WASHINGTON -- Federal immigration agents were prepared to arrest an illegal immigrant and registered sex offender days before the November elections but were ordered by Washington to hold off after officials warned of "significant interest" from Congress and news organizations because the suspect was a volunteer intern for Sen. Robert Menendez, according to internal agency documents provided to Congress.

The Homeland Security Department said last month, when The Associated Press first disclosed the delayed arrest of Luis Abrahan Sanchez Zavaleta, that AP's report was "categorically false."

Sanchez, 18, was an immigrant from Peru who has overstayed a visitor visa that allowed him to enter the United States. He eventually was arrested at his home in New Jersey on Dec. 6. He has since been released from an immigration jail and is facing deportation. Sanchez has declined to speak to the AP.

After the AP story, which cited an unnamed U.S. official involved in the case, Sen. Charles Grassley of Iowa and six other Republicans on the Senate Judiciary Committee asked the Obama administration for details about the incident.

According to those documents, U.S. Immigration and Customs Enforcement agents in Newark had arranged to arrest Sanchez at the local prosecutor's office on Oct. 25. That was fewer than two weeks before the election.

Noting that Sanchez was a volunteer in Menendez's Senate office, ICE officials in New Jersey advised that the arrest "had the possibility of garnering significant congressional and media interest" and were "advised to postpone the arrest" until officials in Washington gave approval. The documents describe a conference call between officials Washington and New Jersey to "determine a way forward, given the potential sensitivities surrounding the case."

The senators, in a letter to the Homeland Security Department, said the agency documents showed that Sanchez's arrest "was delayed by six weeks," as AP had reported. They asked for details about the department's review of potentially sensitive, high profile immigration cases when arrests are delayed.

In a letter Monday, Assistant DHS Secretary Nelson Peacock said an allegation that the government delayed Sanchez's arrest "for political purposes" was categorically false. Neither the unnamed U.S. official cited in AP's original story or the senators in their letters to the department had specifically alleged that the arrest had been delayed for political purposes. The official, who spoke on condition of anonymity because this person was not authorized to discuss details of Sanchez's immigration case, told AP last month that the department had instructed federal agents not to arrest Sanchez until after Election Day.

The documents provided to Congress do not indicate why the arrest should have been delayed or whether anyone outside Immigration and Customs Enforcement — such as in the headquarters offices of the Homeland Security Department — was consulted.

J. Scott Applewhite / AP file

Sen. Robert Menendez, D-N.J.

Menendez, D-N.J., who advocates aggressively for pro-immigration policies, was re-elected on Nov. 6 with 58 percent of the vote. Menendez said last month that his staff was notified about the case immediately before AP's story, he learned about the case from the AP and he knew nothing about whether or why DHS had delayed the arrest.

According to police records, Sanchez was 15 when he was arrested on a charge of aggravated sexual assault in 2009. The records show he was accused of sexually assaulting an 8-year-old boy at least eight times and sentenced to two years' probation and required to register as a sex offender. The AP is not reporting the boy's relationship to Sanchez to avoid identifying the victim.

The agency documents show that Sanchez failed to update his sex offender registration, and local prosecutors considered arresting him for that. During the same time, immigration officials learned that Sanchez had applied for the Obama administration's Deferred Action for Childhood Arrivals program, which would have allowed him to stay in the country and legally work for two years. He did not disclose his arrest or status as a sex offender on the application and was eventually denied, according to the documents.

Immigration enforcement operations in New Jersey were largely halted starting Oct. 28 as officials prepared for Hurricane Sandy. By Nov. 29, ICE had planned to arrest Sanchez after Citizenship and Immigration Services had formally denied his deferred action application. The following day, the ICE Office of the Principal Legal Advisor was consulted. The agency's chief counsel was also consulted and the arrest was approved Dec. 5.

Sanchez was arrested the next day.

During the final weeks of President George W. Bush's administration, ICE was criticized for delaying the arrest of President Barack Obama's aunt, who had ignored an immigration judge's order to leave the country several years earlier after her asylum claim was denied. She subsequently won the right to stay in the United States after an earlier deportation order, and there was no evidence of involvement by the White House.

In that case, the Homeland Security Department had imposed an unusual directive days before the 2008 election requiring high-level approval before federal agents nationwide could arrest fugitive immigrants including Zeituni Onyango, the half-sister of Obama's late father. The directive from ICE expressed concerns about "negative media or congressional interest," according to a copy of that directive obtained by AP. The department lifted the immigration order weeks later.

Previous story: Sen. Menendez employed intern who was illegal immigrant, sex offender

N.J. Sex Offender Internet Registry

The poster child of sex offenders who altered their digital identity is Fran Kuni, who changed his name to Jamie Shepard and was able to get a job as a U.S. Census worker in New Jersey before being busted by a mom who recognized him when he knocked on the door of her home.

