Last updated on 2026-11-10 07:52:00 CET.
| Flavor | Version | Tinstall | Tcheck | Ttotal | Status | Flags |
|---|---|---|---|---|---|---|
| r-devel-linux-x86_64-debian-clang | 0.2.0 | 17.24 | 286.76 | 304.00 | OK | |
| r-devel-linux-x86_64-debian-gcc | 0.3.1 | 16.02 | 335.81 | 351.83 | OK | |
| r-devel-linux-x86_64-fedora-clang | 0.3.1 | 18.00 | 315.12 | 333.12 | ERROR | |
| r-devel-linux-x86_64-fedora-gcc | 0.3.1 | 16.00 | 306.50 | 322.50 | ERROR | |
| r-devel-windows-x86_64 | 0.3.1 | 27.00 | 476.00 | 503.00 | OK | |
| r-patched-linux-x86_64 | 0.2.0 | OK | ||||
| r-release-linux-x86_64 | 0.3.1 | 23.27 | 468.42 | 491.69 | OK | |
| r-release-macos-arm64 | 0.3.1 | 6.00 | 112.00 | 118.00 | OK | |
| r-release-macos-x86_64 | 0.3.1 | 16.00 | 419.00 | 435.00 | OK | |
| r-release-windows-x86_64 | 0.3.1 | 35.00 | 508.00 | 543.00 | OK | |
| r-oldrel-macos-arm64 | 0.3.1 | 7.00 | 125.00 | 132.00 | OK | |
| r-oldrel-macos-x86_64 | 0.3.1 | 16.00 | 361.00 | 377.00 | OK | |
| r-oldrel-windows-x86_64 | 0.3.1 | 28.00 | 649.00 | 677.00 | OK |
Version: 0.3.1
Check: tests
Result: ERROR
Running ‘testthat.R’ [159s/162s]
Running the tests in ‘tests/testthat.R’ failed.
Complete output:
> # This file is part of the standard setup for testthat.
> # It is recommended that you do not modify it.
> #
> # Where should you do additional test configuration?
> # Learn more about the roles of various files in:
> # * https://r-pkgs.org/testing-design.html#sec-tests-files-overview
> # * https://testthat.r-lib.org/articles/special-files.html
>
> library(testthat)
> library(colleyRstats)
Loading required package: ggplot2
>
> test_check("colleyRstats")
Fitted: Cumulative Link Mixed Model (CLMM) -- ordinal::clmm(rating ~ cond + (1 | id), data = your_data)
Coefficient covariances computed by hccm()
Fitted: Linear Mixed Model (LMM) -- lmerTest::lmer(y ~ cond * age + (1 | id), data = your_data)
Fitted: Linear Mixed Model (LMM) -- lmerTest::lmer(y ~ cond + (1 | id), data = your_data)
Fitted: Linear Mixed Model (LMM) -- lmerTest::lmer(y ~ a * b + (1 | id), data = your_data)
Fitted: Aligned Rank Transform (ART), repeated measures -- ARTool::art(y ~ a * b + (1 | id), data = your_data)
Fitted: Rank-based repeated measures (nparLD) -- nparLD::nparLD(y ~ cond, subject = "id", data = your_data)
No pairwise post-hoc contrasts are computed for nparLD fits; follow up with a dedicated test (e.g. FSA::dunnTest() after Kruskal-Wallis).
Kruskal-Wallis rank sum test
data: x and g
Kruskal-Wallis chi-squared = 12.9013, df = 2, p-value = 0
Dunn's Pairwise Comparison of x by g
(Holm)
Col Mean-│
Row Mean │ A B
─────────┼──────────────────────
B │ -1.149803
│ 0.2502
│
C │ -3.521840 -2.372036
│ 0.0013* 0.0354*
FWER = 0.05
Reject Ho if adjusted p ≤ FWER with stopping rule, where (unadjusted) p = Pr(|Z| ≥ |z|)
Wrote results to '/tmp/Rtmpbn3PD4/working_dir/RtmppIvlaZ/file87a4b4e477c.tex'.
Wrote results to '/tmp/Rtmpbn3PD4/working_dir/RtmppIvlaZ/file87a4b246b3614.tex'.
Wrote results to '/tmp/Rtmpbn3PD4/working_dir/RtmppIvlaZ/file87a4b5c9e735f.tex'.
boundary (singular) fit: see help('isSingular')
boundary (singular) fit: see help('isSingular')
Wrote '/tmp/Rtmpbn3PD4/working_dir/RtmppIvlaZ/sty-test/colleyRstats.sty'. Add \usepackage{colleyRstats} to your document.
Kruskal-Wallis rank sum test
data: x and g
Kruskal-Wallis chi-squared = 96.9374, df = 2, p-value = 0
Dunn's Pairwise Comparison of x by g
(Holm)
Col Mean-│
Row Mean │ setosa versicol
─────────┼──────────────────────
versicol │ -6.106326
│ 0.0000*
│
virginic │ -9.741784 -3.635458
│ 0.0000* 0.0003*
FWER = 0.05
Reject Ho if adjusted p ≤ FWER with stopping rule, where (unadjusted) p = Pr(|Z| ≥ |z|)
<colleyRstats analysis recommendation>
Outcome : rating (ordinal)
Predictors : cond (categorical, within)
Design : within (cluster: id)
Random effects : (1 | id)
Recommendation : Cumulative Link Mixed Model (CLMM)
Family : cumulative link (logit)
Fit with : ordinal::clmm(rating ~ cond + (1 | id), data = your_data) # outcome must be an ordered factor
Report with : reportCLMM()
Alternative(s) : nparLD (rank-based repeated measures) if proportional odds is untenable
Rationale : the outcome is ordinal and the observations are clustered, so an ordinal (proportional-odds) model with random effects is appropriate
The main effect of \textit{cond} on workload was not significant (\F{2}{38}{0.07}, \p{0.931}).
The contrast \textit{B} vs.\ \textit{A} of \textit{cond} on workload was not significant ($b = 0.12$, 95\% CI $[-0.63, 0.86]$, $t(38) = 0.32$, \p{0.751}).
The contrast \textit{C} vs.\ \textit{A} of \textit{cond} on workload was not significant ($b = 0.12$, 95\% CI $[-0.62, 0.87]$, $t(38) = 0.34$, \p{0.739}).
boundary (singular) fit: see help('isSingular')
boundary (singular) fit: see help('isSingular')
Kruskal-Wallis rank sum test
data: x and g
Kruskal-Wallis chi-squared = 96.9374, df = 2, p-value = 0
Dunn's Pairwise Comparison of x by g
(Holm)
Col Mean-│
Row Mean │ setosa versicol
─────────┼──────────────────────
versicol │ -6.106326
│ 0.0000*
│
virginic │ -9.741784 -3.635458
│ 0.0000* 0.0003*
FWER = 0.05
Reject Ho if adjusted p ≤ FWER with stopping rule, where (unadjusted) p = Pr(|Z| ≥ |z|)
A Shapiro--Wilk test on the within-participant differences between the two levels of g ($n = 20$) indicated no significant deviation from normality ($W = 0.95$, $p = 0.408$); therefore, parametric tests were used.
