Current development release: 0.8.2. First public
release: 0.7.0.
testflow helps choose, run, and report common
statistical tests from the study design. The package returns a
consistent object with assumptions, the recommended test, the null
hypothesis, the result, an effect size when appropriate, a
plain-language report, and an optional plot.
The guiding idea is:
testflow = design + assumptions + test + effect size + report + plottestflow is an R package for teaching and applied
statistical workflows. It keeps the statistical question organized
around the study design, then combines assumption checks, test
selection, effect sizes, visual diagnostics, and ready-to-use reporting
in one object.
It is designed for analysts, students, and instructors who want a reproducible path from a data question to a statistical test, a plot, and a concise written interpretation.
testflow is not on CRAN yet. Install the development
version from GitHub:
install.packages("remotes")
remotes::install_github("ielbadisy/testflow")library(testflow)
cardio <- make_cardio_data()
res <- test_two_groups(sbp_3m ~ sex, data = cardio, plot = TRUE)
res
#> Statistical test workflow
#>
#> Outcome: sbp_3m
#> Group: sex
#> Design: two independent groups
#>
#> Assumptions
#> * Independence of observations: assumed: Assumed from study design.
#> * Normality: sbp_3m (female): acceptable: Approximate normality looks reasonable. (method=Shapiro-Wilk; statistic=0.99; p=0.913)
#> * Normality: sbp_3m (male): acceptable: Approximate normality looks reasonable. (method=Shapiro-Wilk; statistic=0.98; p=0.233)
#> * Variance homogeneity: acceptable: Variance homogeneity looks reasonable. (method=Levene test; statistic=1.57)
#> * Extreme outliers: warning: 4 potential outlier(s) flagged by IQR. (IQR rule, n = 4)
#> * Variance ratio check: acceptable: Variance ratio looks reasonable. (statistic=1.27)
#>
#> Recommended test
#> Student independent t-test
#>
#> Result
#> H0: the population mean or location of sbp_3m is equal across levels of sex.
#> statistic = -1.91, df = 178.00, p = 0.058, 95% CI [-11.22, 0.18]
#>
#> Effect size
#> Cohen's d: -0.29, small
#>
#> Report
#> The two independent groups workflow for sbp_3m did not show a statistically significant result using Student independent t-test, statistic = -1.91, df = 178.00, p = 0.058. The 95% confidence interval was [-11.22, 0.18]. The effect size was small (Cohen's d = -0.29). H0: the population mean or location of sbp_3m is equal across levels of sex.report(res)
#> [1] "The two independent groups workflow for sbp_3m did not show a statistically significant result using Student independent t-test, statistic = -1.91, df = 178.00, p = 0.058. The 95% confidence interval was [-11.22, 0.18]. The effect size was small (Cohen's d = -0.29). H0: the population mean or location of sbp_3m is equal across levels of sex."
as_tibble(res)
#> # A tibble: 1 × 15
#> workflow design outcome group recommended_test null_hypothesis statistic df
#> <chr> <chr> <chr> <chr> <chr> <chr> <dbl> <dbl>
#> 1 two_gro… two i… sbp_3m sex Student indepen… H0: the popula… -1.91 178
#> # ℹ 7 more variables: p <dbl>, conf.low <dbl>, conf.high <dbl>,
#> # effect_size <dbl>, effect_size_name <chr>, effect_size_magnitude <chr>,
#> # decision <chr>plot(res)
Printed results use cli styling in interactive R
sessions. GitHub strips terminal colors, but the structure is the same
in the console. To control colors explicitly, use:
options(testflow.cli_colors = FALSE)
options(testflow.cli_colors = TRUE)A typical analysis starts with the study question, lets
testflow check the assumptions, then uses the returned
object for the report, tidy result, and plot.
paired <- test_paired(sbp_3m ~ sbp_baseline, data = cardio, plot = TRUE)
paired
#> Statistical test workflow
#>
#> Outcome: sbp_3m - sbp_baseline
#> Design: paired measurements
#>
#> Assumptions
#> * Independence of observations: assumed: Paired observations from the same subjects are assumed by design.
#> * Normality: diff: acceptable: Approximate normality looks reasonable. (method=Shapiro-Wilk; statistic=0.99; p=0.557)
#> * Symmetry of paired differences: not checked: Normality made the symmetry check unnecessary.
#> * Extreme outliers: warning: 1 potential outlier(s) flagged by IQR. (IQR rule, n = 1)
#>
#> Recommended test
#> Paired t-test
#>
#> Result
#> H0: the mean or median paired difference (sbp_3m - sbp_baseline) equals 0.
