Open bdesmarais opened 6 years ago
Yes, it did finally work! Our adjacency matrix now has 9841 nodes. I have compared the Fowler citation data with the courtlistener citation data based on degree distribution, number of triangles, number of citations per year, etc. and they look very similar/identical.
However, there were about 200 cases in the SCDB that I couldn't match with the courtlistener data. I simply removed these 200 cases from our model.
Is it the case that those 200 cases could be isolates in the supreme court citaiton network (i.e., and therefore not listed in the courtlistener edgelist)? If so, we should leave them in the network, but as isolates.
Bruce A. Desmarais Associate Professor, Department of Political Science Director, Graduate Programs in Social Data Analytics Pennsylvania State University brucedesmarais.com
On Mon, Feb 19, 2018 at 12:28 PM, Christian notifications@github.com wrote:
Yes, it did finally work! Our adjacency matrix now has 9841 nodes. I have compared the Fowler citation data with the courtlistener citation data based on degree distribution, number of triangles, number of citations per year, etc. and they look very similar/identical.
However, there were about 200 cases in the SCDB that I couldn't match with the courtlistener data. I simply removed these 200 cases from our model.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/desmarais-lab/Supreme_Court_Citation_Network/issues/5#issuecomment-366758541, or mute the thread https://github.com/notifications/unsubscribe-auth/ABYXKT2vth7xCv587fx3KJwXIcfHmgOWks5tWa80gaJpZM4Rpw0G .
If you upload a list of those cases I could try to determine what might be going on.
On 19 February 2018 at 14:59, Bruce Desmarais notifications@github.com wrote:
Is it the case that those 200 cases could be isolates in the supreme court citaiton network (i.e., and therefore not listed in the courtlistener edgelist)? If so, we should leave them in the network, but as isolates.
Bruce A. Desmarais Associate Professor, Department of Political Science Director, Graduate Programs in Social Data Analytics Pennsylvania State University brucedesmarais.com
On Mon, Feb 19, 2018 at 12:28 PM, Christian notifications@github.com wrote:
Yes, it did finally work! Our adjacency matrix now has 9841 nodes. I have compared the Fowler citation data with the courtlistener citation data based on degree distribution, number of triangles, number of citations per year, etc. and they look very similar/identical.
However, there were about 200 cases in the SCDB that I couldn't match with the courtlistener data. I simply removed these 200 cases from our model.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/desmarais-lab/Supreme_Court_Citation_ Network/issues/5#issuecomment-366758541, or mute the thread https://github.com/notifications/unsubscribe-auth/ ABYXKT2vth7xCv587fx3KJwXIcfHmgOWks5tWa80gaJpZM4Rpw0G .
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/desmarais-lab/Supreme_Court_Citation_Network/issues/5#issuecomment-366789971, or mute the thread https://github.com/notifications/unsubscribe-auth/AKKqe1bA7gNiZOb7aRX4yTlDS0mbYLvVks5tWdKigaJpZM4Rpw0G .
I just looked into this a little closer. It looks like these 200 cases did not get a majority vote, i.e. for some reason the votes were split 4-4, meaning that one judge was absent or abstained from voting. This might be the reason why no majority opinion was written and as a result the Courtlistener data has not assigned a opinion ID for these cases. What does usually happen in the SCOTUS if a case is tied?
According to major news outlets, if there is a 4-4 split, the case does not set precedent. Often, they are very short (in the example below).
From the court reports, we see that they are still cited by later cases (even by the SCOTUS). For example: https://www.courtlistener.com/opinion/111402/board-of-trustees-of-scarsdale-v-mccreary/?q=BOARD%20OF%20TRUSTEES%20OF%20SCARSDALE It is cited by, for example, Allegheny County v. Greater Pittsburgh ACLU (1989), which is a supreme court case.
Do these not show up in the data? (This specific case and its two inbound supreme court citations are in the data we were originally using.)
On 19 February 2018 at 16:17, Christian notifications@github.com wrote:
I just looked into this a little closer. It looks like these 200 cases did not get a majority vote, i.e. for some reason the votes were split 4-4, meaning that one judge was absent or abstained from voting. This might be the reason why no majority opinion was written and as a result the Courtlistener data has not assigned a opinion ID for these cases. What does usually happen in the SCOTUS if a case is tied?
