Prepare for the CIA PET by training claim calibration rather than content memorization. Learn to distinguish fact, inference, and assumption; grade source reliability separately from information credibility; and practice timed case analysis with a five-point rubric across a four-week sequence.
Separating the PET from the Polygraph and Background Investigation
The PET is a test within a multi-stage hiring process. The polygraph interview and background investigation are separate security and medical evaluations that come later in the sequence, with entirely different preparation demands.
CIA's careers page describes the sequence as: submit a resume via MyLINK, receive an invitation for a specific position, complete screening, testing, and interviews, receive a Conditional Offer of Employment, submit paperwork such as the SF-86, then complete security and medical evaluations including a background investigation, a polygraph interview, and physical and psychological examinations. Knowing which stage you are in tells you which kind of preparation actually applies.
The practical consequence is how you allocate study time. Reasoning and writing drills belong to the testing stage; document accuracy and full candor belong to the clearance process, and no amount of test practice substitutes for either. Keep advice for other screening processes separate, because blending them wastes the limited preparation window you control.
One eligibility point the same page settles: dual U.S. citizens are eligible, but you must be physically in the United States or its territories when you submit your resume via MyLINK, be at least 18, be willing to move to the Washington, DC area, and be able to complete the evaluations listed above.
| Stage | What it involves | PET-style preparation applies? |
|---|---|---|
| MyLINK resume submission | Initial application; must be filed while physically in the U.S. or its territories | No |
| Screening, testing, and interviews | Testing occurs here; this is where the PET sits | Yes |
| Conditional Offer of Employment | COE paperwork, including the SF-86 | No |
| Security and medical evaluations | Background investigation, polygraph interview, physical and psychological examinations | No |
Fact, Inference, and Assumption: Three Labels Your Prose Must Earn
A fact traces to a stated source, an inference follows from facts without being stated anywhere, and an assumption is taken as given without evidence. Conflating the three produces overclaimed writing that a single contrary detail can collapse.
Apply one test to every sentence: can you point to the line that supports it? 'The report states the port closed Monday' is a fact. 'The closure suggests a supply disruption' is an inference, because no source says disruption. 'Disruptions usually precede policy announcements' is an assumption, a general belief doing evidentiary work it has not earned.
Under time pressure, mark each claim in the margin with F, I, or A while drafting, then revise. Convert assumptions into either sourced facts or explicit conditionals such as 'if the closure holds through Friday.' The revision takes seconds and turns a fragile paragraph into one where each claim's strength is visible and defensible.
This distinction is the core difficulty of case-analysis work: a coherent story built on an unstated assumption reads exactly as convincingly as one built on evidence. Only the deliberate labeling habit reveals the difference before someone else does.
| Statement type | What it asserts | Safer wording | Red-flag phrasing |
|---|---|---|---|
| Fact | Something a stated source records or reports | 'The cable states…' | 'Everyone knows…' |
| Inference | A conclusion that follows from stated facts | 'is consistent with', 'suggests' | 'proves' |
| Assumption | Something taken as given without evidence | 'If X holds, then…' | Unstated 'obviously' |
Scenario One: A Coherent Story Built on a Weak Source
Worked scenario: a harbor shutdown where two unverified posts blame a fuel shortage. The plausible mistake is writing 'a fuel shortage has shut down the fleet' as established fact.
The evidence set contains three items: a fishing fleet has not left harbor for three days; two blog posts attribute the inactivity to a fuel shortage; a government spokesperson denies any disruption. A plausible first draft declares that a fuel shortage has shut down the fleet. The problem: the only attribution comes from unverified posts, an official source contradicts it, and nothing in the set rules out weather, a regulatory hold, or a labor issue. The draft has converted one hypothesis into ground truth.
A better sentence reads: 'The fleet's inactivity is consistent with the fuel shortage reported in two unverified posts; the official denial and unexamined alternatives keep confidence low.' This claim survives new evidence — a reopened harbor, a confirmed delivery record — without collapsing. It also names what would change your mind, which is the mark of a calibrated assessment rather than a bet. Overclaiming matters because your conclusion is only as durable as its weakest load-bearing claim.
