Package {testflow}

Create a survival object

Description

Re-exported from survival so Surv() is available after library(testflow) alone, for use in test_survival() and test_cox() formulas such as Surv(time, status) ~ group. See Surv for full documentation.

Convert a testflow object to a one-row tibble

Description

Convert a testflow object to a one-row tibble

Usage

as_tibble(x, ...)

Arguments

Value

A one-row tibble with the workflow name, design, variables, recommended test, null hypothesis, statistic, degrees of freedom when available, p-value, confidence interval when available, effect-size fields, and decision text.

Simulate a small cardiovascular teaching dataset

Description

Simulate a small cardiovascular teaching dataset

Usage

make_cardio_data(n = 180, seed = 2026)

Arguments

Value

A tibble with example numeric and categorical variables.

Plot a testflow object

Description

Plot a testflow object

Usage

## S3 method for class 'testflow'
plot(x, title = NULL, subtitle = NULL, caption = NULL, ...)

Arguments

Value

A ggplot object stored in the testflow object, with optional title, subtitle, and caption overrides applied. If the workflow was created with plot = FALSE, returns NULL.

Print a testflow object

Description

Print a testflow object.

Usage

## S3 method for class 'testflow'
print(x, ...)

Arguments

Details

Console colors are enabled by default in interactive sessions. Use options(testflow.cli_colors = FALSE) to disable colors, or options(testflow.cli_colors = TRUE) to force colors in non-interactive output.

Value

The input testflow object, invisibly. Called for its side effect of printing a formatted workflow summary to the console.

Return a ready-to-use testflow report

Description

Return a ready-to-use testflow report

Usage

report(x, ...)

Arguments

Value

A length-one character vector containing a plain-language summary of the workflow result, including the design, recommended test, p-value, effect size when reported, and null hypothesis when available.

Return a ready-to-use testflow report

Description

Return a ready-to-use testflow report

Usage

report_test(x)

Arguments

Value

A length-one character vector containing the same report text as report() for a testflow object.

Sample size planning for common trial designs

Description

Planning functions for common trial designs across continuous, binary, survival, and ordinal endpoints.

Usage

sample_size(
  endpoint = c("continuous", "binary", "survival", "ordinal"),
  design = c("parallel", "paired", "repeated"),
  objective = c("superiority", "noninferiority", "equivalence"),
  ...
)

sample_size_continuous(
  design = c("parallel", "paired", "repeated"),
  objective = c("superiority", "noninferiority", "equivalence"),
  delta = NULL,
  sd = NULL,
  sd_diff = NULL,
  expected_difference = 0,
  margin = NULL,
  alpha = 0.05,
  power = 0.9,
  allocation = 1,
  dropout = 0,
  n_time = 2,
  correlation = 0.5
)

sample_size_binary(
  design = c("parallel", "paired", "repeated"),
  objective = c("superiority", "noninferiority", "equivalence"),
  p1,
  p2,
  margin = NULL,
  method = c("pooled", "anticipated"),
  discordant_or = NULL,
  discordance_rate = NULL,
  p10 = NULL,
  p01 = NULL,
  alpha = 0.05,
  power = 0.9,
  allocation = 1,
  dropout = 0,
  n_time = 2
)

sample_size_survival(
  design = c("parallel"),
  objective = c("superiority", "noninferiority", "equivalence"),
  hr,
  margin_hr = NULL,
  lower = NULL,
  upper = NULL,
  survival_a = NULL,
  survival_b = NULL,
  alpha = 0.05,
  power = 0.9,
  allocation = 1,
  dropout = 0,
  method = c("exponential", "ph_only"),
  accrual_duration = NULL,
  follow_up = NULL
)

sample_size_ordinal(
  design = c("parallel"),
  objective = c("superiority"),
  p_superiority = NULL,
  alpha = 0.05,
  power = 0.9,
  dropout = 0
)

sample_size_adjust_dropout(n, dropout = 0)

Arguments

Details

The implementation currently covers:

• Continuous endpoints: parallel and paired superiority, non-inferiority, and equivalence, plus repeated-measures approximation for 3+ time points.

• Binary endpoints: parallel risk-difference planning, paired discordant-pair planning, non-inferiority, and equivalence.

