Run a One-Way ANOVA with Automated Narrative Reporting

Executes a one-way ANOVA using stats::aov(), extracts key metrics via broom, computes eta-squared as an effect-size measure, and generates a plain-language narrative via the Narrative Generator Module.

Usage

easy_anova(formula, data, alpha = 0.05)

Arguments

formula

A formula of the form outcome ~ group_factor, or a character string. Passed directly to stats::aov().

data

A data frame containing the variables in formula.

alpha

Significance threshold for the narrative. Default 0.05.

Value

An object of class "easystat_result" with:

test_type

Character: "anova"

formula_str

Character string of the formula used

raw_model

The raw aov object

coefficients_table

ANOVA table (SS, df, MS, F, p)

model_fit_table

Summary metrics (F-statistic, eta-squared, p-value)

explanation

Plain-language narrative string

Examples

result <- easy_anova(Sepal.Length ~ Species, data = iris)
print(result)
#> 
#> ================================================================================
#>  EasyStat Result :: ANOVA
#> ================================================================================
#> 
#> TABLE 1 — MAIN RESULTS
#> --------------------------------------------------------------------------------
#>     Source  df Sum of Squares Mean Square F Statistic  p-value
#>    Species   2        63.2121     31.6061    119.2645 <0.0001%
#>  Residuals 147        38.9562      0.2650          NA       NA
#> 
#> TABLE 2 — MODEL FIT / SUMMARY
#> --------------------------------------------------------------------------------
#>            Metric    Value
#>       F-statistic 119.2645
#>          Group df        2
#>       Residual df      147
#>   Overall p-value <0.0001%
#>  Eta-squared (η²)   0.6187
#> 
#> TABLE 3 — GROUP DESCRIPTIVES
#> --------------------------------------------------------------------------------
#>       Group  N  Mean     SD     SE CI_Lower CI_Upper
#>      setosa 50 5.006 0.3525 0.0498   4.9058   5.1062
#>  versicolor 50 5.936 0.5162 0.0730   5.7893   6.0827
#>   virginica 50 6.588 0.6359 0.0899   6.4073   6.7687
#> 
#> TABLE 4 — ASSUMPTION CHECKS
#> --------------------------------------------------------------------------------
#>                              Check                                 Result
#>  Residual normality (Shapiro-Wilk)                               21.8864%
#>         Equal variances (Bartlett)                                0.0335%
#>              Recommended next step Consider Welch ANOVA or Kruskal-Wallis
#> 
#> TABLE 5 — TUKEY POST-HOC COMPARISONS
#> --------------------------------------------------------------------------------
#>            Comparison Difference CI_Lower CI_Upper Adj_p_value Significant
#>     versicolor-setosa      0.930   0.6862   1.1738    <0.0001%         Yes
#>      virginica-setosa      1.582   1.3382   1.8258    <0.0001%         Yes
#>  virginica-versicolor      0.652   0.4082   0.8958    <0.0001%         Yes
#> 
#> ================================================================================
#>  PLAIN-LANGUAGE INTERPRETATION
#> ================================================================================
#> 
#> ONE-WAY ANOVA Formula: Sepal.Length ~ Species
#> 
#> A one-way ANOVA revealed a highly statistically significant (p < 0.0001%)
#>   difference across the 3 groups (F(2, 147) = 119.265). The effect size
#>   (eta-squared = 0.6187) indicates a large practical significance of the
#>   group factor, meaning the grouping variable accounts for approximately
#>   61.9% of the total variance in the outcome. Post-hoc tests (e.g., Tukey
#>   HSD) are recommended to determine which specific group pairs differ
#>   significantly.
#> 
#> ================================================================================
#>