A number is not
yet an answer.
Most tools stop at the output table and leave the reading to you. Every analysis in Skari comes with guidance on how to read the result, which assumptions it rests on, and what the number does — and does not — let you claim.
Before the numbers,
know if you can trust them.
Every result opens with a quick, traffic-light read of significance, fit, sample size, and assumption checks — so you know how much weight to put on it before reading any further.
- Significance, fit, sample size, and assumptions flagged green or red at a glance
- Four key numbers surfaced before the full table
- A small sample or a violated assumption is never buried in fine print
- The same four checks on every analysis, so you learn to read them once
Significance (F, p)
p < .001Model fit (R²)
0.574 — moderateSample adequacy
n = 19 (small)Residual assumptions
3 violatedWhy these indicators matter
Slope (β₁)
1.512
SE = 0.316
R²
0.574
Moderate fit
F p-value
<0.001
Significant
RMSE
142.66
Avg. prediction error
The result,
read out loud.
Every analysis gets a plain-language pass — what the model found, which finding matters most, what it assumes, and what to do next — written the way a colleague would explain it, not the way a textbook would.
- Statistical findings translated into a short narrative summary
- The single most important result is named, not buried in a table
- Assumption violations explained in terms of what they change
- Concrete next steps, not just "further research is needed"
The model predicts pressure from temperature, with a high and statistically significant explanatory power (R² = 0.574).
Statistical findings
Model fit: R²=0.574, F(1,17)=22.93, p<.001
Temperature: β=1.51, SE=0.32, p<.001
Result summary
- •The model explains 57.4% of the variance in pressure — a "large" effect size.
- •Temperature is the strongest, statistically significant predictor.
- •A 1-unit rise in temperature lifts pressure by 1.51 units on average.
- •A strong, positive linear relationship holds between temperature and pressure.
Assumptions
Residuals violate normality (Shapiro-Wilk p = .037), and heteroscedasticity is detected (Breusch-Pagan p = .045).
Recommendations
- •Predictive performance is strong, but the normality and homoscedasticity violations need further attention.
- •Consider weighted least squares (WLS) or robust standard errors for the heteroscedasticity.
- •Explore potential non-linearity or omitted confounders to improve the model.
Every conclusion,
shows its work.
Interpretation isn't one paragraph — it's the four questions you'd actually ask, answered in order: so what is this analysis, so what came out of it, so how do I read that, and what's next. You can follow the reasoning, not just read the ending.
- 1. "So what is this analysis?" — the method and what the p-value rests on
- 2. "So what came out of it?" — the numbers, without editorializing
- 3. "So how do I read that?" — the same numbers, translated into a claim
- 4. "What's next?" — the concrete follow-up action
So what is this analysis?
- •Purpose — describe how one predictor moves the outcome, and check whether the relationship is real.
- •Method — least squares finds the line minimizing squared distances, then a t-test on the slope (t = slope ÷ SE); p < .05 rejects "slope is zero."
- •Slope — the expected change in the outcome when the predictor moves by 1 unit.
- •Residual assumptions — errors should be independent, equal-variance, and roughly normal; the p-value rests on this.
1 / 4 — scroll to advance
See the path,
not just the destination.
When assumptions are violated, Skari doesn't quietly swap in a different test. It shows which checks ran, which failed, which result is actually being reported, and which alternative it would recommend instead.
- Every diagnostic test that ran is listed, pass or fail
- The reported result and the recommended alternative are both named
- A key-findings summary ties model fit, effect, and error into one read
- Nothing is decided for you behind the scenes
Key findings
- •Model fit: R² = 0.574, adj. R² = 0.549 — explains 57.4% of the variance in pressure.
- •Overall model: F(1,17) = 22.93, p < .001 — the model as a whole is statistically significant.
- •Slope (β₁): 1.512 (SE = 0.316) — a 1-unit rise in temperature lifts pressure by 1.512.
- •Regression equation: pressure = -147.899 + 1.512 × temperature — used for prediction.
- •Prediction error: RMSE = 142.656 (Pearson r = 0.758) — the average miss, in the original scale.
1 / 2 — scroll to advance
Read every result
with confidence.
Run an analysis and see the quality check, the narrative, the four-step reasoning, and the recommended path — together, every time.