an agency can actually do: writing the guidance document, drafting the rule and its preamble, assembling the regulatory impact analysis, answering the congressional inquiry, running the clearance process that sends a document through six offices and a general counsel. The work is heavily textual, deadline-bound, and constrained by process you did not design. Federal titles for this work are often not policy analyst at all: Management and Program Analyst, Program Analyst, Social Science Analyst, Policy Advisor, Economist.

In a think tank or a research firm, a policy analyst is producing evidence. At the Urban Institute, Brookings, RAND, Mathematica, Abt, Westat, RTI, NORC, MITRE, the Pew Charitable Trusts or a university policy center, the entry role is usually titled research assistant or research analyst, and it is quantitative in a way the title policy analyst does not signal. You will spend real weeks cleaning data, writing code, checking somebody else's regression output, and building the exhibits for a report with your name third on it. The senior title, policy analyst or senior researcher, often requires a PhD, and at RAND in particular the title Policy Analyst sits in a research career ladder rather than an administrative one.

In an advocacy organization, a policy analyst is arguing. You write the comment letter on a proposed rule, the fact sheet a legislator reads in the elevator, the testimony, the state-by-state table that makes a national argument local. The organization has a position and you are expected to advance it honestly, which is a different discipline from being neutral: your numbers must survive hostile checking precisely because somebody hostile will check them.

In a legislative office, at either the federal or state level, the job is speed and triage. Congressional staff titles run staff assistant, legislative correspondent, legislative assistant, legislative director, and a legislative assistant covering health or tax is doing policy analysis under a clock measured in hours. State legislative service agencies are the quieter and often better version of this: the California Legislative Analyst's Office, the Texas Legislative Budget Board, the Colorado Legislative Council, Florida's Office of Economic and Demographic Research and the fiscal and research offices in most other statehouses hire nonpartisan analysts to write fiscal notes and bill analyses, and they hire on written exercises rather than on connections. Executive budget offices (a governor's budget division, Washington's Office of Financial Management, a city budget office) are the executive branch equivalent and hire the same way.

In a consulting firm or a federal contractor, a policy analyst is billable. Deloitte Government and Public Services, Booz Allen, Guidehouse, ICF, Accenture Federal and dozens of smaller firms staff the analysis that agencies are not resourced to do themselves. The hiring is faster than the government's, the pay is usually higher, the work can be excellent or can be slide production, and the question to ask in the interview is which contract you would sit on and how many option years remain on it.

Two practical consequences follow. First, apply to at least two of these categories at once, because their clocks differ by an order of magnitude and a search built only on federal applications can leave you unemployed for a quarter waiting on a certificate. Second, name the category in your own materials. An opening line that says you want to do regulatory analysis at a health agency beats one that says you are passionate about public policy, because the first is a claim a hiring manager can evaluate.

• Federal agency: implementation and process. Titles include Management and Program Analyst (GS-0343), Program Analyst, Social Science Analyst (GS-0101), Economist (GS-0110), Policy Advisor. Hired through USAJOBS, scored by HR, slow.
• Nonpartisan legislative support: GAO, the Congressional Budget Office, the Congressional Research Service, state legislative fiscal and research offices. Hired on writing and demonstrated neutrality, often with a written exercise. GAO and CBO both run their own assessments rather than relying on a resume alone.
• Think tank and research firm: evidence production. Entry titles are research assistant and research analyst. Quantitative tests are normal. A PhD is common above the entry rung.
• Advocacy and philanthropy: argument and coalition work. Comment letters, testimony, state fact sheets, legislative tracking. Among the fastest hiring of the lot, with the strongest mission-fit screen.
• Government affairs inside a company or trade association: monitoring, position papers, regulatory comment, and translating a legislature for executives. Usually the best paid of these tracks, and it values political judgment and industry fluency over econometrics.
• Contractor and consultancy: billable analysis for an agency client. Faster hiring, contract-dependent stability, sometimes a clearance requirement.
• Local government: a city or county budget, performance or policy office. Smallest teams, widest scope, the fastest route to owning something whole. Most city and county postings sit on GovernmentJobs.com rather than on a national board.
• International and multilateral: the World Bank, the IMF, the OECD, the UN system and the development consultancies around them. A separate system with its own young professional programs, language requirements and long timelines, usually requiring a master's and often a specific regional or sectoral background.

Credentials: no license, a de facto master's, and the federal rules that are the real gate

Policy analysis is not a licensed profession anywhere in the United States. There is no board, no registration, no protected title, and no exam you must pass before someone may pay you to analyze policy. That is worth saying plainly because advice written about other professions leaks into this one, and candidates waste money on certificates no hiring manager has heard of.

What exists instead is a strong hiring convention. For an analyst title at a federal agency above the entry grades, at a national think tank, at a research firm, or at a well-funded advocacy organization, a master's degree is the normal expectation. The Master of Public Policy is the canonical one and is usually the most quantitative of the public sector master's degrees. The MPA sits closer to management and is the better fit for local government, nonprofit leadership and program administration. A disciplinary master's is often a stronger signal than either for a specialized job: economics for anything with a price or a tax in it, statistics or data science for evaluation work, public health for health policy, social work for human services, planning for land use and transportation.

