Study this exam as a set of decision rules rather than a definition list: convert scores through a single z-chain, read reliability and validity as two separate questions, judge item statistics against the test's purpose, and route interpretation to psychologist-level practice. Drill short paper scenarios weekly and gate yourself with the five-point readiness rubric in the final section.
Reliability vs. validity: which question is the scenario actually asking?
Reliability is score consistency; validity is accuracy for a stated purpose. Read each scenario for its signal: stability, equivalence, or agreement points to reliability; prediction, diagnosis, or fitness-for-purpose points to validity.
A test can be highly reliable yet invalid for the decision at hand, because consistency and accuracy are separate properties. Name the subtypes as you study: test-retest stability, parallel-forms equivalence, internal consistency through split-half or Cronbach's alpha, and inter-rater agreement. A scale that weighs the same person identically every morning is reliable; whether it actually measures body fat is validity. If a scenario reports a high alpha and asks whether the test suits a diagnostic decision, alpha alone never answers it — reliability is necessary but not sufficient.
Turn this into a tagging drill. Underline the cue sentence: 'yields the same results on retesting' means test-retest; 'correlates with grades next year' means predictive criterion-related validity; 'the items hang together' means internal consistency; 'covers every content domain' means content validity. When both properties appear in one stem, decide which one the question's conclusion actually depends on. Practice by labeling ten statements in your reviewer with R or V plus the subtype before you read the explanation, and correct your tags in pen.
Converting standard scores quickly: z, T, deviation IQ, stanine, percentile
Master one conversion chain — raw score to z, then z to any standard score — plus a mean and standard-deviation lookup. Every scale differs only in its center and spread.
Compute z as (X − M) / SD, then translate: T = 10z + 50, with mean 50 and SD 10; deviation IQ has mean 100 and SD 15; stanines span nine bands centered on 5 with a spread of roughly 2. Percentile rank is not a standard score at all — it is a rank, so percentiles bunch near the middle of a normal distribution. The classic error is reading the 84th percentile as '84 percent correct' instead of 'scored higher than 84 percent of the norm group.'
Worked example: Maria earns 115 on a test with mean 100 and SD 15. Her z is +1.00, so her T is 60 and her stanine is 7. On a second instrument scaled with mean 50 and SD 10, an equivalent standing is 60 — the same z produces different raw numbers on different scales. The mistake to avoid is comparing raw scores across two tests; compare z's. The drill is mechanical: twenty conversions daily until the chain z to T to stanine runs in under a minute without notes.
Use this table as your lookup until the conversions are automatic.
| Score type | Center (mean) | Spread (SD) | Where you meet it | Same standing |
|---|---|---|---|---|
| z-score | 0 | 1 | Standard normal unit, the conversion anchor | z = +1.00 |
| T-score | 50 | 10 | Personality and clinical reports | T = 60 |
| Deviation IQ | 100 | 15 | Cognitive ability reports | IQ 115 |
| Stanine | 5 | about 2 | Nine-band school reporting | Stanine 7 |
| Percentile rank | 50 (median) | varies | Rank within a norm group, not distance | 84th percentile |
Choosing validity evidence when a test's norms don't fit your examinees
Match validity evidence to the decision: content for coverage, criterion-related for prediction, construct for theory. Then check whether the norm group resembles the people you are actually testing.
Worked scenario: a guidance team screens Grade 1 pupils for a fast-track program using a reasoning test praised for a .92 internal-consistency coefficient, but its norm sample was drawn entirely from a foreign school system. The plausible mistake is endorsing the instrument because 'it is reliable.' The better decision is to ask whether any validity evidence — criterion correlations or construct studies — covers Filipino children of that age, language, and schooling background before anyone is placed.
The reason it matters: a high consistency coefficient says nothing about whether the items measure the intended construct for this population, and vocabulary-heavy items can end up measuring exposure to a particular curriculum instead. The stronger route is to flag the norm-relevance gap to the supervising psychologist, request locally relevant norms or a better-fitting instrument, and document the basis for the choice. Decision rule: confirm norm-group resemblance before interpreting; when it is absent, the report should carry that limitation explicitly.
Item analysis numbers: difficulty index, discrimination, and distractor flaws
Compute p as the proportion answering correctly and D as the gap between upper and lower scorer groups. Read them together: a very easy item can still be a poor discriminator.
An item with p = .90 and D = .05 is passed by nearly everyone, and high and low scorers answer it alike, so it contributes almost no ranking information. Reviewers commonly treat D of roughly .30 or higher as acceptable and a low or negative D as a red flag; these cutoffs are rules of thumb, so anchor on the logic rather than the exact number. A negative D means low scorers outperformed high scorers — suspect a miskeyed answer or an ambiguous stem before suspecting the examinees.
