Tuesday, July 16, 2013

Sexual Assault

A Navy Times website article (1) discusses aspects of the sexual assault issue confronting the military.  Normally, ComNavOps would not find this type of media-hyped, politically frenzied “news” to be worthy of comment.  In this case, however, the article cites a glaring inconsistency. 

The article states that 3374 cases of sexual assault were reported in 2012 and goes on to state that the military believes the actual number of cases was 26,000.  There are several things wrong with this seemingly simple statement of “facts”.

First, the supposed 26,000 cases is based on extrapolation from surveys rather than actual data.  Extrapolation of data is perilous, at best, and hugely misleading at worst.  Let’s assume, though, for sake of discussion, that the extrapolation was done under the most rigorous conditions of statistical analysis and projection.  Simple arithmetic tells us, then, that 22,626 cases of sexual assault went unreported.  Why would 22,626 military personnel, trained and toughened by boot camp, many, presumably, combat veterans, exemplifying the pinnacle of honesty, integrity, and, most importantly, courage, fail to report a grievous case of criminal misconduct?  Only two reasons come to mind:  fear of retribution of some sort or the belief that the incident did not rise to the level of criminal misconduct.

The second of these reasons for not reporting an incident, the belief that the incident did not rise to the level of criminal misconduct, leads to the second problem with both the reported and unreported cases.  Having seen, and taken, many surveys about matters of this nature and related issues, ComNavOps has noted that the surveys are usually written to promote a desired outcome.  For instance, a sexual assault survey might ask if one has ever seen a pin-up type picture during the course of one’s duties.  If one answered yes, the survey might lump that response into a general category of unwanted sexual atmosphere and, hence, classify it as a sexual assault.  Most of us would agree that there is no comparison between racy pictures and physical assault.  So, the second point is that reporting statistics without defining the criteria for those statistics, is irresponsible at best and willfully misleading at worst.

Regardless, let’s say, again for sake of discussion, that every reported case of assault and every extrapolated case of assault is non-trivial, and of a serious criminal nature.  That leads us back to the question of why 22,626 service members did not report such a criminal incident.  Our other reason for non-reporting was fear of retribution of some sort.  It could be fear of reprisal from the assailant, fear of reprisal from the chain of command, or simple fear of the unpleasantness of the entire investigative and legal process.  While all would be valid fears, fear is not the determiner of one’s actions.  Everyone in combat is afraid and yet all are expected to execute their duties and responsibilities in spite of that fear. 

I have to pose the question, “If fear, in whatever form, is preventing 22,626 people from reporting a serious criminal incident, are these people suited for their chosen profession?”  Will a person who is too afraid to report a criminal incident be brave and steadfast on the battlefield?  Will a person who lacks the courage and fortitude to stand up for themself, stand up for their fellows on the battlefield?  Will a person who can be intimidated in daily life demonstrate the determination, courage, and integrity to function properly in combat?

I am not at all trivializing the trauma of a sexual assault or the difficulty associated with pursuing such an incident to its legal conclusion.  I am, however, pointing out that the harsh reality is that we either have 22,626 service personnel who do not have the courage and intestinal fortitude required for combat and should not be in the military or the military is generating false statistics based on extrapolations of suspect criteria in order to pursue a purely political agenda.  I don’t know which is the case but I don’t see a third alternative. 



18 comments:

  1. Remember Organizations take their cues from their "Leaders". Refer to this article http://www.navytimes.com/article/20080920/NEWS/809200305/Drunkenness-sexual-advances-derail-1-star
    That details how Adm Goddard is a drunken letch. Everyone knew about his behavior and did nothing about it. Are all of the NavSea personnel cowards that can't report bad behavior? Note also that he got hired by LM and now runs the Wis Shipyard that builds the LCS-1 class of ships.

    Take a hard look in the mirror before arguing about numbers. One incident of abuse of power in and organization that is charged with defending our citizens is one too many!

    ReplyDelete
    Replies
    1. Anon, you are right on the money. The characteristics of an organization flow from the top. An excellent observation!

      You lost me, though, with the sentence about looking in the mirror and numbers. What are you suggesting? You're making a point but I've missed it. Want to try restating it?

      Delete
    2. My point is that critizing the numbers gives the appearance of not believing there is a problem. Just one Admiral doing this makes it a problem, a BIG problem and it that is the only thing that should be addressed. The Services have a problem in this area, as do civilian organizations, let's address the problem. I feel one is too many, 2600 for certain is too many. Why talk about the numbers? Talk about finding these low lifes and getting them out!

      Delete
    3. Anon, I have to disagree somewhat. While one assault is too many, the military is no different from the general populace in the sense that there are a certain percentage of bad people (one would hope that the military has a lower percentage, actually) who will commit bad acts. - Side note: I wonder what the percentage of assaults is in the military compared to the general populace? - Short of instituting a total, 24 hour per day police state there is no practical way to totally eliminate such acts. Even if the military were to cease its warfighting responsibilities and totally devote itself to preventing assaults as its only reason for existence, it still couldn't succeed. We're on the verge of another Tailhook overreaction, I suspect.