A master's is not required, and treating it as required is itself a mistake. Three routes in without one are well worn. The congressional ladder: a staff assistant or legislative correspondent position, two or three years, then a legislative assistant portfolio, then a move downtown to an association or a firm that wants a person who knows a committee. The state route: a fiscal analyst or program evaluator position in a legislative service agency or an executive budget office, which hires on a written test and promotes on competence. The data route: a data or research analyst job anywhere that has administrative data, then a lateral move into the policy shop that consumes your tables.

There is also a set of federal on-ramps that exist specifically for people without federal experience, and they are easy to miss because they are announced differently. Pathways internships hire current students; the Recent Graduates Program hires people within a limited window after completing a degree (OPM has long set that window at two years, extended for veterans, so confirm the current eligibility text before you rely on it). Both can convert to a permanent position. Agency use of them varies by year and by budget, so check whether the agency you want is actually running them rather than assuming.

If you do pursue the master's, two details matter more than rank. First, does the program put you in front of a real client? A capstone or practicum that produces a deliverable for an actual agency gives you the sole-authored or clearly attributed writing sample this field hires on, and a transcript does not. Second, is the quantitative sequence real? Compare syllabi, not brochures: a program whose required statistics course ends before regression leaves you unable to pass a research firm's data exercise.

Funding deserves a blunt paragraph. Policy analysis pays less than law, consulting or finance for a comparable amount of schooling, and a debt-financed MPP at full price is a genuinely risky trade. Chase funded offers. Ask every program for the share of students receiving aid and the median award, ask about assistantships that come with tuition remission, and compare a funded offer at a good state school against an unfunded offer at a famous one with clear eyes. Public Service Loan Forgiveness has been the standard answer to this problem for public sector workers; its rules and its administration have been contested and amended repeatedly, so verify the current terms with your servicer and the Department of Education rather than with a blog post or with this page.

Certificates are mostly noise, with a few exceptions that are specific rather than general. For analysts who will touch confidential federal statistical data, the training and the data use agreements that come with restricted data access are real and worth naming on a resume. For anyone doing budget work, the Government Finance Officers Association and the National Association of State Budget Officers publish the vocabulary your interviewers use, and their conferences are where state budget people actually meet. In evaluation, visible participation in the American Evaluation Association does more than any paid certificate. A generic online certificate in data analytics will not move a policy hiring decision, although the skills it teaches will, if you can show the output.

One more credential-shaped item: a security clearance. Some agency and contractor policy jobs require one, and an active clearance is a genuine hiring advantage because it removes months from the employer's timeline. You cannot sponsor yourself. If a posting requires an active clearance and you have none, it is usually not worth your application unless the announcement says the employer will sponsor.

• MPP: the canonical degree, strongest quantitative core, two years, best fit for analyst roles in agencies, think tanks and advocacy.
• MPA: management and administration emphasis, better for local government and nonprofit operations than for evaluation work.
• Disciplinary master's (economics, statistics, public health, social work, planning): often the stronger signal for a specialized policy desk.
• PhD: expected on the senior researcher track at RAND, Brookings, the Urban Institute and most university policy centers; usually unnecessary and occasionally a mild negative for an implementation job in an agency.
• JD: valuable for regulatory and legislative drafting work, and not a substitute for the quantitative half of the job.
• No master's: enter through a legislative office, a state fiscal or evaluation office, a local budget office, a Pathways or Recent Graduates appointment, or a data job adjacent to policy, and expect roughly two to four years to the analyst title.
• Fellowships and structured on-ramps worth researching: the AAAS Science and Technology Policy Fellowships (advanced degree required), congressional and state capital fellowships such as California's Capital Fellows and the Illinois Dunn Fellows, the Emerson National Hunger Fellows, ORISE appointments, and think tank research assistant programs.
• The Presidential Management Fellows program was for decades the main federal on-ramp for master's graduates and has since been the target of executive action directing its termination. Confirm with OPM whether it is operating before you build a plan around it.

Different hiring machines, different clocks, and the one that will waste your year

These employers do not run the same process. Running one search against all of them as though they did is the most common structural error in a policy job hunt. Here is what each one actually does.

Federal agency hiring through USAJOBS is a rating process disguised as an application. An announcement goes up with a closing date and sometimes a cap on applications. You answer an occupational questionnaire in which you rate yourself on each competency, usually with the top option phrased as having performed the task as a regular part of a job and having trained others in it. A human resources specialist then verifies that your resume supports the ratings you claimed and that you meet the specialized experience statement for the grade, which is normally one year of relevant experience at the next lower grade level. If the questionnaire ratings and the resume do not match, you are rated ineligible and no policy person ever sees you. If you clear, you land on a certificate of eligibles and the hiring manager selects from it. Then come interviews, a tentative offer, a suitability and background investigation, and a start date. Two to nine months end to end is normal, with veterans' preference, direct hire authorities and funding lapses all capable of changing the arithmetic. Apply, then keep applying to other things.

