Name this session to save it to your account. Renames it if it already exists.
Saved sessions
Click a session to load it. Your current work will be replaced.
Share
Share with a group, or get a public read-only link.
Public link
Anyone with this link can view a read-only summary — no account required. They won't be able to edit it.
Public link is off
Share with group
Export graph
Choose format and resolution.
PNGLossless · web & slides
SVGVector · journals & print
PPTXSlide-ready · PowerPoint
PDFPrint-ready · vector
TIFFHigh-res · journals
Download allZIP: figure, chart, blot images & CSV
Figure legend
Auto-generated from current graph settings. Edit freely before copying.
Export analysis script
Methods & citation text
Drop these straight into your manuscript's methods section.
AutoBlot Studio methods text
Copy this reference into your methods section or bibliography.
Power analysis
Calculate required sample size or achieved power for common tests.
Test
α (significance level)
Effect size (Cohen's d)
Groups (k)
Desired power (1-β)
n per group
Tails
—
Sensitivity analysis
Runs the current test twice — with and without Grubbs-flagged outliers — and shows both results side by side.
Multi-panel layout
Grid:
Quick start
Stats Suite in 3 steps: create a dataset, enter or paste your data, then run an analysis. Blotty can guide you through each step.
Create a dataset. Click ＋ New dataset in the toolbar. Give it a name and choose column or row format.
Enter data. Type or paste values into the table. Each column is a group. You can also import CSV files.
Run analysis. Switch to the Analyse tab and choose a test, or ask Blotty "which test should I use?" and it will recommend and run the right one.
Export. Download your graph as PNG/SVG, or export results as CSV.
Dataset formats
Column format
Each column is a group. Best for comparing means across independent groups (e.g. control vs treated).
Row format
Each row is a subject, columns are measurements. Use for repeated-measures or paired data.
Choosing the right test
Not sure which test to use? Ask Blotty "which test should I use?" and it will ask about your data structure and recommend the right one.
2 groups, unpaired
Unpaired t-test (Welch's) for normally distributed data. Mann-Whitney U for non-normal or small n.
2 groups, paired
Paired t-test for before/after or matched pairs. Wilcoxon signed-rank for non-normal data.
3+ groups, unpaired
One-way ANOVA with Tukey or Bonferroni post-hoc. Kruskal-Wallis for non-normal data.
3+ groups, repeated
Repeated-measures ANOVA. Use when the same subjects appear in every group.
Normality testing
Shapiro-Wilk is the recommended normality test for n < 50. If p < 0.05, the data is unlikely to be normal — consider a non-parametric alternative. For n > 50, use visual inspection (Q-Q plot) or the Kolmogorov-Smirnov test.
Choosing a graph
Bar graph
Shows mean ± SD or SEM. Good for comparing group means when n is large enough. Consider showing individual points too.
Box plot
Shows median, IQR, and whiskers (1.5× IQR). Points beyond are outliers. Better than bar charts for skewed data.
Dot / scatter plot
Shows every individual data point. Strongly recommended when n < 15. Most transparent representation.
Violin plot
Shows the full distribution via kernel density estimation. Best for n ≥ 15 per group.
Before-after plot
Lines connect paired observations. Essential for paired data — shows individual changes, not just group means.
CI plot (forest)
Shows mean ± 95% CI for each group. Non-overlapping CIs roughly indicate significance.
Best practices
Always report exact p-values. "p < 0.05" alone is insufficient. Report the exact value (e.g. p = 0.032) alongside your test statistic and effect size.
Pair p-values with effect sizes. A significant p-value with a tiny effect size (e.g. Cohen's d = 0.1) may be statistically real but biologically trivial. Always interpret magnitude alongside significance.
Check outliers carefully. Before excluding outliers, understand why they're there. Biological outliers are real data points. Use Grubbs' test to identify statistical outliers, but document your decision.
Asking Blotty for help
Blotty can run tests, switch graph types, and explain results directly from chat. Try: "Run a t-test", "Change to box plot", "What do my results mean?", or "Normalise to log10".
