Three Sources, One Unread
Dev is drafting a 1,200-word review of after-school tutoring programs. He has three sources: (A) a 2019 meta-analysis of 42 studies reporting a modest positive effect; (B) a 2023 randomized study of 300 students finding no effect; (C) a widely shared practitioner blog post that summarizes A and calls the finding "settled science." Dev has read A and B in full; he has only skimmed C's summary of A. His instructor requires at least one direct quotation framed inside his own sentence, one paraphrase with a citation, no citation of anything he has not read, and a synthesis that represents disagreement rather than averaging it.
1. What can Dev legitimately do with source C?
Read the worked answer
C cannot support any claim about tutoring, because Dev has not read what C summarizes and A is already in hand. Citing C for the effect size would both break the instructor's rule and misrepresent provenance.
C remains usable in two narrower ways:
- As an object of analysis, not evidence. Dev may cite C for the claim that a practitioner audience has received the finding as "settled science" - a statement about reception, clearly marked as such: One widely shared practitioner account describes the result as "settled science." This is a real and interesting observation, and it is supportable because Dev has read the post itself.
- As a lead. C may point him to primary work he then reads and cites directly.
The test: is the claim being used as evidence about tutoring or as evidence about how people talk about tutoring? Only the second is available to him.
2. Write a two-sentence synthesis of A and B that represents disagreement rather than averaging. Name the move you used.
Read the worked answer
Synthesis: "A 2019 meta-analysis of 42 studies reported a modest positive effect of after-school tutoring, while a 2023 randomized trial of 300 students found none. The divergence is not necessarily a contradiction: the meta-analysis pooled heterogeneous programs, whereas the trial tested a single implementation, so the open question is whether modest average effects survive in any specific program."
The move: state both findings at full strength, then locate the axis of disagreement (pooled versus single-program evidence) instead of splitting the difference.
The averaged version - "research shows mixed but generally modest benefits" - is the failure mode. It is not neutral; it is false to B, which found no effect, and it hides the design difference that actually explains the divergence. Averaging competing findings is a way of making a disagreement disappear rather than reporting it.
3. Dev wants to quote a sentence from A that begins "Because the effect sizes were small..." Show one integrated version and one dropped-in version, and explain what changes.
Read the worked answer
Dropped in: The study has limits. "Because the effect sizes were small, the gains may not translate into measurable achievement."
Integrated: The authors frame their caveat causally: "Because the effect sizes were small," they write, "the gains may not translate into measurable achievement."
What changes: in the dropped-in version the quoted sentence sits alone, so nothing outside the quotation marks receives the because-clause; the caveat reads as Dev's own assertion, and the attribution evaporates. In the integrated version the framing clause ("they write") supplies the main clause the subordinator needs, and the quotation attaches to Dev's syntax instead of floating beside it.
The practical rule: a quotation beginning with a subordinating conjunction must either be quoted whole, with its main clause intact, or be completed outside the quotation marks. Trimming a sentence to its most dramatic clause and dropping it in is how a hedged finding becomes an unhedged one.