15. APA figures
statAPA produces publication-ready statistical
tables formatted according to the 7th edition of the American
Psychological Association (APA) style guidelines. All functions return
an invisible list so results can be stored, piped into
apa_to_flextable() for Word export, or left to print to the
console via message().
result <- apa_descriptives(
mtcars,
vars = c("mpg", "wt", "hp"),
group = "cyl"
)
#> Descriptive Statistics by cyl
#>
#> Variable 6: M (SD) 4: M (SD) 8: M (SD) Total: M (SD) N
#> mpg 26.66 (4.51) 19.74 (1.45) 15.10 (2.56) 20.09 (6.03) 32
#> wt 2.29 (0.57) 3.12 (0.36) 4.00 (0.76) 3.22 (0.98) 32
#> hp 82.64 (20.93) 122.29 (24.26) 209.21 (50.98) 146.69 (68.56) 32
#>
#> Note. M = Mean; SD = Standard Deviation; N = number of observations.The returned list has $descriptives_df (the formatted
data frame) and $note (the APA table note).
# Two-sample Welch t-test: mpg by transmission type
auto <- mtcars$mpg[mtcars$am == 0]
manual <- mtcars$mpg[mtcars$am == 1]
res <- apa_t_test(auto, manual, output = "console")mtcars2 <- mtcars
mtcars2$cyl <- factor(mtcars2$cyl)
res <- apa_anova(lm(mpg ~ cyl, data = mtcars2), es = "partial_eta2")
#> ANOVA Results
#>
#> Source SS df MS F p partial_eta2
#> cyl 824.78 2 412.39 39.70 < .001 0.732
#> Residuals 301.26 29 10.39 <NA> <NA> <NA>
#>
#> SS = Sum of Squares; MS = Mean Square; partial_eta2 = partial eta squared. Type II sums of squares.mtcars2$gear <- factor(mtcars2$gear)
res <- apa_twoway_anova(
mpg ~ cyl * gear,
data = mtcars2,
factorA = "cyl",
factorB = "gear",
simple_effects = TRUE
)
#> Two-Way ANOVA Results
#>
#> Source SS df MS F p partial_eta2
#> cyl 349.79 2 174.90 15.60 < .001 0.565
#> gear 8.25 2 4.13 0.37 .696 0.030
#> cyl:gear 23.89 3 7.96 0.71 .555 0.082
#> Residuals 269.12 24 11.21 <NA> <NA> <NA>
#>
#> Note. Two-Way Analysis of Variance (ANOVA). SS = Sum of Squares; MS = Mean Square; partial_eta2 = partial eta squared . Type II sums of squares. Simple effects of cyl within levels of gear computed via emmeans::joint_tests().
#> Simple Effects of cyl Within gear
#>
#> Source df1 df2 F p
#> cyl | gear=3 2 24 3.08 .065
#> cyl | gear=4 1 24 12.24 .002
#> cyl | gear=5 2 24 7.46 .003res <- apa_ancova(
formula = mpg ~ cyl + wt,
data = mtcars2,
covariate = "wt",
focal = "cyl",
es = "partial_eta2"
)
#> ANCOVA Results
#>
#> Source SS df MS F p partial_eta2
#> cyl 95.26 2 47.63 7.29 .003 0.342
#> wt 118.20 1 118.20 18.08 < .001 0.392
#> Residuals 183.06 28 6.54 <NA> <NA> <NA>
#>
#> Note. Analysis of Covariance (ANCOVA) with covariate(s): wt . SS = Sum of Squares; MS = Mean Square; partial_eta2 = partial eta squared . Type II sums of squares. Adjusted means estimated at covariate mean(s) via emmeans.
#> Covariate-Adjusted Means for cyl
#>
#> cyl Adjusted M SE 95% CI
#> 4 23.678 1.043 [21.54, 25.81]
#> 6 19.422 0.969 [17.44, 21.41]
#> 8 17.607 0.903 [15.76, 19.45]The second table shows covariate-adjusted marginal means for each
level of cyl, evaluated at the mean of wt.
res <- apa_manova(
cbind(Sepal.Length, Petal.Length) ~ Species,
data = iris
)
#> MANOVA Results
#>
#> Effect Test Stat approx F num df den df p eta2
#> Species Pillai 0.9885 71.83 4 294.00 < .001 0.494
#> Species Wilks 0.0399 292.56 4 292.00 < .001 0.800
#> Species Hotelling-Lawley 23.3647 846.97 4 290.00 < .001 0.921
#> Species Roy 23.3342 1715.06 2 147.00 < .001 0.959
#>
#> Note. Pillai = Pillai's trace; Wilks = Wilks' lambda; Hotelling-Lawley = Hotelling-Lawley trace; Roy = Roy's largest root. approx F = approximate F-statistic; num df = numerator degrees of freedom; den df = denominator degrees of freedom. eta2 = eta-squared, approximated from F-ratio and degrees of freedom. Type II sums of squares.All four multivariate test statistics are reported: Pillai’s trace, Wilks’ lambda, Hotelling-Lawley trace, and Roy’s largest root.
