Compare development measures and competing explanations
Can higher average income coexist with worse measured health?
Your task: Make a three-indicator comparison using separate axes and units. Write two rival explanations for the pattern and identify evidence that distinguishes them, including within-region inequality.
Open the investigation · Read the explanation · Your saved evidence
Starting recap and vocabulary
Read counts with their units. A percentage divides a part by its named whole and multiplies by 100. A model assumption is not an observed fact.
- Purchasing power parity
- A comparison method accounting for differences in the prices of comparable goods and services across economies.
- Assumption
- A condition accepted for a model or argument; it needs justification before applying the result elsewhere.
Can higher average income coexist with worse measured health?
Make a three-indicator comparison using separate axes and units. Write two rival explanations for the pattern and identify evidence that distinguishes them, including within-region inequality.
Make a three-indicator comparison using separate axes and units. Write two rival explanations for the pattern and identify evidence that distinguishes them, including within-region inequality. Use the supplied records and show one intermediate step.
Make a provisional claim
Loading saved evidence…
Inspect evidence and test an explanation
The investigation controls appear when your local evidence is ready.
Connect evidence and mechanism
Compare development measures and competing explanations
Question and record
Can higher average income coexist with worse measured health?
Fictional Region A has income per person20000 international-comparison units, life expectancy70 years and expected schooling14 years. Region B has income15000, life expectancy78 and expected schooling15. No composite index is supplied.
The quantities and graphs are teaching assumptions, not observed policy outcomes.
Mechanism and worked calculation
Development analysis compares economic, health, educational and institutional outcomes. Composite indices require explicit transformations and weights; different dimensions cannot simply be added in raw units. Stage-based modernization and dependency/world-systems approaches emphasize different causal histories and relationships. Treat them as arguments to test with evidence, not inevitable destinies.
A has one-third more income than B:20000/15000−1≈33.3%. B has eight more years of life expectancy and one more expected schooling year. These comparisons do not produce a valid composite index without its method.
Challenge the interpretation
A report ranks A first using income alone, then calls that the ranking of a human-development index. Purchasing-power comparisons adjust for different price levels, while market exchange conversions answer another question. Means conceal internal disparities and informal activity can be difficult to measure.
Your evidence task
Reconstruct a development index before interpreting it
The opening region comparison lacks mean years of schooling, so it cannot produce an official HDI calculation. This separate investigation uses UNDP's published worked example, then fictional changes. Predict whether doubling income doubles the composite score, and whether equal arithmetic means always give equal geometric means.
Inspect the official example and its scope
UNDP Human Development Report2025, Technical Note1, printed pages3–4 supplies the Benin2023 example: life expectancy60.8 years, expected schooling10.4 years, mean schooling3.2 years and GNI per person3,806 in2021 PPP dollars. It labels the inputs rounded and prints HDI0.515. This frozen example is not a current-country ranking or a complete microdataset.
UNDP's indicator explanation distinguishes expected schooling for children entering school from mean schooling among adults25 and older. Neither is the percentage attending school or a direct measure of learning quality. GNI is gross national income; PPP adjusts purchasing-power comparisons. These units are not a market exchange-rate conversion.
HDI summarizes aggregate dimensions. It does not by itself measure internal inequality, political empowerment or every aspect of development. This is public aggregate geography and indicator analysis, not biological research, diagnosis or an individual assessment. Method/source checked29 October2026.
Four inputs, three dimensions
Convert unlike units into dimensionless indices using stated goalposts. Health=(life expectancy−20)/(85−20). Expected-schooling index=years/18; mean-schooling index=years/15. Education is their arithmetic mean. Income index=ln(GNI/100)/ln(75,000/100), with GNI capped at75,000. Here ln is the natural logarithm: equal proportional income changes give equal increments before the cap, rather than equal dollar increases giving equal increments.
The final HDI is the cube root of health × education × income. The cube root finds the equal dimension value with the same product. This differs from adding the three values and dividing by3. Goalposts and aggregation rules are methodological choices, not a ranking of people's worth.
Selected case: reconstruction of published rounded inputs. All other inputs are held fixed by assumption. These changes are not observed Benin outcomes, policy forecasts or equal-cost interventions.
| Quantity | Selected input | Transformed index |
|---|---|---|
| Life expectancy | 60.8 years | 0.627692 |
| Expected schooling | 10.4 years | 0.577778 |
| Mean schooling | 3.2 years | 0.213333 |
| Education: two schooling indices averaged | Separate age groups, normalized first | 0.395556 |
| GNI per person | 3,806 in2021 PPP$; capped input3,806 | 0.549717 |
Selected geometric HDI calculation: 0.514870. Arithmetic alternative: 0.524321 (a comparison rule, not official HDI). Missing mean schooling prevents the education index and composite; it is not replaced with zero or omitted from the denominator.
Published example HDI: 0.515. Recomputed from the displayed rounded inputs without intermediate rounding: 0.514870. Keep the source value and reconstruction separate. Extra decimal places in our calculation do not make the source estimates more precise.
Transfer: the same mean, a different balance
These three dimension indices are invented mathematical profiles, not real countries or people.
Values: 0.7, 0.7, 0.7. Arithmetic mean 0.700000; geometric mean 0.700000. Explain why the first two profiles share arithmetic0.7 yet have different products. A zero dimension makes the product zero; a missing dimension instead prevents calculation.
Compare calculations and improve the conclusion
Doubling only income produces about0.545679, not twice the original composite. The logarithm and geometric aggregation both matter. The uneven normalized profile has geometric mean0.624025 despite arithmetic0.7, while the balanced profile gives0.7 under both rules. Higher values in other dimensions only partly offset a low dimension under the product rule.
