• What makes “An Unbelievable Story of Rape” so good?

    Aaron’s right. Go read An Unbelievable Story of Rape, by T. Christian Miller and Ken Armstrong. Read it for the content, but also read it for the style.

    As one who wants to improve writing (my own and that of those I advise), I’ll highlight four of the piece’s stylistic strengths.

    1. It’s inherently a good story. Yeah, this is not quite style. It’s arguably content. But I’m putting it partially in the style bucket because there’s an art to finding a good story. Or, more precisely, there’s an art to noticing that there’s a good story within the relative cacophony of a subject area or general topic.

    An author can’t just say, “I’m going to write about rape. Go!” There’s a ton of work before the writing happens making that far more specific. In any space, there are myriad directions to shine one’s flashlight. It takes a sense of style to find the right one.

    How does one find the right one? I don’t know, exactly. I have no idea how others do it. One way I know when I have something worth writing about (or, rule out things not worth writing about) is to see if I can write one to three sentences that are compelling enough to me and others (e.g., editors) that they demand more. Here’s three sentences from the lede of their piece that I imagine Miller and Armstrong might have (or could have) pitched:

    She had reported being raped in her apartment by a man who had bound and gagged her. Then, confronted by police with inconsistencies in her story, she had conceded it might have been a dream. Then she admitted making the story up.

    2. It’s told in parallel threads, until the end. Until close to the end, the piece bounces back-and-forth between the case of the recanted rape and another set of rapes across the country and several years later. Both stories are compelling. All but the most naive readers know they must have something to do with each other — otherwise, what’s the point? But how? That suspense is useful. It makes one want to keep reading.

    The art isn’t employing this device, which is fairly common. Anyone can do that once they know it’s possible. The art is doing it well. I imagine the authors spent some time figuring out just when and how to switch back and forth, how many hints to leave, as well as how to end the suspense and come clean with the connection.

    I noticed one, repeated stylistic element in the switching. Just before many of the switches, the authors wrapped things up with a paragraph consisting of a single, short, often declarative sentence. It’s nice. It prepares the reader for a switch. Here they are:

    • “She took the deal.”
    • “‘We have one just like that,’ he said.”
    • “‘It was awesome.'”
    • “‘Am I in trouble?’ Marie asked the detective.”
    • “She emailed a crime analyst at another police department, ‘I so want to see this guy’s leg! BAD.'”
    • “Lynnwood, Washington.”
    • “He left the room, and walked to the front door, and he was gone.”

    3. The language is simple. The sentences are short. This is a very clean piece. It’s not larded up with lots of fancy words. It’s an easy read. This is good style! The authors are not jumping up and down shouting, “Hey, look at how many big words we know. We’re so smart!” Readers who notice should be impressed with how much work they did in getting out of the way and letting the story shine through. Readers who don’t notice are doing what they should be, just enjoying the work, not fighting through long sentences and fancy language.

    A sample:

    After a pause, O’Leary opened the door. He looked confused and shocked as he stepped out into the bright winter sun. Two dogs, a small pit bull and a Shar-Pei, tumbled out ahead of him. He wore a gray hoodie, baggy gray sweatpants and gray slip-on houseshoes. He was alone.

    How many of us would write this so simply? “He was alone.” That’s a great sentence to end this paragraph. I confess, I might have thought to work in that he was alone earlier. Good, simple-sounding style is not so simple!

    4. Quotes do a lot of work. The piece is threaded with quotes. They work so well because the authors weaved them in so nicely. It’s so well done that one hardly notices. That’s excellent, but hard!

    After rushing to the apartment that morning, Peggy found Marie on the floor, crying. “But it was so strange because I sat down next to her, and she was telling me what happened, and I got this — I’m a big Law & Order fan, and I just got this really weird feeling,” Peggy says. “It was like, I felt like she was telling me the script of a Law & Order story.” Part of it was what Marie was saying. Why would a rapist use shoelaces to tie her up? And part of it was how Marie was saying it: “She seemed so detached and removed emotionally.”

    Peggy’s words are used to move the story forward. Great!

