![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
Bad science
Mar. 7th, 2009 12:48 pmI always feel sorry for people wheeled out for radio debates who are obliged to argue a rational point against a person whose fate belies the statistics. One such instance occurred yesterday when we had someone trying to explain how a parole board works and what they are obliged by law to consider, versus a woman whose daughter had been killed and an interviewer who was, unsurprisingly, determined to take the side of the mother*.
As the poor man, who had been saying that the duty of the parole board was to consider likelihood to reoffend, not the gravity of the crime or the suffering of the victim/the victim's families, was hectored by the (female) presenter about those who are released on parole and go on to re-offend, I felt like shouting at the guy "FFS, just say "are you suggesting that no-one convicted of any crime ever be let out, just in case they re-offend?", but of course he didn't. He just muttered stuff about jails being full to bursting and other such party-line hokum.
But, the sad fact is, if you release 100 people and one of those people reoffends, then one family will be related to a victim who would not have suffered otherwise. That becomes news. That there are 99 people out there, rather than rotting inside a jail, who have harmed no-one, who may be contributing postively to society -- those guys won't be wheeled into the debate; not least because such releasees are not bundled up into groups of 100.
Perhaps that might be an idea. All people released on parole are placed into artificial but real "groups". If one of them reoffends, then the media would be obliged to bring in (in addition to the victim or victim's relatives of the reoffender) two or three of those people who were released as part of this group of 100 (plus their families?). Pointing at a good upstanding citizen and his loving kids, the interviewer would then have to ask: "Are you suggesting that this man, who has gone on to help hundreds of people, should still be rotting in jail?". Because you can't say "release the ones who won't reoffend, but don't release the ones who will. That's like that episode in nathan Barley where it was suggested that two stories be printed rather than one, so that people could just read the one that was good.
+++++++++++++++++
British Airways is stiffing people who haven't flown with them for a while by threatening to take away their air miles. Air Miles Director Andrea Burchett said that "it's very simple. All you need to do is collect at least one Airmile in 24 months and their account will remain open".
I've got another idea, one that's even fucking simpler. Keep the accounts open anyway.
+++++++++++++++++
One of the great trends in recent years that I have spotted is that interviewees will cite "evidence", apparently hard numbers, at a questioner, knowing that the questioner almost certainly won't have done his or her research on the matter. This guess is, sadly, true. Interviewers have badly paid researchers, get given some Cliff Notes, and run it from there. Journalists, meanwhile, are usually too lazy to question something that is said with such specificity and confidence. It's a bit like the Excel Spreadsheet. Show a business plan on an Excel Spreadsheet and it has far more chance of getting through a numbers man than if it was on the back of an envelope. You can make the worst mistakes in the world on an Excel Spreadsheet, but the sheet's mere existence adds veracity.
Step forward Mark Ciavarella, a judge who sentenced juveniles to a correctional facility in Pennnsylvania and, it is alleged, received kickbacks from the builders and maintainers of the facility for providing the "customers". Ciavarella admits accepting the money, but denies that he sentenced more juveniles as a result. He has plea-bargained down to accepting a "conflict of interest" and tax evasion (and faces seven years' jail). Ciavarella told journalists that one thing that showed there was no link between the payments and the sentencing was that the percentage of children he sentenced to custody remained constant before and after he began receiving payments.
Fine, except that, according to one journalist who didn't take this statement at face value, in the two years before the payments started, Ciavarella sentenced 4.5% of cases to custody. By 2004 it had risen to 26%, said reporter Ed Pilkington. Ciavarella looks to have simply relied on journalists not checking the stats. Given that many non-financial journalists appear proud of being mathematically challenged, that looks to be a reasonable gamble.
_________________________
* A standard schlock start to a bad newspaper piece (points if you can come up with examples) is "It's every mother's worst nightmare".
As the poor man, who had been saying that the duty of the parole board was to consider likelihood to reoffend, not the gravity of the crime or the suffering of the victim/the victim's families, was hectored by the (female) presenter about those who are released on parole and go on to re-offend, I felt like shouting at the guy "FFS, just say "are you suggesting that no-one convicted of any crime ever be let out, just in case they re-offend?", but of course he didn't. He just muttered stuff about jails being full to bursting and other such party-line hokum.
But, the sad fact is, if you release 100 people and one of those people reoffends, then one family will be related to a victim who would not have suffered otherwise. That becomes news. That there are 99 people out there, rather than rotting inside a jail, who have harmed no-one, who may be contributing postively to society -- those guys won't be wheeled into the debate; not least because such releasees are not bundled up into groups of 100.
Perhaps that might be an idea. All people released on parole are placed into artificial but real "groups". If one of them reoffends, then the media would be obliged to bring in (in addition to the victim or victim's relatives of the reoffender) two or three of those people who were released as part of this group of 100 (plus their families?). Pointing at a good upstanding citizen and his loving kids, the interviewer would then have to ask: "Are you suggesting that this man, who has gone on to help hundreds of people, should still be rotting in jail?". Because you can't say "release the ones who won't reoffend, but don't release the ones who will. That's like that episode in nathan Barley where it was suggested that two stories be printed rather than one, so that people could just read the one that was good.
+++++++++++++++++
British Airways is stiffing people who haven't flown with them for a while by threatening to take away their air miles. Air Miles Director Andrea Burchett said that "it's very simple. All you need to do is collect at least one Airmile in 24 months and their account will remain open".
I've got another idea, one that's even fucking simpler. Keep the accounts open anyway.
+++++++++++++++++
One of the great trends in recent years that I have spotted is that interviewees will cite "evidence", apparently hard numbers, at a questioner, knowing that the questioner almost certainly won't have done his or her research on the matter. This guess is, sadly, true. Interviewers have badly paid researchers, get given some Cliff Notes, and run it from there. Journalists, meanwhile, are usually too lazy to question something that is said with such specificity and confidence. It's a bit like the Excel Spreadsheet. Show a business plan on an Excel Spreadsheet and it has far more chance of getting through a numbers man than if it was on the back of an envelope. You can make the worst mistakes in the world on an Excel Spreadsheet, but the sheet's mere existence adds veracity.
Step forward Mark Ciavarella, a judge who sentenced juveniles to a correctional facility in Pennnsylvania and, it is alleged, received kickbacks from the builders and maintainers of the facility for providing the "customers". Ciavarella admits accepting the money, but denies that he sentenced more juveniles as a result. He has plea-bargained down to accepting a "conflict of interest" and tax evasion (and faces seven years' jail). Ciavarella told journalists that one thing that showed there was no link between the payments and the sentencing was that the percentage of children he sentenced to custody remained constant before and after he began receiving payments.
Fine, except that, according to one journalist who didn't take this statement at face value, in the two years before the payments started, Ciavarella sentenced 4.5% of cases to custody. By 2004 it had risen to 26%, said reporter Ed Pilkington. Ciavarella looks to have simply relied on journalists not checking the stats. Given that many non-financial journalists appear proud of being mathematically challenged, that looks to be a reasonable gamble.
_________________________
* A standard schlock start to a bad newspaper piece (points if you can come up with examples) is "It's every mother's worst nightmare".