      The reasons I'm looking at the numbers is because I don't believe they're real. If they are real then we're doing a poor job of selecting and producing combat soldiers as I suggested in the post. I suspect, however, but cannot prove, that the statistics include a lot of things that don't count as assaults by any reasonable definition. I think this is largely political correctness rum amok.

      Delete
  2. A new Commanding General speaking to his staff years ago noted that he had absolute faith that if he ordered an assault on a machine gun nest, he was certain that 90%+ of the officers would execute the attack and display great physical courage in doing so. The general went on to note, that of the same group of officers, he thought that less than 10% would have the moral courage to disagree with him and tell him he was wrong.

    There are a lot of reasons that military forces seem to suffer from this phenomena, but it is a reality.

    As a side note, there is no evidence that officers in the Wehrmacht who refused to execute civilians, or failed to carry the infamous Commissar Order (Kommissarbefehl) requiring the summary execution of Soviet political officers, was ever reprimanded or punished. Sadly, a lot of human behavior is governed by what we think others expect of us, not what is actually required.

    GAB

    ReplyDelete
    Replies
    1. GAB, your point is valid but possibly somewhat misapplied. It's one thing to stand up to a direct order - soldiers are sworn to obey lawful orders with severe consequences for failing to do so. It's a completely different situation with respect to reporting an assault. There is no order to disobey - in fact, service members have been "ordered" to report such incidents and are "disobeying" by not reporting. Assaults are, by definition, criminal acts. Failing to report a criminal act is quite different from refusing to obey a direct order.

      Your final sentence is an astute observation and there is, undoubtedly, an element of "don't rock the boat" at play that influences the failure to report (if, indeed, there really is a widespread failure to report - I'm not at all convinced that there is - I suspect there is not).

      Delete
  3. The article states that 3374 cases of sexual assault were reported in 2012 and goes on to state that the military believes the actual number of cases was 26,000

    You misread the article. There were 3374 cases of sexual assault reported in the military. The 26,000 number is of "unwanted sexual contact" which may or may not rise to the level of sexual assault. So, no, the military is not saying it thinks the actual number of rapes is 26,000.

    That negates much of your analysis. As to this:

    So, the second point is that reporting statistics without defining the criteria for those statistics, is irresponsible at best and willfully misleading at worst

    You could, you know, actually read the survey. It's here:

    http://www.sapr.mil/index.php/research (first link)

    And they have an entire section devoted to their methodology.

    ReplyDelete
    Replies
    1. Total, the underlying survey has this to offer in the way of definition:

      "The 2012 WGRA survey includes a measure of unwanted sexual contact (i.e., sexual assault) originally developed ... For the purposes of the 2012 WGRA survey, the term "unwanted sexual contact" means intentional sexual contact that was against a person's will or which occurred when the person did not or could not consent, ..."

      Thus, we see that the report lumps all forms of unwanted sexual contact together under the generic term "sexual assault". You might want to read the survey, yourself. I see nothing that breaks down the 3374 reported cases by type of unwanted contact. Hence, by the report's definition, all 26,000 extrapolated cases are sexual assaults of an equally severe and criminal nature. My analysis stands.

      Regarding methodology, the report states that around 24% of the sample population responded. Is it hard to imagine that the liklihood of response increases proportional to bad experiences or other strong motivating factors? A reasonable, though unproven, assumption would be that the 76% who did not respond had no bad experiences. Thus, 6.1% of women WHO RESPONDED and claimed to experience unwanted sexual contact represent only around 1.3% of the total sample population. Quite a difference! Of course, who knows whether the claim rate would have stayed the same or not had every member of the sample population responded. The point is that the methodology was biased towards a higher than actual number of claims, in all probability. It's akin to surveying shoppers for satisfaction. The only ones who respond tend to be those who are very displeased or very happy. The vast majority are ambivalent and don't respond.

      Delete
  4. The 26,000 figure has no relationship to the 3374 figure. The 26,0000 figure was extrapolated from the percentages of those responding to the survey who claimed unwanted sexual contact (6.1% for women, 1.2% for men) to the total number of people in the military. The report itself, in fact, never mentions the 3374 cases you claim. The "unwanted sexual contact" includes, but is not limited to sexual assault, as the report specifically mentions.

    Is it hard to imagine that the liklihood of response increases proportional to bad experiences or other strong motivating factors?

    No, it's not hard to imagine at all, WHICH IS WHY THE REPORT AUTHORS USED A WEIGHTED SAMPLE OF THE RESPONDENTS, not the entirety. Shockingly, statisticians actually think about these problems ahead of time.

    Your analysis is wrong.

    ReplyDelete
    Replies
    1. Actually, there are a number of problems with the survey, the principal one being selection bias.

      What stands out to me at first blush.

      - There was no attempt to get a representative sample of the military population.

      - The response rate on this survey is appalling low at 24% - Did the non-respondents not submit responses because they had no issues and wanted to spend their time on the beach instead of filling out polls; or were they in absolute living fear of retribution for reporting on abusive behavior?

      - There is no way to gage the veracity of responses - there are people who grind axes, and others who deny even the worst abuse.

