Analysis of Modern Music Criticism
Introduce the topic and motivation
Our group was interested in non-academic subjects and we found two datasets related to music journalism that piqued our interest.
One of the datasets is from a well-known music publication called Pitchfork, which has been reviewing albums since the 90s. The other is from Anthony Fantano, a music blogger and YouTuber who has been reviewing music since the early 2010s.
The research questions we have decided to explore are:
Is the source of a review a significant factor in determining the review’s score?
What’s the difference in pitchfork people’s taste over time for folk/country music and rock music?
Introduce the data
Each observation contains an album review and information such as when the review was published, the score the album received, the artist name and album genre, et cetera.
The datasets differ somewhat: for example, the Pitchfork data set has an author column, but the Fantano one doesn’t, as he wrote every review.
Highlights from EDA
Mean score for Folk/Country is 7.26 and mean score for Rock is around 7.00
Peak year for Folk/Country is around 1998, and peak year for Rock is around 1962 (the earliest year in the dataset)
Both genres exhibit a downward trend in popularity indicated by pitchfork scores
Inference/modeling/other analysis 1
- Point estimate was about -1.02, which is entirely outside of generated distribution
- P-value is 0
Reject null hypothesis: source affects score
Fantano is more likely to rate albums lower
Inference/modeling/other analysis 2
- Point estimate is about .76, also outside of generated distribution
- P-value is 0
Reject null hypothesis: Difference between genre scores is significant
Folk / Country albums are more likely to score higher.
Conclusions + future work
Conclusion:
After performing our two analyses, we found that our data provided convincing evidence that the score of a review is not independent (therefore, significant) from what source it originates from and that the difference between the mean scores of Rock and Folk/Country albums, before and after 1990, is significant.
Future work:
An interesting suggestion that we received from our peers for future work is exploring sources of music reviews from smaller communities, such as from college campuses, because we may able to see how the sources of this kind of data may affect ratings of certain genres. For example, we can explore the ratings of certain genres and their relationships between variables such as major, dorm, where they’re from, etc.