Zombies in the Federal Courts: Medications, Corporations and Criminals
Let me straight up acknowledge that this post is some weak social science and I’m not sure the stats are correct. Nevertheless, I found some significant differences regarding Zombies in the Federal Courts and so it seemed worth sharing. And this is a blog not my dissertation so ya know why not click publish…
A search of lexisnexis academic legal search of all Federal Court opinions with the search phrase: opinions(zombi!) will result in 310 Federal Court opinions that include the words zombi, zombie or zombies in the text of the opinion (hereafter referred to as: the zombie opinions). ZombieLaw has previously posted a graph of these cases over time, compared to zombie movies on wikipedia. Here’s that graph again:
Funny enough, when I found Daniel Drezner’s “Theories of International Politics and Zombies” and wrote about it in the ZombieLaw post entitled “Critical Zombie Theory – cognitive linguistics and construction of the non-creative” – I discovered he uses a similar graph of zombie movies listed on wikipedia for his opening argument about why zombies are a good topic for international policy theories. I’m not sure that makes the wiki list any more credible but it does situate my study of legal zombies in good company. Here’s Drezner’s graph available in the free preview of his book on google:
Ok so we know zombies are popular at the movies and the prior ZombieLaw graph showed the spike in Federal Court opinions comparable to Drezner’s line for scholarly publications. But what are all these judicial opinions about?
It turns out, the bulk of these judicial opinions (91.6%, 284 of 310) include one of three other words: medication, corporation or criminal.
Of the 310 zombie opinions, 158 (51%) include the word Medication, 124 (40%) include the word Corporation and 99 (32%) include the word Criminal. If you are quick with math you can tell these are not mutually exclusive categories. In fact, 91 (29%) of the zombie opinions include at least two of the associated words (including only six opinions with all three words). Below, a Venn diagram shows the overlap:
There is a significant negative correlation between use of the word medication and use of the word corporation (r=-0.450, p<0.05). There is also a significant negative correlation between use of the words medication and criminal (r=-0.242, p<0.05). There is no apparent relationship between the words corporation and criminal by themselves; but if the word medication is accounted for then there is a small but significant relationship between the words criminal and corporation when there is no use of the word medication (r=-0.171, p<0.05).
What if we do some quick and dirty Chi-Square tests for differences. .
|No Corporation||Yes Corporation||Total|
Med*Corp chi-sq=62.9 p<0.000
|No Criminal||Yes Criminal||Total|
Med*Crim chi-sq=18.1 p<0.000
|No Criminal||Yes Criminal||Total|
Crim*Corp chi-sq=0.121 p=0.728
Therefore, in the zombie opinions, there is a significant association between the word medication and both the words criminal and corporation but there is no significant relationship between the word corporation with the word criminal. A significant negative relationship between criminal and corporation can be found only amongst the nearly half (49%) of zombie opinions that do not use the word medication. (r=-0.171, p<0.05).
|NO MEDICATION||No Criminal||Yes Criminal||Total|
No Med/ Corp*Crim chi-sq=4.463 p=0.035
Similarly, there are group differences in the relationship between the words medication and criminal in the subset of cases that use or do not use the word corporation. The overall effect of medication and criminal (reported above) is only apparent in the subset not using the word corporation. There is a large group (48%) of corporation-zombie opinions that use neither of the other words and only a few (4.8%) that use both, whereas contrast the non-corporation-zombie opinions, where the number of opinions using both other words is about equal to the number using neither other word (both about 14%). The inverse relationship between corporation and medication is more robust, and the significant overall relationship (see above) remains significant in both the criminal and non-criminal subsets.
In linear regression analysis, all three interaction terms were tested but none were significant. Therefore the main effects are interpretable. These effects are consistent with the chi-square analysis shown above.
Use of the word medication is a significant predictor of both the other words (criminal F=19.1, corporation F=78.4, both p<0.00). Inversely, predicting use of the word medication from both the other words finds both are significant predictors (F=53.1, p<0.00) and each contributes significantly above and beyond the other (criminal t=-4.7, corporation t=-9.1). Also, predicting criminal, from both corporation and medication provides a significantly better model than predicting criminal from medication alone (17% increase in reduced error), as such the word corporation is is an important non-significant mediating factor on the prediction of criminal from medication. The reverse however, is not true and criminal is not a mediating factor in the prediction of corporation from medication as medication is already a strong predictor of corporation.
Assuming I am not making some zombie-logical error (?), there are three groups of zombie opinions: medication-zombies, corporation-zombies and criminal-zombies (and some miscellaneous others). There are overlaps amongst the three types but the the words can provide small but significant predictive information about use of the other words.
Aside from small effect sizes, a major weakness of this study is that I did not actually read the case opinions and just used a computerized search. This would make for very bad legal analysis but in terms of a word study or content analysis, I hope some will find it insightful. The computer searches only the text and there is no way to know exactly how the word is being used in each usage. A future more thorough study will require reading each of these cases and coding the types of usages. For example, are the entities referred to as zombie the same as those referred corporation, medication-user or criminal? Also these are all the judicial opinions so there are redundancies as some cases have multiple opinions as the case moves through the system. Also it would be ideal to see if these categories predict to anything else like the legal outcomes of the case. For example, I expect that most cases using the word zombie result in the zombie losing the case but I'm not so sure and I have not coded nearly enough cases to make that claim.
I have much more research to do but when I found these three words (I found medication and corporation ranked high in a word frequency and criminal was just a good guess) I wanted to share. It is interesting to see how much of the set is captured by these three words and that they are not likely to appear together. More to come…