analysis of manual threat evaluatiok sand … research was to evaluate human operator performance...
Post on 05-May-2018
213 Views
Preview:
TRANSCRIPT
Technical Report 419 L
ANALYSIS OF MANUAL THREAT EVALUATIOkSAND WEAPONS ASSIGNMENT (TEWA) IN THEC AN/TSQ-73 AIR DEFENSE SYSTEM
0 Charles C. Jorgenien and Michael K. Strub
ARI FIELD UNIT AT FORT BLISS, TEXAS
SS-P2
U. S. Army
Research Institute for the Behavioral and Social Sciences
~ j October 1979
, Approved for public rmleese; distrlbulon unlimited.
_ 80 9 25 064ý4• : . . . .. , , , .
U. S. ARMY RESEARCH INSTITUTEFOR THE BEHAVIORAL AND SOCIAL SCIENCES
A Field Operating Agency under the Juri-ýdiction of the
Deputy Chief of Staff for Personnel
~ I FRANKLIN A. HARTJOSEPH ZEIDNER Colonel, US ArmyTech niCL Director Commander
.................... . ........
[ NOTICES
DISTRI BUTi ON: Primary distributior. of this report him been made by ARI. Please address correspondenceconcernin% distribution of reports to: U. S. Army Research Institute for the Behavioral and Social Sciences,ATTN: P-:RI-TP 5001 Eisenhower Awint.., Alexandria, Virginia 22333.
F INAL DISPOSITION: This report may be destroy'ed when it is no longer needed. Please do not return it tothe U. S. Army Research institute for the Behavioral and Social Sciences.
PjQT The findings in !his report are not to be construed as en official Department of the Army position,
unless so designated by other authorized documents.
~,h.
UnclassifiedSECURITY CLASSIFICATION OF THIS PAGE (WFMan De. Bnt*er0
REPORT DOCUMENTATION PAGE BEFORE C TO.
7. REPORT NUMBER I MGOVT ACCESSION NO. S. RECIPIENT'S CATALOG NUMRERTechnical Report 419 -/ )"'
NALYSIS OF-MANUAL D REAT IDVALUATION ANDr,•. /-APONS-.SSMNIMENT IZEW IN THE A•NITS73 Z To r"- • .R F'.p],
WiR DEFEMRE SYSTEM . • . ... r-.'"-=- , _"•..""•* • '""" • "'lm 1. CONTRACT ORGRANT NUMmEP00'o)
S. PErRFORMING O11GANIzATION NAME AND ADDRE[SS 10. PROGRAM E[LEME[NT.r PROJE[CT. TS4KU.S. Army Research Institute for the Behavioral A -A"• UNO IUm[
5001 Eisenhower Avenue, Alexandria, VA 22,f33 XQ; ~~~11. CONTRLKI.NG OFFICE N AME[ AND ADDRESS$ 131'• ' '"9. .....
7O 1/979~R
U.S. Amy Air Defense Schoo • •/ C _97/ "
Fort Bliss, TX 79916 ••:;•-"'"'e--23
4MONITORING AGENCYE NAME AORF.S(II differen from ContrainO Olf) 15. SECURITY CI.ASS. (o fl thisrepot)
"- 0• • ,•::I--T "-/-:I-i /Unclassified
' ~~1S*. DECL.ASSI FI CATION/ DOWN GRADINGSCHEDrULE -
IS. DISTRIBUTION STATYSENT (of this Report)
Approved for public release; distribution unlimited.
17. PISTRIBUTION STATEMINT (of the, absrct AnD tred In Block 20, PO differet from Report)
IAR. SUPPL.OM.NTAIY NOTTS
r .IUS. K AY WORmS (Continue fn tovorso sid o It ncessary ad Identify by block number)
U.S. Air defense systemsFot lisHuman factors
i')l• Mnull weapons assignmentM.DITRI 1UI threat evaluation
20""J(. A" ACT renhw as ara eneswebb N noem aind 1l 6hl7 by loc* nueebni,)
As tactical ai.r threats have increased, so has the need to efficientlycoordinate ground defense. With the advent of computer aided command .Andcontrol systems, the identification of human strengths and limitations has
become critical to the success of the Army air defense mission.
7 SThe research reported in this paper was conducted in 2075 during thef
" ~~early development of the AN/TSQ-73 missile minder. The missile minder is \S, , I(Continue]) ,
L% I " FORM 3 am -nO nO F ,I o v e ,sisO mW .•E-'T U n c la s s if ie dg SECURITY CLAYO CiCATRODR OF THiIS PAGE (libra. Dat Ennered)
Huma fator
UnclassifiedSICUMITY CL.ASSIFICATION OF THIS PAGE(I~ Deita Bnrm*
Item 20.(Continued)
representative of a class cf emerging air defense systems which place unusualdemands on the processing capabilities of the human operator. The purpose of
the research was to evaluate human operator performance under realistic taskloading, aircraft threats, and manning configurationsi. As a result of thisstudy, procedures were developed which can effectively be applied to assessoperator performance ,"r.dr a wide variety of emerging air defense systems.The procedure can also be utilized to aid in firing doctrine development,assist human factors specification, and improve interoperability decisionsin linked air defense systems.