Scale for x is already present.
Adding another scale for x, which will replace the existing scale.
Scale for x is already present.
Adding another scale for x, which will replace the existing scale.
Scale for x is already present.
Adding another scale for x, which will replace the existing scale.
Scale for x is already present.
Adding another scale for x, which will replace the existing scale.
A Shapiro--Wilk test on the residuals of the additive model with participant and cond as factors ($n = 57$) indicated no significant deviation from normality ($W = 0.99$, $p = 0.915$); therefore, parametric tests were used.
An ANOVA estimation for factorial designs using 'afex' found a significant effect of \cond{} on y (\F{1.86}{33.48}{110.56}, \pminor{0.001}, $\omega_{p}^{2}$ = 0.03).
A Student's t post-hoc test (Holm-adjusted) found that B was significantly higher (\m{1.94}, \sd{4.82}) in terms of y compared to A (\m{0.88}, \sd{4.60}; \padjminor{0.001}).
A Student's t post-hoc test (Holm-adjusted) found that C was significantly higher (\m{2.98}, \sd{4.64}) in terms of y compared to A (\m{0.88}, \sd{4.60}; \padjminor{0.001}).
A Student's t post-hoc test (Holm-adjusted) found that C was significantly higher (\m{2.98}, \sd{4.64}) in terms of y compared to B (\m{1.94}, \sd{4.82}; \padjminor{0.001}).
CAUTION: item numbers, order and polarity depend on how the questionnaire
was administered -- survey tools renumber items, translations reorder them,
and short forms drop them. This package applies the PUBLISHED key. If your
sheet differed, the scores will be wrong without any error being raised.
Check the mapping against the survey your participants actually saw, and
double-check any number before it goes into a paper.
Assumed response range: 1-5 (the instrument's own; pass `scale` if your survey differed)
Observed responses: 3 to 3
Item mapping (verify against the survey your participants saw):
item code column subscale reverse observed_min observed_max n_missing
1 sus1 sus_1 Usability FALSE 3 3 0
2 sus2 sus_2 Usability TRUE 3 3 0
3 sus3 sus_3 Usability FALSE 3 3 0
4 sus4 sus_4 Learnability TRUE 3 3 0
5 sus5 sus_5 Usability FALSE 3 3 0
6 sus6 sus_6 Usability TRUE 3 3 0
7 sus7 sus_7 Usability FALSE 3 3 0
8 sus8 sus_8 Usability TRUE 3 3 0
9 sus9 sus_9 Usability FALSE 3 3 0
10 sus10 sus_10 Learnability TRUE 3 3 0
Notes:
- The SUS score is the sum of the ten recoded items multiplied by 2.5, giving 0-100. That is a percentage of the maximum, NOT a percentile: the mean SUS across studies is about 68, so 68 is average rather than poor.
- Usability (8 items, x 3.125) and Learnability (2 items, x 12.5) follow Lewis & Sauro (2009) and are likewise on 0-100. The two are strongly correlated; report them only if the study has a reason to separate them.
- Missing items: Brooke (1996) instructs a respondent who cannot answer an item to mark the centre point of the scale, so when `min_valid` < 1 lets an incomplete row be scored, each missing item counts as the centre (3 on 1-5). One missing item in an otherwise perfect response therefore gives 95, not 100. Pass impute = "prorate" to scale up from the answered items instead, and say which rule you used.
- Learnability has only two items. score_reliability() reports the Spearman-Brown coefficient for it, which Eisinga, te Grotenhuis & Pelzer (2013) recommend over alpha for two-item scales.
item code column subscale reverse observed_min observed_max n_missing
1 sus1 sus_1 Usability FALSE 3 3 0
2 sus2 sus_2 Usability TRUE 3 3 0
3 sus3 sus_3 Usability FALSE 3 3 0
4 sus4 sus_4 Learnability TRUE 3 3 0
5 sus5 sus_5 Usability FALSE 3 3 0
6 sus6 sus_6 Usability TRUE 3 3 0
7 sus7 sus_7 Usability FALSE 3 3 0
8 sus8 sus_8 Usability TRUE 3 3 0
9 sus9 sus_9 Usability FALSE 3 3 0
10 sus10 sus_10 Learnability TRUE 3 3 0
item code label subscale reverse
1 ueqs1 obstructive - supportive Pragmatic Quality FALSE
2 ueqs2 complicated - easy Pragmatic Quality FALSE
3 ueqs3 inefficient - efficient Pragmatic Quality FALSE
4 ueqs4 confusing - clear Pragmatic Quality FALSE
5 ueqs5 boring - exciting Hedonic Quality FALSE
6 ueqs6 not interesting - interesting Hedonic Quality FALSE
7 ueqs7 conventional - inventive Hedonic Quality FALSE
8 ueqs8 usual - leading edge Hedonic Quality FALSE
item code column subscale reverse observed_min observed_max n_missing
1 sus1 sus_1 Usability FALSE 3 3 0
2 sus2 sus_2 Usability TRUE 3 3 0
3 sus3 sus_3 Usability FALSE 3 3 0
4 sus4 sus_4 Learnability TRUE NA NA 2
5 sus5 sus_5 Usability FALSE 3 3 0
6 sus6 sus_6 Usability TRUE 3 3 0
7 sus7 sus_7 Usability FALSE 3 3 0
8 sus8 sus_8 Usability TRUE 3 3 0
9 sus9 sus_9 Usability FALSE 3 3 0
10 sus10 sus_10 Learnability TRUE 3 3 0
CAUTION: item numbers, order and polarity depend on how the questionnaire
was administered -- survey tools renumber items, translations reorder them,
and short forms drop them. This package applies the PUBLISHED key. If your
sheet differed, the scores will be wrong without any error being raised.
Check the mapping against the survey your participants actually saw, and
double-check any number before it goes into a paper.
See the full mapping with check_questionnaire(data, "sus", ...); silence this note with options(colleyRstats.quiet_questionnaires = TRUE).
CAUTION: item numbers, order and polarity depend on how the questionnaire
was administered -- survey tools renumber items, translations reorder them,
and short forms drop them. This package applies the PUBLISHED key. If your
sheet differed, the scores will be wrong without any error being raised.
Check the mapping against the survey your participants actually saw, and
double-check any number before it goes into a paper.
See the full mapping with check_questionnaire(data, "sus", ...); silence this note with options(colleyRstats.quiet_questionnaires = TRUE).
Scoring System Usability Scale (SUS) from 10 columns (sus_1 ... sus_10, matched by item number); responses observed 3 to 3 on the assumed range 1 to 5; reverse-coded: sus2, sus4, sus6, sus8, sus10.
See the full mapping with check_questionnaire(data, "sus", ...); silence this note with options(colleyRstats.quiet_questionnaires = TRUE).
Scoring System Usability Scale (SUS) from 10 columns (sus_1 ... sus_10, matched by item number); responses observed 3 to 3 on the assumed range 1 to 5; reverse-coded: sus2, sus4, sus6, sus8, sus10.