    Probably all these stylistic elements are obvious to good journalists and writers. They’re not to me. I have to look for them. When I read a piece of good writing I try to identify why I found it so good. When applicable, I try to emulate. I suspect that’s how to improve.


    Comments closed
  • How often should I get dental x-rays? (Take 2)

    I asked this question over a year ago. Earlier this week the New York Times’ Well Blog answered:

    People who see a dentist regularly and have good oral hygiene and no current dental problems might need bitewing X-rays of molars only every two to three years to check for early cavities, said Dr. Aruna Ramesh, director of the oral and maxillofacial radiology division at Tufts University School of Dental Medicine in Boston. More extensive X-rays may be needed every three to five years to check the health of roots and adjacent bone.

    The post goes on to say kids need x-rays more frequently (every year or two) because their teeth are changing more rapidly. But it doesn’t say which kinds of x-rays they needed every year or two, bitewing or “more extensive.”

    My problem with all of this is that it comes off as eminence based. Are there studies that test the benefits of x-ray intervals? That said, it’s all I need to reject the annual offer of x-rays from my dentist. That’s too frequent, except possibly for kids.

    What about harms? The post reassures that risk from x-rays is very small, but points to a study that found a connection to brain cancer for people with bitewing x-rays at intervals of one year or more frequent.

    Exposure to some dental x-rays performed in the past, when radiation exposure was greater than in the current era, appears to be associated with an increased risk of intracranial meningioma.

    The Well Blog sensibly suggests that “patients may want to inquire whether dentists are using the fastest possible exposure to minimize radiation.” Still, note to self, stay away from annual bitewings. It’s overkill.


    Comments closed
  • Subtle but important things academic journals do incorrectly: a case study

    Yesterday, at 5:44PM, the table of contents of the winter issue of the Journal of Policy Analysis and Management hit my inbox. The issue includes a paper by Kerri Raissian titled “Hold Your Fire: Did the 1996 Federal Gun Control Act Expansion Reduce Domestic Homicides?” Perhaps, in time, I’ll read and comment on it, but I want to first get a few things off my chest about how it was released.

    From time to time I’m asked by the editors of academic journals how to increase the reach and impact of the work they publish. Sometimes I feel like they’re looking for that secret thing they can do with almost no effort that will suddenly boost their journals’ relevance.

    The truth—as far as I know—is there is no such thing. There’s no magic. There’s just hard work, which begins with paying attention to some very basic things.

    Let’s take Raissian’s article as a case study. Right now, the subject of her paper—gun control and its relationship to homicides—is hot. That’s an understatement. It’s burning with the heat of 1,000 suns. This is exactly the time to maximally promote this paper.

    But, what did the journal do? I see three mistakes.

    1. A 5:44PM Friday release of a table of contents is a good way to get people to ignore it. There’s a reason administrations release controversial or embarrassing news late on Fridays. People, including journalists, are less likely to pay attention. That’s not what a journal should want.

    Tip #1: Release tables of contents in the first half of the work week.

    2. As best I can tell, this journal does no outreach to journalists, which is a fairly common oversight. Was there a press release? Is there a way for journalists to get on a list to get them automatically? Was an embargoed copy circulated? From what I can see, the answers to all these questions is “no.” (If someone shows me otherwise, I’ll update this. Suffice it to say, using my press affiliation, I’ve tried to get on every media distribution list of every health policy relevant journal. Many don’t have one. This is bad.) Even if a journal does not want to invest in the infrastructure to do these things regularly, it should be obvious that making an exception in this case (or similar ones) would help the journal achieve greater reach. Again, the subject of this paper is red hot. Now is the time to make sure it gets noticed. It’s when people care.

    Tip #2: Promote research when people care, even if it isn’t in the latest issue.

    3. The paper is gated. Clicking on the link for the PDF gets you this:

    gated 2

    Hey, journalists and policymakers, want to read our red hot paper? Pay up! This is not a good approach. The journal actually ungated a few articles in this issue, just not this one. The paper just after this one is ungated, for instance. So close! (Perhaps authors pay for ungating. I don’t know. If a journal wants to make ungating a revenue center, OK. But it won’t maximize reach and influence if the right papers aren’t ungated.)


    Tip #3: Ungate hot papers, even if temporarily.

