Tag Archives: Statistics

January Links: A Genuine Surprise in a Request for Plain English, no Free Grant Writing Lunches, and More on Specious Statistics

* We argued that There is no Free Grant Writing Lunch and You Won’t Find Writers for Nothing, and the New York Times in part explains why in When to Work for Nothing (answer: almost never). In addition, the article says you should seldom work for getting “paid in exposure.”

* Many of you probably read the disturbing article in the Wall Street Journal and elsewhere about how “Murders of Black Teens Are Up 39% Since 2000-01:”

The data confirm a pattern identified earlier this year by The Wall Street Journal, which found that while most communities in the U.S. were seeing a decline in homicides, many African-American neighborhoods were continuing to see an increase. The Northeastern University research shows that the pattern is more pronounced among juveniles.

What the article doesn’t tell you is that said murders have substantially—probably by more than half—since 1993. However, Freakonomics points out what’s wrong with the scaremongering implied in the WSJ:

This figure presents homicide rates by age for blacks from 1976 to 2007. The dominant pattern in this picture is the huge spike in black youth homicides in the early 1990’s. The phenomenon captured in the scary New York Times graphic above corresponds to the barely perceptible rise in the black circles at the far right of the figure.

[…]

According to U.S. Census data, the number of blacks aged 15 to 19 rose by about 15 percent between 2000 and 2007.

So even if any individual black teen’s propensity for crime was unchanged over this time period, the aggregate amount of black-teen crime would have risen by 15 percent. In other words, in that New York Times graphic on perpetrators, just based on changes in population, the number of perpetrators would have been expected to rise from a little over 800 to nearly 1,000. Knowing that, the actual rise to roughly 1,150 doesn’t seem that noteworthy.

Nonetheless, if you’re writing a proposal, you’d do well to ignore the sensible Freakonomics pieces and quote the WSJ or NYT liberally, since they are authoritative and your chief responsibility is making sure that your grant story gets the money.

* In April 2008, we wrote a post on FEMA Tardiness, Grants.gov, and Dealing with Recalcitrant Bureaucrats, in which I described FEMA’s failure to use Grants.gov to announce the Assistance to Firefighters Grants program; the post illustrates the problems discussed in Grants.gov Lurches Into the 21st Century.

To FEMA’s credit, an administrator named R. David Paulison responded to a letter I sent, and Paulison said that FEMA disseminated information about the Assistance to Firefights Grants program through other channels. But, if I send him the firefighter department I’m associated with, he’d consider allowing a late application. Obviously I’m not associated with a fire department, but perhaps this means FEMA will issue the announcement on Grants.gov in a timely manner next year. Still, the letter Paulison sent me is dated December 5, indicating that perhaps FEMA’s tardiness problems haven’t exactly been solved, given that I sent my letter in April.

* Like the businesses they’re bailing out and the nonprofits they’re funding, federal and state governments are not looking good. Note in particular the last line of the linked post.

* RFPs are normally written in bizarre doublespeak, as we’ve amply documented. But in the DARPA Broad Agency Announcement NanoThermal Interfaces (NTI) MTO (warning: .pdf link), Isaac found something he’s never seen over 35 years and thousands of RFPs—a request for simplicity:

Statement of Work (SOW) – In plain English, clearly define the technical tasks/subtasks to be performed, their durations, and dependencies among them.

Presumably, the Defense Advanced Research Projects Agency (DARPA)—the same guys who laid the foundation for the Internet—has among the brightest technical lights in the U.S. government working for it. If they request sections of highly technical proposals in plain English, perhaps the Department of Education should learn something from them, instead of ceaselessly requiring proposals written in educrat. Foundation proposals will occasionally request summaries or even proposals in relatively clear language, but more often than not, their guidelines look as though HUD rejects got into the review process.

Sometimes one will find proposal narrative guidelines almost as long as the page limits on the narrative itself; I can’t think of an immediate example save YouthBuild, as the narrative section of SGA is about 17 double spaced pages for 20 pages of required narrative. Finding another example would require digging through the voluminous (albeit digitally so, these days) RFP archives that make even seasoned grant writers blanch. Regardless, when you can find a request for real writing, savor it: you’ve got a rare dish you won’t taste often.

