Building a Relational Database

If I had to write an essay on my summer vacation, this year's topic would be "Building a Relational Database."  As I noted a few months ago, the LSAC funded our proposal for a large-scale longitudinal analysis on the production and content of legal scholarship (my co-PI's include Funmi Arewa, Ken Dau-Schmidt, Peter Hook, and Andy Morriss).  The most difficult task in this endeavor is cost-effectively digitizing biographic information on every law professor who ever appeared in an AALS directory (from 1922 to 2007) and linking these individuals to the detailed bibliographic records contained in the Index of Legal Periodicals from 1928 to the present.

How does someone tackle a job this big?  My legal training has negligible value here.  It it really all about planning and collaboration among a team with a wide array of skill sets.  With the input of a talented database expert, Mark Newton, and our co-PI, Peter Hook, who is a law librarian with graduate training in information sciences, we are developing and implementing a process that--we theorize--will (a) maximize the capture of information and complex relationships, and (b) minimize key strokes and errors. 

So far, one of our greatest assets is the excellent workforce here in Bloomington.  We currently have a research team that includes four graduate students from the School of Library and Information Sciences (SLIS), two excellent college graduates who are headed to law school in the fall, and a brilliant University of Chicago undergraduate with an eagle eye for detail.   These talented folks are closest to the data and provide enormous insights that affect our planning process. 

Although the relational database itself will not produce a scholarly paper, I am really grateful for the opportunity learn a lot of new technical and managerial skills.  I am also developing a deeper respect for expert knowledge and team work.      

Comments

[Tracy Lightcap]

Ahhh, relational databases! Glad to see someone is actually thinking about creating them in this field. Perhaps the biggest problem looming over the horizon in ELS and pol sci is that most of the data analysis engines favor square table, non-relational data structures. Fine for the past, ok for the present, BIG trouble for the future. It's getting to where you have to either a) know how to do SQL first to get data into shape or b) work with humongous, inefficient datasets.

Why is this a problem? Don't we all have computing power of unprecedented power sitting on our desks? Isn't that enough? Today, yes. But think about the iPhone for a second. Isn't it obvious that the day of the personal computer is just about over? What about when we are all walking around with flash RAM lecturns that run off Web 2 applications accessing said humongous databases of the future? I.e. about 5 - 8 years from now? We'd better get to where we can work more easily with relational data. Fast.

[Dwight Hines]

Professor Henderson:
I’m was most unhappy to read in your post on ELSBLOG oday that your work on creating a relational database will not produce a scholarly paper. There were good people and good events between the years 1922 to 2007, much of what they did and what happened has remained unexplored in depth because the information was not in easily accessible locations by people who know at a gut level that Barzun’s criticism of empirical historical studies is that they tended to be superficial collections of numbers rather than good history. It won’t take a McClellan to see that the legal scholarship database you’re building could be used in studies of need for achievement in children and cultures, or that much of the variance that needs to be accounted for in predicting changes in ethnic and gender disparities is likely to be due, significantly due, to need for justice rather than need for power. Your database will make questions like that those approachable, if not answerable.

The ubiquity of relational databases in the United States is constantly noted in the non-legal literature and the ability to obtain the software as open source, or as simple free software from Microsoft, means we need to have more people explaining what relational databases can and can not achieve. You don’t have to make many public records requests in Florida about a town’s relational databases to realize that those who maintain the databases are not using them for exploration, or simple statistics, much less for evaluation of programs or practices. Many, if not most, of the attorneys who are defending, or will be defending organizations with the massive relational databases, are also not aware of the good, solid, valid information available in relational databases that can convict, or completely or partially exonerate, their clients. Given the new federal rules on electronic discovery, the practical value of assigning at least one attorney, with support, for the legal dimensions of relational databases must be considered a prudent practice, even in relatively small agencies. Obviously, the auditors would have mixed feelings, but I think it’s inevitable.

Now, since you raised the issue of how you spent your summer, when you were on the beach with your eyes closed, not to hot or cold, just being all dreamy and comfy, did you think of any ways to explain relational databases to non-math, non-computer, and non-auditor people, using familiar items from every day life? Did you then morph into how that collection of different sized and shaped Rubik’s Cubes (or whatever) were used by MacDonalds to go into video leasing, based on conjoint analyses, analyses that controlled for liability in several areas?

Ah yes, I look forward to reading your papers.

Dwight