    There may be very fine, business reasons for all these things. I’m not contesting that. But, if an editor comes to me asking how her journal can be more influential, my first response will be, “Pay attention!” If you don’t know that an issue is smoking hot AND you have a paper on that issue coming out AND you don’t make it accessible for free AND you don’t tell journalists about it AND you send your announcement out on Friday night, I give up. This is on you.

    Final tip: You get the influence you work for. Do little work and the meager influence you earn is what you deserve.

    PS: Listen to a podcast about this paper here. I have not listened to it myself.

    PPS: It has not escaped my mind that, perhaps, publishers, not editors are responsible for some of these dissemination problems. Fine, editors need to push their publishers. Publishers need to pay attention to their editors.


    Comments closed
  • Alphabet game

    Via 101.7 MIKE FM on Facebook:

    alphabe game

    If you don’t get this, rules for some variants of the alphabet game are here.


    Comments closed
  • Ben Carson’s “Prescription for Empowerment”

    Earlier this week Ben Carson released his “Prescription for Empowerment,” which is an outline of a health reform plan. For the American Journal of Managed Care, I answered a few questions about it. Here’s a teaser quote:

    [R]aising the Medicare age only “saves” money if by “save” you mean “ignore how much more it costs,” which is a lot.

    Go read the rest.

    Meanwhile, here are a few other things I noticed in the proposal that I care to comment on (I’m letting a lot of other things go because I don’t have time):

    “[M]inimal essential coverage” deemed satisfactory to Washington […] required [consumers] to pay for things like marriage therapy and acupuncture, whether they wanted them or not. [Cites this.]

    Actually, essential health benefits are driven by state-level decisions and markets.

    In critiquing Medicare Dr. Carson wrote,

    Overall, the percentage of doctors who closed their practices to Medicare or Medicaid by 2012 had increased by 47 percent in just the four years since 2008. [Cites this.]

    I don’t have time to fact check this (go ahead and click through to the citation to see for yourself), but I just wanted to flag that this is not a specific critique of Medicare. It conflates Medicare and Medicaid. Moving on,

    [T]he Medicare Hospital Insurance Trust Fund will be depleted just 15 years from now, in 2030. [Cites this.]

    The trust fund is always within some number of years of depletion, and 15 is fairly long as these things go. Typically, policy action occurs when we’re within seven years of depletion. So, if history is any guide, now is probably not the best time to succeed at Medicare reform. Still (and I mean this without implying endorsement), best of luck to Dr. Carson!


    Comments closed
  • AcademyHealth: Colonoscopy reference pricing by CalPERS

    Last month, James Robinson and colleagues published a new study has come out on reference pricing by the California Public Employees’ Retirement System (CalPERS). I summarize it, as well as round-up some other reference pricing posts, on the AcademyHealth blog.



    Comments closed
  • An important message*

    This post is pinned to the top of TIE. If you’re on the home page, scroll down for new posts. We’re still posting daily!

    TIE’s daily email is not working (again). If you rely on the email but still want to read TIE, you can do so on TIE’s website, and follow in other ways too. You do not need to unsubscribe and resubscribe to the email list. It’s our problem, not yours.

    The problem, if you want to know, is that our RSS feed is invalid because of a “blank line before XML declaration”. We haven’t found a permanent fix yet, but we’re still working on it.

    * Yes, I’m aware that those who rely on email subscriptions may not see this post.

    Comments closed
  • Algorithms + physicians = ?

    The following originally appeared on The Upshot (copyright 2015, The New York Times Company). It also appeared on page A27 of the December 9, 2015 print edition.

    Can machines outperform doctors? Not yet. But in some areas of medicine, they can make the care doctors deliver better.

    Humans repeatedly fail where computers — or humans behaving a little bit more like computers — can help. Even doctors, some of the smartest and best-trained professionals, can be forgetful, fallible and prone to distraction. These statistics might be disquieting for anyone scheduled for surgery: One in about 100,000 operations is on the wrong body part. In one in 10,000, a foreign object — like a surgical tool — is accidentally left inside the body.

