clean the section below. fix the language and check for ai or general plagiarism. The literature review is attached as an excel. There is also the paper that I described the equation for handling ceiling effect. I need you clean the section below. fix the language and check for ai or general plagiarism. and you can see the attachments as reference:
To investigate how psychological and educational researchers have statistically handled ceiling/floor data in post-hoc methods for multiple comparisons, a brief literature review was conducted.
I collected 200 English articles published within a
five-year span that mentioned “ceiling effects” or “floor effects,” illustrating the presence of ceiling and floor effects in
the literature. Among the articles, we focused on reviewing
those that were published in journals with higher impact factors (i.e., five-year impact factor > 2). As examples, we
reviewed articles from the Journal of Experimental
Psychology, Psychological Science, American Educational
Research Journal, and Child Development.
I collected 200 English articles published within a
five-year span that mentioned “ceiling effects” or “floor effects,” illustrating the presence of ceiling and floor effects in
the literature. Among the articles, we focused on reviewing
those that were published in journals with higher impact factors (i.e., five-year impact factor > 2). As examples, we
reviewed articles from the Journal of Experimental
Psychology, Psychological Science, American Educational
Research Journal, and Child Development.
After excluding papers focused on methodology and literature review, the review of post-hoc methods for multiple comparisons and handling of ceiling/floor effects in the 60 sources revealed diverse approaches. Most studies utilized specific post-hoc methods to address multiple comparisons. Bonferroni corrections were prominently used, appearing in 15% (9 studies) of the cases. Notable examples include \cite{peterson2019saccadic} and \cite{herde2022retinotopically}, who applied Bonferroni corrections to correct for Type II error. Other methods, such as Tukey’s HSD, were employed in 13.33% (8 studies) of the cases. Holm-Bonferroni corrections were used in 21.6% (13 studies). Notably, the studies using Holm-Bonferroni and Bonferroni corrections did not specify the tests used, such as Scheffé, Tukey HSD, or pairwise t-tests. Additionally, methods like Sidak-adjusted pairwise comparisons and Student–Newman–Keuls tests were also observed, reflecting tailored strategies to meet the unique demands of each study.
Handling ceiling and floor effects varied significantly across the studies. 8.33 % (5 studies) noted their ceiling/floor data percentages, with \cite{newman2019effects} mentioning a potential ceiling effect in 3% of the data. Approximately 76.66% (46 studies), including \cite{huggins2021autistic}, chose to ignore potential ceiling or floor effects, either due to their minimal impact or through acknowledgment without specific adjustments. In contrast, 24.33% (14 studies) took active measures to mitigate these effects. Specifically, 6.66% (4 studies) removed ceiling/floor data. For instance, \cite{chierchia2020prosocial} employed data truncation to address skewed results. Furthermore, 13.33% (8 studies), including \cite{senftleben2021stay} and \cite{samuel2020reduced}, repeated their experiments to validate their findings. Notable examples include \cite{adams2021introspective}, who addressed floor effects by increasing the overall magnitude of oculomotor capture effects by color singletons in Experiment 2, and \cite{wiesmann2022makes}, who repeated the experiment with more controlled viewing conditions and added various stimulus conditions and SOAs to diversify task difficulty levels. Despite the changes, ceiling effects was still observed but at a reduced rate. Therefore, they also used generalized linear mixed models (GLMMs) for detailed statistical analysis to compare performance across different conditions, allowing for more meaningful findings despite the remaining presence of ceiling effects. Additionally, 2 studies employed other methods to handle ceiling/floor effects. \cite{shepherdson2018working} address ceiling effects indirectly by using a hierarchical drift diffusion model (HDDM) that handles near-ceiling accuracy by including both accuracy and reaction time data, ensuring reliable parameter estimates despite high accuracy levels. The mathematical reasoning for choosing this method was not described.
One notable study by \citet{multisensory2020} has handled the ceiling effect by using a method that \citet{macmillan1985detection} recommends which is to adjustments based on Signal Detection Theory (SDT). The sensitivity index \(d’\) is calculated using the z-transformed hit rate \(z(H)\) and false-alarm rate \(z(FA)\) with the formula:
\begin{equation}
d’ = z(H) – z(FA)
\end{equation}
When proportions are 0 or 1, adjustments are made by replacing 0 with \(\frac{1}{2N}\) and 1 with \(1 – \frac{1}{2N}\), where \(N\) is the number of trials. This ensures the proportions remain within a calculable range, avoiding infinite values in the z-transformation. Using these adjusted hit and false-alarm rates, the recalculated sensitivity index \(d’\) remains finite and accurate. This method was applied in the study \citet{multisensory2020} to handle ceiling effects and ensure robust statistical analysis.
\begin{equation}
d’ = z(H) – z(FA)
\end{equation}
When proportions are 0 or 1, adjustments are made by replacing 0 with \(\frac{1}{2N}\) and 1 with \(1 – \frac{1}{2N}\), where \(N\) is the number of trials. This ensures the proportions remain within a calculable range, avoiding infinite values in the z-transformation. Using these adjusted hit and false-alarm rates, the recalculated sensitivity index \(d’\) remains finite and accurate. This method was applied in the study \citet{multisensory2020} to handle ceiling effects and ensure robust statistical analysis.