• Survival endpoints: event-based superiority, non-inferiority, and equivalence.

• Ordinal endpoints: Noether's superiority approximation.

The helper sample_size_adjust_dropout() applies \lceil n / (1 - d) \rceil.

See the individual function help pages (e.g. ?sample_size_continuous) for the underlying formulas and their literature references.

Value

A sample_size object with the raw sample size, dropout-adjusted sample size, method label, assumptions, report text, and a plot-ready summary. Formulas and literature references are documented in the individual function help pages, not returned in the object.

References

Julious SA. Sample Sizes for Clinical Trials. Chapman & Hall/CRC; 2010.

Examples

sample_size_continuous(
  design = "paired",
  objective = "superiority",
  delta = 5,
  sd_diff = 10,
  alpha = 0.05,
  power = 0.90
)

sample_size_binary(
  design = "paired",
  objective = "superiority",
  p10 = 0.2,
  p01 = 0.1,
  alpha = 0.05,
  power = 0.90
)

sample_size_continuous(
  design = "repeated",
  n_time = 4,
  correlation = 0.5,
  objective = "superiority",
  delta = 5,
  sd_diff = 10,
  alpha = 0.05,
  power = 0.90
)

Sample size for average bioequivalence

Description

Sample size for average bioequivalence

Usage

sample_size_bioequivalence(
  design = c("crossover", "parallel"),
  gmr = 1,
  cv_within = NULL,
  cv_between = NULL,
  lower = 0.8,
  upper = 1.25,
  alpha = 0.05,
  power = 0.9,
  allocation = 1,
  dropout = 0,
  method = c("iterative_tost", "normal_approx")
)

Arguments

Details

On the log scale, with \theta_0=\log(GMR), \theta_L=\log(L), \theta_U=\log(U), and d_{BE}=\min(\theta_0-\theta_L,\ \theta_U-\theta_0) (must be positive: the anticipated GMR must lie strictly inside the bioequivalence bounds):

method = "iterative_tost" searches for the smallest n such that the exact two-one-sided-test power is at least the target:

Power(n) = \Phi\left(\frac{\theta_U-\theta_0}{SE(n)}-z_{1-\alpha}\right) + \Phi\left(\frac{\theta_0-\theta_L}{SE(n)}-z_{1-\alpha}\right) - 1

with SE(n) = \sqrt{2\sigma_w^2/n} for a crossover design (total n) or SE(n_A) = \sigma_b\sqrt{(r+1)/(rn_A)} for a parallel design (per-arm n_A, n_B=rn_A). Coefficients of variation are converted to log-scale standard deviations via \sigma = \sqrt{\log(1+CV^2)}.

method = "normal_approx" uses the closed-form approximation

n_{total} \approx \frac{2\sigma_w^2(z_{1-\beta}+z_{1-\alpha})^2}{d_{BE}^2}

(crossover) or, per arm,

n_A \approx \frac{(r+1)\sigma_b^2(z_{1-\beta}+z_{1-\alpha})^2}{rd_{BE}^2}

(parallel), switching to z_{1-\beta/2} in place of z_{1-\beta} when GMR=1 exactly, matching the analogous special case in sample_size_continuous(). iterative_tost is preferred because this approximation can meaningfully differ from the exact TOST power away from GMR=1.

Value

A sample_size object.

References

Julious SA. Sample Sizes for Clinical Trials. Chapman & Hall/CRC; 2010.

Phillips KF. Power of the two one-sided tests procedure in bioequivalence. Journal of Pharmacokinetics and Biopharmaceutics. 1990;18(2):137-144.

Adjust an individually randomized sample size for cluster randomization

Description

Adjust an individually randomized sample size for cluster randomization

Usage

sample_size_cluster_adjust(n_ind, m, rho, cv_m = NULL)

Arguments

Details

Equal cluster size:

DE = 1 + (m-1)\rho, \qquad n_{clustered} = \lceil n_{ind} \cdot DE \rceil

Unequal cluster size (approximate), with coefficient of variation CV_m:

DE \approx 1 + \left((1+CV_m^2)m - 1\right)\rho

Value

The cluster-adjusted sample size.

