The Anchored-Scale Panel Answer Teardown, One Example Rescored From 3 to 5

This is a teardown for anyone facing a structured panel interview where each answer gets a number against a published grade descriptor. That covers UK Civil Service behaviour interviews scored 1 to 7 and US federal structured interviews scored against proficiency levels. I carry one representative answer, a Delivering at Pace example for an SEO post, all the way from a first draft that three markers scored at 3 to a version they moderated to 5. That includes the rewrite that did not move the number at all. By the end you can mark your own example the way a panel member does, rewrite it past the pass mark, and walk in with a prepared answer for every likely probe.

How are civil service and federal structured interviews scored?

Each panel member scores each answer on their own against a fixed anchor scale for the advertised grade, then the panel settles on one number per behaviour. In UK Civil Service behaviour interviews that scale usually runs 1 to 7, with 4 as the usual pass mark. US federal panels typically use a proficiency scale such as OPM's five levels.

An anchor is the written description attached to each point on the scale. It is what a marker compares your answer against, so you are not scored on how impressive you sounded. The two scales line up like this:

Three things about this table matter when you are marking yourself:

• The UK wording for 3 and 4 is the gap you are closing. A 3 is "Moderate positive evidence but some negative evidence demonstrated". A 4 is "Adequate positive evidence and any negative evidence would not cause concern". Going from 3 to 4 means your answer has to stop giving the marker anything to worry about.
• The labels for 5 to 7 are unverified. They come from a coaching site. I could not retrieve GOV.UK's own definitions for those points, so check your candidate pack for the wording your panel will use.
• OPM levels step up by difficulty and guidance, not effort. The step from Intermediate to Advanced goes from "occasional guidance" to "little or no guidance" in a harder situation. The Commerce Department's template describes Expert as someone who acts as "a key resource and advises others".

UK strengths (questions about what you enjoy and do naturally) use a different scale: each strength is rated 1 to 4. In the US, OPM's guide tells agencies to "Decide on one proficiency level range for all" competencies, typically with 3 to 7 levels. So a federal panel may not use five points. Find out which scale you face before you start counting.

What the pass mark means

A 4 is the usual UK benchmark, but some campaigns raise it to 5. Pass marks are set per campaign, so your invitation letter or candidate pack is the authority. I could not find an official cross-government document stating a single pass mark.

Coaching sites also report that appointable candidates are ranked by total score, in merit order. That is not official guidance either. Still, it is the reason to aim above the pass mark instead of stopping once you reach it.

How long you have

You get one question per behaviour, and the time allowed varies by source:

Coaching sites recommend 2 to 3 minutes per answer, which conflicts with the figures above. The UK HR guide caps critical behaviours at four or fewer so the panel can "sufficiently test the criteria at interview stage". That cap exists to leave time for probing. So whatever time your invitation gives you, plan to use some of it answering follow-up questions.

The worked example: a Delivering at Pace answer for an SEO post

The worked example is a candidate applying for an SEO post that lists four behaviours. I follow only Delivering at Pace. The same method applies to each of the others.

Step 1, decode the advert. SEO maps to Level 3. Level 3 covers both "Higher Executive Officer (HEO) and Senior Executive Officer (SEO) or equivalent", and Level 4 covers Grade 7 and Grade 6. HEO and SEO therefore share one descriptor set. One practitioner puts the practical difference this way: "you'll need a more complex example for an SEO role than an HEO role". Keep that sentence in mind, because it is what keeps this example stuck at a 3 later on.

Step 2, build the descriptor grid. The candidate copies the Level 3 Delivering at Pace bullets from the Civil Service Careers Behaviours page, plus the Level 2 bullets below and the Level 4 bullets above. I have not reproduced GOV.UK's Level 3 wording here. In the walkthrough I use four short working labels of my own, P1 to P4, standing in for the real bullets. Your grid should hold the exact published text.

• P1: plans and prioritises the work against the deadline
• P2: spots and removes barriers to delivery
• P3: owns delivery decisions and accounts for them
• P4: keeps quality up while moving at pace

The level below gives the useful contrast. One published set of EO-level examples for this behaviour includes "regularly review the success of activities in the team to identify barriers to progress or challenging objectives". If your strongest action reads like that sentence, you are describing EO-level work.

