Here is the first try via datasummary_skim function from the modelsummary package. Considering the nature of our subset data, it doesn’t show much details about teacher participant in the year of 2018 and 2020 respectively.

At here, I added output = "gt" to apply functions from the gt package within datasummary_skim. Surprisingly, it worked! gt package gives me flexibility to customize my table. However, this version of summary table looks still robust.

I then decided to subgroup the dataset by country and year to show the counts and percentage for each category. To highlight the highest level of ICT use frequency in the table, I used tab_style and the location argument to highlight everyday use. I also choosed a blind friendly color in this process. This version was almost to what I need, but the whole table still needed further adjustment.

The initial attempt helps me to map as much information as possible on a stack bar plot. However, it looks very busy and doesn’t deliver the message efficiently.

In my second attempt, I adjust the position of bars and applied scale_fill_OkabeIto() to make it color-blind friendly.

I tried coord_flip()in this attempt mainly because the level labels on the x-axis took too much of the lower space. However, we can see all the key information mapped on this plot are either clustered or separated. The overall flow is not good.

My first attempt was to map the recoded school priority with a stack bar to make it vertically visuable, but it was difficult to tell the difference across different levels of ICT use frequency.

I applied scale_fill_viridis_d to ajust the color for my plot. In addition, I use scale_y_log10 to adjust the scale of y-axis. However, the change of scale does’t show the disparity between school ICT priority.

My first thought for this plot was to using geom_tile to construct a heatmap, but as it is shown abobe, the darkest purple shades are hard to differentiate. In addition, the heatmap doesn’t give much details on the difference among each countries. A good practice in this version is that I tried scale_fill_viridis_c to change the color.

I then tried geom_bar to separate the ICT use in different subjects by country and years. It is important to show the difference between 2018 and 2020 because thses two years demonstrate the years before and after the pandemic. The labels on x-axis are clustered, so I applied ggplotly for interaction and use axis.text.x = element_blank() to remove the labels. There would be some trade-off, though, if audience can only see this plot on a non-accessible screen. Therefore, I choose to not apply it.

In this plot, I converted counts to percentages and flips bars horizontally. Instead of using scale_fill_viridis_c, I tried scale_fill_manual to adjust the colors.