Official Numbers, Honest Questions: A Method for Reading Government Releases

Introduction

Government statistics are the base of economic debate. They cover populations no private survey can reach, and they follow consistent definitions over time. They are also dense, and the press summaries of them vary. This guide uses three real releases to illustrate five questions. We read press coverage of each, not the official documents, and where it matters we say so.

Five Questions to Ask of Any Official Release

Question 1: What is the unit? The Periodic Labour Force Survey monthly bulletin for July 2026, as reported by the Free Press Journal on 17 August 2026, gave an unemployment rate of 5.1% for persons aged 15 and above, a labour force participation rate of 55.4% and a sample of 3,71,021 persons. The unemployment rate is the share of the labour force without work who are seeking it. The labour force participation rate is the share of the population aged 15 and above that is working or seeking work. They are different denominators. A rise in participation can leave the unemployment rate unchanged, or even raise it, if more people enter the labour force and not all find work.

Question 2: Who is counted, and how? The August 2026 bulletin, as reported by Civilsdaily on 16 September, said monthly estimates use the Current Weekly Status approach, under which a person is classified by their activity during the previous seven days. That is a specific definition. A person who worked in the previous month but not in the last week is counted differently. Do not compare monthly CWS figures with annual figures based on a different reference period without checking the definitions.

Question 3: What changed, and compared with what? July's overall unemployment rate of 5.1% compared with 5.5% in June. The August rate of 5.0% was described as broadly stable. Female labour force participation was 34.8% in August against 33.7% in August 2025, and female worker population ratio 33.0% against 32.0%. Rural female participation was 39.4% and urban female participation 25.4%. These year-on-year comparisons are more reliable than month-on-month for series with seasonal patterns. A fall of 0.4 percentage points in a month may be within the survey's sampling error. We do not have the margin of error in our sources and so cannot say whether it is.

Question 4: Is the number a level, a rate or a count? The UDISE+ 2024-25 report, as covered by Careers360 on 28 August 2025, gave enrolment of 24.69 crore (a level), a secondary dropout rate of 8.2% (a rate), and secondary retention of 47.2% (a different rate). A dropout rate and a retention rate describe different things at the same stage, and both can be true. Our school-data piece explained this.

Question 5: What does it not cover? The PLFS bulletin pages we read gave no youth figures and the Free Press Journal article gave no worker population ratio. The tourism compendium's visits are not unique persons, and travel flows for 2025 sit beside GDP-share figures for 2023-24 and employment for 2024-25, which should not be combined. Most official releases state their scope. Read it.

“A rise in participation can leave unemployment unchanged, or even raise it.”

A Worked Example (Invented Numbers, Ours)

An invented state, Hillshire, reports that its labour force participation rate rose from 50% to 52% and its unemployment rate rose from 4.0% to 4.5%. Is this good or bad news? Participation rose, so more people are in the labour force. If the working-age population is 100 and the labour force rose from 50 to 52, workers rose from 48 (96% of 50) to 49.66 (95.5% of 52). So the number of workers rose by about 1.7 even though the rate of unemployment rose. The headline "unemployment rises" is true; the story "more people working" is also true. Hillshire and its figures are invented.

Habits of a Careful Reader & Finding Originals

Habits of a careful reader: 1. Read the note on method. 2. Use the same series. Compare like with like. 3. Prefer year-on-year for seasonal data. 4. Look at sample size for small groups. 5. Read two outlets. Differences show what was lost in summarising. 6. Go to the source. The ministry's website has the tables.

What this guide cannot do: It cannot fix a poor survey or a misleading summary. It reduces the chance that you misread a good one.

Where to find the originals: Ministry and statistical office websites publish the bulletins, tables and methodology notes for each release. Press summaries are a good way to learn that a release exists, but a table from the original lets you check the unit, the sample and the footnotes for yourself. If you quote a figure in a report, cite the original release by name and date as well as the article you read.