## What this test is actually measuring This is a **Situational Judgment Test (SJT)** — the format used in Google's Hiring Assessment and similar pre-interview screens. There is no code, no behavioral story, and no interviewer. You read a workplace scenario and pick which action you are **MOST likely** and **LEAST likely** to take. The scoring rewards a consistent, well-calibrated professional persona, and two principles win almost every item: 1. **Be "extreme" on the dimensions that matter.** On ethics, confidentiality, security, and respect, take the clearly principled stance every time. There is no partial credit for being pragmatically lax with a policy. 2. **Be consistent across items.** The test deliberately rephrases the same underlying value in different scenarios (and, in the Likert section, with near-identical statements). Answering "compliant" in one item and "convenient" in another is a large red flag. Pick a persona — the analyst who never bends a confidentiality rule, who addresses disrespect directly, who escalates through proper channels — and hold it across the whole assessment. A reliable decision rule for the MOST/LEAST format: - **MOST likely** = the option that *contains* the problem without creating a new violation: comply with policy, route through the approved channel, manage expectations, intervene directly and respectfully. - **LEAST likely** = the option that *creates a new violation* or harm: moving sensitive data into personal/unauthorized hands, blaming the victim, taking the convenient shortcut around a stated rule. When two options both look bad, the **worse** one is usually the one that (a) discloses sensitive data to an *additional person*, or (b) causes *direct interpersonal harm* — both are hard to undo and create an audit trail. --- ## Part 1 — Sensitive employee data, deadline, broken work laptop **The binding constraint:** policy says sensitive data may *only* be accessed on company equipment. The deadline is real, but it does not override a confidentiality control. Three of the four options breach that control; only one does not. | Option | What it does | Verdict | |---|---|---| | A. Email data to personal laptop | Transfers sensitive data to non-company equipment | Policy breach | | B. Send to a coworker who lacks access | Breaches least-privilege *and* discloses to an unauthorized person | Worst | | C. Tell the leader you'll miss the deadline | Slow, but the only choice that breaks no rule | Compliant | | D. Strip "confidential parts," then use personal laptop | Still a breach: de-identification can be reversible, you're self-adjudicating what's confidential, and policy typically bans any derived transfer | Policy breach | **MOST likely: C.** It is the only option that doesn't immediately violate the confidentiality control. In a real answer you would pair it with action — contact IT / on-call support to fix or replace the work laptop, and propose a safe, slightly-later timeline to the leader — but among the four given, C is the principled choice. **LEAST likely: B.** Sending sensitive employee data to someone not authorized to see it is the most severe failure here: it expands access (violating least-privilege), discloses to a third party, and is exactly the event a security/compliance audit flags. A and D are also breaches, but B is the one that pulls another, unauthorized person into the violation, which is harder to contain and undo. --- ## Part 2 — Inappropriate comment, respectful-workplace policy **The binding constraint:** every employee is accountable for a safe, respectful environment, and a coworker has told you the comment ("you don't look like someone who works in this industry") made them uncomfortable. The right response addresses the *behavior at its source* while supporting the person who was harmed — and never asks the harmed person to change. | Option | What it does | Verdict | |---|---|---| | A. Suggest the coworker raise it themselves | Supportive, but shifts the burden onto the person who was harmed | Weaker | | B. Talk to the commenter about why it was inappropriate | Direct, timely, corrects behavior at the source | Best | | C. Send a general team email about respect | Indirect, performative, doesn't correct the specific behavior | Weaker | | D. Advise the harmed coworker how to "fit in" | Victim-blaming; reinforces the bias the comment expressed | Worst | **MOST likely: B.** You witnessed it, so you can address it directly and proportionately — a private, non-accusatory conversation with the new analyst about why the remark landed badly. This corrects the behavior and sets expectations without escalating prematurely or exposing the affected coworker. (If the behavior recurred or were severe, you'd escalate to a manager/HR — but the first, proportionate move is the direct conversation.) **LEAST likely: D.** Telling the person who was made uncomfortable that *they* should change to "fit in" blames the victim, validates a biased comment, and directly contradicts the respectful-workplace standard. It is harmful, not merely ineffective, which makes it clearly the worst option. --- ## Part 3 — Confidential report, coworker has only a personal laptop **The binding constraint:** the report must not be shared on non-company equipment, and your coworker only has a personal laptop. The question pits *urgency* against a confidentiality control. As in Part 1, the control wins — but here the four options differ mainly in *how* the information leaves your hands. | Option | What it does | Verdict | |---|---|---| | A. Read the relevant section to the coworker | Conveys information verbally — no file, no copy, nothing rendered on the personal laptop | Best available | | B. Tell the stakeholder to wait until next week | Safe and rule-preserving, but ignores "urgent"; over-delays | Strong runner-up | | C. Email the report to a personal email account | Creates a persistent, unsecured copy off company systems | Worst | | D. Video call + screen-share the section | The coworker *views* confidential content rendered on a personal laptop | High-risk | **MOST likely: A — the genuine close call.