    Something as simple as a checklist — a very low tech-type of automation — can reduce such errors. For example, in a wide range of settings, surgical complications and mortality fell after implementation of a basic checklistincluding verification of patient identity and body part for surgery, confirmation of sterility of the surgical environment and equipment, and post-surgical accounting for all medical tools. Though simple procedureswould all but eliminate certain sources of infections in hospitals, thousands of patients suffer from them in American hospitals every year.

    Limits on how much information we can process and manipulate make it hard or impossible for even the smartest and most adept doctors to keep up with new evidence. In 2014 alone, more than 750,000 additional medical studies were published. Granted, a physician might need to keep up only with the evidence in her specialty, but even at a fraction of this rate, it isunrealistic to expect even the best physicians to assimilate every new development in their fields. In cancer alone, 150,000 studies are published annually.

    Computers, on the other hand, excel at searching and combining vastly more data than a human. I.B.M.’s Watson — the computer that won Jeopardy! — is among the best at doing so. Teams of physicians at Memorial Sloan Kettering Cancer Center in New York, the University of Texas MD Anderson Cancer Center in Houston, and the Cleveland Clinic are helping to train Watson to apply humanity’s huge store of cancer knowledge to the delivery of more personalized treatment.

    At Boston Children’s Hospital, Watson will help diagnose and treat a type of kidney disease. It will team up with Apple to collect health care data;with Johnson & Johnson to improve care for knee and hip replacements; with medical equipment manufacturer Medtronic to detect when diabetespatients require adjustments to insulin doses; and with CVS to improve services for patients with chronic conditions. Another computer-assisted approach to cancer treatment is already in place in the vast majority of oncology practices. Other automated systems check for medication prescribing errors.

    To many patients, the very idea of receiving a medical diagnosis or treatment from a machine is probably off-putting. Apart from the sense that it just doesn’t feel right to some, there’s a fundamental question of whether medicine is or can be purely data driven. If the only thing between your illness and its diagnosis and cure is the manipulation of evidence, then, in principle, a computer should one day be able to deliver care as well or better than a human.

    But healing may rely on more than the mere processing of data. In some cases, we may lack data, and a physician’s judgment might be the best available guide. A good deal of health care’s benefits may also be in the human interaction between doctor and patient. Placebo effects can be real and strong. Many people engage the health system for reassurance and hope, even when no cure is available. Studies show that patients with close,personal bonds with their doctors and shared engagement with their careare more likely to follow their prescribed treatments. To the extent medical treatment relies on the human touch, on the trust of patients in their doctors and on physicians’ embodiment of authority, a computer-delivered cure may never feel complete.

    Patients also may be skeptical that a computer can deliver the best care. A2010 study published in Health Affairs found that consumers didn’t believe doctors could deliver substandard care. In contrast, they thought that care strictly based on evidence and guidelines — as any system for automating medical care would be — was tailored to the lowest common denominator, meeting only the minimum quality standards.

    But algorithms can be put to good use in certain areas of medicine, ascomplements to, not substitutes for, clinicians. A Princeton University economics professor, Janet Currie, and colleagues developed a simple algorithm to improve care for heart attack patients. Their analysis found that about one-fifth of patients arriving in Florida hospital emergency departments with heart attacks between 1992 and 2011 received treatments not ideally suited to their condition, increasing their chances of dying in the hospital. Their algorithm based on patient characteristics could be used to indicate when treatments were not well matched to patients, potentially improving their outcomes.

    Just because algorithms can assist in making decisions doesn’t mean humans should check out and play no role. It is important not to over-rely on data and automation. Bob Wachter, a physician, relates a story about how automated aspects of an electronic medical system contributed to the overdose of a child at the University of California San Francisco Medical Center.

    Notwithstanding the cases in which reliance on automation contributes to harm, automation can improve safety, and has. Our ambition to protect us from ourselves inevitably leads to automation, even in medicine. The reason is simple: Some things humans do better than algorithms, but not all things.


    Comments closed
  • How to write, part 4

    The first three posts in this series are herehere, and here. This post reflects review comments and edits suggested by Ishani Ganguli.

    This sentence should make me want to write the next one. It should make you want to read it. In a nutshell, that’s my best answer to a question I receive frequently: “How do you write?”

