Exploring Alt-Text Contents

### write file to enter rubric data
write.csv(annotatedL2, file="../data/annotated.csv", row.names=FALSE)

I started with the file “data/AltTextSubset_Annotated.rds” from Silvia. I kept only the columns I needed, excluding the hashtag variable that contained lists that couldn’t be written easily to .csv. I also removed rows where AltText == “ScrapingError”. I ended up with 215 observations.

I added columns for the rubric because I was having trouble adding rows to the spreadsheet where. I evaluated each alt-text on five items:

This is a subjective evaluation in that it is made without any access to the graphic, and I’m sure I’m using my knowledge of different kinds of graphs to fill in some of the gaps. Also I’m not familiar with some of the newer, fancier types of plots (waterfall?).

I saved this as a .csv and filled the rubric item fields in Numbers. I saved my results as a .csv and read it back into R.

After removing duplicate tweets and tweets that seemed to contain only photos or screenshots that didn’t contain data visualizations, there were 196 tweets that contained alt-text. The following table contains frequencies of presence (1) or absence (0) of alt-text rubric items.

Characteristic N = 1961
conveysMeaning
0 130 (66)
1 66 (34)
NA 0
axisInfo
0 121 (72)
1 48 (28)
NA 27
axisOrValueScale
0 148 (88)
1 21 (12)
NA 27
plotType
0 86 (44)
1 109 (56)
NA 1
repeatTitle
0 177 (91)
1 18 (9.2)
NA 1

1 n (%)