References

Donner A, Klar N. Design and Analysis of Cluster Randomization Trials in Health Research. Arnold; 2000.

Sample size for confidence-interval precision

Description

Sample size for confidence-interval precision

Usage

sample_size_precision(
  endpoint = c("continuous", "binary"),
  design = c("one_sample", "two_sample", "odds_ratio"),
  width,
  sd = NULL,
  p = NULL,
  p1 = NULL,
  p2 = NULL,
  alpha = 0.05,
  allocation = 1,
  dropout = 0,
  conservative = FALSE
)

Arguments

Details

Unlike the other ⁠sample_size_*()⁠ functions, precision-based planning is driven by a target confidence-interval half-width w rather than power against an effect size.

Continuous, one sample:

n = \left(\frac{z_{1-\alpha/2}\sigma}{w}\right)^2

Continuous, two independent means:

n_A = \frac{(r+1)\sigma^2z_{1-\alpha/2}^2}{rw^2}

Binary, one proportion:

n = \frac{z_{1-\alpha/2}^2p(1-p)}{w^2}

(or n = z_{1-\alpha/2}^2/(4w^2) for the conservative p=0.5 case)

Binary, two independent proportions (equal allocation):

n_{per\ group} = \frac{z_{1-\alpha/2}^2\left[p_A(1-p_A)+p_B(1-p_B)\right]}{w^2}

Binary, log-odds-ratio half-width:

n_A = \frac{z_{1-\alpha/2}^2}{w^2} \left[\frac1{p_A}+\frac1{1-p_A}+\frac1{rp_B}+\frac1{r(1-p_B)}\right]

Value

A sample_size object.

References

Julious SA. Sample Sizes for Clinical Trials. Chapman & Hall/CRC; 2010.

Summarize a testflow object

Description

Summarize a testflow object.

Usage

## S3 method for class 'testflow'
summary(object, ...)

Arguments

Details

Console colors follow the same testflow.cli_colors option used by print.testflow().

Value

A summary.testflow list containing the workflow metadata, descriptives, assumptions, recommended test, primary and alternative test results, post-hoc results when available, effect size, decision, and report text.

Build a compact descriptive summary table

Description

Builds a compact table of descriptive summaries using a formula interface. Numeric variables are summarized as mean (SD); median [Q1, Q3]; n. Categorical variables are summarized as n (percent).

Usage

sumtab(
  formula,
  data,
  p_value = FALSE,
  overall = TRUE,
  digits = 1,
  p_digits = 3,
  alpha = 0.05,
  fisher_threshold = 5,
  na.rm = TRUE
)

Arguments

Details

When p_value = TRUE and a grouping variable is supplied, sumtab() chooses the p-value test automatically. Numeric variables use Student t-test, Welch t-test, or Wilcoxon rank-sum test for two groups, and one-way ANOVA, Welch ANOVA, or Kruskal-Wallis test for more than two groups. Categorical variables use a chi-square test unless expected counts fall below fisher_threshold, in which case Fisher's exact test is used.

Value

A tibble with one row per numeric variable and one row per categorical level.

Examples

dat <- make_cardio_data(80, seed = 1)
sumtab(~ age + sex | treatment, dat, p_value = TRUE)

Inter-rater agreement workflow (Cohen's kappa)

Description

Inter-rater agreement workflow (Cohen's kappa)

Usage

test_agreement(data, rater1, rater2, alpha = 0.05, plot = TRUE, na.rm = TRUE)

Arguments

Details

With observed agreement p_o (the proportion of subjects on which both raters agree) and chance-expected agreement p_e (from the marginal category proportions):

\kappa = \frac{p_o - p_e}{1 - p_e}

The standard error uses the large-sample formula of Fleiss, Cohen & Everitt (1969), which accounts for the full marginal covariance structure (not the simpler \sqrt{p_o(1-p_o)}/[(1-p_e)\sqrt n] approximation sometimes seen, which understates the variance when categories are unevenly distributed).

Value

A testflow object with class testflow_agreement. The object is a list containing the cleaned data, categorical descriptives, the test of kappa against 0 as the primary result with null hypothesis, the agreement (confusion) table, Cohen's kappa as the effect size, optional ggplot, original call, and report text.

