Grad Cafe Political Science data stuff

[Clintarius]

Hi all!

I've been stuck home because of a health-related issue, so (in addition to watching a lot of Brooklyn Nine-Nine) I scraped the tables from the Grad Cafe political science results page since its beginning in 2006 and here are a few things I thought I'd share! I'm trying to set up a GitHub page to make the scraping code and data accessible if you guys want to play with it.

Disclaimers:

1. I only used data that matches "political science" and "PhD". So 1) these things don't apply to Master's degree applicants, and 2) the data is "biased against" (includes less posts from) Harvard, Princeton, NYU, etc. because they use other degree names (Government, Politics, etc.).
2. I only kept US schools that are in the top 100 based on the current USNWR, because cleaning the data would have taken me too much time otherwise. This is not an assessment of the quality of any school!
3. I know that things change fast, so the data from early years might not mean much. This is not meant to give any lesson, I just thought it might be interesting to some people. ?
4. Data is from this morning in Europe (Feb 22nd, 2020).

First: the average grades since 2006 (for the GRE, it only includes years with the new system):

• Average reported GPA: 3.75
• Average Verbal GRE: 163.4
• Average Quant GRE: 160.9
• Average Writing GRE: 3.8

Second: the distribution of the posts between A/I/U and decisions:

Third: the distribution of the post between schools. I don't have enough space left so I'll upload these pictures in a comment. 

Any thoughts based on this? I can also look at other metrics if you guys think it'd be interesting! (Next thing planned is visualising the dates at which decisions are received).

PS 1: thanks again for all the support and positivity on this forum! ?
PS 2: there probably are a some coding mistakes, so once again I'm not pretending this gives any lesson!

Edited February 22, 2020 by Clintarius
Adding the comment about reliability.

[Clintarius]

[ihatedecisions]

I remember coming across a paper or a blog post that did similar analysis a couple of years back. Cant find it anymore though

[Dwar]

Thanks so much for this! it is super interesting to see all this data in one format like that.  

I also think it's pretty interesting how this data puts the overall website user bias on display so clearly. For example, the average GRE scores that you found are pretty far ahead of the average GRE scores for political science that ETS reports (157v 152q). I think it just goes to show that people should not take this site as a general representation of the field. 

This isn't meant to disparage your work or anything like that, just making an observation. 

[Clintarius]

45 minutes ago, Dwar said:

Thanks so much for this! it is super interesting to see all this data in one format like that.  

I also think it's pretty interesting how this data puts the overall website user bias on display so clearly. For example, the average GRE scores that you found are pretty far ahead of the average GRE scores for political science that ETS reports (157v 152q). I think it just goes to show that people should not take this site as a general representation of the field. 

This isn't meant to disparage your work or anything like that, just making an observation. 

Yeah I agree. The treemap also interestingly shows how the admission rate on the GC result page is much higher than the true value (which makes sense, I guess most people prefer sharing their successes), and that can make people feel insecure about rejections even though they are in reality part of a much larger majority.

[Dwar]

14 minutes ago, Clintarius said:

, I guess most people prefer sharing their successes), and that can make people feel insecure about rejections even though they are in reality part of a much larger majority

100% 

[Theory007]

2 hours ago, Clintarius said:

Hi all!

I've been stuck home because of a health-related issue, so (in addition to watching a lot of Brooklyn Nine-Nine) I scraped the tables from the Grad Cafe political science results page since its beginning in 2006 and here are a few things I thought I'd share! I'm trying to set up a GitHub page to make the scraping code and data accessible if you guys want to play with it.

Disclaimers:

1. I only used data that matches "political science" and "PhD". So 1) these things don't apply to Master's degree applicants, and 2) the data is "biased against" (includes less posts from) Harvard, Princeton, NYU, etc. because they use other degree names (Government, Politics, etc.).
2. I only kept US schools that are in the top 100 based on the current USNWR, because cleaning the data would have taken me too much time otherwise. This is not an assessment of the quality of any school!
3. I know that things change fast, so the data from early years might not mean much. This is not meant to give any lesson, I just thought it might be interesting to some people. ?
4. Data is from this morning in Europe (Feb 22nd, 2020).

First: the average grades since 2006 (for the GRE, it only includes years with the new system):

• Average reported GPA: 3.75
• Average Verbal GRE: 163.4
• Average Quant GRE: 160.9
• Average Writing GRE: 3.8

Second: the distribution of the posts between A/I/U and decisions:

Third: the distribution of the post between schools. I don't have enough space left so I'll upload these pictures in a comment. 

Any thoughts based on this? I can also look at other metrics if you guys think it'd be interesting! (Next thing planned is visualising the dates at which decisions are received).

PS 1: thanks again for all the support and positivity on this forum! ?
PS 2: there probably are a some coding mistakes, so once again I'm not pretending this gives any lesson!

Even though I agree that this probably does not give that much useful information, this post deserves a like due to all the work you must have put into it! I wish I was better at coding and scraping myself. If you still plan to upload this on GitHub I will be waiting for it! 

[Clintarius]

On 2/22/2020 at 6:03 PM, Theory007 said:

Even though I agree that this probably does not give that much useful information, this post deserves a like due to all the work you must have put into it! I wish I was better at coding and scraping myself. If you still plan to upload this on GitHub I will be waiting for it! 

Hey I remembered you had asked that, here's the link to the Github repo that shows the code for scraping and other things.