Supported statistical tests

GraphHelix supports 30+ statistical tests across parametric, non-parametric, Bayesian, regression, survival, and advanced methods. Every test reports the complete set of statistics: test statistic, degrees of freedom, p-value, effect size with magnitude label, and 95% confidence interval, formatted in APA 7th edition notation.

Compare two groups

Test	When to use	Reports
Unpaired t-test	Compare means of two independent groups (normal data)	t, df, p, Cohen's d, 95% CI
Paired t-test	Compare means of two related measurements (before/after)	t, df, p, Cohen's d, 95% CI
One-sample t-test	Compare a sample mean to a known value	t, df, p, Cohen's d, 95% CI
Mann-Whitney U	Compare two independent groups (non-normal or ordinal data)	U, p, rank-biserial r, 95% CI
Wilcoxon signed-rank	Compare two related measurements (non-normal data)	W, p, rank-biserial r
Bayesian t-test	Bayesian evidence for/against a group difference	BF10 (Jeffreys scale), 95% credible interval

Compare 3+ groups

Test	When to use	Reports
One-way ANOVA	Compare means across 3+ independent groups	F, df, p, η2; Tukey/Bonferroni/Holm post-hoc
Two-way ANOVA	Factorial design with two grouping variables	F, df, p, partial η2 for each effect + interaction
Repeated measures ANOVA	Compare means across 3+ conditions (same subjects)	F, df, p, partial η2; Mauchly's + Greenhouse-Geisser
Kruskal-Wallis H	Compare 3+ independent groups (non-normal data)	H, df, p; Dunn's post-hoc
Friedman test	Compare 3+ conditions, same subjects (non-normal data)	χ2, df, p
Bayesian ANOVA	Bayesian evidence for group differences	BF10, 95% credible intervals

Relationships and regression

Test	When to use	Reports
Pearson correlation	Linear association between two continuous variables	r, p, 95% CI
Spearman correlation	Monotonic association (ordinal or non-normal data)	rs, p, 95% CI (Fisher z-transform)
Linear regression	Predict a continuous outcome from one or more predictors	R2, adj. R2, F, p, β, VIF
Logistic regression	Predict a binary outcome	Odds ratios, 95% CI, AUC, classification table
Bayesian linear regression	Bayesian evidence for predictor effects	BF10, posterior coefficients, 95% credible intervals

Categorical data

Test	When to use	Reports
Chi-square test	Test association between two categorical variables	χ2, df, p, Cramér's V
FDR correction	Correct for multiple comparisons (Benjamini-Hochberg)	Adjusted p-values

Survival analysis

Test	When to use	Reports
Kaplan-Meier survival	Estimate survival probabilities over time	Survival curve with 95% CI, log-rank test for group comparisons
Cox proportional hazards	Covariate-adjusted survival regression	Hazard ratios, 95% CI, concordance index, Schoenfeld PH test

Advanced methods

Test	When to use	Reports
Linear Mixed Model (LMM)	Clustered or nested data (random intercepts)	Fixed effects table, ICC with magnitude label, AIC/BIC
PCA	Reduce dimensionality of 3+ variables	Scree plot, component loadings, biplot
Exploratory Factor Analysis	Identify latent factors from observed variables	Factor loadings, communalities, scree plot
Mediation analysis	Test whether variable M mediates X → Y	Direct, indirect (bootstrap CI), and total effects
Moderation analysis	Test whether variable Z changes the X → Y relationship	Interaction coefficient, simple slopes at M−1SD/M/M+1SD

Custom scripts

Need something not listed above? Write Python scripts with numpy, pandas, scipy, and statsmodels in a sandboxed environment. Scripts run in an isolated subprocess with resource limits for safety.

Automatic features with every test

• Assumption checking — Shapiro-Wilk normality, Levene's equal variances, Mauchly's sphericity (where applicable), with Q-Q plots
• One-click non-parametric fallback — when assumptions are violated, switch to the appropriate alternative in one click
• APA 7th edition formatting — one-click copy of the complete formatted result string
• Effect size with magnitude label — Cohen's benchmarks applied automatically
• AI interpretation — plain-language explanation of what the results mean