#> statistic = -9.20, df = 179.00, p = <0.001, 95% CI [-9.53, -6.16]
#>
#> Effect size
#> Cohen's dz: -0.69, moderate
#>
#> Report
#> The paired measurements workflow for sbp_3m - sbp_baseline showed a statistically significant result using Paired t-test, statistic = -9.20, df = 179.00, p = <0.001. The 95% confidence interval was [-9.53, -6.16]. The effect size was moderate (Cohen's dz = -0.69). H0: the mean or median paired difference (sbp_3m - sbp_baseline) equals 0.
report(paired)
#> [1] "The paired measurements workflow for sbp_3m - sbp_baseline showed a statistically significant result using Paired t-test, statistic = -9.20, df = 179.00, p = <0.001. The 95% confidence interval was [-9.53, -6.16]. The effect size was moderate (Cohen's dz = -0.69). H0: the mean or median paired difference (sbp_3m - sbp_baseline) equals 0."
as_tibble(paired)
#> # A tibble: 1 × 15
#> workflow design outcome group recommended_test null_hypothesis statistic df
#> <chr> <chr> <chr> <chr> <chr> <chr> <dbl> <dbl>
#> 1 paired paire… sbp_3m… <NA> Paired t-test H0: the mean o… -9.20 179
#> # ℹ 7 more variables: p <dbl>, conf.low <dbl>, conf.high <dbl>,
#> # effect_size <dbl>, effect_size_name <chr>, effect_size_magnitude <chr>,
#> # decision <chr>plot(paired)
Different designs produce different visual summaries. These examples use the same teaching dataset so the function call, statistical result, and plot stay together.
groups <- test_groups(sbp_3m ~ treatment, data = cardio, plot = TRUE)
repeated <- test_repeated(cardio, c(sbp_baseline, sbp_3m, sbp_6m), id = id, plot = TRUE)
correlation <- test_correlation(sbp_3m ~ age, data = cardio, plot = TRUE)
cor_matrix <- suppressWarnings(test_correlation_matrix(cardio, c(age, sbp_3m, ldl), plot = TRUE))
categorical <- test_categorical(treatment ~ controlled_3m, data = cardio, plot = TRUE)
outliers <- suppressWarnings(test_outliers(c(sbp_3m, ldl, crp), data = cardio, plot = TRUE))plot(groups)
You can keep the default plot style while changing labels:
plot(
groups,
title = "Systolic blood pressure by treatment",
subtitle = "Automatically selected test and p-value are still available in the result object",
caption = "Teaching dataset generated by make_cardio_data()"
)plot(repeated)
plot(correlation)
Cell labels show the pairwise correlation coefficient
(r).
plot(cor_matrix)
plot(categorical)
For IQR screening, points below Q1 - 1.5 x IQR or above
Q3 + 1.5 x IQR are flagged. The dashed lines show those
fences and red points are detected outliers.
plot(outliers)
sumtab() creates descriptive tables and can add
automatically selected p-values based on variable type and grouping
structure.
sumtab(~ age + sex + sbp_3m | treatment, cardio, p_value = TRUE)
#> # A tibble: 4 × 8
#> variable level `Overall (n = 180)` `usual care (n = 55)` `lifestyle (n = 71)`
#> <chr> <chr> <chr> <chr> <chr>
#> 1 age <NA> 57.9 (10.7); 58.0 … 56.3 (10.0); 56.0 [5… 59.4 (10.4); 60.0 […
#> 2 sex female 96 (53.3%) 32 (58.2%) 39 (54.9%)
#> 3 sex male 84 (46.7%) 23 (41.8%) 32 (45.1%)
#> 4 sbp_3m <NA> 139.3 (19.5); 139.… 144.1 (19.3); 146.0 … 140.1 (18.2); 139.4…
#> # ℹ 3 more variables: `medication (n = 54)` <chr>, p.value <chr>, test <chr>Use formulas when there is an outcome and a grouping, predictor, or repeated measure structure. Use tidyselect-style column selection for multi-column workflows such as repeated measures, correlation matrices, and outlier screens.