Step 3, the first draft. The candidate wrote this, condensed here from roughly four spoken minutes:

"In my last role our team had a backlog of correspondence cases and a hard deadline from the private office. We sat down and agreed a plan. We split the cases by type, and we held a short meeting each week to review progress and see what was holding us up. When colleagues went off sick, we redistributed their work. I kept a tracker of where each case was. We cleared the backlog before the deadline, the director thanked the team in an all-staff email, the private office said it was the best turnaround they had seen, and the approach was later picked up by other teams."

Before any marking, count what is there:

• Action clauses: 5 (agreed a plan, split the cases, held weekly reviews, redistributed work, kept a tracker)
• First-person action clauses: 1 of 5 (kept a tracker)
• Outcome clauses: 4, and together they run longer than the actions

It is a decent story, and those counts already predict the score.

If your history is scattered across old CVs and appraisal notes, choosing examples is the slow part of this step. If you applied through Refolk, the resume it tailored to this posting and its fit score show which advertised requirements your history only thinly supports. Those behaviours need the most work on their examples.

How three markers scored the first draft

Marked independently, the draft scored 3, 3 and 4, and the panel discussion settled it at 3. The only first-person action mapped to no Level 3 indicator, and the strongest team action matched the level below.

Here is the sequence the markers followed. OPM documents it explicitly. UK guidance says only that "Variances in scoring amongst panel members can be discussed".

The three marks and the reasons written against them:

• Marker A, 3. Credited P1 and P2, but only to the team. The one "I" action, keeping a tracker, is administration rather than delivery. Moderate positive evidence, with the plural voice as the negative evidence.
• Marker B, 3. Read the weekly review as the closest thing to evidence, then noticed it matches the EO-level example almost word for word. Marked it as level-below evidence. P3 (owning decisions) is absent.
• Marker C, 4. Assumed the candidate ran the weekly review and the reallocation, since they kept the tracker. Gave the benefit of the doubt.

In discussion, Marker C was asked which stated sentence showed the candidate making the reallocation decision. There was none. The 4 rested on an inference, and the panel settled at 3. This is the mechanism to plan around: in a consensus discussion, a generous mark has to be backed by something you actually said. OPM states that "Final scores or ratings should be based on the consensus of the panel". Whether UK panels average and round instead is not publicly established. Either way, one marker's 4 did not carry the answer.

Rewriting the answer: what moved the score and what did not

Three rewrites followed. The first added more outcome and left the score at 3. The second turned "we" into "I" and reached a borderline 4. The third lifted the scope to SEO level and stated the decisions taken without sign-off, which earned a moderated 5.

Rewrite 1, the wrong turn: more outcome

The candidate's instinct was that the result was not impressive enough. They added that the work fed into the department's annual reporting and that a deputy director nominated the team for an internal award. Marker C actually dropped to 3, because the longer result made the thinness of the actions more obvious.

The score depends on actions matched to indicators, not on the result. A bigger outcome attached to the same five team actions is still five team actions.

Rewrite 2: the first person

The candidate went back over what they had actually done and found that most of the "we" actions were theirs:

"I proposed splitting the cases by type so the most experienced caseworkers took the complex ones. I ran a short weekly review where I asked each person what was blocking them, and I cleared the blockers I could that day. When colleagues went off sick, I reallocated their cases by complexity, not by volume."

Now 4 of the 5 action clauses are first person, and they map to P1, P2 and P4. Markers A and B moved to 4. Marker C stayed at 3, and gave a reason worth taking seriously: every reallocation had been agreed with the candidate's line manager first. In OPM terms that is "occasional guidance" in a "difficult" situation, which is Intermediate. In UK terms, it is a solid HEO story. The agreed 4 passes, but only just. If appointment runs in merit order, a just-passing 4 is a weak place to stop.

Rewrite 3: lifting the scope

The candidate had left two true facts out of every draft because they felt like detail:

1. The deadline covered their own team and a policy team they did not manage, and they had to get both moving.
2. Partway through, they decided on their own authority to pause a lower-priority workstream. They told the deputy director afterwards, with their reasons, and renegotiated the deadline for one category of case directly with the private office.

