t-Tests: tadaa_t.test

On of the first implemented functions, tadaa_t.test, automagically checks for homogenity of variance via car::leveneTest, and if the resulting p-value is below .1, homogenity of variance is assumed in the following call to stats::t.test. Afterwards the effect size \(d\) is is calculated with pooled/weighted variances to ensure accuracy, and the power of the test is calculated via the pwr package (also keeping in mind whether the test is paired or not). It then formats the output according to broom::tidy, sprinkles it with pixiedust and prints either to console, markdown or whatever printing method is passed via the print argument to pixiedust::sprinkle_print_method.

tadaa_t.test(ngo, stunzahl, geschl, print = "markdown")

Table 1: Two Sample t-test with alternative hypothesis: \(\mu_1 \neq \mu_2\)

Or, if you’re having a non-parametric day, try this:

tadaa_wilcoxon(ngo, stunzahl, geschl, print = "markdown")

Table 2: Wilcoxon rank sum test with continuity correction with alternative hypothesis: \(M_1 \neq M_2\)

ANOVA: tadaa_aov

The function knows 3 types of sums of squares, adjustable via the type argument and defaulting to type 3 sums of squares. Additionally for type 3, the function also automatically checks the contrasts associated with the factor variables (only for unordered factors), and if check_names = TRUE, the contrasts are set to contr.sum.

tadaa_aov(stunzahl ~ jahrgang * geschl, data = ngo, print = "markdown")

Table 3: Two-Way ANOVA: Using Type III Sum of Squares

Or, if you’re still unsure about the parametricity of your day:

tadaa_kruskal(stunzahl ~ jahrgang, data = ngo, print = "markdown")

Table 4: Kruskal-Wallis Rank Sum Test

Pairwise tests

tadaa_pairwise_t(ngo, response = deutsch, group1 = jahrgang, p.adjust = "bonf", print = "markdown")

tadaa_pairwise_t(ngo, response = deutsch, group1 = jahrgang, group2 = geschl, p.adjust = "sidakSD", print = "markdown")

Tests for normality: tadaa_normtest

Supported methods: - ad: Anderson-Darling - shapiro: Shapiro-Wilk - pearson: Pearson’s chi-square test - ks: Kolmogorov-Smirnov

library(dplyr)
print("test")
cols <- ngo[c("deutsch", "englisch", "mathe")]

tadaa_normtest(data = cols, method = "shapiro", print = "markdown")

Summaries

• tadaa_nom: All the nominal stats in one table.
• tadaa_ord: All the ordinal stats in one table.

Intervals and recoding

• generate_recodes: To produce recode assignments for car::recode for evenly sequenced clusters.
• interval_labels: To produce labels for clusters created by cut.
• delete_na: Customizable way to drop NA observations from a dataset.
• pval_string: Shamalessly adapted from pixiedust::pvalString, this will format a p-value as a character string in common p < 0.001 notation and so on. The difference from the pixiedust version is that this function will also print p < 0.05.

Plotting helpers

• mean_ci_t: Returns a data.frame with y (mean), ymin and ymax for the CI bounds.
  • confint_t: For the underlying function to get the CI width. Returns a single value.
  • confint_norm: Similar, but baes on normal distribution. Returns a single value.
• mean_ci_sem: Standard error and CI, you guessed it, in one table.

library(ggplot2)

ggplot(data = ngo, aes(x = jahrgang, y = deutsch)) +
  stat_summary(fun.data = "mean_ci_t", geom = "errorbar") +
  theme_tadaa()

As a convenience, we added tadaa_mean_ci to quickly plot means with errorbars to get a quick glance at your data.

tadaa_mean_ci(data = ngo, response = deutsch, group = jahrgang) +
  theme_tadaa()

library(ggplot2)
if (!("cowplot" %in% installed.packages())) install.packages("cowplot")

tadaa_int(data = ngo, response = stunzahl, group1 = jahrgang, group2 = geschl, grid = T, brewer_palette = "Set1")