Do not infer which intervention causes the greatest real improvement from this arithmetic. The worksheet holds inputs fixed mechanically and supplies no costs, causal estimates or within-country distributions. Propose a companion distributional indicator and a source that could challenge a claim of broad improvement. Preserve both the official example and your hypothetical changes in the final brief.
Copy your development-index report
Feedback opened this visit: no. Reset clears writing and restores source inputs, complete data and balanced transfer; exposure remains until reload. Copy temporary work before leaving. No mastery or policy judgment is automatically certified.
Adjusting for inequality still leaves an overlap question
A fictional briefing says: “An inequality-adjusted index tells us how many people face several disadvantages together.” Test that claim using a published calculation and a separate four-person example. This extends the ordinary HDI activity; an adjusted aggregate is not a household record.
Inspect the official example and calculation discrepancy
UNDP2025 Technical Note2, printed pages5–7 gives Nepal2023 HDI0.622, inequality-adjusted HDI (IHDI)0.437 and relative loss29.7%. It prints dimension indices0.775/0.534/0.582 and inequality measures0.150/0.398/0.319. The measures come from microdata; they cannot be reconstructed from national average income or schooling alone. The page labels values rounded.
The adjusted dimension is I×(1−A): I is the dimension index and A its inequality measure. Combine the three adjusted dimensions with a geometric mean. The printed adjusted values are0.659/0.320/0.397. Direct products from the displayed inputs differ, particularly education0.534×0.602=0.321468 versus printed0.320. Both are retained here; rounding alone is not assumed to explain the discrepancy. The publisher's underlying microdata/calculation files are not supplied.
The same note explicitly says the IHDI does not capture overlapping inequalities. This packet uses public aggregate facts and an invented matching exercise, not biological research or personal data collection. Sources and the printed table were checked29 October2026; the PDF page was visually inspected.
| Dimension | Printed index I | Inequality A | Calculated I×(1−A) | Printed adjusted index |
|---|---|---|---|---|
| Health | 0.775 | 0.150 | 0.658750 | 0.659 |
| Education | 0.534 | 0.398 | 0.321468 | 0.320 |
| Income | 0.582 | 0.319 | 0.396342 | 0.397 |
Calculated from displayed dimensions: HDI 0.622189, IHDI 0.437834, relative loss 29.630007%. The loss is100×(1−IHDI/HDI), an index comparison—not the percentage of people who are poor or the amount of money lost.
Published example:0.622 →0.437, loss29.7%. The distinct arithmetic average of its three inequality measures is28.9%; that is not the product-based loss formula. Withholding education inequality prevents the adjusted aggregate, while the ordinary HDI remains computable. Missing is not zero inequality.
Transfer: same separate distributions, different people matched
Four invented people have positive income-achievement and schooling-achievement points. These are dimension-specific teaching units, not money, actual schooling years or Nepal observations. Change only which person gets each schooling value. Both marginal lists remain1,1,9,9.
| Person | Income points | Schooling points | Low in both? |
|---|---|---|---|
| P | 1 | 1 | Yes |
| Q | 1 | 1 | Yes |
| R | 9 | 9 | No |
| S | 9 | 9 | No |
Each separate list has arithmetic mean5 and geometric mean3. For these positive, equally weighted values, the Atkinson form A=1−geometric/arithmetic gives1−3/5=0.4. The real methodology has additional data-handling rules; this toy calculation does not reconstruct Nepal's inequality measures.
For illustration only, stipulate base dimensions0.7/0.7/0.7, with health inequality0 and schooling/income inequality0.4 each. The product adjustment gives 0.497965 under either matching. These base dimensions are stipulated, not derived from the point values. People low in both: 2/4 = 50%.
Compare the overlap claim and revise it
The aligned case has two people low in both; the reversed case has none. Both separate distributions and their inequality measures remain identical, so a formula using only those marginal summaries cannot distinguish the matching. This is an information limitation, not a claim that inequality adjustment is useless.
State exactly what joint records would be needed to investigate overlap, including common person identifiers, compatible populations and measurement periods, and appropriate consent/access. Propose a public aggregate or safely de-identified source rather than collecting classmates' sensitive information. No real Nepal overlap, poverty rate or causal improvement follows from the toy example.
In your revised brief, distinguish the publisher's displayed figures, arithmetic reconstructed from those figures, missing microdata and a separate fictional counterexample. Preserve the unexplained source discrepancy rather than silently replacing the printed values.
Copy your inequality and overlap report
Feedback opened this visit: no. Reset clears writing and restores complete source inputs/aligned toy records; exposure remains until reload. Copy temporary work before leaving. No interpretation is automatically graded.
The official worked-example reconstruction above is separate from the opening fictional regions, whose incomplete inputs still do not establish a composite ranking.
Make a three-indicator comparison using separate axes and units. Write two rival explanations for the pattern and identify evidence that distinguishes them, including within-region inequality.
Revise your answer to “Can higher average income coexist with worse measured health?” using both source IDs. Separate the checked calculation from a judgment, and identify evidence that could reverse your judgment.
Vocabulary
Purchasing power parity: A comparison method accounting for differences in the prices of comparable goods and services across economies.
Assumption: A condition accepted for a model or argument; it needs justification before applying the result elsewhere.
Independent extension
Independently extend the investigation “Compare development measures and competing explanations” with a documented real dataset or scholarly source. Preserve definitions, recreate one calculation or evidence table, compare a rival explanation and report why the teaching example cannot establish the real-world conclusion.
Worked example
A has one-third more income than B:20000/15000−1≈33.3%. B has eight more years of life expectancy and one more expected schooling year. These comparisons do not produce a valid composite index without its method.
Check the reasoning, then transfer it
Loading response history…
Keep your evidence and choose a next step
Loading saved lesson status…