If you have any questions or comments, please put them below.
2 Comments
meetings
May 11, 2021
Hello! What an interesting and timely project! I have some questions:
1) I am not a statistician. Can you explain what a p-value is, and its importance, in addressing your research questions?
2) I like your idea of moving to auditory "transcripts" next. I also find myself wondering whether the same video might get one political classification based on its written transcript (say, minimal left), but a different one based on listening to it, and especially to seeing its visuals (for instance, extreme visuals might change the rating to extreme left). Do you have any insight into whether this kind of thing can happen?
3) I also wonder about your thoughts in regard to actual human responses to the algorithm.…
Like
Michael Lutz
May 16, 2021
Replying to
Thank you for your response!
1) To answer your first question, we used the p-value to determine the statistical significance of our results. There are many ways to determine p-values, but we used a two-sample t-test for our first experiment to evaluate whether the top video search bias values are, on average, different from the value of the entire scraped data set (which had an average of 0). Since our p-value is less than 0.01, there is a 99% chance that the average bias in the top 3 search results is different than the average bias in the top 200 videos. For our second experiment, we used the same test to compare the bias values of the first and last…
Hello! What an interesting and timely project! I have some questions:
1) I am not a statistician. Can you explain what a p-value is, and its importance, in addressing your research questions?
2) I like your idea of moving to auditory "transcripts" next. I also find myself wondering whether the same video might get one political classification based on its written transcript (say, minimal left), but a different one based on listening to it, and especially to seeing its visuals (for instance, extreme visuals might change the rating to extreme left). Do you have any insight into whether this kind of thing can happen?
3) I also wonder about your thoughts in regard to actual human responses to the algorithm.…