Blotty
Your stats assistant · Powered by AI
Hi! I'm Blotty — your stats assistant. Ask me to run tests, change graphs, or explain your results.
Blotty makes suggestions, not decisions — verify important calls with a statistician.
Help & Reference
No results matching your search.
Interactive tour
New here? The tour walks you through every part of Stats Suite step by step, highlighting each element as you go.
Quick start
3 things you need: a dataset, a graph, and an analysis. That's it.
Click + New dataset and paste your data from Excel — or enter it manually.
Your graph appears automatically in the centre panel.
Switch to the Analyse tab on the right, then click Run analysis.
Export your graph as PNG or publication-quality SVG using ⬇ Export…
Keyboard shortcuts
⌨
Navigation
Tab Move right one cell Enter Move down one cell Escape Cancel editing Delete Clear active cell
⌨
Data actions
Ctrl+V Paste data (Excel-compatible) Ctrl+Z Undo last edit Ctrl+Y Redo Ctrl+N New dataset
Templates
Try a template: Click + New dataset → Column for 3 groups, then paste your own values over the placeholder data.
📋
Pasting from Excel
Select a range in Excel, copy (Ctrl+C), then paste into the data table or the "Paste data" box in the new dataset modal. First row is auto-detected as headers if all values are non-numeric.
No results matching your search.
Choosing the right table
📊
Column
Column
Each column = one group; rows = individual replicate measurements. The most common table type. Use for: t-tests, one-way ANOVA, bar charts, box plots, dot plots, violin plots.
📉
XY
XY
Column 1 = X values (e.g. time or concentration); subsequent columns = Y values for each replicate or series. Use for: scatter plots, line charts, dose-response curves, linear regression.
🧮
Grouped
Grouped
Rows = row-factor levels (e.g. time points 0h, 24h, 48h); columns = column-factor levels (e.g. Control, Drug A, Drug B). Each cell holds replicate values separated by commas. Use for: two-way ANOVA, grouped bar charts comparing multiple factors simultaneously.
🔢
Contingency
Contingency
A grid of frequency counts. Rows and column labels are editable. Default is 2×2 (e.g. Disease Yes/No × Treatment Yes/No). Use for: chi-square test, Fisher's exact test.
📈
Survival
Survival
Two columns per group: Time (numeric) and Event (1 = event occurred, 0 = censored). Multiple groups are supported by adding Time+Event column pairs. Use for: Kaplan-Meier survival analysis, log-rank test.
No results matching your search.
Parametric tests (assume normal distribution)
⚖️
Unpaired t-test (Welch's)
2 groups
Compares means of two independent groups. Welch's variant does not assume equal variances — use this by default. Assumptions: roughly normal distribution, independent samples. n ≥ 5 per group recommended. Output: t statistic, degrees of freedom (Welch), p-value, Cohen's d effect size.
🔗
Paired t-test
2 groups
Compares means of two related groups (e.g. before vs after in the same subjects). More powerful than unpaired when pairing is valid. Assumptions: differences are normally distributed, data are paired. n ≥ 5 pairs recommended. Output: t statistic, mean difference, Cohen's d.
📊
One-way ANOVA + Tukey HSD
≥3 groups
Tests whether any group mean differs from the others. Tukey HSD post-hoc test identifies which specific pairs differ, while controlling family-wise error rate. Assumptions: normality, roughly equal variances (homoscedasticity), independent groups. Output: F statistic, p-value, η² effect size, pairwise post-hoc comparisons.
🔄
Repeated measures ANOVA
≥3 conditions
Like one-way ANOVA but for related samples (same subjects measured multiple times). Removes between-subject variability, increasing power. Use when: same subjects at each time point / condition. Assumes sphericity (use Greenhouse-Geisser correction for violations).
🔀
Two-way ANOVA
2 factors
Tests effects of two independent factors and their interaction. E.g. "Does drug treatment × time point interact?" Use for: Grouped table data. Output: F and p for Factor A, Factor B, and A×B interaction.