fit <- aov(mpg ~ cyl, data = mtcars2)
res <- apa_posthoc(fit, by = "cyl", adjust = "tukey")
#> Post-hoc Pairwise Comparisons
#>
#> Contrast Estimate SE t p 95% CI
#> cyl4 - cyl6 6.92 1.56 4.44 < .001 [3.07, 10.77]
#> cyl4 - cyl8 11.56 1.30 8.90 < .001 [8.36, 14.77]
#> cyl6 - cyl8 4.64 1.49 3.11 .011 [0.96, 8.33]
#>
#> Note. Estimates are pairwise differences of marginal means. Contrast method = pairwise; adjusted using tukey; confidence level = 0.95.# Independence test
m <- matrix(c(30, 10, 20, 40), nrow = 2,
dimnames = list(c("Group A", "Group B"),
c("Yes", "No")))
res <- apa_chisq(m)fit <- lm(mpg ~ wt + hp + factor(cyl), data = mtcars)
res <- apa_table(fit)
#> Linear Model Predicting mpg
#>
#> Predictor b SE t p 95% CI
#> (Intercept) 35.85 2.04 17.56 < .001 [31.66, 40.03]
#> wt -3.18 0.72 -4.42 < .001 [-4.66, -1.70]
#> hp -0.02 0.01 -1.93 .064 [-0.05, 0.00]
#> factor(cyl)6 -3.36 1.40 -2.40 .024 [-6.24, -0.48]
#> factor(cyl)8 -3.19 2.17 -1.47 .154 [-7.64, 1.27]
#>
#> Note. Unstandardized coefficients reported. SE = Standard Error; CI = Confidence Interval. p-values based on t tests. N (Level 1) = 32.res <- apa_robust(fit, type = "HC3")
#> Linear Model (HC3 robust SEs)
#>
#> Predictor b SE t p 95% CI
#> (Intercept) 35.85 2.71 13.22 < .001 [30.53, 41.16]
#> wt -3.18 0.81 -3.93 < .001 [-4.77, -1.60]
#> hp -0.02 0.01 -1.83 .078 [-0.05, 0.00]
#> factor(cyl)6 -3.36 1.38 -2.43 .022 [-6.07, -0.65]
#> factor(cyl)8 -3.19 2.52 -1.26 .217 [-8.13, 1.76]
#>
#> Note. Unstandardized coefficients reported. Standard errors are HC3 robust via sandwich/lmtest; 95% CIs use normal approximation.res <- apa_hetero(fit)
#> Heteroscedasticity Diagnostics
#>
#> Test Stat df p
#> Breusch-Pagan 6.17 4 .187
#> Non-Constant Variance (NCV) 3.37 1 .066
#>
#> Note. Breusch-Pagan uses studentized residuals.
res <- apa_homoskedasticity(fit)
#> Homoskedasticity Check
#>
#> Test Stat df p
#> Breusch-Pagan 6.17 4 .187
#>
#> Note. Breusch-Pagan and White tests are sensitive to functional form. Levene/Brown-Forsythe not run (no 'group' provided). At alpha = 0.05, no test detected heteroskedasticity (results are consistent with homoskedastic errors).library(lme4)
data(ECLS_demo)
m0 <- lmer(math ~ 1 + (1 | schid), data = ECLS_demo, REML = FALSE)
m1 <- lmer(math ~ SES + (1 | schid), data = ECLS_demo, REML = FALSE)
m2 <- lmer(math ~ SES + gender + (1 | schid), data = ECLS_demo, REML = FALSE)
res <- apa_multilevel(
m0, m1, m2,
model_names = c("Null", "+ SES", "+ SES + Gender")
)
#> Multilevel Model Results: Null - Fixed Effects
#>
#> Term b SE t p 95% CI
#> (Intercept) 501.27 1.02 491.13 < .001 [499.27, 503.27]
#> Null - Random Effects
#>
#> Group Var Variance SD
#> schid (Intercept) 18.24 4.27
#> Residual <NA> 103.58 10.18
#> Multilevel Model Results: + SES - Fixed Effects
#>
#> Term b SE t p 95% CI
#> (Intercept) 501.41 1.02 490.99 < .001 [499.41, 503.41]
#> SES 2.73 0.36 7.60 < .001 [2.03, 3.44]
#> + SES - Random Effects
#>
#> Group Var Variance SD
#> schid (Intercept) 18.44 4.29
#> Residual <NA> 96.44 9.82
#> Multilevel Model Results: + SES + Gender - Fixed Effects
#>
#> Term b SE t p 95% CI
#> (Intercept) 500.86 1.08 465.32 < .001 [498.75, 502.97]
#> SES 2.76 0.36 7.69 < .001 [2.06, 3.47]
#> genderMale 1.17 0.70 1.67 .095 [-0.21, 2.55]
#> + SES + Gender - Random Effects
#>
#> Group Var Variance SD
#> schid (Intercept) 18.54 4.31
#> Residual <NA> 96.09 9.80
#> Model Comparison (Likelihood Ratio Tests)
#>
#> Model df AIC BIC logLik Chi-sq Chi-sq df p
#> Null NA 6030.3 6044.4 -3012.15 <NA> NA <NA>
#> + SES 1 5976.6 5995.3 -2984.29 55.72 NA < .001
#> + SES + Gender 1 5975.8 5999.2 -2982.90 2.77 NA .096
#>
#> Note. Multilevel model (lme4). b = unstandardized fixed-effect coefficient; SE = Standard Error; CI = Confidence Interval (Wald). p-values use Satterthwaite degrees of freedom (lmerTest). ICC = intraclass correlation coefficient. R2m/R2c via MuMIn::r.squaredGLMM(). R2m = marginal R-squared (fixed effects only); R2c = conditional R-squared (fixed + random effects).The function reports fixed effects, random effects, the intraclass correlation coefficient (ICC), marginal and conditional R-squared, and a likelihood ratio model-comparison table.