Identity Theft and Statistics 101
"Facts are stubborn things," Mark Twain once wrote, "but statistics are more pliable." Not more so than public understanding of statistics, if one is to judge from initial reactions to the recent Bureau of Justice Statistics report on identity theft-related survey data (at http://www.ojp.usdoj.gov/bjs/pub/pdf/it04.pdf). Here are some of the more egregious errors in reporting on the BJS data:
- According to ConsumerAffairs.com, "The repport [sic] found that 3.6 million Americans were affected by identity theft in 2004, a significant drop from a similar report issued conducted [sic] by the Federal Trade Commission (FTC)." (http://www.consumeraffairs.com/news04/2006/04/id_theft_stats02.html) It is hard to imagine a more careless misreading of the data. First, the BJS bulletin plainly refers to 3.6 million households -- not individuals. Second, it notes that in those households, "at least one member of the household had been the victim of identity theft during the previous 6 months." Thus, 3.6 million is the minimum (not the maximum) possible number of victims, based on the survey data. Third, if the article is referring to the only identity theft survey conducted for the FTC, Synovate issued that survey in September 2003, using random-digit-dialing survey data from March and April 2003. (http://www.ftc.gov/os/2003/09/synovatereport.pdf) In that survey, 4.6 percent of survey respondents, which the survey extrapolates to nearly 10 million Americans, reported that they had been victims of identity theft within the past year. By contrast, the BJS survey dealt only with identity theft victimization within the previous six months. It is highly probable (to say the least) that the number of people who have had any kind of experience -- identity theft, a meeting with a college roommate, a traffic accident -- within one year's time will be greater than the number of people who have had that same kind of experience within six months' time. At any rate, there is no valid basis for concluding, on the basis of these two surveys, that the BJS data represent "a significant drop" from the Synovate data.
- The headline for a SecurityProNews article on the BJS survey data declared in red letters: "Identity Theft 20X Bigger Problem Than Reported." (http://www.securitypronews.com/news/securitynews/spn-45-20060410IdentityTheft20XBiggerProblemThanReported.html) Actually, no, on two grounds. First, if the calculation is based on the 3.6 million households divided by the 246,847 identity-theft complaints filed with the FTC for 2004, identity theft is only a "14.58X Bigger Problem." Nothing else in the article supports the 20X figure. Second, if -- instead of mixing apples and oranges by dividing the number of households by the number of individuals who filed complaints -- the article had divided the 246,847 complaints into the 9.3 million individuals who reportedly were identity-theft victims in 2004, based on a Better Business Bureau - Javelin Strategy survey (see http://www.javelinstrategy.com), the headline should have read: "Identity Theft 37.68X Bigger Problem Than Reported." As for the article itself, it stated that the BJS report "reveals that the [FTC's] initial 2004 identity theft report missed severely missed [sic] the mark, according to the National Crime Prevention Center [sic] (NCPC)." Again, no. If a report accurately states the number and types of complaints to a government agency about a particular problem, that report does not "miss the mark" merely because the majority of people who experience that problem do not complain about it to that agency. The complaint data are what they are. Because the FTC has never claimed that the complaints it receives reflect the true incidence of identity theft, the "misses the mark" comment is unwarranted.
- An E-Commerce Times article states that the BJS numbers "suggest that the incidence of identity theft might be lower than what has been reported in the past." (http://www.ecommercetimes.com/story/Cr3GF4pG2kKwjx/DoJ-Identity-Theft-Touches-Millions-in-US.xhtml) This statement is more carefully couched, but still subject to misinterpretation. Because the survey addresses households rather than individuals, and the number of individuals per household can range from one to ten or more, there is a range of possibilities as to the prevalence of identity theft. If one takes the low end of that range, then 3.6 million individuals would be victims of identity theft. But if one takes the average number of persons per household, 2.59, based on 2000 U.S. Census data (http://quickfacts.census.gov/qfd/states/00000.html), and multiplies that by the 3.6 million households, there could have been as many as 9.3 million identity-theft victims -- coincidentally, the same number of victims that the BBB - Javelin survey found for 2004. The only thing that the BJS data "suggest" is what can be directly inferred from those data. Followup studies by BJS for later periods will be necessary before reliable conclusions can be drawn about possible trends in the prevalence of identity theft.
