WSJ : Avocados, Gorillas and Other Favorites

THE NUMBERS GUY By CARL BIALIK
July 28, 2005

[ABOUT THIS COLUMNThe Numbers Guy examines numbers and statistics in the news, business, politics and health. Some numbers are flat-out wrong, misleading or biased. Others are valid and useful, helping us to make informed decisions. As the Numbers Guy, I will try to sort through which numbers to trust, question or discard altogether. And I'd like to hear from you at numbersguy@wsj.com. I'll post and respond to your letters. WSJ.com subscribers can sign up to receive email when new columns are published (nonsubscribers click here to sign up), and you can read more columns at WSJ.com/NumbersGuy. ]

Workers who used the Internet for non-work-related tasks cost their employers a whopping $178 billion last year in lost productivity, according to a new study sponsored by Websense Inc., a company that sells software to monitor employee Internet use.
But I'm hoping that as faithful Numbers Guy readers, you focused not on that giant dollar amount but on "sponsored by," the two dirty words of market research. I wrote about those dangerous little words in a column back in March, one of many I've written about how numbers – even good ones -- only tell part of the story. This is the 25th Numbers Guy column, and I'm marking the silver anniversary by revisiting some of my favorite topics.
First up in the Numbers Guy time machine is a set of two studies that attempt to estimate the cost companies suffer because of employees' time-wasting. These studies highlight two subjects dear to my skeptical heart: sponsored research, and efforts to put a dollar amount on lost productivity.
Websense's study, which focused on idle Web surfing, worked like this: The company, working with research firm Harris Interactive, surveyed workers with Internet access, and half said they use the Web at work for personal reasons. The survey also asked information-technology managers to estimate how much time their workers wasted on Web surfing. Their average answer: 5.9 hours. Multiply 5.9 hours by the 68 million U.S. workers with Web access, then multiply by 0.5 because only half of those surveyed admitted to engaging in Web procrastination, multiply the result by 50 work weeks in a year and then multiply by the average U.S. salary (still with me?) and you get $178 billion. The study was picked up by IDG News and Red Herring.
A large number is good for Websense, because it is trying to sell products that help companies monitor and improve worker productivity. The company made several choices that might have helped produce a larger number. One worth noting: It used IT managers' estimates of 5.9 wasted hours per week per worker, even though workers themselves had given a lower average number -- 3.4 hours -- in response to the survey. (For more detailed critiques of Websense's study, see the blog ars technica, ZDNet and Computer World.)
But the bigger problem is that these lost-productivity studies assume that eliminating nonwork activity will translate into a corresponding savings. Wages aren't that simple. Many jobs include some built-in downtime. Also, many bosses don't care how their workers spend their time as long as they successfully complete their given tasks. Furthermore, mixing work and fun online cuts both ways; as Ken Fisher wrote on the ars technica blog, "I'm still waiting to see a good study done on how many hours the average employee spends doing work-related stuff online, at home." (I also took a look at a silly estimate of lost productivity due to the Super Bowl in a February column.)
Leo Cole, vice president of marketing for San Diego-based Websense, said he felt comfortable with the $178 billion number, based on anecdotal reports from Websense customers that squared with the 5.9-hour estimate and his feeling that workers underreported their personal Web usage. Mr. Cole acknowledged that some might question the survey because of Websense's bias; nonetheless, he said, "we try to be a thought leader" in areas including time wasted online at work.
Others have come up with even bigger estimates for productivity lost due to time-wasting at work. A study jointly conducted by America Online and Salary.com, also released this month, pegged the loss to U.S. businesses at $759 billion. (Numbers Guy reader William Walker suggested I examine the study.) Its methodology was similar to Websense's, and it, too, found that workers were wasting about six hours a week that their bosses didn't know about. AOL referred questions to Salary.com, which declined to grant me an interview.
The Associated Press ran several local stories citing AOL's rankings of the states with the most and least time wasting. Missouri was ranked the biggest time-wasting state, with 2.8 hours lost each day. Missouri governor Matt Blunt's office took time out to respond to the survey in a statement, according to the AP: "Nobody can match the work ethic of Missourians. This survey, which our busiest citizens did not want to waste their time on, cannot undermine decades of experience. Missouri workers are among the most productive in the world."
* * *
In a public-service ad campaign, viewers are told a scary stat: One in five children has been sexually solicited online. But as I wrote back in January, the statistic is misleading. Among other problems, it was based on research that was five years old, it only covered children who spent time online, and it used a broad definition of sexual solicitation. Yet the stat was revived last month in a fresh set of Ad Council spots sponsored by the National Center for Missing and Exploited Children, as reported by the New York Times. Why continue to use an outdated, misleading stat? "There isn't a more updated number available," a spokeswoman for the National Center told me. "We're hoping to have new research soon, hopefully by the end of the year."
* * *
Did Americans really consume 43.8 million pounds of Hass avocados during the Super Bowl, as the Hass avocado board, an industry group, predicted before the big game in February? I pointed out that the projection assumed all avocadoes purchased within several days of the game would be consumed during the game, which surely was an exaggeration. Nonetheless, the projection turned out to be pretty good, for what it was: A Hass spokeswoman told me that the actual number of avocadoes sold in the relevant two-and-a-half-week period before the game, as reported in direct sales from members, was 43.75 million pounds. (Of course, these are all self-reported numbers, so should be taken with a grain of salt, like any good avocado.)
