We believe in showing our work. This page explains exactly how we gather data, run our analysis, and produce the statistics you see on TaxFacts. Questions? Contact us anytime.
Data Sources
1 Official IRS Data
Our foundation is official government data. We use:
- IRS Data Book — Annual statistics on audits, refunds, collections, and enforcement actions
- IRS Publications — Pub 17, 334, 463, 502, 503, 505, 525, 526, 535, 587, 590-A/B, 596, 970
- Revenue Procedures — Annual inflation adjustments, standard deduction amounts, bracket changes
- Form Instructions — Official guidance directly from the IRS
Every factual claim we make is verified against these official sources.
2 Taxpayer Experience Data
To understand what happens in practice, we analyze anonymized discussions from public tax forums where taxpayers share their experiences:
- Refund processing timelines (how long people actually waited)
- Deduction outcomes (approved, denied, or asked for documentation)
- Audit experiences (what triggered them, how they resolved)
- Tax software experiences (satisfaction, issues, recommendations)
Important: We never quote or identify individual users. All data is converted into aggregate statistics.
Our Analysis Process
3 Data Collection & Cleaning
Raw data goes through several cleaning steps:
- Remove duplicates — The same story posted multiple places is counted once
- One vote per person — Each user contributes one data point per topic, regardless of how many times they posted
- Filter outliers — Extreme values are flagged for manual review
- Verify dates — Tax year attribution is confirmed before counting
4 Statistical Analysis
Our analysts apply standard statistical methods:
- Sample size requirements — We only publish statistics with at least 30 data points
- Median over average — For dollar amounts and timelines, we use median (less affected by extreme cases)
- Confidence intervals — We calculate and report the range of uncertainty
- Year-over-year comparison — We track changes across tax years when possible
5 Writing & Fact-Checking
Our editorial team transforms analysis into clear content:
- Plain language — We explain complex tax topics without jargon
- IRS citations — Every factual claim includes a link to the official source
- Visual presentation — Charts and tables make data easy to scan
- Editorial review — All content is reviewed before publication
What Our Data Can (and Can't) Tell You
Understanding Our Limitations
Forum data has built-in bias. People who have problems are more likely to post about them. This means:
- Negative outcomes may be overrepresented in our data
- Wait times may skew longer (people with fast refunds post less often)
- Audit stories may seem more common than they actually are
That's why we're careful about how we present statistics. Instead of saying "10% of home office deductions are denied," we say "Among taxpayers who discussed home office deductions, 10% reported denial." This accurately describes our data without overstating what it means.
Other important limitations:
- User-reported data may contain mistakes or misunderstandings
- Everyone's tax situation is different — averages may not apply to you
- Tax law changes — historical patterns may not predict next year
- State taxes vary — our data focuses on federal taxes
Quality Controls
6 Verification Process
Before anything goes live:
- Source check — Every IRS citation is verified against the current publication
- Math check — All percentages and calculations are double-checked
- Common sense check — Does this statistic match what we know from other sources?
- Tax year check — Is this information current?
7 Keeping Content Current
Tax information requires regular updates:
- Annual review — All articles are reviewed when new IRS guidance comes out
- Inflation updates — Dollar amounts updated when IRS announces changes
- Quick corrections — Errors are fixed immediately with visible notes
- Date stamps — Every article shows when it was last updated
Privacy in Our Research
We take privacy seriously in our data collection:
- No names or identifiers — We don't store usernames, emails, or any personal information
- Generalized details — Specific numbers become ranges (exact salary → income bracket)
- No direct quotes — We analyze patterns, not individual posts
- Data deleted after analysis — Raw text is removed; only statistics remain
Read our full Privacy Policy for more details.
Questions About Our Methods?
We're committed to transparency. If you have questions about how we calculated a specific statistic, spotted an error, or want more details about our process, please reach out. We're happy to explain.