erwan/phd-thesis
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Complete comments

Browse Source
This commit is contained in:
Erwan Normand
2025-04-18 11:21:19 +02:00
parent 3de6ad37df
commit 0a0e1ff4b5
11 changed files with 35 additions and 38 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
3-perception/vhar-textures/2-experiment.tex
View File

@@ -122,5 +122,5 @@ After each of the two tasks, participants answered to the following 7-item Liker
In an open question, participants also commented on their strategy for completing the \level{Matching} task (\enquote{How did you associate the tactile textures with the visual textures?}) and the \level{Ranking} task (\enquote{How did you rank the textures?}).
\comans{JG}{I suggest to also report on [...] the software packages used for statistical analysis (this holds also for the subsequent chapters).}{This has been added to all chapters where necessary.}
The results were analyzed using R (v4.4) and the packages \textit{afex} (v1.4), \textit{ARTool} (v0.11), \textit{corrr} (v0.4), \textit{FactoMineR} (v2.11), \textit{lme4} (v1.1), and \textit{performance} (v0.13).
\comans{JG}{I suggest to also report on [...] the software packages used for statistical analysis (this holds also for the subsequent chapters).}{This has been added to all chapters where necessary.}

2
3-perception/vhar-textures/3-results.tex
View File

@@ -7,10 +7,10 @@
\paragraph{Confusion Matrix}
\label{results_matching_confusion_matrix}
\comans{JG}{For the two-sample Chi-Squared tests in the matching task, the number of samples reported is 540 due to 20 participants conducting 3 trials for 9 textures each. However, this would only hold true if the repetitions per participant would be independent and not correlated (and then, one could theoretically also run 10 participants with 6 trials each, or 5 participants with 12 trials each). If they are not independent, this would lead to an artificial inflated sample size and Type I error. If the trials are not independent (please double check), I suggest either aggregating data on the participant level or to use alternative models that account for the within-subject correlation (as was done in other chapters).}{Data of the three confusion matrices have been aggregated on the participant level and analyzed using a Poisson regression.}
\figref{results/matching_confusion_matrix} shows the confusion matrix of the \level{Matching} task with the visual textures and the proportion of haptic texture selected in response, \ie the proportion of times the corresponding \response{Haptic Texture} was selected in response to the presentation of the corresponding \factor{Visual Texture}.
To determine which haptic textures were selected most often, the repetitions of the trials were first aggregated by counting the number of selections per participant for each (\factor{Visual Texture}, \response{Haptic Texture}) pair.
An \ANOVA based on a Poisson regression (no overdispersion was detected) indicated a statistically significant effect on the number of selections of the interaction \factor{Visual Texture} \x \response{Haptic Texture} (\chisqr{64}{180}{414}, \pinf{0.001}).
\comans{JG}{For the two-sample Chi-Squared tests in the matching task, the number of samples reported is 540 due to 20 participants conducting 3 trials for 9 textures each. However, this would only hold true if the repetitions per participant would be independent and not correlated (and then, one could theoretically also run 10 participants with 6 trials each, or 5 participants with 12 trials each). If they are not independent, this would lead to an artificial inflated sample size and Type I error. If the trials are not independent (please double check), I suggest either aggregating data on the participant level or to use alternative models that account for the within-subject correlation (as was done in other chapters).}{Data of the three confusion matrices have been aggregated on the participant level and analyzed using a Poisson regression.}
Post-hoc pairwise comparisons using the Tukey's \HSD test then indicated there was statistically significant differences for the following visual textures:
\begin{itemize}
\item With \level{Sandpaper~320}, \level{Coffee Filter} was more selected than the other haptic textures (\ztest{3.4}, \pinf{0.05} each) except \level{Plastic Mesh~1} and \level{Terra Cotta}.