Nearly one in six convicted sex offenders is using sophisticated techniques invented by identity thieves to avoid their legally mandated registration requirements, a new study has found. These digital absconders might be able to avoid post-incarceration restrictions by living near schools and playgrounds, and could possibly gain employment working with children.

The study, conducted by Utica College and funded by the U.S. Justice Department, estimates that roughly 92,000 of the 570,000 registered sex offenders across the country are systematically manipulating their names, birthdays, Social Security numbers and other personal identifiers so they can live as they want while appearing to satisfy court-imposed or statutory restrictions.

"These are offenders who are flying under the radar and authorities don't know it," said Don Rebovich, the Utica professor who directed the study. "The authorities really don’t have the resources to keep on checking on these people. Offenders find where the vulnerabilities are in the system and exploit them."

These digital absconders create two obvious problems. Communities expend energy and resources dealing with offenders who aren't really there -- local police knock on doors and send notices to warn neighbors; public listings are published on the Internet. And sex offenders live where they please as normal adults, without any protective measures kicking in.

"In the worst-case scenario, by thwarting registration requirements, they could potentially have easier access to children," said Staca Shehan, director of case analysis at the Center for Missing and Exploited Children, who is familiar with the study. "(In) those jurisdictions that have residency restrictions that would not allow (offenders) to live within distance of a school, daycare or park, (they) could avoid that type of requirement."

While the study found that an average of 16.2 percent of sex offenders manipulate their identities nationally, some states fared worse: Louisiana, Washington, D.C., Nevada, Tennessee and Delaware all had digital absconder rates of higher than 25 percent.

Officials in Tennessee, Nevada and Delaware challenged the study's conclusions and complained that they had not been contacted by the researchers for additional information that might have clarified the results; officials in the other states did not immediately respond to requests for comment.

'Strategic' manipulation
Shehan said there are generally two kinds of sex offender absconders: those who simply fail to keep their records current, and hope they fall through the cracks; and those who are more systematic in their evasion, intentionally altering their identities so they can circumvent the restrictions. 

"That takes a lot more thought," she said. "They are much more strategic about what they are doing ... and so that's much more concerning."

In one celebrated case of sex offender identity manipulation, a convict named Frank Kuni changed his name to Jamie Shepard and was able to get a job as a U.S. Census worker in New Jersey. Kuni was recognized by a mom after he knocked on the door of her Pennsauken home, and he was later sentenced to three years in prison. Kuni’s case attracted national headlines because of the fear it created surrounding temporary Census workers.

The Utica study, believed to be the first attempt to quantify these more strategic absconders, was conducted by Utica College's Center for Identity Management, set up to examine a variety of identity issues in the digital age. Rebovich is director of the center.

It's well known that some sex offenders neglect their registration requirements, dropping off the grid and accepting only cash-paying jobs to remain hidden. But the Utica study found something more subtle, and perhaps more disturbing -- sex offenders who appear to be satisfying their registration requirements while living a digital double life.

In a parallel survey of 223 law enforcement agencies from 46 states, Utica found that awareness of ID-theft style registry evasion was low -- only 5 percent of respondents said they knew of an identity manipulation case within their jurisdiction. 

And nearly 40 percent of the agencies responded that they had zero absconders, suggesting some law enforcement agencies are unaware of the problem.

The power of the Utica study lies in the use of sophisticated algorithms developed by private firm ID Analytics, a fraud-fighting company used by many large banks and other financial institutions. ID Analytics receives more than 1 billion credit applications and other credit-related events from clients every year. It uses sophisticated software to track the behavior of identity thieves across the credit system, and can find fraud that individual firms miss. It knows, for example, if a criminal uses a systematic series of birthdays or addresses on a set of credit card applications at various banks in an attempt to evade fraud detection. The ID Analytics tool has enough data that it can generally tell the difference between honest typographical errors and systematic fraud attempts. 

ID Analytics ran sex offender data through its massive database of credit-related events, and found evidence of rampant identity manipulation among the offenders.

Kristin Helm, a spokeswoman for Tennessee's sex registry, challenged the study's findings, saying that fewer than 1 percent of that state's sex offenders are absconders. Criminals have always used false identities to try to evade police, but law enforcement systems are geared to handle that issue, she said. "Fingerprints obtained by law enforcement identify individuals regardless of a name or Social Security number," she said, adding that names sometimes change for legitimate reasons, too, such as marriage. 

But Stephen Coggeshall, chief technology officer for ID Analytics, said his technology is well-versed in screening out mundane reasons for identity changes and finding patterns that specifically indicate active evasion is taking place.