📏
Linear regression
XY data
Fits a straight line Y = a + bX. Tests whether the slope is significantly different from zero. Output: slope, intercept, R², t statistic, p-value for slope. Use for: continuous X and Y, when you expect a linear relationship.
Non-parametric tests (no normality assumption)
When to use non-parametric: Small samples (n < 10), non-normal distributions, ordinal data, or when Shapiro-Wilk rejects normality.
🔀
Mann-Whitney U
2 groups
Non-parametric alternative to unpaired t-test. Ranks all values, then tests if the rank distributions differ between groups. Note: Tests difference in medians / rank distributions, not means. Better for skewed data or ordinal measurements.
🔗
Wilcoxon signed-rank
2 paired
Non-parametric alternative to paired t-test. Tests whether the median difference between pairs is zero. Use when: data are paired and you cannot assume normality of differences.
📊
Kruskal-Wallis
≥3 groups
Non-parametric one-way ANOVA. Tests whether any group has a different median rank. Use Dunn's test for post-hoc pairwise comparisons. Use when: ANOVA assumptions are violated and you have ≥3 independent groups.
Correlation
🔁
Pearson correlation
XY
Measures linear association between two continuous variables. r ranges from −1 to +1. r² = proportion of variance explained. Assumes: bivariate normality, linear relationship. Use Spearman if these are violated.
📈
Spearman correlation
XY
Rank-based correlation; measures monotonic (not necessarily linear) association. More robust to outliers and non-normal distributions than Pearson.
Other tests
✔️
Shapiro-Wilk normality test
Tests whether a sample comes from a normally distributed population. p < 0.05 suggests non-normality. Use before deciding between parametric and non-parametric tests. Limitation: Very sensitive with large n — even trivial deviations become significant. Always visualise with a histogram too.
🎯
Chi-square / Fisher's exact
Contingency
Test independence of two categorical variables. Use Fisher's exact when any expected cell count < 5 (common in 2×2 tables with small totals). Output: χ² statistic, p-value (chi-square); exact p-value (Fisher's).
📈
Kaplan-Meier survival
Survival
Estimates the survival function from time-to-event data including censored observations. Log-rank test compares survival curves between groups. Key concept: Censored = subject did not experience the event before study end (still contributes information).
💊
Dose-response (4PL)
XY
Fits a sigmoidal 4-parameter logistic curve: Bottom + (Top − Bottom) / (1 + (IC₅₀/X)^Hill). Estimates IC₅₀ (or EC₅₀), Hill slope, top and bottom plateaus. Use for: pharmacology, inhibitor curves, dose-response with log-scale X axis.
No results matching your search.
Choosing a graph type
📊
Bar chart
Shows mean (or median) with error bars. Clear and familiar for comparing group averages. Best for ≥4 replicates per group so error bars are meaningful.
📦
Box plot
Shows median, IQR, and whiskers (1.5×IQR). Excellent for showing data distribution shape and outliers. Use when n ≥ 8.
⚫
Dot plot (strip chart)
Plots every individual data point. Most transparent display — recommended by many journals for small n (3–15). Shows actual spread, not just summary statistics.
🎻
Violin plot
Kernel density estimate mirrored on both sides. Shows distribution shape including multimodality. Use when n ≥ 20 for a smooth estimate.
↕
Before/After (slope graph)
Connects paired values with lines to show individual trajectories. Ideal for paired data where direction and magnitude of change matter for each subject.
⟷
CI plot (forest plot)
Horizontal lines showing mean ± 95% CI for each group. Publication-ready. Instantly shows both effect size and precision of estimation. Ideal for comparing multiple groups at a glance.
📈
Scatter / Line
Scatter: individual (X,Y) pairs — great for correlation. Line: connects points in order — use for time-series or ordered X values. Both require an XY table.
📈
Kaplan-Meier
Step-function survival curve. X axis = time, Y axis = survival probability (0–100%). Use with a Survival table. Censored observations shown as tick marks.
No results matching your search.
Common mistakes to avoid
⚠️
SD vs SEM — which to show?