See the full mapping with check_questionnaire(data, "sus", ...); silence this note with options(colleyRstats.quiet_questionnaires = TRUE).
System Usability Scale (SUS) -- Brooke (1996), Usability Evaluation in Industry; Lewis & Sauro (2009), HCII
Assumed response range: 1-5 (the instrument's own; pass `scale` if your survey differed)
Observed responses: 3 to 3
Item mapping (verify against the survey your participants saw):
item code column subscale reverse observed_min observed_max n_missing
1 sus1 sus_1 Usability FALSE 3 3 0
2 sus2 sus_2 Usability TRUE 3 3 0
3 sus3 sus_3 Usability FALSE 3 3 0
4 sus4 sus_4 Learnability TRUE 3 3 0
5 sus5 sus_5 Usability FALSE 3 3 0
6 sus6 sus_6 Usability TRUE 3 3 0
7 sus7 sus_7 Usability FALSE 3 3 0
8 sus8 sus_8 Usability TRUE 3 3 0
9 sus9 sus_9 Usability FALSE 3 3 0
10 sus10 sus_10 Learnability TRUE 3 3 0
Notes:
- The SUS score is the sum of the ten recoded items multiplied by 2.5, giving 0-100. That is a percentage of the maximum, NOT a percentile: the mean SUS across studies is about 68, so 68 is average rather than poor.
- Usability (8 items, x 3.125) and Learnability (2 items, x 12.5) follow Lewis & Sauro (2009) and are likewise on 0-100. The two are strongly correlated; report them only if the study has a reason to separate them.
- Missing items: Brooke (1996) instructs a respondent who cannot answer an item to mark the centre point of the scale, so when `min_valid` < 1 lets an incomplete row be scored, each missing item counts as the centre (3 on 1-5). One missing item in an otherwise perfect response therefore gives 95, not 100. Pass impute = "prorate" to scale up from the answered items instead, and say which rule you used.
- Learnability has only two items. score_reliability() reports the Spearman-Brown coefficient for it, which Eisinga, te Grotenhuis & Pelzer (2013) recommend over alpha for two-item scales.
System Usability Scale (SUS) -- Brooke (1996), Usability Evaluation in Industry; Lewis & Sauro (2009), HCII
Response range 1-5; 10 items in 2 subscales.
- The SUS score is the sum of the ten recoded items multiplied by 2.5, giving 0-100. That is a percentage of the maximum, NOT a percentile: the mean SUS across studies is about 68, so 68 is average rather than poor.
- Usability (8 items, x 3.125) and Learnability (2 items, x 12.5) follow Lewis & Sauro (2009) and are likewise on 0-100. The two are strongly correlated; report them only if the study has a reason to separate them.
- Missing items: Brooke (1996) instructs a respondent who cannot answer an item to mark the centre point of the scale, so when `min_valid` < 1 lets an incomplete row be scored, each missing item counts as the centre (3 on 1-5). One missing item in an otherwise perfect response therefore gives 95, not 100. Pass impute = "prorate" to scale up from the answered items instead, and say which rule you used.
- Learnability has only two items. score_reliability() reports the Spearman-Brown coefficient for it, which Eisinga, te Grotenhuis & Pelzer (2013) recommend over alpha for two-item scales.
item code
1 sus1
2 sus2
3 sus3
4 sus4
5 sus5
6 sus6
7 sus7
8 sus8
9 sus9
10 sus10
label
I think that I would like to use this system frequently
I found the system unnecessarily complex
I thought the system was easy to use
I think that I would need the support of a technical person to be able to use this system
I found the various functions in this system were well integrated
I thought there was too much inconsistency in this system
I would imagine that most people would learn to use this system very quickly
I found the system very cumbersome to use
I felt very confident using the system
I needed to learn a lot of things before I could get going with this system
subscale reverse
Usability FALSE
Usability TRUE
Usability FALSE
Learnability TRUE
Usability FALSE
Usability TRUE
Usability FALSE
Usability TRUE
Usability FALSE
Learnability TRUE
Scoring System Usability Scale (SUS) from 10 columns (sus_1 ... sus_10, matched by item number); responses observed 3 to 3 on the assumed range 1 to 5; reverse-coded: sus2, sus4, sus6, sus8, sus10.
See the full mapping with check_questionnaire(data, "sus", ...); silence this note with options(colleyRstats.quiet_questionnaires = TRUE).
item code column subscale reverse observed_min observed_max n_missing
1 sus1 sus_1 Usability FALSE 3 3 0
2 sus2 sus_2 Usability TRUE NA NA 1
3 sus3 sus_3 Usability FALSE 3 3 0
4 sus4 sus_4 Learnability TRUE 3 3 0
5 sus5 sus_5 Usability FALSE 3 3 0
6 sus6 sus_6 Usability TRUE 3 3 0
7 sus7 sus_7 Usability FALSE 3 3 0
8 sus8 sus_8 Usability TRUE 3 3 0
9 sus9 sus_9 Usability FALSE 3 3 0
10 sus10 sus_10 Learnability TRUE 3 3 0
CAUTION: item numbers, order and polarity depend on how the questionnaire
was administered -- survey tools renumber items, translations reorder them,
and short forms drop them. This package applies the PUBLISHED key. If your
sheet differed, the scores will be wrong without any error being raised.
Check the mapping against the survey your participants actually saw, and
double-check any number before it goes into a paper.
See the full mapping with check_questionnaire(data, "sus", ...); silence this note with options(colleyRstats.quiet_questionnaires = TRUE).
item code column subscale reverse observed_min observed_max n_missing
1 fms fms FMS FALSE 0 12 0
The NPAV found a significant main effect of \Video{} on mental workload (\F{1}{10}{6.12}, \p{0.033}, $\eta_{p}^{2}$ = 0.38, 95\% CI: [0.00, 0.68]).
The NPAV found a significant interaction effect of \gesture{} $\times$ \eHMI{} on mental workload (\F{1}{10}{5.01}, \p{0.045}, $\eta_{p}^{2}$ = 0.33, 95\% CI: [0.00, 0.66]).
The ART found a significant interaction effect of \gesture{} $\times$ \eHMI{} on mental demand (\F{1}{10}{5.01}, \p{0.045}, $\eta_{p}^{2}$ = 0.33, 95\% CI: [0.00, 0.66]).
The nparLD analysis found no significant effects on TLX1.
Saving _problems/test-reporting-180.R
tlx\_B: \m{8.00}, \sd{1.58}
A Games-Howell post-hoc test (Holm-adjusted) found that 4 was significantly higher (\m{26.66}, \sd{4.51}) in terms of mpg compared to 8 (\m{15.10}, \sd{2.56}; \padjminor{0.001}).
A Games-Howell post-hoc test (Holm-adjusted) found that 6 was significantly higher (\m{19.74}, \sd{1.45}) in terms of mpg compared to 8 (\m{15.10}, \sd{2.56}; \padjminor{0.001}).
A Games-Howell post-hoc test (Holm-adjusted) found that 4 was significantly higher (\m{26.66}, \sd{4.51}) in terms of mpg compared to 8 (\m{15.10}, \sd{2.56}; \padjminor{0.001}).