The nonpartisan legislative support agencies run their own assessments and are worth separating out. GAO hires analysts into a pay band structure and uses a structured process that typically includes a writing assessment and a competency-based panel; the work is engagement-based, team-based, and famously good training. The Congressional Budget Office and the Congressional Research Service hire on demonstrated analytical writing and demonstrated nonpartisanship, and both will read what you have published. If you intend to work in any of these, be careful what you put your name on publicly: a sharp partisan opinion piece is an asset in advocacy and a liability at CBO, CRS, GAO and most state legislative service agencies, where your entire value is that both parties trust you.

Think tanks and research firms move faster and they test. A recruiter screen, then a take-home or timed exercise, then a panel, then a director conversation. The exercise is the heart of it and comes in two flavors: a data task (here is a messy extract, produce these three estimates and tell us what you would caveat) or a writing task (here is a study and a press release about it, write a two-page memo for a nontechnical principal). Three to eight weeks is typical. Firms with federal contracts (Mathematica, Abt, Westat, RTI, NORC, ICF) also hire in waves tied to contract awards, so a rejection in March and an offer in September from the same firm is unremarkable and worth re-applying for.

Advocacy organizations and philanthropies hire the way a small team hires: a hiring manager who will be your boss, a work sample, two conversations, references, offer. Two to six weeks. The screen that eliminates most candidates is not skill but fit with the organization's theory of change. Read their last three publications before the first call and be able to say what they are trying to accomplish and where you would contribute. Many of these organizations have moved to structured, scored interviews and to publishing salary ranges, partly because of pay-transparency laws and partly because small teams have learned that unstructured interviews hire people who sound like the interviewer.

Legislative offices hire on referral and availability. For Congress, the practical channels are the House and Senate vacancy and placement services, the Tom Manatos and Brad Traverse job lists, committee and leadership networks, and alumni of your school's Washington program. Applications are frequently a resume, a cover letter, two writing samples and references in a single PDF, emailed to an address that gets hundreds of them. Turnaround can be a week. Pay at the bottom of the ladder used to be indefensible and has improved in the House through a staff salary floor and larger office allowances, but the top of the ladder is still well below private sector equivalents and the hours during session are long. State legislative offices run on the legislative calendar: hiring happens between sessions, and the fiscal office of a statehouse is one of the best training grounds in the field.

There is also a side door most candidates never try. Policy organizations consume more research help than they can budget for and routinely hire people who have already done visible work in their space: a public comment you filed on a docket under your own name, a state-by-state dataset you assembled and published, a short analysis you posted that a program officer quoted. This is not advice to work for free. It is advice to produce one public artifact in the specific policy area you want to be hired into, because it converts a cold application into a warm one. Filing a comment on a proposed rule at regulations.gov costs nothing, is permanently public, and is a legitimate line on a resume.

If you are not applying in the United States, the structure differs enough to matter. The United Kingdom hires policy people through the Civil Service recruitment system against the Success Profiles framework, with the Fast Stream as the graduate entry route and separate professions for the Government Economic Service and Government Social Research, both of which run technical assessments including a numerical test and a presentation. Canada hires into pools through GC Jobs with standardized tests and official language requirements, and its Policy Analyst Recruitment Program has historically been the main graduate on-ramp. The European institutions hire through EPSO competitions that take the better part of a year. Australia runs an annual Australian Public Service graduate program intake. In all four, the written assessment is a bigger share of the decision than in the United States, and the deadlines are annual rather than rolling, so a missed window costs twelve months.

• Federal (USAJOBS): questionnaire, HR rating, certificate of eligibles, panel, tentative offer, background investigation. Two to nine months. Resume length is an asset, not a liability.
• GAO, CBO, CRS and state legislative service agencies: their own written assessments, competency panels, and a real check on your public partisanship.
• Think tank and research firm: recruiter screen, timed data or writing exercise, researcher panel, director. Three to eight weeks.
• Advocacy and philanthropy: hiring manager, work sample, two conversations, references. Two to six weeks, heavy mission-fit screen.
• Legislative office: referral-driven, single-PDF applications, can close in a week, hires between sessions at state level.
• Contractor: fast, contract-dependent, sometimes clearance-gated. Ask which contract and how many option years remain.
• Where the postings actually live: USAJOBS for federal; Tom Manatos and Brad Traverse for the Hill and downtown; your school's and NASPAA's career boards; Idealist and Work for Good for nonprofits; individual statehouse and state agency HR portals plus NCSL and NASBO listings for state work; GovernmentJobs.com for most city and county roles; and the career pages of Mathematica, Abt, Westat, RTI, NORC and ICF directly.
• Run federal and non-federal tracks simultaneously. The federal track rewards patience and the others reward availability, and you cannot wait on one while the others close.

The writing sample: what to send, how to build it, and what gets you cut

This is the part of the application that decides policy analyst hiring, and it is the part candidates get wrong most often. Nearly every serious employer in this field asks for a writing sample, and most of them are using it to answer one narrow question: can this person take something complicated and give a busy decision maker what they need in two pages without being either vague or wrong.

Send a decision memo or an issue brief. Two to five pages, single-authored, written in the last two years, and about a policy question rather than about a literature. If you only have academic work, write something new; it takes a weekend and it is the single best weekend you can spend on this search. Do not send a thesis chapter, a literature review, a research proposal, a journalism clip, or a twenty-page seminar paper with the recommendation on page fourteen. A reviewer reads the first paragraph and the headings, and if your conclusion is not in the first paragraph they conclude you do not know the genre.