      - There is no attempt to correlate responses with actual criminal statistics (UCMJ proceedings, NJP proceedings, Civilian courts, JAGMAN investigations, medical reports, etc.).

      In short, with no representative sample, and a low response rate, there is much concern for the validity of extrapolating the results to the general population.

      GAB

      Delete
    2. The 3374 reported cases comes straight from the article that I cited. Follow the link and you'll see it.

      Delete
  5. - There was no attempt to get a representative sample of the military population.

    Oh, Good Lord. You don't know anything about how surveys are done, do you? They sent the questions out to 100,000+ people and then built a representative sample out of this. That's standard survey methodology and as straightforward as adding 2+2 to get 4.

    The response rate on this survey is appalling low at 24%

    24% is actually remarkably high for a survey. Political surveys are luckily to break 10% these days.

    So, still striking out in your criticisms. Want to try with something more convincing?

    ReplyDelete
    Replies
    1. Everyone, feel free to discuss and disagree but let's keep it polite and respectful.

      Delete
    2. Total,

      I have a MS in applied mathematics (Ops Research), and have worked with Pew, Gallup, BRAC, IOM, and others in developing, implementing, and analyzing professional polls and focus groups. BTW, I am a retired officer who actually filled out previous iterations of the survey in question. I note that I have no financial interest in this study.

      I am well aware that there is an appalling number of poorly implemented polls, and slack methodology: the infamous Lancet Study in Iraq that falsely predicted almost half a million Iraq civilian deaths and timed to be released right before the U.S. presidential elections. I believe that this is the case with this poll.

      You have yet to address my principal concern: there is no attempt to correlate responses with actual criminal statistics (UCMJ proceedings, NJP proceedings, civilian courts data, JAGMAN investigations, medical reports, etc.) with the sexual assault data from the poll. I note that an actual criminal conviction sets a much higher standard than a poll; even so, DNA evidence has exonerated a fantastically large number of people convicted of rape/sexual assault. It is one thing to claim rape, or assault, it is another for that to be fact. Not comparing actual crime statistics with observed survey results makes for flawed research. People keep confused poll results with the actual sexual assault metrics as if they were the same: they are not.

      I also harp on the issue of selection bias – which you have not addressed. The results of the poll are not disaggregated by demographics (aka the population pyramid), but by rank and gender and service. While rank generally correlates with age, it is not always so. You cannot extrapolate results to the general population without pulling crosstabs and weighting data accordingly. For example, how did the racial makeup of respondents correspond with the active duty population? What about ethnicity and religion. This should all be clearly disaggregated in the appendices: but that data is not there. That is sloppy work. One would expect a much more robust description of the methodology and an actual copy of the survey questions to be included in an appendix.

      Frankly, this poll is targeted at military personnel and civil servants, I believe that the response rate should be much higher than 50% based on experiences with other military specific polls (command surveys, etc.). Troops are generally quite candid. I further note two points regarding non response in this survey: 1) there was no attempt to capture non-contact, and refusal rates to help to characterize the survey, and 2) no subjective discussion of the response rate in said poll.

      I await your response...

      GAB

      Delete
  6. By the way, as best I can determine the military's rate of sexual assault appears to be far less than the civilian population. I say "appears" because it's very difficult to determine what acts are included in the statistics, both civilian and military, so comparing apples to apples is difficult. If this is true, the military is expending a lot of time and energy on a problem that is not very significant compared to the civilian world. And, yes, I understand that it's quite significant to the person being assaulted.

    ReplyDelete
    Replies
    1. COMNAVOPS,

      You are correct: the survey shows that the rate of sexual assault in the military is be far less than the civilian population.

      The caveat to remember is that the military is an exclusive organization: ~75% of youths do not qualify for the military due to a history of violent crime, drug abuse, psychological disorder, or medical problems. Had the researchers actually done their work, you would see some interesting disparities comparing the rates of assault between military and civilian demographics.

      The UCMJ includes a large number of offenses bundled under the sexual assault umbrella that might generally fall under a different civilian statute. There is also the issue of how charges are brought, and the disparity between plea bargaining. Also, the bar for non-judicial punishment (article 15) is much lower in the military. Military Commanding Officers can punish a number of offenses that a prosecutor would not waste their time with, by using NJP.

      GAB

      Delete
    2. GAB, good comments. One of the points I made was that I'm highly dubious about the extent of the problem and I strongly suspect that it's mainly an issue of political correctness and public relations run amok. It very much is looking like a repeat of Tailhook, in that sense.

      You make a good point about the exclusivity of the military population. With that in mind, I wonder if the lower incident rate compared to the general population is still a "good" thing or is it excessive relative to the higher quality of people involved?

      Delete
    3. "You make a good point about the exclusivity of the military population. With that in mind, I wonder if the lower incident rate compared to the general population is still a "good" thing or is it excessive relative to the higher quality of people involved?"
      ================================================================

      You are addressing another reason why this survey falls hugely short - we do not know how the two populations compare broken out across demographics. I suspect it is not a “good thing.”

      GAB

      Delete