Notes · Updated
What changed, what we found, what the numbers say
Short, dated entries. Data from the bank, changes to the outline, findings in public sources, and a few tactical guides. Every entry names its sources and the date they were read.
- 2 minThe 150 question simulator is served for the first timeThe bank now covers all seven domains at the depth a full sitting draws. What that took, and the one outline section that still has no source.All domains
- 2 minCorrection: 27 questions were filed under the wrong domainWe had been filing questions by the source they came from instead of by where the exam examines them. Reading the outline in full moved 27 of them, and made our coverage look worse.All domains
- 2 minCorrection: our score report used a 125 question split we had decided not to publishThe report said three domains carry about 26 questions each. Applying the published weights to a 150 question paper gives 31 to 32. The old figures came from a number we could not verify with HSPA.All domains
- 1 minCorrection: the right answer was the longest option in 58% of our itemsChance would give 25%. We measured our own bank against the Haladyna and Downing item-writing flaw taxonomy, rewrote the distractors, and now screen every item before release.All domains
- 2 minHSPA says to read its own domain feedback with cautionThe retake page's own words, and why that is the argument for measuring before you book.All domains
- 2 minSix weeks is a calendar, not advice: the retake waitHSPA sets a mandatory wait between attempts. How to spend it against the domains that cost you points instead of all seven.All domains
- 1 minThe passing score is not published, and that is policyAngoff, an expert panel, and pass or fail. Why no line is drawn on our report.All domains