- According to ConsumerAffairs.com, "The repport [sic] found that 3.6 million Americans were affected by identity theft in 2004, a significant drop from a similar report issued conducted [sic] by the Federal Trade Commission (FTC)." (http://www.consumeraffairs.com/news04/2006/04/id_theft_stats02.html) It is hard to imagine a more careless misreading of the data. First, the BJS bulletin plainly refers to 3.6 million households -- not individuals. Second, it notes that in those households, "at least one member of the household had been the victim of identity theft during the previous 6 months." Thus, 3.6 million is the minimum (not the maximum) possible number of victims, based on the survey data. Third, if the article is referring to the only identity theft survey conducted for the FTC, Synovate issued that survey in September 2003, using random-digit-dialing survey data from March and April 2003. (http://www.ftc.gov/os/2003/09/synovatereport.pdf) In that survey, 4.6 percent of survey respondents, which the survey extrapolates to nearly 10 million Americans, reported that they had been victims of identity theft within the past year. By contrast, the BJS survey dealt only with identity theft victimization within the previous six months. It is highly probable (to say the least) that the number of people who have had any kind of experience -- identity theft, a meeting with a college roommate, a traffic accident -- within one year's time will be greater than the number of people who have had that same kind of experience within six months' time. At any rate, there is no valid basis for concluding, on the basis of these two surveys, that the BJS data represent "a significant drop" from the Synovate data.
- The headline for a SecurityProNews article on the BJS survey data declared in red letters: "Identity Theft 20X Bigger Problem Than Reported." (http://www.securitypronews.com/news/securitynews/spn-45-20060410IdentityTheft20XBiggerProblemThanReported.html) Actually, no, on two grounds. First, if the calculation is based on the 3.6 million households divided by the 246,847 identity-theft complaints filed with the FTC for 2004, identity theft is only a "14.58X Bigger Problem." Nothing else in the article supports the 20X figure. Second, if -- instead of mixing apples and oranges by dividing the number of households by the number of individuals who filed complaints -- the article had divided the 246,847 complaints into the 9.3 million individuals who reportedly were identity-theft victims in 2004, based on a Better Business Bureau - Javelin Strategy survey (see http://www.javelinstrategy.com), the headline should have read: "Identity Theft 37.68X Bigger Problem Than Reported." As for the article itself, it stated that the BJS report "reveals that the [FTC's] initial 2004 identity theft report missed severely missed [sic] the mark, according to the National Crime Prevention Center [sic] (NCPC)." Again, no. If a report accurately states the number and types of complaints to a government agency about a particular problem, that report does not "miss the mark" merely because the majority of people who experience that problem do not complain about it to that agency. The complaint data are what they are. Because the FTC has never claimed that the complaints it receives reflect the true incidence of identity theft, the "misses the mark" comment is unwarranted.
- An E-Commerce Times article states that the BJS numbers "suggest that the incidence of identity theft might be lower than what has been reported in the past." (http://www.ecommercetimes.com/story/Cr3GF4pG2kKwjx/DoJ-Identity-Theft-Touches-Millions-in-US.xhtml) This statement is more carefully couched, but still subject to misinterpretation. Because the survey addresses households rather than individuals, and the number of individuals per household can range from one to ten or more, there is a range of possibilities as to the prevalence of identity theft. If one takes the low end of that range, then 3.6 million individuals would be victims of identity theft. But if one takes the average number of persons per household, 2.59, based on 2000 U.S. Census data (http://quickfacts.census.gov/qfd/states/00000.html), and multiplies that by the 3.6 million households, there could have been as many as 9.3 million identity-theft victims -- coincidentally, the same number of victims that the BBB - Javelin survey found for 2004. The only thing that the BJS data "suggest" is what can be directly inferred from those data. Followup studies by BJS for later periods will be necessary before reliable conclusions can be drawn about possible trends in the prevalence of identity theft.
posted by Jon Rusch at 7:40 AM
0 Comments:
Post a Comment
<< Home