This is one of those things that I’ve been wondering about for a while, but haven’t asked.
It’s become quite common in the criminal justice/prison/probation/parole world to perform risk assessments of people convicted of crimes. Here is just one example. Many of the risk assessments are quite quantitative in nature while others rely significantly on the judgment of professionals but still provide a framework for assessment that isn’t just R. McIdiot Probationofficer’s opinion. The results of these assessments are used to make decisions as to sentencing, eligibility for parole, and whether or not to revoke probation or parole release.
Is it possible to get a general assessment of the risk that a person will commit a crime and/or a serious antisocial or dishonest act not technically amounting to a crime? Assume that you can get the permission of the subject of the investigation (e.g. , “If you expect to date my daughter and remain in my good graces, you’ll go through the evaluation.”) I’m especially looking if there is anything that would go beyond a basic background check and inquire into the possible reasons for and conditions surrounding prior convictions, and also include an inquiry into current behavior patterns, emotional states, etc. to provide a more broad picture of a person.
Some possibilities that I considered:
All you need is a private investigator. However, I was under the impression that PI’s normally specialize in digging up information, not on analyzing it and drawing conclusions. A PI could discover that a potential boyfriend visits strip clubs once a week, owns a nearly complete collection of Archie Comics, has a subscription to the Porno of the Month DVD Club, and also has two convictions of aggravated second degree misdemeanor mopery with intent to play hopscotch within ten meters of a national park from 2005, but isn’t really trained to tell me whether that is good news or bad. Maybe a PI could contribute findings to another professional doing the final assessment.
Have the subject undergo a general psychological or psychiatric investigation. This might help to provide additional information, but iirc mental health professionals are primarily oriented toward diagnosing and treating mental health conditions and not assessing general character or honesty. Some people with mental illness are quite honest and safe while some people just want to break the rules.
Are you saying you want to create a test (or series of tests) that will spit out a probability that X person will do one of the things on Y list of acts?
Interesting. I think you may have some fun researching military occupational/psychological testing. Try the works of Raymond E. Christal in particular. I bet some of it is in the public domain; it’s from WWII or thereabouts. Anyway, from what I recall, he got paid to figure out if the geniuses who got put in special forces were relatively stable, or were smart crazy dudes who might snap under pressure. Even now, I know the U.S. military, particularly the Air Force and Navy, spend a lot of resources on this sort of intelligence.
It’s been quite a while since I’ve looked at any of this stuff, but bear with me. Some tests in particular that have been sometimes deemed predictive include the Thematic Apperception Test (TAT) (a kind of storytelling, “tell me about this picture” assesment that requires a psychologist/analyst to interpret–much more oomph than a Rorshach, though) or the Implicit Associative Test (IAT) (kind of like a complicated word-association test). These can, ostensibly, reveal latent sociopathic tendencies and things like that.
Then there’s the grandaddy written assessment, the MMPI (the Minnesota Multiphasic Personality Inventory) and its variants . Lots of folks put a great deal of store in this, from the military, to schools, to hospitals, and major corporations on hiring. Unlike the others I mentioned, it is a multiple choice exam and can be graded even by a computer (not that that’s ideal or anything). It’s been around in some form since the 40s and is supposed to reveal/predict all kinds of psychological data such as anxiety, antisocial behavior, homosexual and related, depression/mania, you name it. It’s an interesting test in that it (somewhat annoyingly), asks essentially the same question in several ways to attempt to trip up anyone trying to defeat the intent of the exam…and even gives a guess on how the test subject is attempting to cheat (trying to act sick, trying to act “perfect” etc.). It can also be a very long test (300 or so questions, I believe).
Anyway, I can’t answer your question directly or exhaustively, but maybe you don’t know much about the assessments I listed above…it may be a different place to start answering your curiosity. Oddly enough, while regular medicine is more “evidence-based”, lab tested, and less dependent on the skill and critical judgement/assessment of the physician (detrimentally!!!), I think psychiatry in the sense you describe is moving away from that earlier sense of academic rigor (as compared to the mid 20th century) and moving towards both too many individual judgment calls and bookkeeping-level, “after-the-fact”, QA in order to make policy decisions etc.
I skimmed v. briefly your pamphlet on recidivism in Virginia, and I’m going to look especially at the section on Cox regression (p. 84) tomorrow, when I’m not so brain dead.
Good ideas, thanks. Anyway, the major issue with “recidivism” instruments in the criminal justice world is that they are “recidivism” instruments that are indicated for use only with convicted criminals. In many cases, they only work with a specific type of crime such as sex offenses, alcohol or drug offenses, or property crimes and aren’t rated to work for e.g. taking a person convicted of Jury Tampering and assessing to what extent they are at risk to committing Providing Material Support to a Terrorist Organization and whether that risk is higher or lower than their risk of committing Rape of a Child Under 13, Trespass at Night on Private Cemetery Grounds, or Aggravated DUI Homicide?