* Occasionally we’ll post examples of bureaucratic silliness and obtuseness, and I ran into a great example with the The Service Area Competition – Additional Service Areas funding opportunity. If you read the “Additional Information on Eligibility” section, you’ll see that it defines eligible applicants as “Public or nonprofit private entities, including tribal, faith-based and community-based organizations; and Organizations proposing to serve the same service area and/or populations identified in Appendix F,” without saying what the program is actually designed to do. And two paragraphs mention Appendix F at least half a dozen times. So I downloaded the 153-page file and searched for Appendix F, expecting a cornucopia of possible applicants but instead found four: La Pine, OR, Charleston, SC, Marchester, NH, and Miles City, MT.

Wouldn’t it have been easier simply to write those four applicants in the description on the website? And what makes these incredibly narrow areas important enough to justify their own funding announcement? I don’t know for sure, but if I were to wager, I would guess that HRSA, for whatever reason, wants to wire money for specific organizations in each area, and that whichever organizations know they’re getting the money just need to turn in something mostly correct to collect.

* A point Isaac has made many times in private now finds expression on a blog: senior bureaucrats, not political appointees, really run things in Washington.

* The New York Times discusses the “Evidence Gap” in “Drug Rehabilitation or Revolving Door?“, with the article strongly implying “revolving door.” Note this piece:

Yet very few rehabilitation programs have the evidence to show that they are effective. The resort-and-spa private clinics generally do not allow outside researchers to verify their published success rates. The publicly supported programs spend their scarce resources on patient care, not costly studies.

And the field has no standard guidelines. Each program has its own philosophy; so, for that matter, do individual counselors. No one knows which approach is best for which patient, because these programs rarely if ever track clients closely after they graduate.

(Emphasis added. We’ve discussed why that is in Studying Programs is Hard to Do: Why It’s Difficult to Write a Compelling Evaluation and, to a lesser extent, in What to do When Research Indicates Your Approach is Unlikely to Succeed: Part I of a Case Study on the Community-Based Abstinence Education Program RFP. The whole article illustrates the problems with evaluations that we describe in the two posts above.)

* New York Times columnist Nick Kristof describes what might be called the charity paradox, whereby those who do good deeds are supposed to be utterly saintly while those in business are supposed to be utterly rapacious, in The Sin in Doing Good Deeds. The column, naturally, attempts to reconcile the two. We discuss similar issues in Foundations and the Future, which was published about a year ago.

* More on questionable abstinence studies, this time from the Washington Post, which says “Premarital Abstinence Pledges Ineffective, Study Finds; Teenagers Who Make Such Promises Are Just as Likely to Have Sex, and Less Likely to Use Protection, the Data Indicate.” Read What to do When Research Indicates Your Approach is Unlikely to Succeed: Part I of a Case Study on the Community-Based Abstinence Education Program RFP for more on the smoke surrounding abstinence education, whether in favor or against. Remember too that, if you’re writing a proposal for an abstinence program, Your Grant Story Needs to Get the Money—so if the data don’t support the RFP you’re writing for, don’t use them.

Writing Needs Assessments: How to Make It Seem Like the End of the World

Almost every grant proposal requires some form of needs assessment. More or less, the sentiment one must get across it that “It’s the end of the world as we know it and I feel fine,” as REM says. Essentially, the object is to make problems look overwhelming, but solvable with just a dollop of grant funds. So, how does a grant writer do this?

Start by making the end appear nigh, which requires a needs assessment. Look at the Census data available at American Fact Finder, which has a variety of geographic choices (e.g. county, city, zip code, census tract, etc.). It is almost always best to match the project target area with a census data geographic area to make assembling data easier, regardless of whether the census area perfectly matches the area you want to serve. Try not to make the target area, “the Westside of Dubuque,” unless that happens to conform with four census tracts. Most geographic areas have 2000 Census data, as well as estimates for 2005. Pick the date that is to your advantage, and being to your advantage means making the situation look worse. For example, if incomes have been trending downward and unemployment upward due to plant closings, the 2005 data may be better. Announce that, if current trends continue, Dubuque may be abandoned completely in 2010 because there are too few jobs, but the situation can be improved with the requested grant.