"This goes way beyond typos," he said. "These are people who have slightly adjusted or substantially adjusted their personally identifiable information for a reason. They are actively doing so, and we are observing them use these aliases relatively recently."

Nevada spokeswoman Julie Butler also questioned the validity of the study, which she had not seen. She said that Nevada uses fingerprints to track sex offenders, so identity manipulation techniques would be ineffective.

"Our registry is fingerprint-based. We don't base it on date of birth, or Social Security number, or name," Butler said. "They can put down their name as whatever and we still have them in the database."

But Coggeshall responded that even in states which use fingerprint identification, an identity manipulator would only be discovered when trying to engage in an activity – such as becoming an elementary school teacher – which triggers a fingerprint evaluation. 

"In general it doesn't help you track where they are or if they're living under an alias at an unregistered location," he said. "It can help to find sex offenders as they enroll in certain groups, but many … groups don't routinely fingerprint new enrollees."

SSNs connected to multiple people
Two years ago, using this tool on a database of Social Security numbers, ID Analytics found that rampant evidence of identity theft: 5 million SSNs were connected to three or more U.S. adults in credit applications, and 140,000 were associated with five or more people, indicating almost certain fraud. The tool can also track individual identity manipulators, as ID Analytics calls them, as they attempt various frauds across an array of credit issuers.

This tool was turned on the sex offender registry problem at the invitation of Utica College in Utica, N.Y., beginning last year. ID Analytics took a large sample -- nearly 100,000 -- of the 570,000 active registered sex offender records and ran them through its credit application database, looking for signs of manipulation.

The findings were disturbing. In Louisiana, the study found, nearly two-thirds of offenders' records showed signs of manipulation. Rebovich theorized that Louisiana's problem might stem from the aftermath of Hurricane Katrina, which gave some people a golden opportunity to drop off the grid.

Officials in Louisiana did not immediately respond to requests for comment.

In many cases, the study found, the steps criminals take are subtle -- changing an address from "440 Monroeville Road" to "434 Monroeville Road," for example. In fact, in the majority of cases, digital absconders were much more likely to move across town than across the country. Absconders who fake their address are six times more likely to remain in the same state than to cross state lines, the study found, and 90 percent of those who remain in state stay within 40 miles of their original registered address. In many cases, the data shows, those addresses belong to a family member. That might allow absconders to show up on a moment's notice at their registered address in case local police do a random check, Rebovich said.

But the address change could also allow them to apply for jobs and housing they would otherwise be unable to qualify for, he said.

While half of the manipulations involve bad addresses, plenty of other types of evasion are going on, the study found. One subject studied had five names, three Social Security numbers and four dates of birth, for example.

About 10,000 offenders had used at least four different Social Security numbers, Rebovich said. The evidence indicates this was usually done to evade the court registration requirements rather than commit financial identity theft, the study found.

One reason sex offenders seem to get away with evasion is that registration requirements are set by states and vary widely. In some states, convicts merely send updates through the U.S. mail to state officials, and are subjected to little, if any, verification. In others, officers try to check on sex offenders, but ofter are assigned hundreds, or even thousands of offenders, to track.

In other states, such as Florida, there are strict requirements and frequent random inspections, Rebovich said. That shows up in the data -- Florida's digital absconder rate is about half the national average, at 9.4 percent.

The study was funded by the Justice Department's Bureau of Justice Assistance, which plans on issuing a comprehensive report later this fall. Requests for comment from the Department of Justice went unanswered.

'System is never going to be perfect'
Shehan, of the Center for Missing and Exploited Children, said she didn't believe that the potentially high rate of digital absconders means the entire sex offender registry program is broken. In fact, she said the situation has improved since passage of the Child Safety and Protection Act of 2006, which instituted some national standards on offender registries.

Still, she said it's important that states move to biometric identifiers, such as fingerprints, to maintain more accurate records of offenders and their whereabouts.

"Criminals are constantly thinking of ways to beat the system," she said. "The system is never going to be perfect."

Rebovich is hoping the study will spur new methods for checking up on sex offenders, including techniques that would seem familiar to those who work in financial fraud. In a model developed by Utica and ID Analytics, offenders could be given a score, similar to a credit score, which would rate the likelihood that identity manipulation was occurring. 

"We are trying to develop a predictive model," he said. "So we can turn it into an alert system, so states can do this in real time, if they want to."  

Coggeshall said such an alert system would have helped police track down Frank Kuni before he was able to get a job with the Census Bureau.

"In retrospect, we know there are things we would have been able to observe" he said.

http://on.msnbc.com/topnewsemailsignup">Click here to sign up to receive our Top News email each day.

 *Follow Bob Sullivan on Facebook.

*Follow Bob Sullivan on Twitter.