** A is the narrowest way to stay responsive: it gets your coworker the context without sending a file or rendering the document on non-company equipment. The judgment call is whether "shared on non-company equipment" covers *verbal* disclosure at all — A conveys the same facts as D, just without a file or an on-screen copy. A strict reader could argue that any disclosure of the report's contents to a person on a personal device is off-limits and pick **B** (wait, then respond through compliant means) as the safer MOST. That is defensible; treat **A vs. B** as the real decision, with A edging it because it honors *both* the "company-equipment" wording and the "urgent" requirement, while B sacrifices urgency. In practice you'd first confirm the coworker is authorized to hear it and that verbal disclosure is permitted; if a colleague with company equipment is reachable, routing the answer through them beats all four — but that option isn't offered. **LEAST likely: C.** Emailing the report to a personal account creates an uncontrolled, persistent copy outside company systems — the most concrete and least-reversible breach on the list. D is also problematic (the content is still rendered on a personal device), but C is worse because it produces a lasting artifact you can't take back, while D's exposure ends when the call ends. --- ## Quick-reference answers | Scenario | MOST likely | LEAST likely | |---|---|---| | 1 — Sensitive data / personal device | **C** (accept the delay, escalate to IT) | **B** (send to an unauthorized coworker) | | 2 — Inappropriate comment | **B** (address the commenter directly) | **D** (tell the affected coworker to "fit in") | | 3 — Confidential report | **A** (read the section verbally) | **C** (email it to a personal account) | --- ## How to generalize this to any SJT item - **Classify the scenario** by the value it tests — Ethics, Confidentiality/Security, Respect/Inclusion, Collaboration, Decision-making — and answer from a fixed, principled persona for that category. - **A real deadline never licenses a real breach.** If one option keeps the rule and the rest break it, the compliant one is almost always MOST. - **Rank the "bad" options.** Disclosing sensitive data to an extra person and causing direct interpersonal harm are the worst, because they are hard to undo and create records. - **Stay consistent across the whole test.** The same value reappears, reworded; pick your stance once and don't waver. Internal contradiction is widely reported to be penalized more heavily than any single "imperfect" answer. --- ## Addressing the follow-up questions **1. If the coworker were the only person who could answer Part 3, and verbal disclosure were explicitly prohibited.** Then both A and D are off the table and you can no longer satisfy *both* urgency and the policy. The honest move is to be transparent with the leader rather than manufacture a workaround: tell them you can't safely retrieve the answer until the report can be accessed on company equipment, give a concrete time the author/access returns, and offer any *non-confidential* context you can share in the meantime. Manufacturing an exception under pressure is exactly the failure mode the test probes — a clearly-communicated, slightly-late compliant answer beats a fast non-compliant one. **2. Keeping answers consistent across a reworded assessment.** Decide your persona *once*, up front, in each value category (ethics, confidentiality, respect, collaboration, decision-making) and apply it mechanically. When a later item reads like an earlier one with the labels swapped, answer the *value*, not the surface wording. For Likert statements, "I'd bend an ethical rule if it helped the company" should get the same strong-disagree posture as "I always follow ethical standards" gets strong-agree — the contradiction between those two is the trap. Reading carefully for negations and qualifiers ("never," "only if," "even when") prevents accidentally flipping your own stance. **3. The better real-world action outside the four options, per scenario.** Part 1: call IT/on-call support to repair or swap the work laptop, or request a short, manager-approved extension — neither is offered because the test wants to see whether you'll *break* a rule when the easy fixes are removed. Part 2: depending on severity, loop in a manager or HR after (or instead of) the direct conversation — omitted because the test is checking your *first instinct*, not the full escalation chain. Part 3: route the leader's question through any colleague who has company equipment, or wait for secure access — omitted for the same reason as Part 1: the scenario is engineered to strip away the safe path and see what you do. **4. The tie-breaker when two options both clearly break a rule.** Pick the one with the **larger, less-reversible blast radius** as LEAST likely. Concretely, in order of severity: (a) discloses sensitive data to an *additional unauthorized person* (Part 1, option B); (b) creates a *persistent off-system artifact* you can't recall (Part 3, option C); (c) causes *direct interpersonal harm* (Part 2, option D). Transferring data to your own device or screen-sharing is bad but more contained — the exposure is narrower and ends sooner — so those rank as breaches but not the *worst* breach.