A Games-Howell post-hoc test (Holm-adjusted) found that 6 was significantly higher (\m{19.53}, \sd{1.47}) in terms of mpg compared to 8 (\m{15.10}, \sd{2.56}; \padjminor{0.001}).
A Games-Howell post-hoc test (Holm-adjusted) found that FourCyl was significantly higher (\m{26.66}, \sd{4.51}) in terms of mpg compared to 8 (\m{15.10}, \sd{2.56}; \padjminor{0.001}).
A Games-Howell post-hoc test (Holm-adjusted) found that 6 was significantly higher (\m{19.74}, \sd{1.45}) in terms of mpg compared to 8 (\m{15.10}, \sd{2.56}; \padjminor{0.001}).
Kruskal-Wallis rank sum test
data: x and g
Kruskal-Wallis chi-squared = 96.9374, df = 2, p-value = 0
Dunn's Pairwise Comparison of x by g
(Holm)
Col Mean-│
Row Mean │ setosa versicol
─────────┼──────────────────────
versicol │ -6.106326
│ 0.0000*
│
virginic │ -9.741784 -3.635458
│ 0.0000* 0.0003*
FWER = 0.05
Reject Ho if adjusted p ≤ FWER with stopping rule, where (unadjusted) p = Pr(|Z| ≥ |z|)
A Dunn post-hoc test (Holm-adjusted) found that Sepal.Length for the \Species{} virginica was significantly higher (\mdn{6.50}, \iqr{0.67}) than for setosa (\mdn{5.00}, \iqr{0.40}; \padjminor{0.001}, \rankbiserial{0.97}) and versicolor (\mdn{5.90}, \iqr{0.70}; \padjminor{0.001}, \rankbiserial{0.58}).
% latex table generated in R 4.7.0 by xtable 1.8-8 package
% Tue Nov 10 04:17:17 2026
\begin{table}[ht]
\centering
\caption{Dunn post-hoc comparisons for independent variable \Species{} and dependent variable Sepal.Length. Positive Z-values mean that the first-named level has the higher mean rank; for negative Z-values, the second-named level does. p-values are Holm-adjusted. Effect size reported as rank-biserial correlation (r).}
\label{tab:posthoc-Species-Sepal.Length}
\begingroup\small
\begin{tabular}{lrll}
\hline
Comparison & Z & p-adjusted & r \\
\hline
setosa - versicolor & -6.1063 & $<$0.001 & 0.87 \\
setosa - virginica & -9.7418 & $<$0.001 & 0.97 \\
versicolor - virginica & -3.6355 & $<$0.001 & 0.58 \\
\hline
\end{tabular}
\endgroup
\end{table}
Kruskal-Wallis rank sum test
data: x and g
Kruskal-Wallis chi-squared = 96.9374, df = 2, p-value = 0
Dunn's Pairwise Comparison of x by g
(Holm)
Col Mean-│
Row Mean │ setosa versicol
─────────┼──────────────────────
versicol │ -6.106326
│ 0.0000*
│
virginic │ -9.741784 -3.635458
│ 0.0000* 0.0003*
FWER = 0.05
Reject Ho if adjusted p ≤ FWER with stopping rule, where (unadjusted) p = Pr(|Z| ≥ |z|)
% latex table generated in R 4.7.0 by xtable 1.8-8 package
% Tue Nov 10 04:17:17 2026
\begin{table}[ht]
\centering
\caption{Dunn post-hoc comparisons for independent variable \Species{} and dependent variable Sepal.Length. Positive Z-values mean that the first-named level has the higher mean rank; for negative Z-values, the second-named level does. p-values are Holm-adjusted. Effect size reported as rank-biserial correlation (r).}
\label{tab:posthoc-Species-Sepal.Length}
\begingroup\small
\begin{tabular}{lrll}
\hline
Comparison & Z & p-adjusted & r \\
\hline
setosa - versicolor & -6.1063 & $<$0.001 & 0.87 \\
setosa - virginica & -9.7418 & $<$0.001 & 0.97 \\
versicolor - virginica & -3.6355 & $<$0.001 & 0.58 \\
\hline
\end{tabular}
\endgroup
\end{table}
NOTE: Results may be misleading due to involvement in interactions
An ART-C post-hoc test (Holm-adjusted) found that score for the \mode{} Hand was significantly higher (\mdn{6.12}, \iqr{1.04}) than for Both (\mdn{1.96}, \iqr{1.05}; \padjminor{0.001}, \rankbiserial{1.00}) and Eye (\mdn{3.86}, \iqr{1.37}; \padjminor{0.001}, \rankbiserial{1.00}).
% latex table generated in R 4.7.0 by xtable 1.8-8 package
% Tue Nov 10 04:17:17 2026
\begin{table}[ht]
\centering
\caption{Post-hoc ART-C contrasts for independent variable \mode{} and dependent variable \score{}. Positive t-values mean that the first-named level is higher than the second-named on the aligned-rank scale; for negative t-values, the second-named level is. p-values are Holm-adjusted. Effect size reported as rank-biserial correlation (r).}
\label{tab:artcon-mode-score}
\begingroup\small
\begin{tabular}{lrlll}
\hline
Comparison & t & df & p-adjusted & r \\
\hline
Both - Eye & -6.64 & 36 & $<$0.001 & 0.97 \\
Both - Hand & -13.95 & 36 & $<$0.001 & 1.00 \\
Eye - Hand & -7.31 & 36 & $<$0.001 & 1.00 \\
\hline
\end{tabular}
\endgroup
\end{table}
NOTE: Results may be misleading due to involvement in interactions
NOTE: Results may be misleading due to involvement in interactions
An ART-C post-hoc test (Holm-adjusted) found that score for the \mode{} Hand was significantly higher (\mdn{6.12}, \iqr{1.04}) than for Both (\mdn{1.96}, \iqr{1.05}; \padjminor{0.001}, \rankbiserial{1.00}) and Eye (\mdn{3.86}, \iqr{1.37}; \padjminor{0.001}, \rankbiserial{0.93}).
NOTE: Results may be misleading due to involvement in interactions
Saving _problems/test-reporting-1108.R
A Games-Howell post-hoc test (Holm-adjusted) found that A was significantly higher (\m{5.50}, \sd{0.71}) in terms of val compared to B (\m{1.50}, \sd{0.71}; \padj{0.010}).
[ FAIL 2 | WARN 0 | SKIP 19 | PASS 1637 ]
══ Skipped tests (19) ══════════════════════════════════════════════════════════
• On CRAN (19): 'test-latex-output.R:422:3', 'test-plotting.R:146:3',
'test-reporting.R:923:3', 'test-reporting.R:1052:3',
'test-reporting.R:1089:3', 'test-reporting.R:1111:3',
'test-reporting.R:1135:3', 'test-reporting.R:1159:3',
'test-reporting.R:1178:3', 'test-reporting.R:1190:3',
'test-reporting.R:1259:3', 'test-reporting.R:1302:3',
'test-reporting.R:1394:3', 'test-reporting.R:1437:3',
'test-reporting.R:1451:3', 'test-setup.R:88:3', 'test-setup.R:115:3',
'test-study-project.R:34:3', 'test-study-project.R:243:3'
══ Failed tests ════════════════════════════════════════════════════════════════
── Failure ('test-reporting.R:177:3'): reportNparLD reports a real nparLD fit ──
Expected `suppressWarnings(reportNparLD(model, dv = "TLX1"))` to throw a message.