Source Reliability and Information Credibility: Grade Them Separately
Source reliability asks whether a source is positioned and honest enough to know; information credibility asks whether this particular item checks out. A reliable source can pass along a false claim, so the two judgments must never merge.
Think of two independent axes. Reliability concerns the channel: does this source typically have access to the subject, and does it have a record of accuracy or of distortion? Credibility concerns the item: is it corroborated, internally consistent, and plausible against what else you know? A seasoned diplomat relaying an unfounded rumor is a reliable source passing along low-credibility information.
In practice, sort the evidence set by these two axes before deciding how much weight each item carries. A low-credibility item from a high-reliability source still needs corroboration before it anchors a conclusion. Working the axes separately also stops you from dismissing information simply because the messenger is unofficial — the two blog posts in the harbor scenario were unreliable in reach but their claim deserved a credibility check on its own terms, not a dismissal.
Scenario Two: Choosing Between Alternative Explanations Under Time Pressure
Worked scenario: three border-negotiation officials are removed in one reshuffle. The mistake is anchoring on the first coherent explanation; the better answer enumerates alternatives and names evidence that would discriminate among them.
A candidate notices the timing — the removals follow a contentious negotiation round — and concludes the officials were purged over a policy disagreement. That is one story, but at least two others fit the same facts: an undisclosed corruption inquiry, or a scheduled career rotation. The first draft never asks what evidence would separate them, so the answer rests entirely on how plausible one narrative feels.
A stronger response takes two sentences. First: 'Current evidence, limited to timing, modestly favors a policy disagreement, but confidence is low.' Second, one line of discriminating evidence per alternative: announced audits would favor the corruption explanation, while replacements drawn from the same career track would favor routine rotation. Enumerating alternatives costs seconds even under time pressure, and it prevents the anchoring that makes a single contrary fact fatal to an otherwise reasonable analysis.
A Twenty-Minute Writing Drill with a Five-Point Rubric
Take a 150-word news paragraph, extract its facts, and write a 120-word assessment in twenty minutes. Score five checks, each worth zero to two points, and track your pattern across at least four drills.
Score five checks: (1) every factual claim is traceable to the source paragraph; (2) inferences are flagged with hedged verbs such as 'suggests' or 'is consistent with'; (3) at least one alternative explanation appears; (4) no unsupported quantities or absolutes, such as 'most' or 'widespread', appear without basis; (5) one sentence names evidence that would change your conclusion. Ten points is the maximum.
Expected observations: early attempts summarize heavily and score lowest on the alternative-explanation and discriminating-evidence checks. By the fourth drill, the summary should shrink to a sentence and the total should stabilize at seven or above. These scores are learning milestones for the drill itself, not predictions of any outcome on the test. If a partner is available, have them sort your draft's claims into fact, inference, and assumption independently; disagreements pinpoint exactly where your prose overclaims.
- Check 1 — Traceability: every factual claim points to a line in the source paragraph.
- Check 2 — Hedged inference: conclusions use 'suggests', 'is consistent with', not 'proves'.
- Check 3 — Alternatives: at least one competing explanation is stated, not implied.
- Check 4 — No unsupported quantifiers or absolutes.
- Check 5 — A stated line of evidence that would change your conclusion.
A Four-Week Sequence and Concrete Readiness Checks
Spend weeks one and two compressing dense articles, week three sorting claims into fact, inference, and assumption, and week four running full timed drills. Close by testing yourself against four readiness checks drawn from those exercises.
Weeks one and two: read one dense article daily and compress it into three sentences — what happened, what is claimed, and what remains uncertain. Week three: build or find ten-item evidence lists and sort every item into fact, inference, or assumption, checking against a partner where possible. Week four: run the twenty-minute drill plus a professional-standards review, since objectivity rather than advocacy is the posture the Directorate of Analysis's own mission statement emphasizes for all-source analysis.
Readiness checks: you can explain the difference between source reliability and information credibility without notes; you can sort a ten-item list with strong agreement against a partner; you can produce a 120-word assessment meeting at least four of the five rubric checks within twenty minutes; and you know that administrative specifics — scheduling, logistics, current requirements — are confirmed on the issuer's own careers page rather than inferred from secondary sources. Treat each check as a study milestone, not a passing prediction.
References and further reading
Use these references to explore the concepts and check the latest information from the relevant organizations.