I
.I
Unclassified
JSjECUNITY CLASSIFICATION OF THIS PAGE(Wten Date Bnte-d)
I .. L.¶a . . .. " .
Tinhclnial Report 419
ANAYLSIS OF MANUAL THREAT EVALUATIONAND WEAPONS ASSIGNMENT (TEWA) IN THE
AN/TSQ-73 AIR DEFENSE SYSTEM
Chaeles C. Jorgensen and Michael H. Strub
Submitted by:Michael H. Strub, Chief
ARI FIELD UNIT AT FORT BLPSS, TEXAS
Approved by:
E. Ralph DueekPt.RSONNEL AND THAININGCESEARCH LABORATORY
U.S. ARMY RESEARCH INSTITUTE FOR THE BEHAVIORAL AND SOCIAL SCIENCES5001 Eisenhower Avenue, Alexandria, Virginia 22333
Office, Deputy Chief of Staff for PersonnelI' Department of the Army
October 1979
Army Project Number Human Factors In2Q762722A?65 Air Defense Systems
Approved for public release; distribution unlimited.t ! iii
ARI Research Reports and Technical Reports are intended for sponsors ofR&D tasks and for other research and military agencies. Any findings readyfor implementation at the time of publicationare presented in the last partof the Brief. Upon completion of a major phase of the task, formal recom-mendations for official action normally are conveyed to appropriate militaryagencies by briefing or Disposition Form.
iv
'A
FOREWORD
The Army Research Institute Field Unit at Fort iliss is actively engagedin a research program designed to improve the evaluation of human per-formance in air defense 3ystems. As ta:tic..fl air threats have increased,so has the need to efficiently coordinate gro~und defense. With the ad-vent of computer-aided command and control systems, the identificationof human stv'engths and limitations has become critical to the success ofthe Army air defense mission.
The research reported in this paper was conducted in 1975 during theearly development of the AN/TSQ-73 missile minder. The missile minderrepresents a class of emerging air defense systems which place unusualdemands on the processing capabilities of the human operator. The pur-pose of the research was to evaluate human operator performance underrealistic task loading, aircraft threats, and manning configurations.As a result of this study, procedures were developed which can effec-tively be applied to assess operator performance under a wide varietyof emerging air defense systems. The procedure can also be utilized tohelp develop firing doctrine, assist human factors specification, andimprove interoperability decisions in linked air defense systems. Thisresearch is in response to requirements of Amy project 2Q762722A765and special needs of the U.S. Army Air Defense School (USAADS), FortBliss, Tex.
EPH ZNER
hnical Director
Ii
II
!, V
WqII 1
Analysis of Manual Threat Evaluation and Weapons Assig ient in theAN/TSQ-73 Air Defense System
BRIEF
Requirement:
To investigate human performance under varying air defense threats.
To determine human overload points requiring transition from manual toautomatic operational modes.
Procedure:
Air defense crews were evaluated over a six-day period under changingconditions of crew configuration, number of available fire units, andattacking aircraft threat. Switch action times were reccrded by theAN/TSQ-73 computer and matched against event times in predeterminedthreat scenarios. Analysis of operator actions utilized switchactuation latencies and operator errors (identified via incorrectswitch sequences). Results were subjected to advanced multiple regress-ion analysis modeling in order to generate predictive formulas foroperator performance under load.
Findings:
Hostile aircraft engagement latency averaged 3,47 minutes for systemcontact, identification, and weapon assignment during the simulatedcombat exercise. The chances were at least 50 percent that the air-craft would complete its run if it was in - major assault wave. Amanning configuration in which the officer implement, decisions directlythrough the weapon assignment console significantly improves assignmenttimes. Wave size of aircraft had the greatest effect on the weaponassignment operator although the aircraft identification operator timesincreased linearly as a function of the number of aircraft. Manualweapon assignment operators tended to overload at eight Pircraft andreaction times increased rapidly. Greater numbers of fire units improvedweapon assignment times but only if eight or more were available.Operator fatigue may be a siqnificant factor in weapon assignment underheavy task demands.
