Critical Thinking Tool for Students

Data
Detective

Learn to read charts like a scientist. Spot misleading data, evaluate conclusions, and write your own analysis — step by step.

Step 1 of 4 — Tutorial

What Makes a Good Chart?

Here's a well-crafted chart about monkey species populations. Click each criterion below to learn what makes it trustworthy — and what good critical thinkers always look for.

Primate Population Estimates by Species, 2015–2023
Wild population counts (thousands) across four monitored primate species · Sample: 847 survey sites · Date range: Jan 2015 – Dec 2023
Population data: Howler Monkey 220k, Spider Monkey 140k, Capuchin 195k, Squirrel Monkey 165k.
Y-axis: Estimated wild population (thousands)  |  X-axis: Species  |  Error bars: ±1 standard deviation
Source: International Union for Conservation of Nature (IUCN) Red List Database, 2023 Annual Survey. See: iucnredlist.org  |  Data collected by wildlife biologists across 12 countries.
0 of 6 explored

① Source & Reference

Does the chart name who collected the data — with a verifiable link or citation?

② Sample Size & Date Range

Is it clear how much data was collected, and over what time period?

③ Appropriate Labels

Are all axes, units, and data series clearly labeled so nothing is ambiguous?

④ Proportional Dimensions

Do bar heights, slice sizes, and areas accurately represent the data values?

⑤ Neutral Colors

Do the colors avoid steering the viewer toward a particular interpretation or emotion?

⑥ Honest Conclusions

Does the stated takeaway match what the data actually shows, without overstating?

Step 2 of 4 — Spot the Problem

What's Wrong With This Chart?

This chart has six problems. Study it carefully, then select every issue you notice from the list below.

⚠ Flawed Chart — Analyze Carefully
Hours of Screen Time Per Day
(a chart with information)
Screen time data with misleading y-axis starting at 4 hours.
Labels Here. No error bars or methodology noted.
Source: A Recent Study (link to be provided)

Select all problems you identify in the chart above:

Step 3 of 4 — Evaluate a Conclusion

Which Conclusion Fits the Data?

Below is a real survey chart showing AI scientists' opinions about AI risk. Read the chart carefully, then choose the conclusion that best represents what the data actually shows.

AI Researchers' Views on Existential Risk from AI
Survey of 2,778 researchers who published at NeurIPS or ICML in 2021 · Conducted: May–June 2022 · Question: "What probability do you assign to the possibility that the long-run effect of advanced AI on humanity will be extremely bad (e.g. human extinction)?"
AI extinction risk survey: 5% say >50% probability; 18% say 10-50%; 58% say under 10%; 19% say not sure or decline to answer.
Survey context:
Respondents: 2,778 researchers
Response rate: 17% (738 replies)
Published in: AI Alignment Forum, 2022
Citation: Steinhardt, J. (2022). Existential risk from AI survey.

Note: Survey asks about perceived probability, not about evidence of danger.

Which conclusion best fits what this chart actually shows?

Option A

AI scientists have proven that artificial intelligence poses a catastrophic danger to human civilization and will likely cause mass extinction within this century. The majority of experts believe this outcome is unavoidable.

Option B

Because most AI scientists don't think AI will be extremely dangerous, there is no reason to study or regulate AI safety. The survey shows concerns are overblown.

Option C

A majority of AI scientists assigned a low probability (under 10%) to AI causing human extinction, but a meaningful minority (roughly 23%) assigned a 10% or higher chance, suggesting that while catastrophic risk is not consensus, it is a concern taken seriously by some experts.

Option D

Since this is only a survey of opinions and not direct evidence, no conclusions whatsoever can be drawn from this data. Opinion data is always meaningless in science.

Step 4 of 4 — Write Your Conclusion

Put It Into Practice

Study this final chart, then write a 50–100 word conclusion in your own words. Be precise: describe what the data shows without overstating it.

Global Average Surface Temperature Anomaly, 1950–2023
Departure from 1951–1980 average baseline (°C) · Annual mean · n = 74 years of global station data
Temperature anomaly data showing warming trend from 1950 to 2023.
Source: NASA Goddard Institute for Space Studies (GISS) Surface Temperature Analysis (GISTEMP v4), 2023. See: data.giss.nasa.gov/gistemp
💡 Tip: A strong conclusion (1) states what the chart shows using data, (2) notes the time frame and source, and (3) avoids claiming more than the chart proves. Do not say "proves" or "causes" — say "shows," "suggests," or "is associated with."
0 words (aim for 50–100)
🔍

Case Closed!

You've completed the Data Detective investigation. You can now identify trustworthy data, spot misleading charts, evaluate conclusions, and write your own analysis.

✓ Source Evaluation ✓ Sample Size Awareness ✓ Label Analysis ✓ Visual Bias Detection ✓ Causal vs Correlation ✓ Written Analysis