The skeleton below is the standard shape of a policy memo and no reviewer will think less of you for following it exactly. What they will notice is whether the options are real. The most common weakness in a sample is three options where one is obviously correct, one is a straw man and one is do nothing. Real options have different losers. If nobody is worse off under your recommendation, you have not found a policy question, you have found a press release.

Every number in the sample must be traceable. Put the source in the sentence or in a footnote with enough detail to re-find it: the dataset and vintage, the table number, the series ID, the year. A Bureau of Labor Statistics series ID rather than according to the BLS. ACS 2023 five-year estimates, table B17001, rather than Census data. You will be asked about one of these numbers in the interview, and the answer that gets you hired is the one that includes where it came from, what the denominator is, and what you would not use it for.

Clearance and confidentiality are non-negotiable and candidates do get eliminated here. If a sample came from a government job, a client engagement or an internship, ask permission, remove anything pre-decisional, embargoed, client-identifying or personally identifying, and say on the first page what you redacted and why. Never send a document marked as a draft deliberative product, never send unpublished client data, and never send anything derived from confidential federal statistical data or from protected health or education records. An employer who sees you hand over somebody else's confidential material learns something true about you.

Label your contribution when work is co-authored. One line at the top: you wrote sections two and four, built the cost model, and did not write the legal analysis. Reviewers assume the worst about unlabeled co-authored work, and they are often right to.

Match the sample to the employer. A nonpartisan office wants a sample that presents a disputed question fairly and does not reveal which side you are on. An advocacy organization wants to see you make an argument with evidence and handle the counterargument honestly. An agency wants to see you write within a process: a rule preamble paragraph, a guidance document, a response to a comment, a clean one-page summary for a principal. A research firm wants to see you describe a method and its limits. Keeping three samples in a folder and choosing per application takes an hour of setup and visibly outperforms one general-purpose sample.

If nobody has yet paid you to write policy, the portfolio items below are all available to anyone with a library card and a laptop. Pick the policy area you want to be hired into and produce two of them.

• The memo skeleton: one-paragraph bottom line with the recommendation. The decision and who makes it. Background in no more than a third of a page. Two or three genuine options with cost, who gains, who loses, implementation risk and legal authority. Recommendation with the reason it beats the runner-up. What would change your mind. Next steps with owners and dates.
• Length discipline: if the memo is for a principal, two pages. Put the detail in an appendix and the method in a footnote. A reviewer who sees four tight pages where a peer sent eighteen rambling ones has already made a judgment.
• Portfolio item: a public comment on a proposed rule, filed under your own name at regulations.gov. Pick a docket in your field, read the proposal and the regulatory impact analysis, find one defensible flaw or one unaddressed consequence, and write three pages. Permanently public, verifiable, and it teaches you how rulemaking actually reads.
• Portfolio item: a fiscal note. Take a bill from your own state legislature that has no fiscal note yet, estimate its cost with public data, document the assumptions in a table, and write the two-page note. This is close to the exact exercise state fiscal offices use to hire.
• Portfolio item: a fifty-state or multi-city comparison table on a narrow question, built from primary sources (statutes, agency rules, published budgets), with a methods note and a date. Tedious, genuinely useful, and the kind of thing other people start citing.
• Portfolio item: a two-page translation of a recent peer-reviewed study for a nontechnical principal, including what the study does not show. This is one of the most common take-home exercises in the field, so having already done one is practice and proof at once.
• Portfolio item: a replicable analysis published with its code, a README and a data citation. One clean repository beats a long list of coursework.
• What to never send: a thesis chapter, a group project without attribution, anything confidential, anything with a fabricated or unverifiable citation, anything over ten pages, anything whose conclusion is not visible in the first thirty seconds.

The quantitative bar: what you must actually be able to do, with the data named

Candidates either overestimate or underestimate this, rarely get it right, and the honest answer depends on the employer. For most government and advocacy analyst jobs, the quantitative bar is lower than applicants fear and the Excel bar is higher than they expect. For research firms, think tanks and evaluation offices, the bar is a genuine applied statistics bar and it is tested.

Start with Excel, because most policy numbers in the world live in a spreadsheet and most analysts who get stuck are stuck there. The working standard is: build a cost model where every assumption is in a labeled input cell and nothing is hard-coded inside a formula; pivot a hundred thousand rows without the file dying; write INDEX and MATCH or XLOOKUP rather than nested IFs; build a three-scenario projection and a one-way sensitivity table; chart something so the point is legible without a caption. If you can also read somebody else's inherited model and find the error, you will be the most useful person in several offices.

Then one language, used properly on one real dataset. R, Stata or Python, and which one depends on where you are going: Stata remains the common language in economics-adjacent policy shops and at several research firms, R dominates in evaluation and public health, Python wins where the work touches engineering, text or scale. SAS persists in some federal statistical agencies and in older state systems. Pick one, use it on something real, and be able to talk about your own code without notes. Knowing three languages badly is worth less than knowing one well, and interviewers can tell the difference in two questions.

