A next step?
The paper groups all the respondents as "people". As the study is focused on the "people's" reaction to human / robot and race, it would be helpful to know the same factors (and gender, too) of the people who responded to the videos.
For instance is there one gender of respondent that is more or less dehumanising towards any particular group (as represented by the robots / celebrities in the videos) and likewise with the respondents other characteristics?
The report says that non-English comments were discarded - presumably for practical reasons. But a deeper drill-down into the age, gender, race, geographical location and probably many other attributes of the people who responded: either positively or negatively would be illuminating.