Page 68 - JSOM Winter 2018

PID Station lower for users, there were no significant differences in per-
Once the course was completed, the Operator received specific formance between the groups in raw time, total time, and
instructions from the THOR3 staff for the kill HVT station (i.e., PID time. There were also small effect sizes between users and
PID station). At this station, Operators were told to face the in- nonusers in penalty time. However, variance in scores between
structor, with each Operator’s back to the HVT faces, and then groups for raw, total, and penalty times was less in the user
asked if they were provided an HVT face at the start of the stress group when compared with that of the nonuser group (Fig-
shoot (all said yes). The Operator was instructed to turn once ures 2, 3, 4 and 5) and had more homogeneous scores when
the beeper sounded and engage the HVT they were shown at compared with those of nonusers. Hence, there was more con-
the beginning of the stress shoot, with only one shot from their sistency among users when compared with score consistency
secondary weapon system (pistol). Failure to positively identify among the nonuser group. Conversely, there appeared to be
and/or accurately engage the HVT in 10 seconds resulted in a more variation in PID time for users (Table 3).
30-second time penalty, whereas successful engagement resulted
in a 10-second time deduction. Upon completion of the Kill TABLE 2 Comparison by Tactical Human Optimization, Rapid
HVT task, the Operator cleared their weapon systems. Proper Rehabilitation and Reconditioning (THOR3) User Classification
User
clearing procedure was evaluated by an SFAUC instructor. Characteristic (n = 25) a Nonuser a Partial
(n = 39)
2
η
Age, y 31.18 ± 4.57 31.39 ± 5.27 0.00
Data Analysis (n = 38)
The collected data were entered in a computer file suitable for
statistical analysis using Statistical Package for Social Sciences, SOF experience, y 2.61 ± 3.27 4.00 ± 4.61 0.03
(n = 18)
(n = 27)
version 24.0. Prior to electronic transmission to researchers at Raw time, s 489.4 ± 68.8 523.2 ± 102.7 0.03
7
the University of Colorado, all data were rendered anonymous by
the training staff. Each individual Operator was given a unique Total time, s 568.8 ± 74.5 613.0 ± 125.9 0.04
number so confidentiality could be maintained. Only members Positive identification 2.9 ± 1.8 3.1 ± 2.1 0.00
(n = 17)
(n= 19)
time, s
of the THOR3 training staff had access to specific identifiers for
the data analyzed and they followed their standard operating Penalty time, s 81.2 ± 34.17 89.6 ± 39.5 0.01
procedures to maintain the confidentiality of this information. Data are given as mean ± standard deviation unless otherwise indi-
cated. Descriptive statistics for all outcome variables are based on
5,000 bootstrap samples.
A descriptive statistical analysis was conducted to determine a Unless otherwise indicated.
the mean scores and standard deviations for the total sample of
Operators. The data were then subdivided by those that used
services provided by THOR3 (i.e., users [n = 25]) and those
that did not use these services (i.e., nonusers [n = 39]). A series
of independent bootstrapped analyses of variance (ANOVAs)
then was conducted to determine if significant mean score dif- FIGURE 2
ferences existed between users and nonusers on raw and total Comparison
course completion times, PID time, and penalties. of raw time
to completion
between users
Although t tests are the intuitive statistical method to analyze versus nonusers.
these data, SPSS, version 24.0, does not currently provide
calculations of effect size for the use of t tests. The program
does provide calculations of effect size for ANOVA, however,
8
which is an equivalent option and allows for more efficient
analysis of the study data. Bootstrapping is an alternative to
more traditional statistical methods that use the median as a
measure of central tendency (i.e., nonparametric statistics). It
is a robust method of analysis that treats samples as popula-
tions and randomly draws smaller samples from them to cor-
rect for non-Gaussian distributions. Before each subsequent
8
sample is drawn, the previously used samples are returned to FIGURE 3
Comparison
the data set. For this analysis, 5,000 bootstrap samples were of total time
used; hence, the process was done 5,000 times to create a new between users
sampling distribution that meets the assumptions of paramet- versus nonusers.
ric statistical methods (i.e., ANOVA).
All tests were measured at the a priori α of < .05. Partial η effect
2
size calculations were also conducted to determine the magni-
tude of the differences between the mean scores of each group.
The thresholds for small (0.01), medium (0.09), and large (0.25) Discussion
effect sizes used in this study have been described by Cohen. 9
The primary purpose of this study was to determine if sig-
nificant mean score differences existed in SOF performance
Results
capabilities between Operators who participate in the THOR3
Descriptive statistics and mean scores differences are displayed program and those who do not. Although there were no statis-
in Table 2. Although performance time means were slightly tically significant differences between the two groups, a small

66 | JSOM Volume 18, Edition 4 / Winter 2018