Organize Evidence Across Sources into a Synthesis View
In the last chapter you saw that synthesis means reading across your sources at once, not one after another. But you cannot read across sources that are scattered through separate pages of notes, one per article, each in its own words. Before you can weigh or reconcile anything, you have to get all the evidence into a single view where the sources sit side by side and line up against the same questions. This chapter shows you how to build that view, and how to read it so that agreement, disagreement, and every quiet “it depends” jump straight off the page.
See Why Per-Source Notes Hide the Answer
When most people gather evidence, they take notes source by source: everything the first article said, then everything the second said, and so on. Each note is fine on its own. Together they are almost useless for synthesis, because the thing you actually care about, how the sources compare on each part of your question, is split across pages and buried in different wording.
Say your question has three parts to it: cost, speed, and size. Source one talks about cost and speed but never size. Source two talks about size and cost but calls cost “value for money.” Source three only reviews speed. To answer “what do my sources say about cost?” you now have to hunt through three sets of notes, translate three different phrasings, and hold it all in your head at once. That is exactly where people give up and fall back on the loudest source.
The fix is not more notes. It is a different shape of notes: one that turns “read across the sources” from a memory feat into a simple left-to-right scan.
Build a Synthesis Matrix
The tool for that is a synthesis matrix: a simple table that lays your sources against the parts of your question, so every source is answering the same set of questions in the same place. Read down a column and you see one source; read across a row and you see what every source says about one thing. That second move, reading across, is the one per-source notes make impossible and the matrix makes effortless.
Each row of the matrix is a factor: one specific part of your question that you want every source to speak to, such as a cost, a sub-question, or a single point of comparison. The columns are your sources. Each cell where a row meets a column holds, in a few words, what that source says about that factor, or a blank if the source is silent on it.
Building one is a short, repeatable process.
List your factors as rows. Pull them straight from the scoped question and success standard you built in Part 1: each measurable element, sub-question, or comparison point becomes one row.
List your sources as columns. One column per source, labeled so you remember what kind of source it is (a report, a review, a set of figures).
Fill each cell briefly. Write what the source actually says about that factor in a few words or a number, not a full paragraph. If the source says nothing about that factor, leave the cell blank on purpose.
Add a short note on what each source is. A one-line reminder of where each column came from, so that when you read across a row you remember you are comparing a lab measurement with a personal opinion, not two things of equal standing.
Read the Matrix for Agreement, Disagreement, and Qualification
A filled matrix is not the answer; it is the view that lets you find the answer. You read it row by row, and each row falls into one of a few patterns. Naming the pattern is what turns a wall of cells into something you can reason about.
Agreement: the cells in a row mostly say the same thing. That factor is looking solid, though you will still weigh how good the agreeing sources are in the next chapter.
Disagreement: two or more cells flatly contradict each other. This is not a problem to hide; it is the most useful thing the matrix shows you, because it tells you exactly where the real work is.
Qualification: the cells do not contradict so much as add conditions, “yes, but only for small amounts,” “true, except in winter.” A row full of qualifications usually means your factor is real but depends on something you have not pinned down yet.
Silence or gaps: a row with several blank cells. That is not nothing; it tells you the evidence on that factor is thin, which matters when you state how confident you are at the end.
Walk Through a Full Synthesis View
Let’s build one from start to finish on a question you could actually face. We will keep coming back to this same evidence set over the next few chapters, so it is worth meeting it properly now.
The scoped question, framed the Part 1 way, is this: “For a typical household cooking an everyday meal, does an air fryer use less energy and less time than a conventional electric oven?” The decision behind it is a household deciding whether an air fryer is worth buying for regular cooking. You have gathered four sources.
Source A, the manufacturer’s product page: claims the air fryer uses “up to 70% less energy” and cooks “twice as fast.”
Source B, an independent consumer-testing lab report: for a small batch (chips for two), it measured the air fryer at 0.6 kWh against the oven’s 1.4 kWh, and the air fryer clearly faster; for a large batch (a meal for five), the air fryer needed two rounds and the gap shrank to about 1.1 kWh against 1.5 kWh, with cooking times about equal.
Source C, a cooking magazine’s hands-on review: says the air fryer is excellent for small portions and reheating but too small for a full family meal, and notes that an oven’s real time cost is preheating.
Source D, aggregated customer reviews on a retail site: mixed, with many one- and two-person households praising the speed and energy savings, and several larger households saying they still use the oven for big meals.
First, pull the factors from the question. It asks about energy and time, and the sources keep raising batch size, so those become the rows: energy use, cooking time, and capacity or batch-size fit. Then the sources become the columns. Here is the filled synthesis matrix.
Factor | A: Manufacturer | B: Lab report | C: Magazine review | D: User reviews |
Energy use | “Up to 70% less” | Less for small batches (0.6 vs 1.4 kWh); gap shrinks when large (1.1 vs 1.5 kWh) | (not measured) | Savings reported by small households |
Cooking time | “Twice as fast” | Faster for small batches; about equal for large | Faster, but oven’s cost is preheating | Faster for small meals |
Capacity / batch fit | (silent) | Needs two rounds for a large meal | Too small for a full family meal | Large households still use the oven |
Now read it row by row and name the patterns.
Energy use: broad agreement that the air fryer uses less, but heavily qualified. B, C, and D all point the same way for small amounts, while B adds the crucial condition that the energy gap narrows for a big meal (1.1 against 1.5 kWh), shrinking from a wide lead to a slim one rather than vanishing. A’s “up to 70%” is the strongest claim and the least conditional.
Cooking time: agreement for small batches (A, B, C, D), with a qualification that for large batches the times are about equal (B) and that the oven’s real delay is preheating (C).
Capacity: a clear pattern, and notice A is silent here. B, C, and D all independently raise the same limit, that the air fryer is too small for a family meal, which A never mentions.
Look at what the view has handed you. You have not yet decided the answer, and you have deliberately not counted cells. But you can now see three things at a glance that were invisible in scattered notes. The sources genuinely agree that the air fryer wins clearly for small amounts. They also agree that for a large meal the picture shifts: the energy lead narrows, the time saving disappears, and the fryer’s capacity becomes a real limit. And the one source claiming the biggest, most unconditional benefit is also the one selling the product and the one silent on that limitation. That last observation, which the matrix surfaced just by placing A’s row next to the others, is precisely the thread the next chapter picks up when you start weighing sources by quality instead of treating them as equal.
Let’s Recap!
Per-source notes hide the answer because they split each factor of your question across separate pages in different wording.
A synthesis matrix lays sources against the parts of your question so you can read across a row and compare every source on one factor at once.
Build the rows (factors) from your scoped question and success standard, add one column per source, and fill each cell briefly or leave it blank on purpose.
Read each row as agreement, disagreement, qualification, or a gap, and treat disagreements and blanks as useful signals rather than problems to hide.
The matrix organizes evidence but never scores it; deciding whose evidence counts for more comes next, and counting cells is the mistake to avoid.
Your evidence now sits in one view where the patterns are visible, including one source that makes the loudest claim while selling the product. The next chapter gives you the tool that view is asking for: a way to weigh each source by its quality and relevance, so one strong source can rightly outweigh several weak ones.