Once you have your target area, find useful socioeconomic indicators like ethnic breakdown, median family income, age cohort percentages, percent of people living below poverty, percent with disabilities, etc. Only include data that supports your case. A winning grant proposal is not like a thesis, so you are under no obligation to use all available data. Also, it is critical that you provide some data on a larger area for comparison purposes, so your readers understand the relative problems. This can be the city, the county, state or even national data—pick whichever makes your situation look worst, meaning with the greatest discrepancy between the target area and the larger sample. It doesn’t really matter which geographic level you compare to, as long as you can say something to effect of, “The target area median family income is just 2/3 that of Los Angeles County.” Depending on the target population, it may be advantageous to compare data for a particular ethnic group to all residents. For example, if the target area includes a significant African American population with lower incomes, you can set up tables showing African American indicators versus white indicators for the same geographic area, in essence comparing the target area to itself. American Fact Finder has a handy tool on the left button bar for “Fact Sheet for a Race, Ethnic or Ancestry Group” that makes this easy to do.

FactFinder

(Click here to see the full image.)

You can also use census data to obfuscate the actual reality in the target area. For example, in many Southern California cities there are high percentages of Asian Americans, who in some communities have higher-than-average incomes. This can be used for statements such as, “over two in five residents is a person of color.” For better or worse, most grant reviewers will usually associate persons of color with lower incomes and higher risk factors whether this is true or not. Grant reviewers seldom have a deep background in statistics and they probably don’t even know statistics for journalists let alone real statistics. Even if they do, everything starts to become a haze after reading a dozen federal proposals that can be onerously long, so most reviewers are apt to begin looking more for conclusions than data not long into the process. Do you somehow fulfill the checkbox that asks whether educational attainment is lower in the target area than the nation? If so, give ’em five points and move on.

Other good sources of data include state and local departments of education. Some states and school districts have better data engines than others. For example, the California Department of Education has a great site, DataQuest, but other states’s data system are, as Borat would say, “not so much”. If a good data engine/warehouse is not available, find the school/district reports cards mandated by the federal “No Child Left Behind” legislation. Many districts try to hide these reports, as they are often unflattering after you get past the mission/vision statement platitudes, but if you dig hard enough you will find them. If necessary, call the statistics unit at the district or state and force the reports out of them.

Once you have data, only use what helps the argument. So, if test scores for certain grades are low relative to the county or state, use those, not all test scores. If you want to use dropout data, use the four-year derived rate, not the single year, which will be much lower. In some states, such as Illinois, drop out data is wildly understated, due to the way the state treats students who are no longer in school, so if you have to use it, underscore this fact. Health data, including disease incidence, mortality, etc., can usually be found at state and local health department web sites, while crime and gang data are typically found at police department web sites.

If you’re having difficulty building your argument with data, a good technique is to call local “experts” for quotes. For example, find and call the police unit responsible for gang suppression in your target area, then ask leading questions. Invariably, the officer will tell horror stories about rampant gang activity. Just ask if you can quote her and she will almost always agree. It’s always fun to include the names of some local gangs in your proposal for a dash of reader titillation. This is particularly important if the reader is on proposal 35 out of 40 and just wants to go find the hotel bar. You can also find the name of any large social service provider or city official in the target area (other than the one for whom you are working) and ask them about local problems with the target population. For example, if you seek information about at-risk youth services and you talk to the local Boys and Girls Club executive director or city parks director, this person will almost always say that new problems are erupting every minute while their funding is declining.

This gives you the opportunity to write something like, according to Conrad Cuttlebone, YMCA Director, “there are many more latch-key kids in the community since the Hindenburg Dirigible Factory closed, and we’re seeing many more cases of domestic violence, while at the same time the county cut our funding by 50%.” When all else fails, you can simply write, “although specific target area level data is not available, the agency knows anecdotally that teen pregnancy is on the rise, mirroring national trends.” Of course, you can do this even if the local area doesn’t match national trends, as most reviewers don’t have the vaguest idea about national trends for anything.

In other words, while it is not a good idea to make up data, it’s perfectly fair to exaggerate problems through obfuscation and specious analysis. You’re generally rewarded for such effects: the worse the target area, the more likely you are to get points, and the more likely you are to be funded.

The gentle art of writing needs assessments really comes down to painting word pictures that combine cherry-picked data with opinions and anecdotes strung together to meet the expectations of reviewers, who assume something terrible must be going on in your community, or you would be doing almost anything other than writing a grant proposal, such as watching my favorite college football team, the KU Jayhawks, trounce Virginia Tech in the upcoming Orange Bowl. Rock! Chalk! Jayhawk! KU!