SQL deserves its own sentence. The moment your work touches administrative data (Medicaid claims, unemployment insurance wage records, student records, licensing systems, 311 calls), someone is going to hand you database access rather than a CSV, and the analysts who can write a join and a group-by get the interesting assignments. It takes a weekend to learn enough and it compounds for a career.

The methods you need are mostly reading methods, not producing methods. Unless you are in an evaluation shop, you will spend far more time judging whether a study supports a claim than running a model. The literacy that matters: what randomization buys you and what it does not; why a difference-in-differences design needs a credible comparison group and what a pre-trend plot is for; what a regression discontinuity and an instrumental variable estimate are identifying and for whom; the difference between statistical significance and policy relevance; why a precisely estimated tiny effect is often a finding of no importance; what a confidence interval covers; and the practical difference between an intent-to-treat and a treatment-on-the-treated estimate when a program has low take-up. Being able to say that a study is well identified but the effect is too small to justify the appropriation is a sentence that marks you as a policy analyst rather than a student.

Weighted survey data is where honest candidates get caught. If you are using ACS, CPS, NHANES, NHIS, SIPP, NSDUH or any complex survey, you must use the person or household weights, and you must use the replicate weights or the design variables to get standard errors that are not nonsense. Reporting an unweighted mean from a complex survey in an interview exercise is a correctable mistake; not knowing the question exists is not. Know how to say what your margin of error is and why a five-year ACS estimate for a small county is not a one-year estimate.

Cost estimation is the most transferable quantitative skill in policy work and the least taught. It is the one thing every employer in this list eventually needs: how many people are eligible, what share will take it up, what it costs per person, over what period, against what baseline, with what second-order effects, and what happens if take-up is double your assumption. Build one of these for a real proposal, document the assumptions, and you will have the artifact that makes a budget office interview easy.

Finally, reproducibility is now a hiring criterion rather than a virtue. A script that runs from raw downloaded data to final table with no manual steps, a README that says where the data came from and when it was downloaded, and a folder structure somebody else can follow. Research firms ask for this explicitly. Government offices increasingly ask for it after being burned by an analysis whose author left and whose numbers nobody can rebuild.

• Excel, to the standard of a labeled, auditable cost model with scenarios and sensitivity, not just to the standard of a chart.
• One of R, Stata or Python, used end to end on a real dataset you can discuss unprompted. Stata for economics-adjacent shops, R for evaluation and public health, Python where text or scale is involved.
• SQL: joins, group-by, window functions if you can. Essential the moment administrative data appears.
• Named data sources worth having actually used: American Community Survey and Current Population Survey microdata via IPUMS; BLS OES, CES, LAUS and JOLTS; BEA regional accounts; the Census Building Permits Survey and Business Formation Statistics; CMS Medicaid and Medicare public files and T-MSIS where you have access; NCES and EDFacts; FBI NIBRS; HMDA; IRS Statistics of Income; USAspending and SAM for federal awards; the Federal Register and regulations.gov for rulemaking; Congress.gov and GovInfo for legislative history; your own state's open data portal and legislative fiscal office archive.
• Microsimulation and scoring tools worth knowing by name: NBER TAXSIM, the Urban Institute's TRIM3, the Policy Simulation Library and Tax-Calculator, PolicyEngine, and your state's own revenue estimating model documentation where it is public.
• Visualization and delivery: one plotting workflow you can repeat (ggplot2, matplotlib or Excel done carefully), plus the discipline of putting the finding in the chart title rather than in a caption nobody reads.
• Geospatial, if your field is land use, transportation, environment or service siting: QGIS or ArcGIS Pro, a shapefile joined to tabular data, and the knowledge that Census geographies change between vintages.
• Reproducibility: version control, a README with data provenance and retrieval dates, a script that runs clean from the top, and no step that exists only in your head.

The resume, and why the federal resume is a different document entirely

A policy resume has one job: to let a reader see, in about fifteen seconds, what you have produced and what decision it informed. Everything on the page that does not do that is taking up space that could have held a deliverable.

So write bullets around artifacts and consequences. Not supported policy research on housing affordability, but wrote the two-page issue brief and ten-state cost table used by the housing coalition in testimony to the Senate committee, and the estimate held up under opposing testimony. Not assisted with data analysis, but cleaned and merged five years of state unemployment insurance wage records in Stata to produce the earnings outcomes in the agency's annual workforce report, and documented the method so the next analyst could rerun it. Name the document, the audience, the data, the tool and the outcome. Policy work generates artifacts constantly and candidates leave almost all of them off the page.

What reliably gets ignored: passionate about public policy, strong written and verbal communication skills, a list of coursework, a GPA more than two years after graduation, generic software lists including Microsoft Office, student organization leadership once you have a real job, and the word stakeholder used four times. What gets read: the publications or products list, the named datasets, the tools with the thing you did in them, the legislative or regulatory process you have actually been inside, language skills if the job touches a community that needs them, and any evidence your work was used by somebody with authority.

Put a short publications and products section on the resume even if nothing is peer reviewed. Include the public comment you filed, the fiscal note, the issue brief, the dashboard, the testimony you drafted, with dates and links. This section is why a candidate with thinner titles beats one with better titles.

