Two Strong Studies Point in Opposite Directions: Building a Balanced Synthesis Before the Deadline

You have two well-designed studies, both credible, both relevant, and their conclusions pull against each other. The deadline is close, the rubric rewards synthesis, and your draft currently reads like a list of findings rather than an argument. This article walks through the repair: compare design, sample, exposure, outcome, timing, and analysis before you write a single synthesis sentence, then turn that comparison into a deliverable you can submit.

Imagine a course paper due in five days. The rubric asks you to "critically compare the evidence" and "synthesize findings across studies." Your draft has a paragraph on Study A and a paragraph on Study B. Each paragraph is accurate. Together they explain nothing, because the reader cannot tell why two strong studies disagree or which one should carry more weight for your question.

The instinct under deadline pressure is to write a sentence like "the evidence is mixed." That sentence is true but empty. It does not show the comparison the rubric is asking for, and it does not help your reader decide anything. The repair is not more reading. It is a structured comparison of the specific features that make two studies differ.

Before adding more papers, write down what counts as evidence for your question. The Cochrane Handbook's guidance on searching for and selecting studies starts from eligibility criteria: plan the search from those criteria, use multiple sources, combine free-text terms with subject headings, document the process, deduplicate records, and screen in stages.

For your paper, that means a short written list: which populations, which exposures or interventions, which outcomes, which study designs, and which date range. This list does two things. It stops you from drifting toward whatever is easiest to find, and it gives you a defensible reason to include or exclude the next paper you encounter.

If you are still assembling the candidate set, run the search through a scholarly interface rather than a general web search. WisPaper search is the entry point for finding academic papers, and WisPaper Scholar Search explains the paper-finding entry point for English readers. Keep your eligibility list next to the search so each result is judged against the same standard.

Search documentation matters more than students expect, because the rubric often rewards a transparent method. PubMed's user guide describes field tags, phrase searching, filters, Advanced Search, saved searches, and a Search Details view that reveals how terms were translated and combined. Google Scholar's help pages describe searching by author and title, date limits and sorting, Cited by and Related articles, citation export, and locating alternative versions of a work.

You do not need to reproduce a full systematic review. You do need a record: which source you used, which terms, which filters, and when. That record lets you say in your methods paragraph exactly how the two conflicting studies were found, and it lets you rerun the search if your supervisor asks for an update before submission.

This is the core of the repair. Create a table with one row per study and six columns: design, sample, exposure, outcome, timing, and analysis. Fill it from the papers themselves, not from memory.

• Design. Was the study experimental, observational, cross-sectional, longitudinal, or a synthesis of other studies? Design constrains what a study can support.
• Sample. Who was studied, how many, and how were they selected? A sample that differs from your population changes how far the finding travels.
• Exposure. How was the exposure or intervention defined and measured? Two studies can use the same label for different things.
• Outcome. Which outcome was measured, and how? A disagreement about outcomes is not the same as a disagreement about effects.
• Timing. When were exposure and outcome measured relative to each other, and over what period?
• Analysis. What comparison did the authors make, and what did they adjust for?

When a cell is unclear, mark it unclear. An honest gap in the table is more useful than a confident guess, and it tells you where to look next.

Once the table is full, read across the rows instead of down the columns. Most apparent contradictions resolve into one of a few patterns:

1. Different outcomes. Study A measured one endpoint, Study B another. Both can be right about different things.
2. Different samples. The studies describe different populations, so the findings apply to different groups.
3. Different exposure definitions. The same term covers different measurements or thresholds.
4. Different timing. Short-term and long-term windows can produce different pictures.
5. Different analyses. Different comparison groups or adjustments can shift a result.

Name the pattern explicitly in your draft. "These studies disagree" becomes "these studies address different outcomes in different populations," which is a synthesis claim you can defend.

When one study seems to settle the question, resist the pull to stop searching. Google Scholar's Cited by and Related articles features let you see what later work built on a paper and what sits alongside it. PubMed's Similar Articles feature does something comparable within its index. Use these to test whether the two studies you have are representative or unusual.

This is also where a literature-focused agent can carry a longer task. Scholar Agent is the WisPaper area for longer research tasks, including literature review work. Give it the comparison you have already framed, not a vague topic, so the output can be checked against your six-column table.

By this point you may have more PDFs than your draft uses. Save the ones you will cite and set aside the ones you will not. My Library is where you save or upload papers and organize them in the WisPaper library. A clean library makes the final citation pass faster, which matters when the deadline is close.

If you need to ask questions across the papers you have collected, Library QA supports asking questions based on papers uploaded to My Library or added from search results. Use it to check a detail in the table, such as how a study defined its exposure, rather than to generate the synthesis for you.

A balanced synthesis paragraph follows the comparison, not the study order. A workable structure:

• State the question in one sentence.
• Report where the studies agree.
• Report the specific feature that explains the disagreement, drawn from your table.
• Say what each study can and cannot support for your question.
• Close with the narrowest claim your evidence allows.

For example, instead of "findings are inconsistent," write a sentence that names the differing outcome or sample and explains why that difference matters for your research question. The rubric's "critical comparison" criterion is usually satisfied by exactly this move.

Suppose your supervisor returns comments: "You present two studies as equal, but they are not measuring the same thing. Also, where is the third study you mentioned in your proposal?"

Do not rewrite from scratch. Return to the table. Add the missing study as a new row, fill the six columns, and check whether it changes the pattern you identified. Then revise only the paragraphs whose claims the new row affects. This keeps the revision proportional to the feedback and protects your remaining time.

If the feedback asks for a broader literature picture, Research Projects is the WisPaper area for creating and working within a research project, which can hold the work as it grows past a single comparison.

Your revised draft now contains: a methods paragraph naming sources, terms, and filters; a six-column comparison table in an appendix or inline; a synthesis paragraph that names the specific reason the studies diverge; and a limitations sentence that follows from the table rather than from a template. The two strong studies still point in opposite directions, but your reader can now see why, and your claim is narrower and better supported than "the evidence is mixed."

With two days left, make a concrete call. If every cell in the table is filled or honestly marked unclear, if the synthesis paragraph names a specific point of divergence, and if every citation in the draft matches a paper in your library, submit. If one cell is still unclear and it carries the weight of your main claim, spend the remaining time resolving that cell and leave the rest of the draft alone. Do not add new studies at this stage unless the supervisor specifically asked for one; adding rows late is how tables become inconsistent with the text.

• Write eligibility criteria before searching, and keep them visible while screening.
• Document sources, terms, and filters as you go, not afterward.
• Compare studies feature by feature before writing a synthesis sentence.
• Treat an unclear table cell as a finding about the evidence, not a personal failure.
• Revise in proportion to feedback: add rows, then revise only affected paragraphs.
• Keep the library and the citation list in sync so the final pass is mechanical.