Mark VanLandingham's office at Tulane University reflects a life before the storm much different from that of today. Colorful, traditional tapestries and sculptures from Southeast Asia decorate the room; books on that region's culture and economics, history and social issues line the walls; and a paper on migration and health among the Vietnamese hangs above his computer. But one detail seems out of place, one that suggests a shift in the focus of his research. When VanLandingham closes his door and sits at his desk, he faces a map of the city of New Orleans, not Southeast Asia. Until the storm, the international public health professor and demographer studied migration patterns and impacts of the HIV epidemic in the rapidly changing Asian corridor. The homicide rate on the mean streets of New Orleans didn't show up on his academic radar screen.
Little by little, however, VanLandingham has immersed himself in the grim statistics of a city emptied of thousands of its people, yet where murder and muggings have become more commonplace than ever.
The work has so consumed him that he's written two academic papers in the past year exploring the city's post-Katrina population and homicide rates. As abysmal as the "official" homicide rate appears on the pages of the daily newspaper and in NOPD statistics, VanLandingham maintains that the reality is even worse.
'What got me into [researching the city's population and homicide rates] is I hate to see things that are inaccurate. It bugs me," he says. "We need to face this head-on and deal with it, not sugar-coat the truth."
In his most recent paper, "Murder Rates in New Orleans 2007," VanLandingham analyzes local voter turnout in the October 2007 and 2003 statewide elections, coupled with extrapolation from the Census Bureau's initial 2006 mid-year population count of 223,388 (a figure that was revised downward last week). He then estimates the city's "highest plausible" mid-year population in 2007 to be 273,382. That figure contrasts with NOPD's "average monthly" figure of 299, 229 as well as City Hall's estimate of 302,000.
One number on which everyone agrees is the number of homicides last year 210. VanLandingham, using his own population estimate for 2007, pegs New Orleans' homicide rate last year at 76.8 per 100,000 a figure he calls "the most conservative measure of the murder rate" (although NOPD's is more conservative at 70.1 homicides "per capita"). VanLandingham's 2007 calculation shows a jump of at least 6 percent from 2006 and a 34 percent spike since 2004.
'We went from really, really bad in 2006," he says, "to unbelievably bad in 2007."
The whole notion of population counts and how they affect the crime rate took on even greater significance last week, when the U.S. Census Bureau released its official 2007 mid-year population estimate of 239,124 a figure that was even lower than VanLandingham's "conservative" estimate for last year.
Reaction at City Hall was swift. Mayor Ray Nagin and several City Council members called a press conference to announce that the city will challenge the Census count. Using figures provided by the Greater New Orleans Community Data Center, Nagin says the city's 2007 population was about 302,000, which would place last year's homicide rate at 69.5 per 100,000. Lower Census counts mean less federal aid to the city. It also makes the crime rate look worse. "Don't use these Census numbers to calculate crime rates," says Allison Plyer, deputy director of the GNOCDC and one of two demographers siding with the city. She adds, "You should wait until we finish this challenge."
By contrast, VanLandingham says he thinks the city's population estimates have been inflated, perhaps intentionally, in an effort to make crime rates appear lower. He adds that population figures used to determine crime rates by NOPD, as well as calculations published last month by The Times-Picayune, are mathematically incorrect and therefore statistically invalid.
While demographers use multiple techniques to count people (and fiercely debate the validity of one technique over another), experts agree that there is only one date July 1, the mid-year point whose count should be used for purposes of determining an annual rate. By using the mid-year figure, a city is neither overly rewarded for a mass inward migration at the end of the year, nor overly penalized for a mass exodus. Simply put, it's the only figure that will provide an accurate annual rate for whatever is being studied, whether it's annual birth, death or homicide rates. Textbooks teach this formula. The U.S. Census Bureau uses it.
But NOPD and The Times-Picayune stubbornly refuse to accept it.
Both NOPD and the T-P derive their own population figures from raw data provided by the GNOCDC, a nonprofit that quantifies the city's recovery by publishing the number of residences receiving mail each month. While the GNOCDC does not publish its own population estimates, its raw data consistently form the basis for the highest local estimates. To complicate matters further, VanLandingham says that while NOPD and the T-P use the raw data incorrectly, they also use it very differently.
Rather than taking a hard number from July 1, NOPD calculates the 12 monthly population figures using GNOCDC's residency figures, adds them together and then divides by 12 to get an average, according to NOPD spokesman Joe Narcisse. Using that method, he says NOPD's population estimate for 2007 was 299,229, which would place the homicide rate at 70.1 per 100,000 still alarmingly high, but not as high as the 76.8 homicides per capita calculated by VanLandingham.
Plyer says that although NOPD's method for determining the city's population is "not standard," it is "reasonable," considering the extreme population shifts in post-Katrina New Orleans. "Is it an accurate population number?" she asks rhetorically. "No one knows. There's always going to be a variety of estimates, especially now."
In post-Katrina New Orleans, there's no shortage of population estimates. Greg Rigamer of GCR & Associates analyzes the number of active utility accounts and residences receiving mail on both the address and block level to calculate population estimates. Last week, his company's Web site placed New Orleans' 2007 mid-year population at 273,598, only a few hundred people more than VanLandingham's estimate. Furthermore, GCR's 2006 estimate of 225,000 was within 1 percent of the Census Bureau's initial tally of 223,388, giving credence to GCR's methodology, VanLandingham says. (Last week, however, the Census Bureau amended its 2006 estimate downward, to 210,198. Calls to the Bureau seeking an explanation were not returned by press time.)
