Read statistical claims in Book Isle: restore denominators, time windows and source tables

Before adopting “increased by fifty percent” as a conclusion, establish whether it refers to counts or a rate relative to its baseline, and whether denominator and time window changed. Use Book Isle reading, bookmarks, search and notes to organise a manual check, calculating on paper or your own calculator. All numbers here are fictional inspection records, not research or product results.

Choose one claim with a location

Choose one numerical sentence in an authorised source and retain surrounding paragraphs, rather than beginning with a detached screenshot. Record title, author, edition, section and anchor text, plus file page for PDF. Bookmark it and label a minimal note unchecked. Avoid marking an entire chapter without a tractable verification task.

Separate the value from what it describes

Record the measured object, value and change wording. For fictional “flagged records increased by half,” distinguish flagged count from flagged proportion. Resolve vague objects from context rather than inventing a preferred interpretation. Separate source wording from your explanation, using available search without assuming automatic statistical-field extraction.

Find numerator and denominator

Suppose twelve of one hundred records were flagged, then eighteen of another one hundred: rates are twelve and eighteen percent. Record both totals and flagged counts. Equal denominators make counts and rates rise together here, but locate actual totals in real tables or methods. Mark a missing denominator unknown rather than inventing a plausible sample size.

Calculate percentage points and relative change separately

In this equal-base example, eighteen minus twelve is six percentage points; six divided by the original twelve is a fifty-percent relative increase. Both can be correct descriptions of different quantities. Recalculate independently and record baseline, endpoint and meaning. Do not replace six percentage points with six percent, or infer practical importance merely from the larger relative number.

Test understanding with a changed denominator

Now compare twelve of one hundred with eighteen of one hundred fifty. Flagged count still rises by half, but both rates are twelve percent. Label this self-created counterexample clearly. It tests which metric the source means rather than proving the author wrong, and must not later be mistaken for data actually reported in the book.

Check time window and population

Check whether columns use the same period, population and flagging rule. Monthly versus quarterly counts or all records versus voluntary responders may change meaning. Record differences before summarising growth, and check any recalculation after rule changes. Search locates definitions; it does not certify comparable columns without contextual review.

Record missing values and rounding

Do not reconstruct falsely precise counts from rounded percentages. Distinguish zero, blank and unreported through table notes, and verify unreadable cells in the source rather than guessing. Record unresolved precision. For an image-only scan, navigate by page without assuming automatic OCR; visible digits and extractable numbers are different capabilities.

Arithmetic agreement does not establish causation

Correct arithmetic verifies only that layer. A causal explanation still needs attention to data collection, comparison and alternatives; two time points alone are not a cause. Record this as a separate question rather than treating the simple calculation as full research appraisal. The exercise supplies no domain-specific medical or investment decision and cannot replace relevant evidence.

Write a traceable checking result

Summarise metric, denominator, period, arithmetic agreement and unresolved questions with section or table anchors. Plain-text statuses suffice; no new fields are required. Avoid an unsupported “bad statistics” label or exempting a famous author from checks. Date later updates separately rather than replacing the evidence behind the earlier judgement.

Finish with a three-part reread

Reread the source sentence, your interpretation and the table to confirm one metric. Choose one next worthwhile claim rather than highlighting every percent sign. Retain provenance when backing up notes separately from books. Open University’s percentage-point explanation supports terminology, while actual data still require original sources. Completion means explaining calculation and scope, not assigning every claim an absolute true/false label.

Related tutorials and sources

Figure and table anchors, Question-led reading.

Terminology reference: Open University: Percentage points. The checking sequence and numerical examples here are original exercises, not research data reported by that course.

Checked on November 3, 2026 against the official Book Isle listing and Apple’s version lookup: version 1.1.3, requiring iOS / iPadOS 15.6 or later. This manual workflow uses public local-reading, search, bookmark and note capabilities. It does not promise automatic judgements of argument truth, statistical correctness or reference quality. Readers supply their sources and check controls in the installed version. No step-by-step physical-device testing is claimed.