── Failure ('test-reporting.R:1108:3'): reportNparLD writes a compilable F(df, infinity) with the fractional ATS df ──
Expected `result` to match string "\\F{1.8}{\\infty}{".
Actual text:
x | The nparLD analysis found no significant effects on tlx\_mental.
[ FAIL 2 | WARN 0 | SKIP 19 | PASS 1637 ]
Error:
! Test failures.
Execution halted
Flavor: r-devel-linux-x86_64-fedora-clang
Version: 0.3.1
Check: tests
Result: ERROR
Running ‘testthat.R’ [153s/154s]
Running the tests in ‘tests/testthat.R’ failed.
Complete output:
> # This file is part of the standard setup for testthat.
> # It is recommended that you do not modify it.
> #
> # Where should you do additional test configuration?
> # Learn more about the roles of various files in:
> # * https://r-pkgs.org/testing-design.html#sec-tests-files-overview
> # * https://testthat.r-lib.org/articles/special-files.html
>
> library(testthat)
> library(colleyRstats)
Loading required package: ggplot2
>
> test_check("colleyRstats")
Fitted: Cumulative Link Mixed Model (CLMM) -- ordinal::clmm(rating ~ cond + (1 | id), data = your_data)
Coefficient covariances computed by hccm()
Fitted: Linear Mixed Model (LMM) -- lmerTest::lmer(y ~ cond * age + (1 | id), data = your_data)
Fitted: Linear Mixed Model (LMM) -- lmerTest::lmer(y ~ cond + (1 | id), data = your_data)
Fitted: Linear Mixed Model (LMM) -- lmerTest::lmer(y ~ a * b + (1 | id), data = your_data)
Fitted: Aligned Rank Transform (ART), repeated measures -- ARTool::art(y ~ a * b + (1 | id), data = your_data)
Fitted: Rank-based repeated measures (nparLD) -- nparLD::nparLD(y ~ cond, subject = "id", data = your_data)
No pairwise post-hoc contrasts are computed for nparLD fits; follow up with a dedicated test (e.g. FSA::dunnTest() after Kruskal-Wallis).
Kruskal-Wallis rank sum test
data: x and g
Kruskal-Wallis chi-squared = 12.9013, df = 2, p-value = 0
Dunn's Pairwise Comparison of x by g
(Holm)
Col Mean-│
Row Mean │ A B
─────────┼──────────────────────
B │ -1.149803
│ 0.2502
│
C │ -3.521840 -2.372036
│ 0.0013* 0.0354*
FWER = 0.05
Reject Ho if adjusted p ≤ FWER with stopping rule, where (unadjusted) p = Pr(|Z| ≥ |z|)
Wrote results to '/tmp/Rtmp5aYASs/working_dir/RtmpjmKU6q/file2417ad22f7e0c3.tex'.
Wrote results to '/tmp/Rtmp5aYASs/working_dir/RtmpjmKU6q/file2417ad2aa6d7b0.tex'.
Wrote results to '/tmp/Rtmp5aYASs/working_dir/RtmpjmKU6q/file2417ad5f97f4b1.tex'.
boundary (singular) fit: see help('isSingular')
boundary (singular) fit: see help('isSingular')
Wrote '/tmp/Rtmp5aYASs/working_dir/RtmpjmKU6q/sty-test/colleyRstats.sty'. Add \usepackage{colleyRstats} to your document.
Kruskal-Wallis rank sum test
data: x and g
Kruskal-Wallis chi-squared = 96.9374, df = 2, p-value = 0
Dunn's Pairwise Comparison of x by g
(Holm)
Col Mean-│
Row Mean │ setosa versicol
─────────┼──────────────────────
versicol │ -6.106326
│ 0.0000*
│
virginic │ -9.741784 -3.635458
│ 0.0000* 0.0003*
FWER = 0.05
Reject Ho if adjusted p ≤ FWER with stopping rule, where (unadjusted) p = Pr(|Z| ≥ |z|)
<colleyRstats analysis recommendation>
Outcome : rating (ordinal)
Predictors : cond (categorical, within)
Design : within (cluster: id)
Random effects : (1 | id)
Recommendation : Cumulative Link Mixed Model (CLMM)
Family : cumulative link (logit)
Fit with : ordinal::clmm(rating ~ cond + (1 | id), data = your_data) # outcome must be an ordered factor
Report with : reportCLMM()
Alternative(s) : nparLD (rank-based repeated measures) if proportional odds is untenable
Rationale : the outcome is ordinal and the observations are clustered, so an ordinal (proportional-odds) model with random effects is appropriate
The main effect of \textit{cond} on workload was not significant (\F{2}{38}{0.07}, \p{0.931}).
The contrast \textit{B} vs.\ \textit{A} of \textit{cond} on workload was not significant ($b = 0.12$, 95\% CI $[-0.63, 0.86]$, $t(38) = 0.32$, \p{0.751}).
The contrast \textit{C} vs.\ \textit{A} of \textit{cond} on workload was not significant ($b = 0.12$, 95\% CI $[-0.62, 0.87]$, $t(38) = 0.34$, \p{0.739}).
boundary (singular) fit: see help('isSingular')
boundary (singular) fit: see help('isSingular')
Kruskal-Wallis rank sum test
data: x and g
Kruskal-Wallis chi-squared = 96.9374, df = 2, p-value = 0
Dunn's Pairwise Comparison of x by g
(Holm)
Col Mean-│
Row Mean │ setosa versicol
─────────┼──────────────────────
versicol │ -6.106326
│ 0.0000*
│
virginic │ -9.741784 -3.635458
│ 0.0000* 0.0003*
FWER = 0.05
Reject Ho if adjusted p ≤ FWER with stopping rule, where (unadjusted) p = Pr(|Z| ≥ |z|)
A Shapiro--Wilk test on the within-participant differences between the two levels of g ($n = 20$) indicated no significant deviation from normality ($W = 0.95$, $p = 0.408$); therefore, parametric tests were used.
Scale for x is already present.
Adding another scale for x, which will replace the existing scale.
Scale for x is already present.
Adding another scale for x, which will replace the existing scale.
Scale for x is already present.
Adding another scale for x, which will replace the existing scale.
Scale for x is already present.
Adding another scale for x, which will replace the existing scale.
A Shapiro--Wilk test on the residuals of the additive model with participant and cond as factors ($n = 57$) indicated no significant deviation from normality ($W = 0.99$, $p = 0.915$); therefore, parametric tests were used.
An ANOVA estimation for factorial designs using 'afex' found a significant effect of \cond{} on y (\F{1.86}{33.48}{110.56}, \pminor{0.001}, $\omega_{p}^{2}$ = 0.03).