At Nagin's press conference last Thursday, Rigamer joined Plyer in supporting the city's challenge to the Census estimate. Rigamer also gave Gambit Weekly a new 2007 estimate up from 273,598 to 281,314. When asked to explain the change, Rigamer said GCR's earlier tally has been updated based on "subsequent revisions" supported by fieldwork.
Meanwhile, VanLandingham says the Census tally "is closer to what the real population is." He adds that "the fact that the official population numbers are coming out much lower means that the NOPD's numbers are even more skewed."
Regarding NOPD's "monthly average" calculations, Rigamer said in a March 3 email that "it is not unreasonable to monitor population on a monthly basis, but longer periods are more important in establishing and evaluating trends." In a follow-up phone interview, he added, "NOPD is in a jam. They are under a lot of pressure to lower the murder rate. Sometimes it's a lot easier to do it mathematically than it is on the street." (Even Rigamer's revised 2007 mid-year population estimate of more than 281,000 gives New Orleans a high homicide rate of 74.7 per capita last year a figure that's much closer to VanLandingham's calculation of 76.8 than it is to NOPD's 70.1.)
New Orleans crime expert Peter Scharf is equally direct in his assessment of NOPD's math. "There are political drivers behind NOPD's numbers, rather than factual drivers," he says. The former University of New Orleans criminologist, who recently was offered a position in Tulane's Department of International Health and Development, adds, "The city is buying time to keep people like Peter Scharf off its back."
But Scharf says NOPD has done itself a disservice over the long term by using population estimates that many believe are inflated. Once the city's population stabilizes, it will be difficult to determine if reform efforts have been successful because the crime rates in 2006 and 2007 were probably understated.
'Until you have hard, baseline [population] numbers you will have no idea if things are getting better or not," Scharf says, adding that a consistent and statistically valid methodology for determining the city's population needs to be agreed upon and employed by consensus.
VanLandingham agrees. He calls for "a nationally recognized and independent organization to work collaboratively with the Census Bureau to provide the city with reliable measures of population size and composition on a regular basis over the next several years."
The RAND Corporation is such an organization. The nonprofit research institution, which boasts several Nobel laureates, established a permanent office in New Orleans last June. The organization has published several reports since the storm, including population estimates for the Bring New Orleans Back Commission, the timeliness of Road Home payments for the Louisiana Recovery Authority and a preliminary report on economic development for the City Council.
'But before RAND publishes anything, it undergoes independent review," says George Penick, the director of RAND's Gulf States Policy Institute, adding that figures must stand up to external scientific scrutiny. Regarding the multiple population estimates and techniques used to derive those estimates, Penick says, "We need a [population] number where nobody is going to say we have an agenda, where objectivity is beyond reproach."
And once a reliable, objective number is derived, the city can accurately calculate its crime rate. "But no matter how you look at [the homicide rate], the numbers are horrible," Scharf concludes. "They look terrible even though there's an absence of precision."
Although experts debate the validity of NOPD's math, there is no debate about the T-P's numbers. Experts describe the paper's use of year-end population estimates rather than mid-year estimates as "invalid" and "an error." The paper uses 318,177, which is derived from GNOCDC's December 2007 data when calculating crime rates, which places last year's homicide rate at 66 per 100,000 a 16 percent increase since 2004, but far less scary than the 34 percent calculated by VanLandingham.
In crime analysis stories published this year, the paper claims that NOPD uses GNOCDC's year-end data, which "are higher than some other estimates from different sources." Therefore, it bases its crime-rate calculations on that number. However, Narcisse says that is "inaccurate." He maintains that NOPD uses an annual monthly average, not a year-end figure.
Although GNOCDC publishes "indicators of recovery" and not hard population estimates, Plyer says GNOCDC provides a mid-year population figure (of about 302,000 for 2007) when estimates are requested. "The Times-Picayune just hasn't reported it that way," she says. "There's a huge misconception that The Times-Picayune is propagating. They use their own figures. They did not use the numbers I gave them." The figures to which she refers are the 2006 and 2007 population figures used in the paper's Feb. 18 story, "Violent Crime in N.O. Soaring, Maybe."
'I was pretty surprised when the article came out," Plyer adds, "because once again, The Times-Picayune used the end-of-the-year figure."
By way of showing how the differences affect real people, VanLandingham says the paper's practice of using a year-end population figure to determine the city's homicide rate incorrectly places the same "threat level" on someone who moved to New Orleans in, say, November or December as a person who has lived in the city all or most of the year.
After the paper's Feb. 18 crime rate story was published, VanLandingham wrote a letter to the editor highlighting the paper's mathematical mistakes and asking for a correction.
In the letter, he wrote, "Dividing the number of murders by an end-of-the-year-population estimate is not a rate, it is an error. [This] guarantees an understatement of murder trends affecting our city (our population now grows throughout the year and a bigger denominator yields a lower rate); more to the point, it is simply wrong."
The paper declined to print his letter, and no correction was printed.
VanLandingham now says the paper published exactly what City Hall wants people to believe.
'They love it when the media print [year-end population figures] because it makes the murder rate look lower than it really is."