Reading a Paper: the Claim, the Evidence and the Assumption
One useful reading route starts with the central claim, follows the results that support it, and checks how those results were produced. Examine the assumptions linking evidence to the claim, how they were tested, and whether the reported uncertainty supports the conclusion.
What a learner can do afterwards
- Extracts the central claim and identifies the results that support it
- Identifies the assumptions the result depends on and how the authors tested them
- Judges whether the stated uncertainty supports the strength of the claim
1 · Read
One useful route through a paper starts with its central claim, follows the supporting results, then checks the method. Return to earlier sections when needed. The claim is the proposed conclusion. Evidence consists of observations or results offered to support it. Several figures, tables or analyses may support one claim.
Match the check to the inference. A claim about a wider population needs support from the sampling method. Comparisons must address relevant differences that could explain the outcome. More repeats alone do not remove selection bias or confounding. One supporting example cannot establish an 'all' claim, but one confirmed exception can refute it. Check whether background facts apply under the stated conditions.
You have examined systematic error before. A scale reads 183 g for a 180 g reference. The 3 g offset could affect a reported result. Measuring a stable reference before and after a run checks whether the offset changed. In a null run, no signal is intended. Compare any nonzero result with expected random noise before treating it as evidence of an offset or background.
Find what the uncertainty means and which sources it includes. Statistical-only bars may omit calibration uncertainty. A 95% confidence interval from -2 to 4 units includes zero, so the interval alone does not establish a positive difference under that analysis. It does not prove that the difference is zero either. A confidence interval is not a guaranteed bound on the true value.
Redshift means that observed light has longer wavelengths. A cosmological redshift arises from the universe's expansion. A paper may use a cosmological model to estimate distance from redshift, then infer luminosity from distance and apparent brightness. Check whether the redshift is mainly cosmological and whether the model is appropriate. For fixed apparent brightness, a larger inferred distance implies greater inferred luminosity in that model.
Objects close together on the sky can lie at different distances. A quasar and galaxy with different redshifts in nearly the same direction may be a chance alignment. Their angular proximity alone neither establishes physical association nor disproves the distance inference. Look for independent evidence of association and a comparison with the chance alignments expected under the study's selection rule.
Trace the claim to its evidence and method. Check the scope, assumptions, tests and uncertainty before deciding what the results support.
Lesson, questions and flashcards written for this dataset. Khan Academy's Types of evidence and the supplied OpenStax astronomy activity informed concept and question-design study; their prose was not reused.
Readings and flashcards written for this dataset. The supplied evidence and astronomy records were studied for concepts and argument structure, not adapted as text.
More to read
Checking the reach of a conclusion
Start by writing the conclusion in one sentence. Include who or what it concerns and the conditions under which it is claimed to hold. Next, list the observations offered in support. Keeping these separate helps you notice a conclusion that reaches beyond its evidence.
Suppose a laboratory recruits its own volunteers but writes a conclusion about all adults. The population claim needs an account of recruitment and any adjustments used to support that extension. Measuring the volunteers repeatedly can improve estimates for those volunteers without settling the population question.
For an experimental comparison, list relevant conditions that changed together. Different materials tested at different room temperatures leave more than one possible explanation for different cooling times. Ask how the design or analysis separates those explanations.
The role of an example also depends on the claim. One supporting specimen does not establish a statement about every specimen. One confirmed exception can refute a statement that admits no exceptions.
Finally, identify the assumption checks and uncertainty definitions. A calibration reference tests particular measurement risks. A null run checks a setting with no intended signal, but its output must still be compared with expected noise. No individual check establishes that every part of the argument is sound.
Direction is not distance
A position on the sky tells an observer which direction to look. Distance is separate information. Two objects can appear close together even when one is far beyond the other.
Consider a proposed quasar-galaxy pair selected because the objects appear nearby on the sky. If their redshifts differ, the reader still needs evidence that they are physically associated. Angular proximity does not establish that association. The study should also address how often unrelated objects would meet its selection rule by chance.
Redshift describes observed light shifted toward longer wavelengths. Cosmological redshift comes from the expansion of the universe. A cosmological model can connect such a redshift to an estimated distance. A paper can then combine distance and apparent brightness to infer luminosity within its model.
Trace those steps separately when assessing the argument. Ask whether the redshift interpretation and model are appropriate, and whether the claimed physical association has independent support. Finding more apparent pairs does not remove the need to examine selection and chance alignment. The number of pairs must be judged against a stated expectation, not treated as decisive by itself.
2 · Watch
Take it off screen
Where it sits
Optional connections
Where this leads
8 questions wait behind this lesson, each with its answer explained. Every answer feeds the sky: stars light as they are learned, and dim when it is time to come back.