A Student's t post-hoc test (Holm-adjusted) found that B was significantly higher (\m{1.94}, \sd{4.82}) in terms of y compared to A (\m{0.88}, \sd{4.60}; \padjminor{0.001}).
A Student's t post-hoc test (Holm-adjusted) found that C was significantly higher (\m{2.98}, \sd{4.64}) in terms of y compared to A (\m{0.88}, \sd{4.60}; \padjminor{0.001}).
A Student's t post-hoc test (Holm-adjusted) found that C was significantly higher (\m{2.98}, \sd{4.64}) in terms of y compared to B (\m{1.94}, \sd{4.82}; \padjminor{0.001}).
CAUTION: item numbers, order and polarity depend on how the questionnaire
was administered -- survey tools renumber items, translations reorder them,
and short forms drop them. This package applies the PUBLISHED key. If your
sheet differed, the scores will be wrong without any error being raised.
Check the mapping against the survey your participants actually saw, and
double-check any number before it goes into a paper.
Assumed response range: 1-5 (the instrument's own; pass `scale` if your survey differed)
Observed responses: 3 to 3
Item mapping (verify against the survey your participants saw):
item code column subscale reverse observed_min observed_max n_missing
1 sus1 sus_1 Usability FALSE 3 3 0
2 sus2 sus_2 Usability TRUE 3 3 0
3 sus3 sus_3 Usability FALSE 3 3 0
4 sus4 sus_4 Learnability TRUE 3 3 0
5 sus5 sus_5 Usability FALSE 3 3 0
6 sus6 sus_6 Usability TRUE 3 3 0
7 sus7 sus_7 Usability FALSE 3 3 0
8 sus8 sus_8 Usability TRUE 3 3 0
9 sus9 sus_9 Usability FALSE 3 3 0
10 sus10 sus_10 Learnability TRUE 3 3 0
Notes:
- The SUS score is the sum of the ten recoded items multiplied by 2.5, giving 0-100. That is a percentage of the maximum, NOT a percentile: the mean SUS across studies is about 68, so 68 is average rather than poor.
- Usability (8 items, x 3.125) and Learnability (2 items, x 12.5) follow Lewis & Sauro (2009) and are likewise on 0-100. The two are strongly correlated; report them only if the study has a reason to separate them.
- Missing items: Brooke (1996) instructs a respondent who cannot answer an item to mark the centre point of the scale, so when `min_valid` < 1 lets an incomplete row be scored, each missing item counts as the centre (3 on 1-5). One missing item in an otherwise perfect response therefore gives 95, not 100. Pass impute = "prorate" to scale up from the answered items instead, and say which rule you used.
- Learnability has only two items. score_reliability() reports the Spearman-Brown coefficient for it, which Eisinga, te Grotenhuis & Pelzer (2013) recommend over alpha for two-item scales.
item code column subscale reverse observed_min observed_max n_missing
1 sus1 sus_1 Usability FALSE 3 3 0
2 sus2 sus_2 Usability TRUE 3 3 0
3 sus3 sus_3 Usability FALSE 3 3 0
4 sus4 sus_4 Learnability TRUE 3 3 0
5 sus5 sus_5 Usability FALSE 3 3 0
6 sus6 sus_6 Usability TRUE 3 3 0
7 sus7 sus_7 Usability FALSE 3 3 0
8 sus8 sus_8 Usability TRUE 3 3 0
9 sus9 sus_9 Usability FALSE 3 3 0
10 sus10 sus_10 Learnability TRUE 3 3 0
item code label subscale reverse
1 ueqs1 obstructive - supportive Pragmatic Quality FALSE
2 ueqs2 complicated - easy Pragmatic Quality FALSE
3 ueqs3 inefficient - efficient Pragmatic Quality FALSE
4 ueqs4 confusing - clear Pragmatic Quality FALSE
5 ueqs5 boring - exciting Hedonic Quality FALSE
6 ueqs6 not interesting - interesting Hedonic Quality FALSE
7 ueqs7 conventional - inventive Hedonic Quality FALSE
8 ueqs8 usual - leading edge Hedonic Quality FALSE
item code column subscale reverse observed_min observed_max n_missing
1 sus1 sus_1 Usability FALSE 3 3 0
2 sus2 sus_2 Usability TRUE 3 3 0
3 sus3 sus_3 Usability FALSE 3 3 0
4 sus4 sus_4 Learnability TRUE NA NA 2
5 sus5 sus_5 Usability FALSE 3 3 0
6 sus6 sus_6 Usability TRUE 3 3 0
7 sus7 sus_7 Usability FALSE 3 3 0
8 sus8 sus_8 Usability TRUE 3 3 0
9 sus9 sus_9 Usability FALSE 3 3 0
10 sus10 sus_10 Learnability TRUE 3 3 0
CAUTION: item numbers, order and polarity depend on how the questionnaire
was administered -- survey tools renumber items, translations reorder them,
and short forms drop them. This package applies the PUBLISHED key. If your
sheet differed, the scores will be wrong without any error being raised.
Check the mapping against the survey your participants actually saw, and
double-check any number before it goes into a paper.
See the full mapping with check_questionnaire(data, "sus", ...); silence this note with options(colleyRstats.quiet_questionnaires = TRUE).
CAUTION: item numbers, order and polarity depend on how the questionnaire
was administered -- survey tools renumber items, translations reorder them,
and short forms drop them. This package applies the PUBLISHED key. If your
sheet differed, the scores will be wrong without any error being raised.
Check the mapping against the survey your participants actually saw, and
double-check any number before it goes into a paper.
See the full mapping with check_questionnaire(data, "sus", ...); silence this note with options(colleyRstats.quiet_questionnaires = TRUE).
Scoring System Usability Scale (SUS) from 10 columns (sus_1 ... sus_10, matched by item number); responses observed 3 to 3 on the assumed range 1 to 5; reverse-coded: sus2, sus4, sus6, sus8, sus10.
See the full mapping with check_questionnaire(data, "sus", ...); silence this note with options(colleyRstats.quiet_questionnaires = TRUE).
Scoring System Usability Scale (SUS) from 10 columns (sus_1 ... sus_10, matched by item number); responses observed 3 to 3 on the assumed range 1 to 5; reverse-coded: sus2, sus4, sus6, sus8, sus10.
See the full mapping with check_questionnaire(data, "sus", ...); silence this note with options(colleyRstats.quiet_questionnaires = TRUE).
System Usability Scale (SUS) -- Brooke (1996), Usability Evaluation in Industry; Lewis & Sauro (2009), HCII
Assumed response range: 1-5 (the instrument's own; pass `scale` if your survey differed)
Observed responses: 3 to 3
Item mapping (verify against the survey your participants saw):
item code column subscale reverse observed_min observed_max n_missing
1 sus1 sus_1 Usability FALSE 3 3 0
2 sus2 sus_2 Usability TRUE 3 3 0
3 sus3 sus_3 Usability FALSE 3 3 0
4 sus4 sus_4 Learnability TRUE 3 3 0
5 sus5 sus_5 Usability FALSE 3 3 0
6 sus6 sus_6 Usability TRUE 3 3 0
7 sus7 sus_7 Usability FALSE 3 3 0
8 sus8 sus_8 Usability TRUE 3 3 0
9 sus9 sus_9 Usability FALSE 3 3 0
10 sus10 sus_10 Learnability TRUE 3 3 0
Notes:
- The SUS score is the sum of the ten recoded items multiplied by 2.5, giving 0-100. That is a percentage of the maximum, NOT a percentile: the mean SUS across studies is about 68, so 68 is average rather than poor.