"Because the deadline covered the policy team as well as mine, I set up a joint weekly review and agreed with their lead which cases they would draft. When it was clear we would miss the date on the complex cases, I paused the lower-priority workstream without waiting for sign-off, told the deputy director that afternoon why I had done it, and agreed a revised date for that one category with the private office so the rest stayed on time."

Count again. The action section now has 8 first-person verbs, 6 of which map to an indicator, including P3, which had been missing entirely. It also states the guidance level outright: no sign-off, accounted for afterwards. The outcome shrinks to one measured sentence about what was delivered on time. Scores came back 5, 5 and 4. Marker C's note on the 4: "nothing on what went wrong or what they'd change". That is a gap a probe will go looking for, and it is covered in the next section.

The four drafts sit on two axes. Each rewrite moved the answer along one of them:

Draft 1 sits bottom-left. Rewrite 2 moved it up to top-left, and rewrite 3 moved it across to top-right. Rewrite 1 left it where it was.

The panel scores what you say you did, at the level you say you did it, and nothing else.

How to answer probing questions in a panel interview

Treat each probe as the panel offering you marks you have not yet claimed. Answer it in two or three first-person sentences that name the action or decision it is fishing for. One coaching guide puts it as "Probes are the panel hunting for marks you've left unclaimed", and a Cambridge careers guide reports that "people often get a higher score after answering the follow-up questions".

The probes are not improvised. OPM's published examples are:

• Who was involved?
• What specific actions did you take?
• What was the outcome?
• What was the most important factor you considered in taking action?
• What is the first thing you did?

Camden's guidance for panel members adds "What was the candidate's individual role in that situation?" and specifically flags candidates who talk in the plural. If your draft says "we", expect that probe.

For the worked example, Marker C's gap was reflection. The likely probes are "what would you do differently" and "what was the first thing you did". The candidate prepared a short answer for each, plus one for "who was involved", because the policy-team detail invites it. Use this skeleton for every behaviour:

BEHAVIOUR: [name] at LEVEL [n]
GAP THE MARKER WILL PROBE: [indicator not yet evidenced, or evidenced only weakly]

Who was involved?
I worked with [roles]. My part was [the decision or action only I took].

What specific actions did you take?
The two that mattered most were [action mapped to indicator], and [action mapped to indicator].

What is the first thing you did?
First, I [concrete action], because [reason tied to the deadline or risk].

What was the most important factor you considered?
The deciding factor was [factor], so I [decision], without waiting for [sign-off or guidance I did not need].

What was the outcome?
[One measured result.] [One sentence on who used it or what changed.]

What would you do differently?
I would [specific change], because [what went wrong or what I learned].

Aim for 4 to 6 filled answers per behaviour. If one of them carries evidence for an indicator your main answer does not mention, move that evidence into the main answer. The probe set is your backup, not the place your main evidence lives.

Step-by-step: rescore your own example

Follow these seven steps for each behaviour in your advert. Budget about two and a half hours per behaviour the first time and less once your grid is built. The done-condition on each step is what lets you move on.

Step 4 works best with a peer who has not heard the story before. They cannot fill in actions you left out, which is what Marker C did. Give them this sheet:

BEHAVIOUR: ____________  LEVEL ADVERTISED: ___  SCALE: 1 to __  PASS MARK: ___

INDICATOR (exact text)        | STATED ACTION (quote it) | I or WE | LEVEL BELOW?
______________________________|__________________________|_________|_____________
______________________________|__________________________|_________|_____________
______________________________|__________________________|_________|_____________
______________________________|__________________________|_________|_____________

Action clauses total: ___   First-person: ___   Mapped to an indicator: ___
Guidance stated (sign-off needed, or decided alone): ________________
Negative evidence heard: ____________________________________________
MY SCORE: ___   REASON IN ONE SENTENCE: _____________________________
GAP A PROBE WOULD TARGET: ___________________________________________

Failure modes: why a decent example stays at a 3

Most answers that stall at 3 are not bad stories. They are good stories with the scorable evidence missing, borrowed from the team, or pitched a level too low. Each failure below includes the false positive (why the answer feels stronger than it scores) and the check that catches it.