test_one_sample(cardio, sbp_3m, mu = 140)
test_two_groups(sbp_3m ~ sex, data = cardio)
test_paired(sbp_3m ~ sbp_baseline, data = cardio)
test_groups(sbp_3m ~ treatment, data = cardio)
test_factorial(sbp_3m ~ sex * treatment, data = cardio)
test_repeated(cardio, c(sbp_baseline, sbp_3m, sbp_6m), id = id)
test_categorical(treatment ~ controlled_3m, data = cardio)
test_paired_categorical(cardio, controlled_baseline, controlled_3m)
test_repeated_categorical(cardio, c(controlled_baseline, controlled_3m, controlled_6m))
test_proportion(cardio, controlled_3m, success = "yes", p = 0.5)
test_multinomial(cardio, treatment)
test_correlation(sbp_3m ~ age, data = cardio)
test_correlation_matrix(cardio, c(age, sbp_3m, ldl))
test_outliers(c(sbp_3m, ldl, crp), data = cardio)
sumtab(~ age + sex + sbp_3m | treatment, cardio, p_value = TRUE)| Study design | Main function | Tests considered |
|---|---|---|
| Summary table | sumtab() |
Student t-test, Welch t-test, Wilcoxon rank-sum, ANOVA, Welch ANOVA, Kruskal-Wallis, chi-square, Fisher exact |
| One numeric sample | test_one_sample() |
one-sample t-test, Wilcoxon signed-rank, sign test |
| Two independent groups | test_two_groups() |
Student t-test, Welch t-test, Wilcoxon rank-sum |
| Paired numeric measurements | test_paired() |
paired t-test, Wilcoxon signed-rank, sign test |
| More than two groups | test_groups() |
one-way ANOVA, Welch ANOVA, Kruskal-Wallis, post hoc comparisons |
| Factorial numeric design | test_factorial() |
factorial ANOVA with main effects and interactions |
| Repeated numeric measurements | test_repeated(), test_repeated_long() |
repeated-measures ANOVA, Friedman test, paired post hoc tests |
| Two categorical variables | test_categorical() |
chi-square independence test, Fisher exact test |
| Paired categorical measurements | test_paired_categorical() |
McNemar test |
| Repeated categorical measurements | test_repeated_categorical() |
Cochran Q test, pairwise McNemar tests |
| One proportion | test_proportion() |
exact binomial test, one-sample proportion test |
| Multinomial categories | test_multinomial() |
chi-square goodness-of-fit, pairwise binomial checks |
| Two numeric variables | test_correlation() |
Pearson, Spearman, Kendall |
| Correlation matrix | test_correlation_matrix() |
pairwise Pearson, Spearman, or Kendall correlations |
| Outlier screening | test_outliers() |
IQR rule, Mahalanobis distance |
Effect-size formulas are documented in
vignettes/effect-size-formulas.Rmd.
Bartlett, M. S. (1937). Properties of sufficiency and statistical tests. Proceedings of the Royal Society of London. Series A, Mathematical and Physical Sciences, 160(901), 268-282.
Clopper, C. J., & Pearson, E. S. (1934). The use of confidence or fiducial limits illustrated in the case of the binomial. Biometrika, 26(4), 404-413.
Cochran, W. G. (1950). The comparison of percentages in matched samples. Biometrika, 37(3/4), 256-266.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Lawrence Erlbaum Associates.
Cramer, H. (1946). Mathematical methods of statistics. Princeton University Press.
Dunn, O. J. (1964). Multiple comparisons using rank sums. Technometrics, 6(3), 241-252.
Fisher, R. A. (1925). Statistical methods for research workers. Oliver and Boyd.
Friedman, M. (1937). The use of ranks to avoid the assumption of normality implicit in the analysis of variance. Journal of the American Statistical Association, 32(200), 675-701.
Kendall, M. G. (1938). A new measure of rank correlation. Biometrika, 30(1/2), 81-93.
Kruskal, W. H., & Wallis, W. A. (1952). Use of ranks in one-criterion variance analysis. Journal of the American Statistical Association, 47(260), 583-621.
Levene, H. (1960). Robust tests for equality of variances. In I. Olkin, S. G. Ghurye, W. Hoeffding, W. G. Madow, & H. B. Mann (Eds.), Contributions to probability and statistics: Essays in honor of Harold Hotelling (pp. 278-292). Stanford University Press.
Mahalanobis, P. C. (1936). On the generalized distance in statistics. Proceedings of the National Institute of Sciences of India, 2(1), 49-55.
Mann, H. B., & Whitney, D. R. (1947). On a test of whether one of two random variables is stochastically larger than the other. The Annals of Mathematical Statistics, 18(1), 50-60.
McNemar, Q. (1947). Note on the sampling error of the difference between correlated proportions or percentages. Psychometrika, 12(2), 153-157.
Pearson, K. (1895). Notes on regression and inheritance in the case of two parents. Proceedings of the Royal Society of London, 58, 240-242.
Pearson, K. (1900). On the criterion that a given system of deviations from the probable in the case of a correlated system of variables is such that it can be reasonably supposed to have arisen from random sampling. The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science, 50(302), 157-175.
Shapiro, S. S., & Wilk, M. B. (1965). An analysis of variance test for normality (complete samples). Biometrika, 52(3/4), 591-611.
Spearman, C. (1904). The proof and measurement of association between two things. The American Journal of Psychology, 15(1), 72-101.
Student. (1908). The probable error of a mean. Biometrika, 6(1), 1-25.
Tukey, J. W. (1949). Comparing individual means in the analysis of variance. Biometrics, 5(2), 99-114.
Welch, B. L. (1947). The generalization of Student’s problem when several different population variances are involved. Biometrika, 34(1/2), 28-35.
Wilcoxon, F. (1945). Individual comparisons by ranking methods. Biometrics Bulletin, 1(6), 80-83.