The Jail Study is invalid. The Gang of Four is about to make another "school bond" mistake that will cost taxpayers millions in wasted dollars | Eastern North Carolina Now

    Publisher's Note: This article originally appeared in the Beaufort Observer.

    If Al Klemm wanted to start a day-care business and took a business plan to the bank to borrow money do you think they would lend him the money if he was using population projections that included people over 20 years of age as potential customers? That is exactly what he is doing with the Jail Study.

    Let the record show that the Beaufort County Board of Commissioners is on notice: The Moseley Study upon which they are basing their decisions to build a new jail is bogus. We have said it before, and we are saying it again, nearly a year later.

    There are a number of troubling issues in the Moseley Study. Many of them are in the technical side. But the part of the study that proffers population projections is the most troubling. Those projections are simply bogus and any objective statistical expert would tell the Gang of Four that, if they got a competent and unbiased assessment.

    The jail population in recent months/years has averaged around 85 prisoners. But the variance is rather significant. The Gang of Four is proposing to build a new jail for nearly twice that many. (Note: we are rounding the numbers here. The accuracy of the Moseley number is so bad that it does not matter whether one is precise or not. They are, as the old saying goes: Off by a country mile.) Here's why

    Our orginal analysis of the Moseley numbers is in the link above. But for simplicity sake  -  so Al Klemm and his cronies might be able to understand it  -  here is an excerpt:

    We would contend that the total population of the county is more or less irrelevant in these computations. But if you are going to use population as an independent variable you would certainly have to compute the correlation between historical changes in the jail population compared to the changes in the total population. He does not offer any such correlation analysis.

    We would suggest that if population is going to be used as an independent variable to predict the number of inmates you would be better to disaggregate the population data. For example, we know that Beaufort County's population is becoming older, on average. And we know that older people end up in jail less often than younger people. Likewise, we know that the male segment of the population is incarcerated much more frequently than the female segment. To assume there are an equal number of males and females is obviously erroneous, but it would make more sense probably to use male population between 18-35 than total population. At least a good study would have compared the correlation with each logical segment of the population if you were going to use population. A valid and reliable study would disaggregate the population segments to achieve a higher correlation. The Rule of Thumb in statistical analysis is that if you are going to use an independent variable to predict an impact on a dependent variable you at least check the correlation between your dependent and independent variables. Davis does not do this in his report. But you don't need to worry about such things if you just ignore correlations (how one variable changes as another variable changes) as this study does.

    It does not make sense to us to even use total population to project inmate populations in the future. Why not simply fit a trend line to the historical inmate population? The answer to that question is obvious. If you expected your total population to vary significantly you would want to correlate total population to inmate population. But Beaufort County's total population is not expected to change significantly enough to be used as the predictive variable for how many inmates we'll have in the future. Given that, it would make more sense to simply project your inmate population in the future based on the historical trends in the actual number of inmates, on an average daily basis and compute the amount a variance around that mean.

    Now let's boil all that down for Mr. Klemm. The jail population is not going to increase by 53% anytime in the foreseeable future. Furthermore, the inmate population over the last three years (used in the Moseley Study) could have been significantly lower had the case load been managed correctly. And a simple assessment of the "supply/demand" for contracted inmate beds shows that the likelihood of being able to contract with other counties and the Feds to make use of the excess beds is very remote. Just look at Pamlico, Hyde and Pitt counties. They can't fill up their beds doing everything imaginable to attract inmates from other counties.

    But the main point we are making here is that the population projections the Gang of Four is using are simply not valid. They are talking about spending millions of taxpayer dollars based on bogus information. Using total county population projections is not legitimate. And it is equally illegitimate to use inmate projections without using the historical trend data of the number of inmates actually incarcerated on a daily basis. And using these variables is not legitimate unless you know the correlation between the two over time (how much a change it took in one variable compared to the other.) Common sense would tell anyone, even the Gang of Four, that the total population (which includes children and old people) is not what drives the inmate census.

    Bottom line: The Moseley Study is just not valid.