References

Cohen J. A coefficient of agreement for nominal scales. Educational and Psychological Measurement. 1960;20(1):37-46.

Fleiss JL, Cohen J, Everitt BS. Large sample standard errors of kappa and weighted kappa. Psychological Bulletin. 1969;72(5):323-327.

Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33(1):159-174. (Magnitude convention.)

Test association between two categorical variables

Description

Test association between two categorical variables

Usage

test_categorical(
  formula,
  data,
  y = NULL,
  alpha = 0.05,
  fisher_threshold = 5,
  plot = TRUE,
  na.rm = TRUE
)

Arguments

Value

A testflow object with class testflow_categorical. The object is a list containing the cleaned data, categorical descriptives, assumption checks, recommended association test, primary test result with null hypothesis, alternative chi-square and Fisher results, effect size, optional ggplot, original call, and report text.

References

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. Philosophical Magazine, 50(302), 157-175.

Fisher, R. A. (1922). On the interpretation of chi-square from contingency tables, and the calculation of P. Journal of the Royal Statistical Society, 85(1), 87-94.

Cramer, H. (1946). Mathematical Methods of Statistics. Princeton.

Test correlation between two numeric variables

Description

Test correlation between two numeric variables

Usage

test_correlation(
  formula,
  data,
  y = NULL,
  method = c("auto", "pearson", "spearman", "kendall"),
  alpha = 0.05,
  plot = TRUE,
  na.rm = TRUE
)

Arguments

Value

A testflow object with class testflow_correlation. The object is a list containing the cleaned complete-case data, numeric descriptives, assumption checks, recommended correlation method, primary correlation test with null hypothesis, Pearson/Spearman/Kendall results, a correlation table, effect size, optional ggplot, original call, and report text.

References

Pearson, K. (1895). Notes on regression and inheritance in the case of two parents. Proceedings of the Royal Society of London, 58, 240-242.

Spearman, C. (1904). The proof and measurement of association between two things. The American Journal of Psychology, 15(1), 72-101.

Kendall, M. G. (1938). A new measure of rank correlation. Biometrika, 30(1/2), 81-93.

Test a correlation matrix

Description

Test a correlation matrix

Usage

test_correlation_matrix(
  data,
  vars,
  method = c("spearman", "pearson", "kendall"),
  alpha = 0.05,
  plot = TRUE,
  na.rm = TRUE
)

Arguments

Value

A testflow object with class testflow_correlation_matrix. The object is a list containing the cleaned data, numeric descriptives, screening assumptions, selected correlation-matrix method, pairwise correlation matrix, pairwise p-value table, maximum absolute correlation as an effect-size summary, optional heatmap ggplot, original call, and report text.

Cox proportional hazards regression workflow

Description

Cox proportional hazards regression workflow

Usage

test_cox(formula, data, alpha = 0.05, plot = TRUE, na.rm = TRUE)

Arguments

Details

With predictors x_1, \ldots, x_p, the Cox model leaves the baseline hazard h_0(t) unspecified:

h(t \mid x) = h_0(t)\exp(\beta_1 x_1 + \cdots + \beta_p x_p)

Coefficients are reported as hazard ratios HR_j = e^{\beta_j}. The overall model test is the likelihood-ratio test against the model with no predictors, on p degrees of freedom (or more, for multi-level factor terms).

The proportional-hazards assumption is checked via the correlation between scaled Schoenfeld residuals and time (Grambsch & Therneau, 1994); a significant test suggests a predictor's effect on the hazard changes over time.

Value

A testflow object with class testflow_cox. The object is a list containing the cleaned data, descriptive statistics, a proportional- hazards assumption check, the overall likelihood-ratio test as the primary result with null hypothesis, the per-term hazard-ratio coefficient table, the concordance index as the effect size, optional ggplot, original call, and report text.

References

Cox DR. Regression models and life-tables. Journal of the Royal Statistical Society: Series B. 1972;34(2):187-202.

Grambsch PM, Therneau TM. Proportional hazards tests and diagnostics based on weighted residuals. Biometrika. 1994;81(3):515-526.