The federal resume is a separate document and treating it as the same one is an expensive mistake. For USAJOBS: four to six pages is normal and often necessary, every position needs month and year start and end dates, hours per week, your salary or grade, and your supervisor's name and whether they may be contacted. Your bullets should mirror the language of the specialized experience statement in the announcement, because a human resources specialist with no policy background is looking for exactly that language and cannot infer that your phrasing means the same thing. If the announcement says experience analyzing program data to make recommendations to senior leadership, then those words, honestly applied to something you actually did, need to appear in your resume. This is not keyword stuffing; it is answering the question that was asked in the vocabulary it was asked in.

The occupational questionnaire needs the same care and gets less of it. Rate yourself accurately and do not be falsely modest: if you have performed a task as a routine part of a job and have shown someone else how to do it, that is the top rating, and claiming less than the truth is how qualified people fail to be referred. But every rating has to be supported somewhere in the resume, because that is precisely what the HR specialist checks, and an unsupported top rating can get your whole application tossed. Read the questionnaire before you write the resume.

Two small mechanical points that cost people interviews. First, submit every document the announcement demands, including transcripts where there is a positive education requirement and the SF-50 if you are a current federal employee claiming status; a missing transcript is an automatic ineligibility for an Economist or Statistician series job. Second, note the closing time, not just the closing date, and the fact that some announcements close early once they hit an application cap. Apply in the first week.

• Lead with artifacts: memos, briefs, fiscal notes, comment letters, testimony, dashboards, reports, rules, guidance documents. Name them.
• Name the data and the tool together: analyzed ACS five-year microdata in R, not proficient in data analysis.
• Show the consequence: who read it, what decision it fed, whether it survived challenge.
• Include a products section with links even for work that is not peer reviewed, with dates.
• Cut: objective statements, coursework lists, GPA after two years, Microsoft Office, the phrase detail-oriented, and any claim with no artifact behind it.
• Federal version: four to six pages, month and year, hours per week, grade or salary, supervisor and contact permission, and the announcement's specialized experience vocabulary used honestly.
• Keep one master document with everything and cut down per application. Writing a fresh resume per job wastes the hours that should go into the writing sample.
• Cover letter: three paragraphs. The job you want and why this organization specifically. Two pieces of evidence you can do it, tied to their actual work. One sentence on what you would want to work on first. Nobody reads a fourth paragraph.

The interview: a case, a timed exercise, and a quiet test of whether you can be trusted

Policy interviews test four things, and only one of them is knowledge. They test whether you can structure an unfamiliar problem out loud, whether your numbers are yours, whether you can be brief, and whether you can be trusted with something contested. Prepare for all four separately.

The structure test usually arrives as a case. The governor wants to cut the childcare subsidy waitlist in half within two years, what do you need to know and what are the levers. Or: a legislator has asked whether our state should adopt the thing the state next door just adopted, you have three days, what do you do. They are not scoring the answer, they are scoring whether you ask what decision is being made and by when, whether you separate the question into eligibility, take-up, cost, capacity, legal authority and implementation, whether you say what data you would pull and from where, and whether you name what you would not be able to answer in the time available. Say your structure out loud before you start solving. End with what you would hand over and when.

The numbers test is an interrogation of your own writing sample, and candidates who did not build the thing they sent get caught here within two minutes. Expect: where did that figure come from, what is the denominator, what year is that, is that nominal or real and which deflator, why does your number differ from the one in the agency's report, what is the margin of error, who is excluded from that universe. Go into every interview having re-read your own sample and having re-derived your own key numbers the night before. If you do not know, say you do not know and say how you would find out, and then actually describe the lookup. That answer scores better than a confident wrong one, every time.

The brevity test is continuous and most candidates fail it without noticing. The practical drill: be able to answer every question at three lengths. Thirty seconds for a principal in a hallway, three minutes for a briefing, fifteen minutes with the method. Start with the thirty-second version and stop. In an interview for a legislative office, the question how would you explain this to a member with four minutes between votes is a literal description of the job.

The trust test is the one nobody prepares for. In nonpartisan offices it sounds like: tell us about a time your analysis did not support the answer your principal wanted. Tell us how you would handle being asked to leave a finding out. What would you do if your estimate was being cited for something it does not support. The answers that work are concrete, show that you escalated through legitimate channels, and do not cast you as either a pushover or a martyr. In advocacy organizations the same test runs in reverse: can you be committed to a position and still tell your own side that a number is weak. In both settings the hiring manager is deciding whether you are going to embarrass them.

Timed exercises are common enough that you should have practiced one. The patterns: a two-hour take-home writing a memo from two attached documents; an on-site writing exercise with a prompt, a packet and ninety minutes; a data exercise with a messy file and three required estimates; an editing exercise where you improve somebody else's bad paragraph; a presentation on a topic sent in advance, usually ten minutes with a hard stop, followed by questions designed to see whether you defend or revise. The two habits that matter in all of them: answer the question asked rather than the question you prefer, and leave a visible note of your assumptions and caveats. A clean, short answer with honest limits beats a longer one that overclaims.

