WASHINGTON -- Last summer, when dozens of nuclear missile officers allegedly cheated on exams, test scores were among the lowest of the year, according to Air Force records obtained by The Associated Press. That is the opposite of what might be expected if answers were being shared as widely as officials allege.
Were they inept cheaters?
Was there, in fact, no sharing of answers during that period?
Were test questions so difficult that even the cheating by some failed to produce higher-than-usual scores for the group as a whole?
The Air Force isn't saying. It notes that tests are not identical each month, and thus score "variances can be expected."
The facts of the tainted testing are still under investigation by the Air Force Office of Special Investigations. It ranks as the worst such scandal in the history of the intercontinental ballistic missile force and is among a series of security lapses and slip-ups that have plagued the ICBM corps over the past year. The missteps prompted Defense Secretary Chuck Hagel to launch two probes of the entire nuclear force to find root causes for leadership lapses and other problems -- steps Hagel deemed necessary to restore public confidence.
Hagel says he believes the nuclear force remains secure and reliable but says "something is wrong."
The alleged cheating has been described as a symptom of mismanagement by commanders who have given too much weight to monthly test scores in determining which launch officers get promoted. More broadly, it reflects a degree of turmoil inside a force responsible for 450 nuclear-tipped Minuteman 3 missiles that stand launch-ready in underground silos in Montana, North Dakota, Wyoming, Colorado and Nebraska.
The alleged cheating was uncovered in January during an Air Force investigation of illegal drug use. Two officers questioned in that probe happened to be members of the 341st Missile Wing at Malmstrom Air Force Base, Montana, and at least one stands accused of having transmitted test answers to colleagues via text message.
The exam in question, known as a T-1, is given monthly and is meant to test knowledge of classified procedures for targeting and launching the Minuteman 3s, the nation's only land-based nuclear missile. Over the course of a year, the tests cover different segments of a long list of launch tasks.
In addition to these and other written proficiency tests, missile launch officers undergo classroom instruction and routine training on launch simulators; most do 24-hour shifts "on alert" in underground launch control centers about eight times a month with a team of two officers responsible for 10 missiles.
The Air Force has focused its investigation on Malmstrom, where officials say the cheating took place during late summer. Notably, in the months after the cheating allegedly ended, scores at Malmstrom improved dramatically.
Neither of these patterns - relatively weaker scores during the period of alleged cheating, and much improved results later - seems to fit with the scenario described by Air Force officials in January when they announced the cheating investigation.
Brian Weeden, who served on Minuteman 3 crews at Malmstrom in 2000-04, said that while he is not privy to inside information about the investigation, one possible explanation for weaker overall scores in August and September is that the test questions - for cheaters and noncheaters alike - may have been more difficult than usual.
"I saw that happen in my time," he said.
Or, Weeden said, the weaker-than-expected results might reflect a slump in the quality of instruction prior to those tests.
It's not clear what, if any, connection there might be to the fact that the Malmstrom wing failed a nuclear security inspection in August and was successfully re-inspected in October. The August failure was related to a problem with security forces, not the performance of launch officers.
Initially the Air Force said 34 officers assigned to the 341st Missile Wing at Malmstrom were implicated; that later was raised to 92. All have been taken off launch duty, creating a shortage that has been filled in part by temporarily augmenting Malmstrom with 10 launch officers each from ICBM bases in North Dakota and Wyoming.
About 40 of the 92 are alleged to have transmitted or received test answers; the rest are accused of knowing but not reporting it.
Gen. Mark Welsh, the Air Force's top general, told reporters on Jan. 15 that "the indications are that this compromise that we're aware of happened in the August-September timeframe." A spokesman, Lt. Col. Brett Ashworth, said it's not clear whether the cheating was only in August or only in September, or in both months.
Test results obtained by the Associated Press under the Freedom of Information Act suggest a different scenario.
- All three of the squadrons that operate the Minuteman 3 force at Malmstrom had average or weaker-than-usual T-1 test scores in August-September. Of the 44 members of the 10th Missile Squadron tested in August, for example, 79 percent recorded perfect scores. That was about the norm during the spring and summer months of 2013 but well below most other months. In September the squadron had 42 percent perfect scores - the weakest of any month in 2013. Perfect scores are not required; to pass the test an officer needs to get 90 percent correct, meaning he or she could not miss more than three out of 30 questions. Only one failing grade in the Malmstrom wing was recorded out of 2,181 T-1 tests completed during 2013.
- All three squadrons did markedly better on the T-1 test in October, November and December, after the period of alleged cheating. In the 490th Missile Squadron, for example, 47 officers were tested in October and 46, or 98 percent, got perfect scores; 45 of 47 were perfect in November and 47 of 51 were perfect in December.
- In January, the month in which the cheating was announced and the first implicated officers were removed from launch duty, test results declined sharply. The 12th Missile Squadron, for example, had 62 percent perfect scores in January, whereas it had about 90 percent perfect scores in each of the preceding three months.
The AP's review of test data provided by each of the three ICBM bases shows widely varying monthly results in 2013.
Records of the 91st Missile Wing at Minot Air Force Base, North Dakota -- where no reports of cheating have surfaced -- show that of 153 officers who took the T-1 test in June, 30 failed. Just six months earlier, in December 2012, 150 in that unit took the test and none failed. What's more, all 150 of those officers got perfect scores -- not a single incorrect answer.