Diagnostic test accuracy workflow

Description

Diagnostic test accuracy workflow

Usage

test_diagnostic(data, test, reference, alpha = 0.05, plot = TRUE, na.rm = TRUE)

Arguments

Details

With true positives TP, false positives FP, false negatives FN, and true negatives TN:

Sensitivity = \frac{TP}{TP+FN}, \qquad Specificity = \frac{TN}{TN+FP}

PPV = \frac{TP}{TP+FP}, \qquad NPV = \frac{TN}{TN+FN}

LR^+ = \frac{Sensitivity}{1-Specificity}, \qquad LR^- = \frac{1-Sensitivity}{Specificity}

Sensitivity, specificity, PPV, NPV, and accuracy each get an exact (Clopper-Pearson) confidence interval. Likelihood ratios get the closed-form log-scale interval of Simel, Samsa & Matchar (1991):

SE[\log LR^+] = \sqrt{\frac1{TP}-\frac1{TP+FN}+\frac1{FP}-\frac1{FP+TN}}

The primary test compares overall accuracy to the no-information rate (the larger of the two reference-class proportions) via an exact binomial test, following the same convention as caret::confusionMatrix().

Value

A testflow object with class testflow_diagnostic. The object is a list containing the cleaned data, categorical descriptives, the accuracy-vs-no-information-rate test as the primary result with null hypothesis, a table of sensitivity/specificity/predictive values/ likelihood ratios with confidence intervals, the confusion matrix, accuracy as the effect size, optional ggplot, original call, and report text.

References

Simel DL, Samsa GP, Matchar DB. Likelihood ratios with confidence: sample size estimation for diagnostic test studies. Journal of Clinical Epidemiology. 1991;44(8):763-770.

Altman DG, Bland JM. Diagnostic tests 1: sensitivity and specificity. BMJ. 1994;308(6943):1552.

Run a factorial ANOVA workflow

Description

Run a factorial ANOVA workflow

Usage

test_factorial(
  formula,
  data,
  factors = NULL,
  alpha = 0.05,
  type = 2,
  plot = TRUE,
  na.rm = TRUE
)

Arguments

Value

A testflow object with class testflow_factorial. The object is a list containing the cleaned data, descriptive statistics, residual and variance assumption checks, recommended factorial ANOVA, primary ANOVA term result with null hypothesis, ANOVA table, effect size, optional ggplot, original call, and report text.

References

Fisher, R. A. (1925). Statistical Methods for Research Workers. Oliver and Boyd.

Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences (2nd ed.). Lawrence Erlbaum.

Compare a numeric outcome across more than two groups

Description

Compare a numeric outcome across more than two groups

Usage

test_groups(
  formula,
  data,
  group = NULL,
  alpha = 0.05,
  posthoc = TRUE,
  plot = TRUE,
  na.rm = TRUE
)

Arguments

Value

A testflow object with class testflow_groups. The object is a list containing the cleaned data, grouped descriptive statistics, assumption checks, recommended omnibus test, primary test result with null hypothesis, alternative omnibus results, post-hoc comparisons when requested, effect size, optional ggplot, original call, and report text.

Intraclass correlation coefficient workflow

Description

Intraclass correlation coefficient workflow

Usage

test_icc(data, measures, alpha = 0.05, plot = TRUE, na.rm = TRUE)

Arguments

Details

Following Shrout & Fleiss (1979), variance components come from one-way (BMS, WMS) and two-way (BMS, JMS, EMS) ANOVA decompositions of the n subjects by k raters:

ICC(1,1) = \frac{BMS-WMS}{BMS+(k-1)WMS}

ICC(2,1) = \frac{BMS-EMS}{BMS+(k-1)EMS+\frac{k}{n}(JMS-EMS)} \quad\text{(two-way random, absolute agreement)}

ICC(3,1) = \frac{BMS-EMS}{BMS+(k-1)EMS} \quad\text{(two-way fixed, consistency)}

ICC(2,1) (absolute agreement, allowing raters to differ systematically) is reported as the primary/effect-size estimate, following the single-measurement, absolute-agreement, two-way random-effects recommendation of Koo & Li (2016) for typical reliability studies. Confidence intervals use the F-distribution-based formulas of McGraw & Wong (1996); ICC(2,1)'s interval uses a Satterthwaite-approximated denominator degrees of freedom.