Your own questions at the end are part of the assessment, so make them operational. What does the first ninety days look like. Who are the two or three people whose decisions my work would feed. What does a good product from this seat look like, and can you show me one. How long is the funding or the contract. How is the team's work used when the political direction changes. Those questions mark a candidate who intends to do the job rather than have the title.

A note on reference checks, which matter more in this field than in most. Policy is a small world and your reference may be called informally by someone who already knows them. Keep the relationships: the supervisor from your internship three years ago is likely to be asked about you by someone you will never know was asking.

• Likely case prompts: cost a proposal with incomplete data; decide whether to adopt another jurisdiction's program; respond to a legislator's one-line question with three days; prioritize five constituent problems with one analyst; evaluate whether a pilot should be scaled.
• Likely behavioral prompts: a time your analysis contradicted what your principal wanted; a time you were wrong in public; a deadline you had to triage; explaining something technical to someone hostile to it; working with a program office that did not want to give you data.
• Likely technical prompts: what does this coefficient mean; is this study causal and how do you know; what is wrong with this chart; which deflator and why; what do you do when the only available data is three years old.
• Have the thirty-second, three-minute and fifteen-minute version of every answer, and offer the shortest first.
• Re-derive your own writing sample's key numbers the night before. It is the highest-yield hour of interview preparation in this field.
• Ask about the first ninety days, who consumes your output, the funding horizon, and to see an example of good work from the seat.

Pay, and what the 2026-27 market actually looks like

Pay first, with sources rather than invented bands. Policy analyst work does not map to one Bureau of Labor Statistics occupation, which is why aggregator numbers for this title are unreliable. Look at OES code 19-3094 for political scientists, 13-1111 for management analysts, 13-2031 for budget analysts, 19-3011 for economists and 19-3099 for social scientists not otherwise classified, read the metro-level percentiles rather than the national median, and remember that the Washington metropolitan area distorts every national figure in this field. For federal jobs the number is knowable exactly: find the grade and step in the announcement, then apply the General Schedule table for the correct locality from OPM. For congressional staff, the House Statement of Disbursements and Senate public reports list actual individual salaries, which is the most transparent pay data in the field and almost nobody uses it. For state jobs, most states publish classified salary schedules and many publish individual employee compensation. For think tanks and larger nonprofits, the IRS Form 990 lists the highest-paid employees and tells you the shape of the ladder you are joining.

Then the market, honestly. The federal policy job market contracted sharply in 2025 through a combination of deferred resignation offers, reductions in force, agency reorganizations and the elimination or curtailment of some agencies and programs, and it did so after a long period in which federal hiring had been the reliable backstop for this profession. Two consequences followed into 2026 and 2027. Experienced former federal analysts are competing for think tank, state, local, contractor and nonprofit jobs that used to be filled by people at your level, which raises the bar on every posting. And the federal track itself has become less predictable in ways you should price in rather than ignore: hiring freezes and funding lapses can suspend a process you are already inside. Verify the current state of hiring at any agency you are targeting rather than trusting any written source including this one, and if the federal track is your preference, run it in parallel with something else rather than waiting on it.

There is also a structural change worth understanding before an interview rather than during one. Rules reviving a schedule of excepted service positions for roles of a confidential, policy-determining, policy-making or policy-advocating character have been promulgated and litigated, and the practical question for a candidate is whether a specific position you are offered carries the competitive service protections you assume it does. Ask what service and what schedule the position sits in, ask about the probationary period, and read the announcement's Conditions of Employment. That question is entirely fair to ask and a good hiring manager will answer it straight.

Where the demand is growing is as important as where it shrank. State and local government has absorbed both policy responsibility and policy hiring: when federal capacity contracts, state agencies, statehouses, big-city governments and the intergovernmental organizations around them (the National Conference of State Legislatures, the National Governors Association, the Council of State Governments, the National Association of State Budget Officers) carry more of the analytic load. Philanthropy and advocacy have moved money toward state capacity for the same reason. For a candidate willing to leave Washington, a statehouse fiscal office or a state agency policy shop is now one of the better places in this profession to get three years of real experience quickly.

The substantive areas showing up most often in 2026 and 2027 postings: artificial intelligence governance and digital regulation across state attorneys general offices, state legislatures, agencies and the advocacy and industry organizations around them; energy, permitting and electricity demand, where the data center build-out has turned a technical question into a political one; health policy, driven by Medicaid financing, pharmaceutical pricing and coverage changes; housing and land use, which has become a state-level legislative fight in a long list of states; immigration and benefit eligibility administration; defense, industrial policy and export controls; and the permanent, unglamorous, reliably hiring world of budget and program evaluation, which exists in every government in every year regardless of politics.

Three tactical conclusions. First, specialize enough to be the obvious candidate for something: a policy area plus a method plus a jurisdiction beats general competence, and the combination of health policy, Medicaid claims data and a specific state is a hireable identity in a way that interested in social policy is not. Second, geography still matters in this field more than in software: Washington, state capitals (Sacramento, Austin, Albany, Tallahassee, Springfield, Lansing, Olympia, Denver) and a handful of large cities hold most of the jobs, and remote policy work exists but is a minority of postings and usually goes to people with a track record. Third, expect your first job to be narrower than your ambition. The analyst who spends two years becoming the person who understands one funding formula completely is far more employable afterwards than the one who spent two years adjacent to everything.

