LectureSift · Learning library
Check PDF study notes: tables, OCR and conclusions
Correct a flawed summary against a worked PDF example. Trace claims to source pages, compare percentage points, answer a quiz and download a checking sheet.
What should you check in a PDF summary?
A short, fluent summary does not necessarily say the same thing as its source. In this exercise you will read a lesson table, correct a misleading summary, and turn the corrected information into a question and a flashcard. No account, upload or credits are required. The lesson text and numbers below were created for this exercise; they are not a real research study or a customer document.
Your goal is not to rewrite every sentence. It is to preserve the numbers, the comparison and the limits of the conclusion. You can use the same checks on one table or short section of your own PDF before reviewing a longer document.
Source: a mini-test in two groups
Page 1 — Activity. Groups A and B each contain twenty students from the same class. Group A completed a worksheet; group B reread their lesson notes. Students were not randomly assigned to groups. They chose which activity to join.
Page 2 — Results. Both groups took the same mini-test the next day. In the table, “successful” means scoring at least 8 points on this mini-test. Twelve students in A and eight in B reached that threshold. Students’ knowledge before the activity was not measured.
Page 3 — Limitation. The result compares the two groups observed in this activity. These data are not sufficient to show that the worksheet alone caused success.
| Group | Students | Scoring at least 8 | Rate |
|---|---|---|---|
| A | 20 | 12 | 60% |
| B | 20 | 8 | 40% |
Correct the flawed summary yourself first
Imagine that a text-extraction or summarization step produced this note:
“Among randomly selected students, the worksheet increased success by 20%. Everyone in group A succeeded. The result proves that this method works better for all students.”
Before opening the answer, mark four issues: how the groups were formed, the success rate in A, how the difference is described, and who the result is generalized to. Write a source page beside each correction. Produce the corrected sentence instead of simply labeling the original ‘wrong’.
Open a worked correction and its reasoning
“In this example, where students chose their activity, 60% of A and 40% of B scored at least 8 on the mini-test. The observed difference is 20 percentage points. Because prior knowledge was not measured and students were not randomly assigned, we cannot attribute the difference to the worksheet alone.” Source mapping: group selection on page 1; rates and the definition of success on page 2; limits of inference on page 3. Twelve of twenty students in A reached the threshold, not everyone.
Separate percent from percentage points
A’s rate is 12 ÷ 20 × 100 = 60%; B’s is 8 ÷ 20 × 100 = 40%. Subtracting the rates gives 60 − 40 = 20 percentage points. If you use B as the comparison baseline, the relative difference is (60 − 40) ÷ 40 × 100 = 50%. These expressions describe different calculations and should not be substituted for one another.
The example also does not measure an increase over time: it compares two different groups on the same test. Consequently, “success increased by 50%” is not a suitable standalone summary either. Correct arithmetic does not automatically make the relationship described by a sentence correct.
Check yourself: what if 10 students in B had succeeded?
B’s rate would be 50%. The gap from A would be 10 percentage points; the relative difference using B as the baseline would be 20%. These calculations belong to the new assumption and do not change the original source table.
Three small checks for a scanned page
- Character: Check whether the zero in “20” became the letter “O”, or “12” became “I2”. Do not guess a correction merely because it looks plausible. Return to the page image.
- Row: Cells from A and B might have shifted. Confirm that the student count and successful count come from the same row.
- Condition: Shortening “at least 8” to “8” changes the definition. Include the table heading and footnote when checking the text.
If a table is unreadable, obtain a clearer copy or mark the information as “source unreadable”. Asking a model to fill the gap is different from verifying the source.
Turn the correction into study material
Note: In this example, A is 60% and B is 40%, a gap of 20 percentage points. Without random assignment, this does not establish a single causal explanation.
Flashcard front: Why are 20 percentage points and a 50% relative difference different expressions in this example?
Flashcard back: The first directly subtracts the rates. The second compares the gap with B’s 40% baseline. State the comparison baseline and context alongside the calculation.
Quiz: Which conclusion does the source support? A) The method works for everyone. B) The two groups’ observed rates differ. C) The groups had equal prior knowledge.
Show the quiz answer
B. A requires a basis for generalization, and C requires a baseline measurement. Neither is supplied by the source.
A checking sheet for your own document
Download the blank checking sheet
For each important claim, record the source page, source sentence, initial note, correction and remaining uncertainty. Resolve the uncertainty before turning the claim into a memorization card. This record will also remind you why you chose a particular wording when you return to your notes several days later.
Use the same method when creating a document study pack, or continue to the AI-note checking guide for broader review steps. The worked solutions here are authored teaching examples, not the live output of a file upload.