aphantasiaCognitiveClustering

Overview

The main dataset used for the analyses is provided in the package as study_data. It contains the data from the study in a tidy format, ready for analysis.

library(aphantasiaCognitiveClustering)
#> Welcome to aphantasiaCognitiveClustering.
#> This is actually a data analysis project 'disguised' as a package.
#> It contains the code and data to reproduce the analyses presented
#> in an article.
#> See https://osf.io/7vsx6/ for the complete study materials.

head(study_data)
#>     id age sex      group      education
#> 1 7210  25   m    Control         Master
#> 2 7213  22   f    Control         Master
#> 3 7238  52   f    Control       Bachelor
#> 4 7242  48   f    Control Post-secondary
#> 5 7254  37   f Aphantasic      Doctorate
#> 6 7257  36   m    Control       Bachelor
#>                                       field field_code
#> 1  Social sciences, journalism, information          3
#> 2  Social sciences, journalism, information          3
#> 3 Natural sciences, mathematics, statistics          5
#> 4             Business, Administration, Law          4
#> 5                        Health and Welfare          9
#> 6 Natural sciences, mathematics, statistics          5
#>                    occupation occupation_code vviq osivq_o osivq_s osivq_v
#> 1                      Health               5   65      45      45      54
#> 2                     Student               3   59      47      45      47
#> 3 Information, Communications               8   43      44      46      44
#> 4    Business, Administration               7   73      46      35      23
#> 5                      Health               5   16      32      30      50
#> 6    Business, Administration               7   55      64      48      47
#>   psiq_vis psiq_aud psiq_od psiq_gout psiq_tou psiq_sens psiq_feel score_raven
#> 1     8.00     7.67       8      7.33     8.33      8.67      8.00          14
#> 2     7.67     9.00       6      8.67     9.00      9.00      7.00          17
#> 3     7.00     7.33       7      6.67     7.00      6.00      7.00          11
#> 4     7.67     8.33       9      7.00     8.67      7.33      8.00           7
#> 5     1.00     1.00       1      1.00     1.00      1.00      1.00          12
#> 6     6.67     8.00       7      3.33     5.00      7.00      8.67          12
#>   score_sri span_spatial span_digit wcst_accuracy score_similarities
#> 1        27         5.43       6.07            78                 30
#> 2        26         3.79       6.14            72                 27
#> 3        22         2.79       3.93            67                 26
#> 4        15         2.14       3.50            25                 20
#> 5        16         3.36       3.07            45                 22
#> 6        26         4.00       3.57            23                 29
#>   score_comprehension
#> 1                  24
#> 2                  23
#> 3                  24
#> 4                  16
#> 5                  20
#> 6                   9

The main analyses can be roughly divided into three steps: modelling observed scores with two groups, clustering the sample based on a selection of variables, and exploring these clusters in the light of their scores. These core steps are taken care of by the following functions:

• get_association_models(): This function computes Bayes Factors for association between the groups (aphantasia/control) and the education, field, and occupation variables in the dataset.

• get_full_model_table(): This function fits Bayesian models to the observed variables, predicting them with the two groups (aphantasia/control) and the age as a covariate. It returns a data frame summarising the model results for each variable.

• correlate_vars(): This function computes the correlation matrix of the observed variables, returning a data frame with the correlations and their significance levels. It allowed to identify the variables that were most relevant for clustering.

• cluster_selected_vars(): This function performs clustering on a selection of variables, returning a data frame with the cluster assignments for each participant. We can then reuse get_association_models() and get_full_model_table() to fit new models with these clusters.

Some transformations of the shape of the data are also required to prepare them for certain steps, which are handled by other functions in the package. Finally, a suite of functions is provided to visualise the results of the analyses, such as plot_score_violins(), plot_score_radars(), plot_score_cor_matrix(), etc.

See the complete data analysis report this package was built for or the reproducible manuscript it led to.