After the game, I questioned the NFL's projection that Jacksonville, Fla., would enjoy $181 million in direct economic benefits from hosting the game. That number was derived from averaging economic-impact numbers for prior Super Bowls, but Jacksonville is far smaller than many of the cities studied. As it turns out, Jacksonville will probably never have a better number, because the NFL didn't commission a study this year. (The league usually studies economic impact every two or three years; the last one came in 2003 for San Diego.) Nonetheless, Cathy Chambers, spokeswoman for the city's chamber of commerce, told me that the game was a "success," citing the growth in "prospect activity" -- the number of businesses seriously considering relocating to Jacksonville -- to 77 today, from 52 before the game.
* * *
A small change in U.S. News and World Report's law-school ranking methods affected several schools, "a tale of how a tweak to methodology can have large, unintended effects," I wrote in April. Until this spring's rankings, the magazine had been asking law schools to report their incoming students' average grade-point averages and Law School Admission Test scores in the form of the median -- the number in the middle of a set of numbers, if you line them up in order. But U.S. News editors started to doubt the self-reported numbers, and so instead started collecting the 25th and 75th percentile numbers that schools report to the American Bar Association. Several law schools that had been working to optimize their median numbers, and were admitting minority students with much-lower scores in the bottom half of their classes, said they were blindsided by the move. There were concerns the change could affect future minority admissions.
The controversy may have a short lifespan. Last month, U.S. News told a law-school conference that if the ABA started to collect median scores, the magazine would "very seriously consider" returning to its old method, U.S. News' director of data research, Robert Morse, told me by email. Afterwards, the ABA said it would start asking for median scores. (An ABA spokeswoman told me the change was made because of "a feeling that additional data would be helpful for people to apply to law schools," and not because of the magazine's statement.) If U.S. News does decide to use those numbers, it could return to the old method in time for the rankings set to come out next spring.
* * *
A quick update on two criminal cases whose numbers were subjects of prior columns:• The arrest of Dean Arthur Schwartzmiller in San Jose., Calif., for the alleged sexual abuse of two boys made headlines when police revealed they found seven log books with 36,000 entries of sex acts with children. I wrote last month that it was unlikely that each entry represented a separate sex act, since that would mean about two sex acts a day for 47 years. In a recent interview, Sergeant Nick Muyo of the San Jose police department told me, "We do know that some are repeat entries. It's not 36,000 different people." He declined to say how many potential victims police suspect were abused, but he did reveal an interesting detail about where the 36,000 figure came from. It wasn't based on a direct count, but instead derived by counting the entries on some pages, and multiplying by the number of pages in the log books. (Mr. Schwartzmiller's attorney postponed entering a plea on Tuesday, the Associated Press reported.) • In February, I wrote about the case of 3.6 million nickels that went missing from an armored truck, and asked whether three million of them could really fit in a four-foot-deep hole, where they were found by police. Last month in Miami, four men pleaded guilty to receiving stolen property that had traveled in interstate commerce. The truck driver and about 900,000 nickels remain at large.
* * *
Most real gorillas weigh no more than 400 pounds, but metaphorical gorillas know no bounds, I wrote in April. Since that column, more powerful people and businesses have received the label in news articles. Here's a recent sampling:• Wal-Mart is the 800-pound gorilla of supermarkets, and "the 800-pound gorilla is getting larger," Business Week Online reported. It's no surprise: Supermarkets are full of calorie-laden comestibles ideal for simian weight-gain. • Senatorial apes are the front-runners in the 2008 presidential election, the Associated Press reported last month. Hillary Clinton and John McCain are the "800-pound gorillas," Democratic consultant Jeff Link told the AP. "They're well-known, well-liked and will be heavy favorites in their respective parties." • A pension shortfall in San Diego is so big, it's the 1,600 pound gorilla, Republican mayoral candidate Steve Francis told Bloomberg. • Circulation is the 500-pound gorilla stalking America's newsrooms, Stan Tiner, executive editor of The Sun-Herald in Biloxi, Miss., told Editor & Publisher in May. It's unclear if continued circulation declines create a larger or smaller gorilla.
* * *
Two readers wrote in to say I was too quick to capitulate to my editor on the point that the word "data" is plural and must be treated as such.
You are not necessarily wrong. I'm a numbers guy, too (M.S. in statistics), and have been fighting this battle for 20 years. "Data" is often used as a collective noun, like "information." See Dictionary.com. Some people like to say that you have to treat data as plural because you can count it. So, when they rob a bank, the money are all gone?
--John C. Nelson
One would use a plural "data" when the data is not seen as a unit. That would be raw data from a measurement. ... But if you rework raw data into another product, or if you refer to a set of data as a whole, the singular is the correct usage. "The data from the weather station tells us that the average daily maximum temperature last month was 78 degrees."
--Fred Pratt
John and Fred make good points. Unfortunately, I'm bound by The Wall Street Journal's style guide, which dictates that "data" is a plural noun that always takes plural verbs and pronouns. I'll try to behave from now on.
Finally, a couple of readers pointed out that a letter I ran last week about a Monday Night Football game the night before the Sept. 11, 2001, terrorist attacks, erroneously said that the New York Giants played the Philadelphia Eagles. The Giants actually faced the Denver Broncos. I should have caught that, as I watched the game, too. It's further demonstration that memories of just four years ago can be faulty, whether they concern the terror attacks or a football game.
Write to me at numbersguy@wsj.com. I'll post and respond to selected letters here soon. WSJ.com subscribers can also sign up to receive an email notice when new Numbers Guy columns are published (nonsubscribers can sign up by clicking here). Read other columns at WSJ.com/NumbersGuy.