1. Team voice ("we"). Markers cannot credit actions owned by the group. False positive: the story sounds senior, but the stated personal actions only reach a 3. Check: count the "I" verbs that map to an indicator. Camden tells panel members to probe when candidates talk in the plural, so a "we" answer also uses up probe time you could have spent elsewhere.
2. Outcome padding. A long, impressive result attached to thin actions. False positive: it ends on a high, so it feels like it went well. Check: actions should make up most of the answer. In the worked example, adding outcome moved nothing.
3. Pitched at the wrong grade. An EO-level story told at an SEO interview. False positive: every indicator gets ticked, but against the level below, and the answer is marked as only partly meeting the advertised level. Check: put your actions next to the indicators for the level above and the level below. If they read like the lower set, lift the scope.
4. Using the wrong anchor set. Preparing for a 1 to 7 scale when the campaign scores 1 to 5, or the reverse, or preparing a strengths answer as if it were a behaviour. Check: the invitation letter, because pass marks and scales are set per campaign.
5. Relying on averaging. Believing a 6 elsewhere rescues a 3. False positive: your total looks healthy. Check: coaching sources say each behaviour is a separate gate. The official rule is unverified, so treat each behaviour as its own gate.
6. Assuming the panel will infer. Leaving evidence unstated in the hope that a probe draws it out, or that one generous marker fills it in. Check: every indicator should be stated in the first answer. Probes are scripted and limited, and a panel member reports that cross-behaviour marking is allowed only in some departments.
7. Over-rehearsed delivery. One panel member says they mark down answers that sound read from a script. Check: practise from bullet points, never a full script, and vary the wording each time you rehearse.
8. Framework drift. Preparing for a framework the campaign no longer uses. Check: confirm the named framework in the advert and candidate pack before you build the grid (see the last section).

Who is likely to be marking you

Expect markers whose consistency varies. In Refolk's index, very few UK public-sector recruiters list structured interviewing as a skill. That suggests many panels are staffed by vacancy managers rather than trained assessors, which is a reason to state every indicator plainly instead of trusting the panel to apply the anchors precisely.

The underlying counts from Refolk's index, using the same title set (Recruitment Manager, Talent Acquisition Partner, Resourcing or Recruitment Business Partner, Resourcing Manager):

Two cautions on reading this:

• The 5.1% is an inference about panels, not a measurement of them. Not listing a skill is not the same as not having been trained, and the people sitting on your panel may not be recruiters at all. Treat the conclusion as unverified. It does fit what happened to Marker C, who filled a gap with a reasonable guess that consensus then threw out.
• The US figure is mostly military recruiting. In a sample of 25 US matches, 19 work at the US Army (13) or US Navy (6). For these titles, the US count mostly reflects enlistment recruiting, not civilian structured-panel staffing, so you cannot learn federal panel practice by searching on title alone. For federal interviews, OPM's own structured interview guide and the CEDAR proficiency definitions are better primary sources. OPM's practical guide is dated September 2008, so check it against your agency's announcement.

The practical response is the same either way. Write answers that a careful marker and a hurried marker would both score the same, which means stating each indicator plainly in the first person.

Keeping your examples current between campaigns

Frameworks and pass marks change between campaigns, so recheck the descriptor and the scale every time you apply, not just the first time. A bank of examples that scored a 5 against one grid can score a 3 against the next.

Re-check these sources before every campaign:

• The advert and candidate pack: which framework, which behaviours, which scale, which pass mark, and how long you get per answer.
• The GOV.UK Success Profiles collection and the Civil Service Careers Behaviours page: whether the level descriptors are still published and still unchanged.
• For federal roles: the job announcement's competencies, and the proficiency range the agency says it uses. OPM allows anywhere from 3 to 7 levels.

When you rebuild a grid, rescore your old examples against it before you reuse them. Rewrite 3 in this guide only worked because the candidate went back to what actually happened and found the cross-team scope and the solo decision. Keeping that detail on record as it happens is cheaper than digging it out the week before an interview. A work history kept current in Refolk is where that raw material lives when the next posting arrives.