Of course, they are basically completely useless if one wants to know the risk that someone will file for divorce, engage in unseemly but legal sexual practices, take excessive vacations, drive a car that gives crappy mileage for no good reason, wear a mullet, live a legal but antisocial second life as a stripper on alternating weekends without telling their spouse, or “talk back” to their elders. All of that is potentially relevant to human relationships but usually isn’t punishable in criminal court.
Can you actually go to a psychiatrist or other mental health professional and say that you aren’t there to get a diagnosis or treatment, but that you just want to find out how likely it is that you will Compound Treason? Will they actually give you an evaluation and say, ah, based on this data blah blah blah your risk of committing that offense within the next ten years is 4%-6% and your risk of committing it anytime in your life is 5%-10%? I’ll use myself as an example here so we can assume that no privacy issues are likely to prevent the therapist/assessor from accessing the data they need.
Basically, yes. It doesn’t need to be all written tests or especially not multiple-choice tests, but I am looking for something that would at least give an educated opinion. I’ve looked at a few of the criminal justice risk assessments and a lot of them are fairly narrow in scope. For example, the Static-99 sex offender risk assessment at first glance looks like it could be used for estimating the risk that a person will commit a sexual crime in the future. The problem is, the instrument is rated only for people who have not only already been convicted of a sexual crime, but convicted of a violent sexual crime. So using it to assess potential marriage partners, babysitters, or soccer coaches for my children is pretty much right out.
Beyond statistical instruments, are there any professionals that specialize in or at least are regarded as having competence in synthesizing data (e.g. criminal records, personality test results, observations from private surveillance, etc.) and develop an informed but holistic opinion that is based on professional judgment?
Given the “big data” push we’ve seen with advertisers monitoring what we buy and correlating it with our Internet browsing data, zip code, marital status, last name (could indicate ethnicity), age, what kind of car we drive, and several other data points and then deciding to send us a coupon for 30 cents off Shake and Bake, this seems to be something that would either have been developed or be in development. I’m sure lots of people would spend $39.95, or maybe even $2,000, to get a customized risk assessment to predict the likelihood of divorce. “Always assess before you say ‘Yes!’”
Interesting idea. The challenge as I see it is that the big data advertising model has the advantage that you reveal your actions by your purchases. And there’s a huge surveillance net out there to track those purchases. And searches and site visits and location and … And there’s zero privacy law preventing the snoops from sharing the data as they will.
But there’s no corresponding surveillance net to detect most people’s anti-social behavior. The cops & courts deal with some arguable fraction of the worst behavior. And much of what they do is covered by privacy law or common government practice which prevents gathering all the data they do have into one place for analysis.
So bottom line, a big data analysis approach could theoretically work, IF the data existed. But the data doesn’t exist. And there’s no practical way to collect it, nor any immediate commercial motivation to do so.
Good points. It’s true that there is no centralized database that keeps track of how often people lie to their spouses and claim they are working late when it fact they are at the bar, and there is no reasonable way to set such a database up that would be remotely accurate.
There are, however, a few data points we can see. Divorces are typically public record. So are bankruptcies. Many licensed professions make official disciplinary acts public, especially more serious ones that result in license suspension or revocation. Civil judgments are also generally public record, and credit reporting agencies today infamously use these to adjust credit scores. In some jurisdictions, including my own, the names and addresses of people who are issued traffic tickets are published on an official County website. The identities of Registered Sex Offenders are infamously public. Some universities publish the names of graduating alumni, so maybe we can find out what grocery store purchases are most strongly correlated with being within a year of college graduation, which could count as a “virtue” that could be relevant to a general risk prediction, since having a degree is associated with lower unemployment and fewer criminal convictions.
So you could have a “big data” scenario where an observation is made that people who buy more than 5kg of Shake and Bake a month, live within 50 miles of Memphis, have a last name that on the American Society for Genealogy’s List of Surnames Typically of Celtic Origin (3rd Edition), drive a Honda Accord, have an American Express card, subscribe to the Porno of the Month DVD club, read at least 10 articles on cnn.com every two months, and perform at least one Google search a year that involves the phrase “My Little Pony” are 5 times more likely to suffer revocation of a license to practice Pharmacy due to an ethical violation than the general population and twice as likely to have a civil monetary judgment issued against them that is over $10,000 than the general population. On the other hand, they may be found to be half as likely to undergo divorce and ten times less likely to be ordered to pay alimony in cases where a divorce does actually happen. But also, maybe the data shows that such people appear to be neither more or less likely to file bankruptcy, unless they also own at least 10 acres of Kentucky land that is classified as “farmland” according to the public land records of the State of Kentucky Office of Land Registration, in which case they are three times as likely to file bankruptcy once but five times less likely to file for bankruptcy a second time within a ten-year period.
Then, people could pay to look others up in the database. Oooh, Behavioral Statistics and Reporting, Inc. won’t tell me exactly what it was due to trade secrets, but they say that something that my sister’s new boyfriend bought within the past year indicates a medium to high likelihood of appearing on the Texas Sex Offender Registry within the next ten years. Better start looking for signs of perversion…