    When I begin a new Upshot-styled piece, my ambition is only to figure out how it starts—the first few sentences or the lede. I don’t write anything else until I have a good lede, one that almost compels me to write more. I can spend as long on the first three sentences as on the next thirty. (Many others prefer to initially dump their thoughts to the page for later refinement and organization. That’s fine. It’s just not my approach.)

    When the writing is going well, the next sentence practically writes itself. A well written piece has a sensible logic and a palpable rhythm. If I can’t feel the next sentence being pulled out of me, something’s wrong with the previous one. It’s time to stop, think, and rewrite. It may be time to stop for the day and try again later.

    When I return to my draft, I always read it from the top and edit as I go. On the second day of working on this piece, here’s how I revised the first few paragraphs. (I’ve revised them further since then.)

    Edits made on day 2 of day 1's text. Click to enlarge.

    Edits made on day 2 of day 1’s text. Click to enlarge.

    Each paragraph is a thought. As is any good sequence of thoughts, they should be logically ordered. I usually make the logic plain by including in the first sentence of each paragraph something that links to the prior one. One word in common is often enough. For example, the second paragraph ends with a mention of sentences. The first sentence of the third paragraph includes “… the next sentence …” The writer (and the reader) is propelled from one thought to the next.

    Writers worth emulating, like Atul Gawande and Lisa Rosenbaum, break this rule sometimes. From them I’ve learned that it’s OK to broach a new thought abruptly. Until the relevance of the new thought is fully revealed—which should be rather quickly—it adds tension and intrigue. (An example is found two paragraphs below.) It works as long as it isn’t over done.

    There are few sentences that can’t be improved by shortening them. The periods in between allow the reader to rest, to consolidate thoughts. I almost always revise to shorten, not lengthen. Still, a longer sentence is worthwhile if it prevents the reader from resting prematurely—before I’ve conveyed enough to be understood.

    On assignment, writer and writing teacher William Zinsser once traveled to Timbuktu to see the salt caravan. In On Writing Well, he described how he wrote about it. His process included an hour-long search for the right words for just one sentence. This type of effort is worth it, though I usually don’t let the wrong word halt my progress for that long. Instead, I take notice of less-than-perfect words as I write and think about them later. An early draft of the third paragraph included “A well written piece has a clear logic and a precise rhythm.” I knew when I wrote them that “clear” and “precise” were lazy, tired, cliché—wrong. I even wrote them in bold, underline, italics, which is how I signal to myself that I need come back and revise.

    On a walk, I came up with the revision: “sensible logic and a palpable rhythm.” For so many reasons, I prefer “sensible” and “palpable” to “clear” and “precise.” They feel fresh. They also rhyme. Best of all, “palpable” exactly conveys my sense of feeling the right, next sentence. In one word, it makes my point. It must be right.

    My secret weapon is writing well ahead of deadlines. I’m not a procrastinator. This leaves plenty of time to find better words, among other refinements. When searching for words, I prefer ones I know well and feel like me. But I’m always on the lookout for new ones that I might grow into. I recall trying out “fealty” and “attendant” (as an adjective), for example. They still don’t feel like my voice, but maybe someday. Then again, maybe not. I can’t imagine ever writing “crestfallen.”

    When my full draft is written, I like to get it reviewed by a subject-matter expert to check my facts and reasoning. I always pick someone I think is likely to be interested in serving this role, like an author I cite. This piece was reviewed by Ishani Ganguli.

    A piece of this length should have one main point. What is it? The lede tells you. I want to make sure that point appears in the concluding paragraph too. This is a useful trick. If you can’t figure out how to end a piece like this, go back and see how you started it. If the first few sentences are right, the rest of them come more easily, including the last.


    Comments closed
  • Upshot extra: Robots as doctors? Not so fast.

    My Upshot post today is about machines (or algorithms) replacing doctors (or not). A bit from my draft that didn’t make the final is worth sharing. So, quoting my unpublished draft:

    2013 study by University of Oxford scholars predicted that health care jobs are among those least likely to be lost to automation. That does not mean that algorithms cannot be put to good use in certain areas of medicine, as complements to, not substitutes for, clinicians.


    Comments closed