Value

A testflow object with class testflow_icc. The object is a list containing the cleaned long-format data, descriptive statistics, the F-test for ICC(2,1) as the primary result with null hypothesis, a table of the one-way (ICC1), two-way random/absolute-agreement (ICC2), and two-way fixed/consistency (ICC3) estimates with confidence intervals, ICC(2,1) as the effect size, optional ggplot, original call, and report text.

References

Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychological Bulletin. 1979;86(2):420-428.

McGraw KO, Wong SP. Forming inferences about some intraclass correlation coefficients. Psychological Methods. 1996;1(1):30-46.

Koo TK, Li MY. A guideline of selecting and reporting intraclass correlation coefficients for reliability research. Journal of Chiropractic Medicine. 2016;15(2):155-163.

Multiple linear regression workflow

Description

Multiple linear regression workflow

Usage

test_linear_regression(formula, data, alpha = 0.05, plot = TRUE, na.rm = TRUE)

Arguments

Details

For predictors x_1, \ldots, x_p and n observations, ordinary least squares fits:

y_i = \beta_0 + \beta_1 x_{1i} + \cdots + \beta_p x_{pi} + \varepsilon_i

The overall model test compares the fitted model to an intercept-only model:

F = \frac{(SS_{total}-SS_{res})/p}{SS_{res}/(n-p-1)} \sim F_{p,\,n-p-1}

R^2 = 1 - \frac{SS_{res}}{SS_{total}}, \qquad R^2_{adj} = 1 - (1-R^2)\frac{n-1}{n-p-1}

Value

A testflow object with class testflow_linear_regression. The object is a list containing the cleaned data, descriptive statistics, residual/homoscedasticity/multicollinearity assumption checks, the overall model F-test as the primary result with null hypothesis, the per-term coefficient table, R squared and adjusted R squared as the effect size, optional ggplot, original call, and report text.

References

Draper NR, Smith H. Applied Regression Analysis (3rd ed.). Wiley; 1998.

Logistic regression workflow

Description

Logistic regression workflow

Usage

test_logistic_regression(
  formula,
  data,
  alpha = 0.05,
  plot = TRUE,
  na.rm = TRUE
)

Arguments

Details

With outcome y_i \in \{0,1\} and predictors x_1, \ldots, x_p:

\log\frac{P(y_i=1)}{1-P(y_i=1)} = \beta_0 + \beta_1 x_{1i} + \cdots + \beta_p x_{pi}

Coefficients are reported on the log-odds scale and, exponentiated, as odds ratios OR_j = e^{\beta_j}.

The overall model test is the likelihood-ratio test against the intercept-only model, using each model's residual deviance D = -2\log L:

LR = D_{null} - D_{model} \sim \chi^2_{p}

McFadden's pseudo R squared:

R^2_{McFadden} = 1 - \frac{\log L_{model}}{\log L_{null}}

Value

A testflow object with class testflow_logistic_regression. The object is a list containing the cleaned data, descriptive statistics, multicollinearity/influential-point assumption checks, the likelihood-ratio test as the primary result with null hypothesis, the per-term coefficient tables on the log-odds and odds-ratio scale, McFadden's pseudo R squared as the effect size, optional ggplot, original call, and report text.

References

Hosmer DW, Lemeshow S, Sturdivant RX. Applied Logistic Regression (3rd ed.). Wiley; 2013.

Test multinomial goodness of fit

Description

Test multinomial goodness of fit

Usage

test_multinomial(
  data,
  outcome,
  p = NULL,
  alpha = 0.05,
  plot = TRUE,
  na.rm = TRUE
)

Arguments

Value

A testflow object with class testflow_multinomial. The object is a list containing the cleaned data, categorical descriptives, assumption checks, recommended goodness-of-fit test, primary chi-square result with null hypothesis, BH-adjusted pairwise binomial checks, effect-size summary, optional ggplot, original call, and report text.

References

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. Philosophical Magazine, 50(302), 157-175.