- Usability (8 items, x 3.125) and Learnability (2 items, x 12.5) follow Lewis & Sauro (2009) and are likewise on 0-100. The two are strongly correlated; report them only if the study has a reason to separate them.
- Missing items: Brooke (1996) instructs a respondent who cannot answer an item to mark the centre point of the scale, so when `min_valid` < 1 lets an incomplete row be scored, each missing item counts as the centre (3 on 1-5). One missing item in an otherwise perfect response therefore gives 95, not 100. Pass impute = "prorate" to scale up from the answered items instead, and say which rule you used.
- Learnability has only two items. score_reliability() reports the Spearman-Brown coefficient for it, which Eisinga, te Grotenhuis & Pelzer (2013) recommend over alpha for two-item scales.
System Usability Scale (SUS) -- Brooke (1996), Usability Evaluation in Industry; Lewis & Sauro (2009), HCII
Response range 1-5; 10 items in 2 subscales.
- The SUS score is the sum of the ten recoded items multiplied by 2.5, giving 0-100. That is a percentage of the maximum, NOT a percentile: the mean SUS across studies is about 68, so 68 is average rather than poor.
- Usability (8 items, x 3.125) and Learnability (2 items, x 12.5) follow Lewis & Sauro (2009) and are likewise on 0-100. The two are strongly correlated; report them only if the study has a reason to separate them.
- Missing items: Brooke (1996) instructs a respondent who cannot answer an item to mark the centre point of the scale, so when `min_valid` < 1 lets an incomplete row be scored, each missing item counts as the centre (3 on 1-5). One missing item in an otherwise perfect response therefore gives 95, not 100. Pass impute = "prorate" to scale up from the answered items instead, and say which rule you used.
- Learnability has only two items. score_reliability() reports the Spearman-Brown coefficient for it, which Eisinga, te Grotenhuis & Pelzer (2013) recommend over alpha for two-item scales.
item code
1 sus1
2 sus2
3 sus3
4 sus4
5 sus5
6 sus6
7 sus7
8 sus8
9 sus9
10 sus10
label
I think that I would like to use this system frequently
I found the system unnecessarily complex
I thought the system was easy to use
I think that I would need the support of a technical person to be able to use this system
I found the various functions in this system were well integrated
I thought there was too much inconsistency in this system
I would imagine that most people would learn to use this system very quickly
I found the system very cumbersome to use
I felt very confident using the system
I needed to learn a lot of things before I could get going with this system
subscale reverse
Usability FALSE
Usability TRUE
Usability FALSE
Learnability TRUE
Usability FALSE
Usability TRUE
Usability FALSE
Usability TRUE
Usability FALSE
Learnability TRUE
Scoring System Usability Scale (SUS) from 10 columns (sus_1 ... sus_10, matched by item number); responses observed 3 to 3 on the assumed range 1 to 5; reverse-coded: sus2, sus4, sus6, sus8, sus10.
See the full mapping with check_questionnaire(data, "sus", ...); silence this note with options(colleyRstats.quiet_questionnaires = TRUE).
item code column subscale reverse observed_min observed_max n_missing
1 sus1 sus_1 Usability FALSE 3 3 0
2 sus2 sus_2 Usability TRUE NA NA 1
3 sus3 sus_3 Usability FALSE 3 3 0
4 sus4 sus_4 Learnability TRUE 3 3 0
5 sus5 sus_5 Usability FALSE 3 3 0
6 sus6 sus_6 Usability TRUE 3 3 0
7 sus7 sus_7 Usability FALSE 3 3 0
8 sus8 sus_8 Usability TRUE 3 3 0
9 sus9 sus_9 Usability FALSE 3 3 0
10 sus10 sus_10 Learnability TRUE 3 3 0
CAUTION: item numbers, order and polarity depend on how the questionnaire
was administered -- survey tools renumber items, translations reorder them,
and short forms drop them. This package applies the PUBLISHED key. If your
sheet differed, the scores will be wrong without any error being raised.
Check the mapping against the survey your participants actually saw, and
double-check any number before it goes into a paper.
See the full mapping with check_questionnaire(data, "sus", ...); silence this note with options(colleyRstats.quiet_questionnaires = TRUE).
item code column subscale reverse observed_min observed_max n_missing
1 fms fms FMS FALSE 0 12 0
The NPAV found a significant main effect of \Video{} on mental workload (\F{1}{10}{6.12}, \p{0.033}, $\eta_{p}^{2}$ = 0.38, 95\% CI: [0.00, 0.68]).
The NPAV found a significant interaction effect of \gesture{} $\times$ \eHMI{} on mental workload (\F{1}{10}{5.01}, \p{0.045}, $\eta_{p}^{2}$ = 0.33, 95\% CI: [0.00, 0.66]).
The ART found a significant interaction effect of \gesture{} $\times$ \eHMI{} on mental demand (\F{1}{10}{5.01}, \p{0.045}, $\eta_{p}^{2}$ = 0.33, 95\% CI: [0.00, 0.66]).
The nparLD analysis found no significant effects on TLX1.
Saving _problems/test-reporting-180.R
tlx\_B: \m{8.00}, \sd{1.58}
A Games-Howell post-hoc test (Holm-adjusted) found that 4 was significantly higher (\m{26.66}, \sd{4.51}) in terms of mpg compared to 8 (\m{15.10}, \sd{2.56}; \padjminor{0.001}).
A Games-Howell post-hoc test (Holm-adjusted) found that 6 was significantly higher (\m{19.74}, \sd{1.45}) in terms of mpg compared to 8 (\m{15.10}, \sd{2.56}; \padjminor{0.001}).
A Games-Howell post-hoc test (Holm-adjusted) found that 4 was significantly higher (\m{26.66}, \sd{4.51}) in terms of mpg compared to 8 (\m{15.10}, \sd{2.56}; \padjminor{0.001}).
A Games-Howell post-hoc test (Holm-adjusted) found that 6 was significantly higher (\m{19.53}, \sd{1.47}) in terms of mpg compared to 8 (\m{15.10}, \sd{2.56}; \padjminor{0.001}).
A Games-Howell post-hoc test (Holm-adjusted) found that FourCyl was significantly higher (\m{26.66}, \sd{4.51}) in terms of mpg compared to 8 (\m{15.10}, \sd{2.56}; \padjminor{0.001}).
A Games-Howell post-hoc test (Holm-adjusted) found that 6 was significantly higher (\m{19.74}, \sd{1.45}) in terms of mpg compared to 8 (\m{15.10}, \sd{2.56}; \padjminor{0.001}).