A closing word on the thing that actually compounds in this career. It is not the degree and it is not the software. It is being the analyst whose numbers hold up. A reputation for being accurate, brief and honest about uncertainty follows you across a small professional world, and it is the reason a program officer emails you about a job that never gets posted. Build it from the first memo.

• Pay sources that publish real numbers: BLS OES (19-3094, 13-1111, 13-2031, 19-3011, 19-3099) at metro level; OPM General Schedule plus locality tables; House Statement of Disbursements and Senate reports; state classified salary schedules; IRS Form 990 for nonprofit and think tank ladders; posted ranges in pay-transparency jurisdictions.
• Ask in a federal process: what grade and step, what service and schedule, what locality, what is the promotion potential, and is this a term or a permanent appointment.
• Areas hiring most visibly in 2026-27: AI and digital governance, energy and permitting, health financing, housing and land use, benefits administration, industrial policy and export controls, and the budget and evaluation work that never stops.
• A hireable identity is policy area plus method plus jurisdiction. Pick all three.

What a screen is looking for

These are the terms that a resume screen, human or automated, is matching against for this role. Use the ones that are true of you, in the words the posting uses.

• Policy analyst
• Policy analysis
• Public policy
• Program analyst
• Management and Program Analyst
• GS-0343
• GS-0110
• GS-0101
• GS-0301
• GS-1530
• GS-1515
• Occupational series
• Specialized experience
• USAJOBS
• Occupational questionnaire
• Federal resume
• Certificate of eligibles
• Veterans preference
• Pathways Program
• Recent Graduates Program
• Presidential Management Fellows
• Security clearance
• Master of Public Policy (MPP)
• Master of Public Administration (MPA)
• NASPAA
• Capstone client project
• Policy memo
• Decision memo
• Issue brief
• Bottom line up front (BLUF)
• Options analysis
• Fiscal note
• Fiscal impact statement
• Legislative analysis
• Bill analysis
• Comment letter
• Public comment
• Notice of proposed rulemaking
• Regulatory impact analysis
• Rulemaking docket
• Regulations.gov
• Federal Register
• Congress.gov
• GovInfo
• Testimony
• Talking points
• Briefing memo
• One-pager
• White paper
• Policy brief
• Guidance document
• Administrative record
• Cost estimate
• Cost model
• Budget scoring
• Baseline projection
• Revenue estimate
• Appropriations
• Authorizing legislation
• Continuing resolution
• Program evaluation
• Performance measurement
• Logic model
• Theory of change
• Needs assessment
• Benefit-cost analysis
• Distributional analysis
• Microsimulation
• TAXSIM
• TRIM3
• PolicyEngine
• Tax-Calculator
• Stata
• R
• Python
• SQL
• SAS
• Excel
• Pivot tables
• XLOOKUP
• Sensitivity analysis
• Tableau
• Power BI
• ggplot2
• QGIS
• ArcGIS Pro
• Git
• Reproducible research
• Codebook
• American Community Survey (ACS)
• Current Population Survey (CPS)
• IPUMS
• Census microdata
• Public use microdata sample (PUMS)
• Survey weights
• Replicate weights
• Complex survey design
• Margin of error
• Bureau of Labor Statistics (BLS)
• OES
• LAUS
• JOLTS
• Bureau of Economic Analysis
• Administrative data
• Medicaid claims
• T-MSIS
• Unemployment insurance wage records
• HMDA
• IRS Statistics of Income
• NCES
• EDFacts
• NIBRS
• USAspending
• Open data portal
• Causal inference
• Randomized controlled trial
• Difference-in-differences
• Regression discontinuity
• Instrumental variables
• Interrupted time series
• Propensity score matching
• Intent-to-treat
• Statistical significance
• Confidence interval
• Econometrics
• Regression analysis
• Data cleaning
• Data documentation
• Literature review
• Evidence synthesis
• Systematic review
• Congressional Budget Office (CBO)
• Congressional Research Service (CRS)
• Government Accountability Office (GAO)
• Office of Management and Budget (OMB)
• Legislative Analyst's Office
• Legislative Budget Board
• State legislature
• Statehouse
• National Conference of State Legislatures (NCSL)
• Intergovernmental affairs
• Government affairs
• Legislative tracking
• Coalition building
• Stakeholder engagement
• Nonpartisan analysis
• Think tank
• Research assistant
• Research analyst
• Federal contractor
• Technical assistance
• AI governance
• AI policy
• NIST AI Risk Management Framework
• ISO/IEC 42001
• EU AI Act
• Algorithmic accountability
• Model card
• Bias audit
• Algorithmic impact assessment
• Structured data extraction
• Document review at scale
• Citation verification
• CIPSEA
• Restricted-use data
• FERPA
• HIPAA
• Personally identifiable information
• Pre-decisional material
• Deliberative process
• FOIA
• Records management
• Health policy
• Housing policy
• Education policy
• Energy policy
• Permitting reform
• Tax policy
• Labor policy
• Immigration policy
• Criminal justice policy
• Social safety net
• Technical writing
• Plain language
• Executive summary
• Data storytelling
• Public speaking

Mistakes that cost people this job

Questions people ask