Distractor analysis adds a third lens. If the correct option is chosen by everyone while one distractor attracts only low scorers, the item behaves normally; a distractor chosen disproportionately by high scorers signals ambiguity or a flawed key. As a practice exercise, write two short purpose statements — a mastery check where a high p is desirable, and a ranking test where moderate p and strong D matter — then take five practice items and justify each item's statistics against the purpose you assigned it. This trains you to check item statistics against the test's purpose, not a single cutoff.
Where psychometrician scope ends and psychologist work begins
Republic Act No. 10029 licenses psychometricians for assessment functions, principally administration and scoring of objective tests under the supervision of a registered psychologist. Interpretation and intervention belong to psychologist-level practice.
Worked scenario: a testing center, pressed by a deadline, asks its licensed psychometrician to score a completed personality inventory, write the clinical interpretation, and recommend a diagnosis. The plausible mistake is complying because the task feels like 'just finishing the report.' The better decision is to complete the scoring as assigned and route the interpretation and any diagnostic language to the supervising psychologist. The boundary turns on who performs interpretive and diagnostic acts; producing scores under supervision sits inside the psychometrician's lane, declaring what the scores mean does not.
Study the statute from its own text and your ethics reviewer's treatment of it: pin down, phrase by phrase, which functions the law assigns to psychometricians and which it reserves for psychologists, rather than relying on paraphrases. Then rehearse boundary questions as rapid triage — does the task end at producing a score, or does it start with 'what does this mean?' Build three of your own boundary scenarios, such as report writing, releasing results, and handling supervision requests, and state for each which lane it belongs to and why.
Ethics in conduct: security, consent, and confidentiality during administration
Ethics scenarios test conduct: safeguarding test materials, securing informed consent, protecting results, and declining tasks beyond your competence. Pair every administration duty with the reason it exists.
Test security means protocols, answer sheets, and scoring keys stay controlled before, during, and after administration; releasing actual items or coaching examinees invalidates the assessment and breaches professional duties. Informed consent means examinees or their guardians understand the purpose of testing, the information gathered, and who will receive results. Confidentiality means results reach only those authorized to receive them, not whoever happens to ask. A teacher demanding a child's raw protocol because 'she is the adviser' is a decision point, not a formality to wave through.
Rehearse these as chained decisions: identify the duty, identify the requesting party, choose the response, and justify each in one sentence. An adviser may receive results through authorized channels once consent is handled properly; she does not receive the protocol itself. When duties collide — an employer pressing for a candidate's item-level responses against confidentiality — a well-built scenario turns on releasing interpretive conclusions through proper channels rather than raw data. Practice until you can state the duty and its justification in a single sentence for any request you can imagine.
A six-week sequence with decision drills and readiness checks
Sequence review by concept clusters: psychometrics math first, assessment and testing second, ethics and scope last, each drilled weekly with paper scenarios rather than recitation alone.
The decision-card drill: take twenty core concepts — four reliability types, three validity types, five standard scores, key item-analysis terms, and three ethics duties. For each, write one exam-style stem ending in a choice (which type? which action?) and a one-line decision rule on the card's back. Shuffle and drill weekly. Expected observation: by the third shuffle you answer stems in one sentence without recalling the full definition; any card that stalls you twice marks a concept you know as a word but not yet as a decision.
Self-check rubric — score each item 0 to 2; these are learning milestones, not passing predictions. One: convert any raw score across every scale in the standard-score table within a minute. Two: tag fifteen reliability-versus-validity statements correctly, including subtype. Three: compute p and D for a twenty-item practice set without formula notes. Four: resolve three ethics boundary scenarios, stating duty and justification for each. Five: explain in two sentences why reliability alone never establishes fitness for a decision. Any item below 2 sends you back to that specific cluster, not the whole syllabus.
- Weeks 1–2: Statistics and psychometrics — the z, T, deviation-IQ, stanine, and percentile chain, reliability and validity types, item analysis; twenty conversion drills daily.
- Weeks 3–4: Psychological assessment, cognitive and personality testing — for each named instrument, record the construct, the method (objective or projective), and one feature separating it from close neighbors.
- Week 5: Ethics, RA 10029 scope, and test administration and scoring procedure — write three boundary scenarios per session and triage them aloud.
- Week 6: Mixed timed sets across all six topic areas; log every error as a missing decision rule, then re-drill that rule until it fires automatically.
References and further reading
Use these references to explore the concepts and check the latest information from the relevant organizations.