SD (standard deviation) describes the spread of your data — how variable individual measurements are. SEM (standard error of the mean) describes how precisely you estimated the mean.
Use SD when you want to show how variable your data are. Use SEM or 95% CI when you want to show precision of your estimate. Never use SEM to make your error bars look smaller to imply significance.
⚠️
Multiple testing problem
If you run 20 t-tests at α = 0.05, you expect 1 false positive by chance. Always use a correction when doing multiple comparisons: • Tukey HSD — built into one-way ANOVA here, controls family-wise error rate • Bonferroni — conservative, divide α by number of tests • FDR (Benjamini-Hochberg) — less conservative, controls false discovery rate
⚠️
p-value interpretation
A p-value is the probability of getting results at least as extreme as yours if the null hypothesis were true. It is not the probability that H₀ is true.
• p < 0.05 → significant at 5% level (*) • p < 0.01 → (**) • p < 0.001 → (***) • p < 0.0001 → (****)
With very small n (e.g. n = 3), even large differences may not reach significance (underpowered). With very large n, tiny trivial differences become significant.
Rule of thumb: n ≥ 5 per group for parametric tests. If n < 5, consider non-parametric alternatives and treat results with caution. Always state n in your figure legends.
✅
Check assumptions first
Before running ANOVA or t-tests: (1) run a normality test (Analyse tab → Normality test); (2) check for outliers (enable Grubbs' test in the Analyse tab); (3) consider whether variances are roughly equal across groups. If assumptions are violated, switch to a non-parametric test.
Reporting best practice
Tip: Use the 📋 Copy results button below the graph to copy formatted statistics text ready to paste into your methods or figure legend.
📝
What to include in a figure legend
1. What the graph shows (mean ± SD, n = X per group) 2. Statistical test used and post-hoc corrections applied 3. Definition of significance symbols (* p<0.05, ** p<0.01, etc.) 4. Any data excluded and why (e.g. Grubbs' outlier)
Plate Reader Import
Bradford protein quantification — 96-well plate
1Upload
2Plate layout
3Standard curve
4Results
Upload your plate reader file. Supported formats: BMG / Clariostar (.txt), BioTek Gen5 (.txt), Tecan i-control (.xlsx), Spectramax (.xml). The wizard will auto-detect the 96-well grid.
Click or drag your plate reader .txt file here
Reads raw absorbance/fluorescence grids (A1–H12)
Click a well-type below, then click wells (or drag across) to assign.
Standards need concentration values. Samples need a name and optional dilution factor.
Blank
Standard
Sample
Unused
PBS / Blank
NCC
NDC (untreated)
Drug well
Unused
Standards — concentrations (µg/mL)
Well
Abs (blank-sub)
Concentration (µg/mL)
Samples
Well
Name
Dilution factor
Drug wells — assign group name & concentration
Each drug well needs a group name (e.g. "Drug 1") and a concentration. Wells with the same group name and concentration are treated as replicates and averaged. A dose-response/IC50 curve needs at least 3 different concentrations per group.
Set concentration for checked wells
Well
Fluorescence
Group name
Concentration
Controls
Well
Type
Fluorescence
Fit type:
Concentration unit
Serial dilution auto-fill (optional)
Enter the highest concentration, fold-dilution factor, the row(s) the dilution series runs across (technical replicates — e.g. B,C,D), and the column range it runs over. Matching wells are automatically marked as drug wells, assigned to the group, and filled in with concentrations.
Group name
Row(s)
Highest conc
Fold dilution
High col → Low col
→
% Viability preview — one table per group
Formula: (well − NCC) / (NDC − NCC) × 100. Replicates are averaged per concentration.
Sample concentrations
Sample
Well(s)
Raw Abs (blank-sub)
Conc (µg/mL)
Dilution factor
True conc (µg/mL)
Gel loading calculator
Sample
True conc (µg/mL)
Sample vol (µL)
LDS (µL)
Water (µL)
Note
Each group below becomes a separate IC50 dataset. Click Send to IC50 Curve on a single group, or Send all to IC50 Curve below to import every group at once.