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/desmarais-lab/Supreme_Court_Citation_Network/issues/5#issuecomment-366804682, or mute the thread https://github.com/notifications/unsubscribe-auth/AKKqe-nhlPv-1XT5MmbbSaUIfn6nLhl0ks5tWeTOgaJpZM4Rpw0G .
Thanks for the information! There are split votes that have an opinion id and thus can be cited by other cases.
I just created a csv file with all 145 cases that have no opinion_id (see R-code_courtlistener folder). About half of them are 4-4 splits, but the other half had a majority. For most cases with a majority the Supreme Court Database indicates that there was no majority opinion writer (see column: majOpinWriter==NA). I guess we can treat these cases as isolates in the network. I checked the data for some of the remaining cases in the courtlistener data set. Interestingly, I could not find any data for these cases, i.e. the entry for these cases was missing. Something I could do is to manually add citations for these cases from the Fowler data set, but it would not cover citations that involve cases after 2001.
Good detective work, both of you. It could be the case that there was no majority opinion for these cases despite having more than 4 justices vote in the majority (it is possible that no opinion attracts a majority of signatories). I think it would be better to omit the cases that cannot be matched, and just put a footnote in about them in the paper. This would be more accurate than treating them as isolates, since they are not really isolates.
Bruce A. Desmarais Associate Professor, Department of Political Science Director, Graduate Programs in Social Data Analytics Pennsylvania State University brucedesmarais.com
On Thu, Feb 22, 2018 at 9:45 PM, Christian notifications@github.com wrote:
Thanks for the information! There are split votes that have an opinion id and thus can be cited by other cases.
I just created a csv file with all 145 cases that have no opinion_id (see R-code_courtlistener folder). About half of them are 4-4 splits, but the other half had a majority. For most cases with a majority the data indicates that there was no majority opinion writer (see column: majOpinWriter==NA). I guess we can treat these cases as isolates in the network. I checked the data for some of the remaining cases in the courtlistener data set. Interestingly, I could not find any data for these cases, i.e. the entry for these cases was missing. Something I could do is to manually add citations for these cases from the Fowler data set, but it would not cover citations that involve cases after 2001.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/desmarais-lab/Supreme_Court_Citation_Network/issues/5#issuecomment-367893247, or mute the thread https://github.com/notifications/unsubscribe-auth/ABYXKQ0MNvmoEcCFojKMiJYLvrnGoJo7ks5tXiZegaJpZM4Rpw0G .
Sounds good. There are two other points I want to mention for our two Martin Quinn Score variables.
Mean MQ score of judges in majority: In case of a 4-4 split I calculated for this variable the mean MS score of all 8 judges that voted.
Absolute difference of MQ scores of judges in the majority): In case of a 4-4 split I simply assigned a '0' for this variable. I know its not ideal, but I am not sure what would be better. We could also indicate it as a missing value.
Please let me know if you have a better idea on how to treat these cases.
1 is fine for a 4-4 split. That would accurately represent the fact that the case did not establish overly liberal or overly conservative precedent.
On Fri, Feb 23, 2018 at 11:32 AM Christian notifications@github.com wrote:
Sounds good. There are two other points I want to mention for our two Martin Quinn Score variables.
1.
Mean MQ score of judges in majority: In case of a 4-4 split I calculated for this variable the mean MS score of all 8 judges that voted. 2.
Absolute difference of MQ scores of judges in the majority): In case of a 4-4 split I simply assigned a '0' for this variable. I know its not ideal, but I am not sure what would be better. We could also indicate it as a missing value.
Please let me know if you have a better idea on how to treat these cases.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/desmarais-lab/Supreme_Court_Citation_Network/issues/5#issuecomment-368061474, or mute the thread https://github.com/notifications/unsubscribe-auth/ABYXKWD1ExK7sfeKuPRyV8xx-oocrXN6ks5tXugngaJpZM4Rpw0G .
--
Bruce A. Desmarais Associate Professor, Department of Political Science Director, Graduate Programs in Social Data Analytics Pennsylvania State University brucedesmarais.com
Did this work, @schmid86? I see the commit that looks like it did.