Test one numeric sample against a reference value

Description

Test one numeric sample against a reference value

Usage

test_one_sample(
  data,
  outcome,
  mu = 0,
  alternative = c("two.sided", "less", "greater"),
  alpha = 0.05,
  plot = TRUE,
  na.rm = TRUE
)

Arguments

Value

A testflow object with class testflow_one_sample. The object is a list containing the cleaned data, descriptive statistics, assumption checks, recommended test, primary test result with null hypothesis, alternative test results, effect size, optional ggplot, original call, and report text.

References

Gosset, W. S. (1908). The probable error of a mean. Biometrika, 6(1), 1-25.

Wilcoxon, F. (1945). Individual comparisons by ranking methods. Biometrics Bulletin, 1(6), 80-83.

Detect numeric outliers

Description

Detect numeric outliers

Usage

test_outliers(
  formula,
  data,
  group = NULL,
  method = c("iqr", "mahalanobis", "both"),
  plot = TRUE,
  na.rm = TRUE
)

Arguments

Value

A testflow object with class testflow_outliers. The object is a list containing the cleaned data, numeric descriptives, screening assumptions, selected outlier-screening method, flagged IQR and/or Mahalanobis rows, outlier-count summary, optional ggplot, original call, and report text. This is a screening workflow, not a single hypothesis test.

Compare paired before and after numeric measurements

Description

Compare paired before and after numeric measurements

Usage

test_paired(
  formula,
  data,
  after = NULL,
  alternative = c("two.sided", "less", "greater"),
  alpha = 0.05,
  plot = TRUE,
  na.rm = TRUE
)

Arguments

Value

A testflow object with class testflow_paired. The object is a list containing the cleaned paired data, paired difference, descriptive statistics, assumption checks, recommended paired test, primary test result with null hypothesis, alternative test results, effect size, optional ggplot, original call, and report text.

Test paired categorical measurements

Description

Test paired categorical measurements

Usage

test_paired_categorical(
  data,
  before,
  after,
  alpha = 0.05,
  plot = TRUE,
  na.rm = TRUE
)

Arguments

Value

A testflow object with class testflow_paired_categorical. The object is a list containing the cleaned paired categorical data, categorical descriptives, assumption checks, McNemar test result, discordant-pair table, optional ggplot, original call, and report text.

Test a one-sample proportion

Description

Test a one-sample proportion

Usage

test_proportion(
  data,
  outcome,
  success,
  p = 0.5,
  alpha = 0.05,
  plot = TRUE,
  na.rm = TRUE
)

Arguments

Value

A testflow object with class testflow_proportion. The object is a list containing the cleaned data, categorical descriptives, assumption checks, recommended exact or approximate one-sample proportion test, primary test result with null hypothesis, alternative exact and approximate results, observed proportion summary, optional ggplot, original call, and report text.

References

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.

Run a repeated-measures workflow from wide data

Description

Run a repeated-measures workflow from wide data

Usage

test_repeated(
  data,
  measures,
  id = NULL,
  between = NULL,
  alpha = 0.05,
  plot = TRUE,
  na.rm = TRUE
)

Arguments

Value

A testflow object with class testflow_repeated. The object is a list containing long-form repeated-measures data, numeric descriptives, assumption checks, recommended repeated-measures ANOVA or Friedman test, primary test result with null hypothesis, alternative repeated-measures results, post-hoc paired comparisons, effect size, optional ggplot, original call, and report text.

References

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.

Wilcoxon, F. (1945). Individual comparisons by ranking methods. Biometrics Bulletin, 1(6), 80-83.

Girden, E. R. (1992). ANOVA: Repeated Measures. Sage.

Test repeated categorical measurements

Description

Test repeated categorical measurements

Usage

test_repeated_categorical(
  data,
  measures,
  id = NULL,
  alpha = 0.05,
  plot = TRUE,
  na.rm = TRUE
)

Arguments

Value

A testflow object with class testflow_repeated_categorical. The object is a list containing the cleaned repeated categorical data, descriptive counts, assumption checks, Cochran Q test result with null hypothesis, pairwise McNemar post-hoc comparisons, effect size, optional ggplot, original call, and report text.

