Our analysis is conducted at the national level and only for the year 2019 in order to focus the scope. There could be limitations to the employment and salary estimates, or even skills needed for jobs, depending on which location a veteran may be looking for a job. Our analysis is also limited by the scope of the ONET crosswalk. The ONET Crosswalk is based on survey data on knowledge, skills, and abilities (KSAs) acquired in the army and matched to civilian jobs, but in further review using Burning Glass data has created skill vectors that include unlikely or strange skills that MOS's might not provide veterans. For example, in the healthcare industry example the skills acquired in matching jobs to matched SOC codes provided that Army officers may know about cement masonry, a skill outside the scope of the Army's needs. The crosswalk also used MOS-specific KSAs to match to unique SOC codes, but from our analysis of unique skills we can see that most highly employable or frequent skills aren't unique to MOS's. Thus, the crosswalk has produced skills and job connections that are unhelpful. This should serve as a warning sign that crosswalks may not often be as informed as is necessary.
The skills that we have parsed and attached to MOS's implies that there are large similarities in skills acquired across each MOS. In general, Army officers are well prepared with most baseline skills no matter their station. There is, of course, work to be done as this is not accurate. The BLS provided clasification of Army members by sub-Army industry and as there are distributions across types of jobs we would expect to see different types of skills from different MOS's.
In the future, we would like to expand our work to focus on more complicated methods of analysis. We spent the first half of the summer investigating methods for understanding Manpower Planning within the Army. Our hope is that in the future we can focus on simulation and optimization to measure how Armed Services Vocational Aptitude Battery (ASVAB) scores can translate into placement within the Army and therefore which skills a military member would acquire and then translate into probabilities of employment. Specifically, we could see how scores right below or right above a cut-off for a job would translate into future outcomes for a member of the Army. This regression discontinuity design is common within statistics and economics to measure how small differences can amount to large future gains in employment.
In the field of manpower planning we also would like to measure an Army officer's fit in their job based on the skills required for the job, which we have uncovered during this project, and the skills or test scores an officer has, which is still yet to be discovered. This future project would help the Army match officers by skills and test scores, which is an extension of this current project's goals of connected veterans to skills acquired in their jobs after placement.