You’ve heard the advice a thousand times: compare the candidates on the issues. Sounds like homework you can knock out in an afternoon. It isn’t.
What you’re actually handed is a curated, focus-grouped, poll-tested set of positions, delivered in language engineered so that no two candidates describe the same thing the same way. One says “reform,” one says “modernize,” one says “protect,” and none of them tell you what would actually change on paper.
So here’s the real process. It’s not hard, but it’s not the process anyone markets to you either.
Step 0: Write your list before you look at a single candidate
This is the whole trick and almost nobody does it.
If you build your issue list after reading campaign material, you’re comparing candidates on their chosen terrain. Campaigns pick wedge topics that poll well with likely voters and quietly bury the ones that split their own base. That’s not a conspiracy, it’s basic marketing — but it means the issues list you’re handed is a sales document.
So before you open anything:
- Write down 5–8 topics you actually care about, in plain language.
- Weight them. Be honest — a 2% difference in one topic isn’t the same as a total reversal on another.
- Add one topic you’d call a dealbreaker. Not a preference. A line.
Now you have a fixed ruler. Everything else gets measured against it, not the other way around.
Stated position vs. revealed priority
Any candidate will tell you they support popular things. That costs nothing. The useful question is what did they spend on it?
Ask these:
- Did they introduce, sponsor, or push anything concrete, or just vote on something someone else wrote?
- Did they take a hard vote when it was unpopular with their own side?
- Did they bring it up in a setting where it wasn’t convenient — a hostile interview, a room full of donors who didn’t want to hear it?
- Did it survive contact with a budget?
A stated position is a claim. A revealed priority is a pattern of behavior with a cost attached. Compare the second one.
Read the record — and note what’s missing from it
Advertisements are not evidence. Records are. The problem is nobody hands you the records, so here’s where they actually live:
- Official legislative archives: roll-call votes, bill text, amendments, committee assignments.
- Committee hearing transcripts and recordings — long, boring, and the single richest source of how someone actually thinks.
- Meeting minutes and agendas for local bodies, which are usually posted and almost never read.
- Financial disclosure and campaign finance filings — more on this below.
- Court and regulatory filings, if the person’s public role touched either.
Then read the gaps, because the gaps are data:
- Absences and abstentions. Who skips the hard votes? Who’s conveniently “not available”?
- Procedural votes. A vote on whether to even debate something tells you more than a vote on the final version everyone knew would pass.
- Sponsorship vs. co-sponsorship. Signing on is cheap. Writing is not.
- Binding law vs. resolutions. One changes things. The other is a press release with a stamp on it.
The one move that separates serious people from everyone else
Read the actual bill text, not the summary. Or read the amendment that changed it. Or the fiscal note. Any of these will, within about 20 minutes, tell you whether a candidate’s description of their own record is accurate or a work of fiction.
You do not need to read 200 pages. Read the definitions section — that’s where the real scope hides — and read what the thing actually appropriates or prohibits. That’s it. That’s the quiet workaround here: the primary source exists, it’s public, and virtually nobody bothers.
Decode the language, because it’s designed to be undecodable
Watch for phrases that can’t be falsified. If a statement can’t be proven wrong, it isn’t a position — it’s decoration.
- “I’ll fight for…” — no mechanism named. Ask: through what, exactly?
- “Common-sense reform” — we’d like to change something and would prefer not to say what.
- “Studies show…” — which study, funded by whom, measuring what?
- “Bipartisan” — at least one person from the other side agreed. That’s the entire bar.
- “Modernize” / “streamline” / “update” — could mean expand, cut, or nothing. Always ask which.
When you hear one of these, don’t argue with it. Just translate it into “what would change in the text?” If there’s no answer, file it as noise and move on.
Follow the structure of the money, not the scary total
Total dollar figures are trivia. Structure is information.
- Look at donor composition by category, not just size. Who is the money coming from, in aggregate?
- Cross-reference donor categories against the committees or offices the person holds or wants. Overlap is a pattern; coincidence is not.
- Look at spending by outside groups that support them. Legally they can’t coordinate, and yet the message alignment is often astonishing.
- Look at where they went after previous public roles. Revolving doors are documented, public, and rarely mentioned in campaign material.
Money doesn’t buy votes in the cartoon sense. It buys access, and access shapes what gets brought up at all. Compare who gets the meeting.
Check the job description before you compare anyone
This one is quietly devastating. A huge share of candidate comparison happens over things the office literally cannot do.
Before you grade anyone, figure out what the position actually controls: what it votes on, what it appoints, what it budgets, what it can veto, what it can only urge. Then compare candidates only on the slice they’ll actually touch. You’ll be stunned how much campaign material evaporates under that filter.
Build the scorecard (this is the whole thing)
- Take your pre-written topics and weights from Step 0.
- For each candidate, score each topic 0–3 — but only using primary sources.
- Next to every score, write the citation: the vote, the bill, the filing, the transcript line.
- Mark your confidence: solid, thin, or guessing. Guessing is allowed; pretending it isn’t is not.
- Add a tiebreaker column for “can this office even do this?”
- Wait a week, then re-check. Positions that shift under mild pressure are worth knowing about.
One page per candidate. If you can’t fill in a box with evidence, leave it empty and note it. Emptiness is a finding.
What doesn’t count as evidence
- Endorsements from people who endorse nearly everyone.
- Viral clips with no context — especially the ones that confirm what you already wanted to believe.
- A single bad vote, isolated from the ten adjacent ones.
- Promises with no mechanism and no cost.
- “Momentum,” “electability,” and crowd sizes. Those are strategy inputs, not issue positions.
- How they handled one hostile interview — that’s temperament data, not policy data. Useful, but don’t confuse the two.
The uncomfortable part
If you do this honestly, you’ll probably end up fully agreeing with nobody. That’s the normal outcome, and it’s the part the “just compare the issues” advice never prepares you for.
So decide your tradeoffs in advance, in writing, while you’re calm. Which of your five topics would you sacrifice for your dealbreaker? Which compromise would you accept? If you make that call after the fact, you’ll make it based on vibes and whoever annoyed you least that week.
Bottom line
Comparing candidates isn’t about finding the one you like. It’s about producing a document you could defend to someone who disagrees with you — with citations, gaps noted, and confidence levels marked.
You won’t get certainty. Nobody does. But there’s a real difference between voting on a paper trail and voting on a feeling someone engineered for you, and almost everyone settles for the second one because the first one takes a weekend.
Spend the weekend. The records are public, the transcripts are boring, and that’s exactly why they still work.