Kruskal-Wallis rank sum test
data: x and g
Kruskal-Wallis chi-squared = 96.9374, df = 2, p-value = 0
Dunn's Pairwise Comparison of x by g
(Holm)
Col Mean-│
Row Mean │ setosa versicol
─────────┼──────────────────────
versicol │ -6.106326
│ 0.0000*
│
virginic │ -9.741784 -3.635458
│ 0.0000* 0.0003*
FWER = 0.05
Reject Ho if adjusted p ≤ FWER with stopping rule, where (unadjusted) p = Pr(|Z| ≥ |z|)
A Dunn post-hoc test (Holm-adjusted) found that Sepal.Length for the \Species{} virginica was significantly higher (\mdn{6.50}, \iqr{0.67}) than for setosa (\mdn{5.00}, \iqr{0.40}; \padjminor{0.001}, \rankbiserial{0.97}) and versicolor (\mdn{5.90}, \iqr{0.70}; \padjminor{0.001}, \rankbiserial{0.58}).
% latex table generated in R 4.7.0 by xtable 1.8-8 package
% Tue Nov 10 06:42:20 2026
\begin{table}[ht]
\centering
\caption{Dunn post-hoc comparisons for independent variable \Species{} and dependent variable Sepal.Length. Positive Z-values mean that the first-named level has the higher mean rank; for negative Z-values, the second-named level does. p-values are Holm-adjusted. Effect size reported as rank-biserial correlation (r).}
\label{tab:posthoc-Species-Sepal.Length}
\begingroup\small
\begin{tabular}{lrll}
\hline
Comparison & Z & p-adjusted & r \\
\hline
setosa - versicolor & -6.1063 & $<$0.001 & 0.87 \\
setosa - virginica & -9.7418 & $<$0.001 & 0.97 \\
versicolor - virginica & -3.6355 & $<$0.001 & 0.58 \\
\hline
\end{tabular}
\endgroup
\end{table}
Kruskal-Wallis rank sum test
data: x and g
Kruskal-Wallis chi-squared = 96.9374, df = 2, p-value = 0
Dunn's Pairwise Comparison of x by g
(Holm)
Col Mean-│
Row Mean │ setosa versicol
─────────┼──────────────────────
versicol │ -6.106326
│ 0.0000*
│
virginic │ -9.741784 -3.635458
│ 0.0000* 0.0003*
FWER = 0.05
Reject Ho if adjusted p ≤ FWER with stopping rule, where (unadjusted) p = Pr(|Z| ≥ |z|)
% latex table generated in R 4.7.0 by xtable 1.8-8 package
% Tue Nov 10 06:42:20 2026
\begin{table}[ht]
\centering
\caption{Dunn post-hoc comparisons for independent variable \Species{} and dependent variable Sepal.Length. Positive Z-values mean that the first-named level has the higher mean rank; for negative Z-values, the second-named level does. p-values are Holm-adjusted. Effect size reported as rank-biserial correlation (r).}
\label{tab:posthoc-Species-Sepal.Length}
\begingroup\small
\begin{tabular}{lrll}
\hline
Comparison & Z & p-adjusted & r \\
\hline
setosa - versicolor & -6.1063 & $<$0.001 & 0.87 \\
setosa - virginica & -9.7418 & $<$0.001 & 0.97 \\
versicolor - virginica & -3.6355 & $<$0.001 & 0.58 \\
\hline
\end{tabular}
\endgroup
\end{table}
NOTE: Results may be misleading due to involvement in interactions
An ART-C post-hoc test (Holm-adjusted) found that score for the \mode{} Hand was significantly higher (\mdn{6.12}, \iqr{1.04}) than for Both (\mdn{1.96}, \iqr{1.05}; \padjminor{0.001}, \rankbiserial{1.00}) and Eye (\mdn{3.86}, \iqr{1.37}; \padjminor{0.001}, \rankbiserial{1.00}).
% latex table generated in R 4.7.0 by xtable 1.8-8 package
% Tue Nov 10 06:42:21 2026
\begin{table}[ht]
\centering
\caption{Post-hoc ART-C contrasts for independent variable \mode{} and dependent variable \score{}. Positive t-values mean that the first-named level is higher than the second-named on the aligned-rank scale; for negative t-values, the second-named level is. p-values are Holm-adjusted. Effect size reported as rank-biserial correlation (r).}
\label{tab:artcon-mode-score}
\begingroup\small
\begin{tabular}{lrlll}
\hline
Comparison & t & df & p-adjusted & r \\
\hline
Both - Eye & -6.64 & 36 & $<$0.001 & 0.97 \\
Both - Hand & -13.95 & 36 & $<$0.001 & 1.00 \\
Eye - Hand & -7.31 & 36 & $<$0.001 & 1.00 \\
\hline
\end{tabular}
\endgroup
\end{table}
NOTE: Results may be misleading due to involvement in interactions
NOTE: Results may be misleading due to involvement in interactions
An ART-C post-hoc test (Holm-adjusted) found that score for the \mode{} Hand was significantly higher (\mdn{6.12}, \iqr{1.04}) than for Both (\mdn{1.96}, \iqr{1.05}; \padjminor{0.001}, \rankbiserial{1.00}) and Eye (\mdn{3.86}, \iqr{1.37}; \padjminor{0.001}, \rankbiserial{0.93}).
NOTE: Results may be misleading due to involvement in interactions
Saving _problems/test-reporting-1108.R
A Games-Howell post-hoc test (Holm-adjusted) found that A was significantly higher (\m{5.50}, \sd{0.71}) in terms of val compared to B (\m{1.50}, \sd{0.71}; \padj{0.010}).
[ FAIL 2 | WARN 0 | SKIP 19 | PASS 1637 ]
══ Skipped tests (19) ══════════════════════════════════════════════════════════
• On CRAN (19): 'test-latex-output.R:422:3', 'test-plotting.R:146:3',
'test-reporting.R:923:3', 'test-reporting.R:1052:3',
'test-reporting.R:1089:3', 'test-reporting.R:1111:3',
'test-reporting.R:1135:3', 'test-reporting.R:1159:3',
'test-reporting.R:1178:3', 'test-reporting.R:1190:3',
'test-reporting.R:1259:3', 'test-reporting.R:1302:3',
'test-reporting.R:1394:3', 'test-reporting.R:1437:3',
'test-reporting.R:1451:3', 'test-setup.R:88:3', 'test-setup.R:115:3',
'test-study-project.R:34:3', 'test-study-project.R:243:3'
══ Failed tests ════════════════════════════════════════════════════════════════
── Failure ('test-reporting.R:177:3'): reportNparLD reports a real nparLD fit ──
Expected `suppressWarnings(reportNparLD(model, dv = "TLX1"))` to throw a message.
── Failure ('test-reporting.R:1108:3'): reportNparLD writes a compilable F(df, infinity) with the fractional ATS df ──
Expected `result` to match string "\\F{1.8}{\\infty}{".
Actual text:
x | The nparLD analysis found no significant effects on tlx\_mental.
[ FAIL 2 | WARN 0 | SKIP 19 | PASS 1637 ]
Error:
! Test failures.
Execution halted
Flavor: r-devel-linux-x86_64-fedora-gcc