References

Cochran, W. G. (1950). The comparison of percentages in matched samples. Biometrika, 37(3/4), 256-266.

McNemar, Q. (1947). Note on the sampling error of the difference between correlated proportions or percentages. Psychometrika, 12(2), 153-157.

Run a repeated-measures workflow from long data

Description

Run a repeated-measures workflow from long data

Usage

test_repeated_long(
  data,
  outcome,
  within,
  id,
  between = NULL,
  alpha = 0.05,
  plot = TRUE,
  na.rm = TRUE
)

Arguments

Value

A testflow object with class testflow_repeated. The object is a list containing the cleaned long-format data, numeric descriptives, assumption checks, recommended repeated-measures ANOVA or Friedman test, primary test result with null hypothesis, alternative repeated-measures results, post-hoc paired comparisons, effect size, optional ggplot, original call, and report text.

Receiver operating characteristic (ROC) curve workflow

Description

Receiver operating characteristic (ROC) curve workflow

Usage

test_roc(data, predictor, outcome, alpha = 0.05, plot = TRUE, na.rm = TRUE)

Arguments

Details

The area under the ROC curve equals the probability that a randomly chosen positive-class observation has a higher predictor value than a randomly chosen negative-class observation, and is computed from the Mann-Whitney U statistic:

AUC = \frac{U}{n_1 n_0}

The standard error uses the closed-form Hanley & McNeil (1982) formula (an asymptotically equivalent, but not numerically identical, alternative to DeLong's method):

SE(AUC) = \sqrt{\frac{AUC(1-AUC)+(n_1-1)(Q_1-AUC^2)+(n_0-1)(Q_2-AUC^2)}{n_1 n_0}}

Q_1 = \frac{AUC}{2-AUC}, \qquad Q_2 = \frac{2AUC^2}{1+AUC}

The Youden's J statistic identifies the threshold maximizing J = Sensitivity + Specificity - 1.

Value

A testflow object with class testflow_roc. The object is a list containing the cleaned data, descriptive statistics by outcome class, the test of AUC against 0.5 as the primary result with null hypothesis, the full ROC curve and the Youden's-J-optimal threshold, AUC as the effect size, optional ggplot, original call, and report text.

References

Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology. 1982;143(1):29-36.

Youden WJ. Index for rating diagnostic tests. Cancer. 1950;3(1):32-35.

Hosmer DW, Lemeshow S, Sturdivant RX. Applied Logistic Regression (3rd ed.). Wiley; 2013. (AUC magnitude convention, p. 177.)

Kaplan-Meier and log-rank survival workflow

Description

Kaplan-Meier and log-rank survival workflow

Usage

test_survival(formula, data, alpha = 0.05, plot = TRUE, na.rm = TRUE)

Arguments

Details

For groups A and B, the log-rank test compares observed to expected events under the null of equal survival distributions:

\chi^2 = \sum_{g \in \{A,B\}} \frac{(O_g - E_g)^2}{V_g} \sim \chi^2_1

The companion effect size is the hazard ratio from a univariate Cox model on the same grouping factor, HR = e^{\hat\beta}, reported with its Wald confidence interval; the hazard ratio is not itself part of the log-rank test and can disagree with it under strongly non-proportional hazards.

Value

A testflow object with class testflow_survival. The object is a list containing the cleaned data, per-group descriptive survival statistics, the log-rank test as the primary result with null hypothesis, the Kaplan-Meier curve data, a companion univariate hazard ratio as the effect size, optional ggplot, original call, and report text.

References

Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association. 1958;53(282):457-481.

Mantel N. Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemotherapy Reports. 1966;50(3):163-170.

Compare a numeric outcome between two independent groups

Description

Compare a numeric outcome between two independent groups

Usage

test_two_groups(
  formula,
  data,
  group = NULL,
  alternative = c("two.sided", "less", "greater"),
  alpha = 0.05,
  plot = TRUE,
  na.rm = TRUE
)

Arguments

Value

A testflow object with class testflow_two_groups. The object is a list containing the cleaned data, descriptive statistics by group, assumption checks, recommended test, primary test result with null hypothesis, alternative test results, effect size, optional ggplot, original call, and report text.