Utilization:
The techniques developed in this report can be applied in any air defensesystem having built in microprocessors that can record operator switchactions, such as the developing PATRIOT system. The resulting improvedestimates of operator actions and re~ction times can be applied tofiring doctrine evaluation, system software development, the man/machineinterface,-and interoperability studies involved in linking togethercommand and control systems.
vii.Preceding Page Blank
TABLE OF CONTENTS
I Introduction.......... ................................................. .1
II Procedure .. .. ....... . ........ .. ... . .. ...... .. ....... 3
III Results and Discussion. .. .. ........ .. ... . ....... . .....
IV Conclusions........ ...... . .... .. . .. .. .. .. .. .. .. ... 14
Appendix. .. ...... .. ... ........ ... ... .................................. 21
Distribution .. . . . . . . . . . . . . . . 25
LIST OF TABLES AND FIGURES
fable Accuracy Data (days 5 and 6).......................... 8
Figure 1 Experimental Design .................................. 5
Figure 2 Mean R'-action1 Times for ID as o Functionof Six Wave Sizaes ................................... 15
Figure 3 Mean Re~action Times~ for WA as a Functionof Six Wave Sizes................................... 16
I 4Figure 4 Mean Reaction Times and Estimated ReactionTimes for Total Reaction to a HostileAircraft as a Function of Wave Size................... 17
Figure 5 Best Fit Estimated ID Reaction Times asY.a Function of Day of Test............................ 18
Figure 6 Bist Fit Estimate of WA Reaction TimesasdFunction of Day of Test..................... 19
ix
I. INTRODUCTION
The Army is fielding a new air defense system for command and
control of Hawk and Hercules Fi,'• Units. Called the AN/TSQ-73 (Missile
Minder), the system contains software which will automatically perform
threat evaluations and weapon assignment (TEWA). The chief advantage
of the automatic TEWA is its speed in assigning hostile aircraft to
available fire units. Given sufficient time, it is likely that the
tactical director, by virtue of being able to take into account complex
information and recognize event patterns, would make more "intelligent"
TEWA than the machine since the machine TEWA is software limited.
The tactical director performing manual TEWA is subject to increased
information loading and stress as the number of aircraft in an attacking
wave increases. The present research sought to investigate human TEWA as
a function of wave size to assist in the determination cf a manual satura-
tion point. Such a point would suggest the level at which the system
should transition from manual/semi-automatic control to automatic control
as the intensity of a battle increases and the number of radar tasks
become too numerous for manual weapons assignment. While some argue for
exclusive manual/semi-automatic mode operation and others for an alli1,"
automatic mode, it is more likely that the system will reside in the manual/
semi-automatic mode until the operators and tactical directors are over-
loaded by the sheer volume of tracks requiring attention. At this point,
the operators and tactical directors would transfer the system to the
automatic mode but still maintain constant monitoring and override
capabilities.
'if
The purpose of this experiment was to collect baseline data from
which overload points could be determined. It was judged that such
points would be a function of the number of hostile aircraft in an
attacking wave or scenario, the number of fire units under direct control
of the Q-73, and the rate at which the attacking wave approaches. The
first two variables were systematically changed in the experiment, while
the third was fixed. Also, a key conmmand and control question concerned
whether the tactical director could coordinate activities better from behind
the console operators or at the weapons assignment console where he could
implement his fire unit assignment decisions directly through the console.
Thus, the third primary variable in the experiment was the manning
configuration. In one configuration the officer was behind the console
op~eraitors, while in the second configuration the tactical director-sat at
the weapons assignment console. Also included for control purposes were
three minor variables. They were the specific raid scenario, the kr*est
day, and the crew.
2
11. PROCEDURE
Personnel participating in the experiment (crews) consisted of
three Air Defense officers and six enlisted men from the US Arm~y Air
Defense School (USAADS) recently detailed to Ft MacArthur. CA, in
connection with the TACS/1ADS effort in 1975. (Pilot data were collected
from an "expert" team of one officer and two enlisted men who had been
assigned to the TACS/TADS project for over a year.) While the crews did
not receive a formal training program on the AN/TSQ-73, their arrival at
the testing site approximately three weeks prior to the start of the
experiment allowed sufficient time for them to become familiar with the
functions and operations which would be required of them in the experiment.
Although some practice effects were expected to be found over the six
experimental sessions, all crews were Judged to be qualified for operating
in the Q-73 at the start of the experiment.
The testing environment consisted of the interior of a prototype
Q-73 van. The principal items of equipment were two operator consoles,
identical in configuration, which were placed side by side at the front
of the van. The console on the left side was used for track identifica-
tions and the right console for weapon assignments. Other equipment
included cc'nuunication lines for cross talk between the simulated Group
73 and Battalion 73, and between the Battalion 73 and the fire units.
An officer from USAADS served as Group 73 and issued directives to the
Battalion 73 as required. Four experimenters rotated in the role of
simulating "Squawk" from the fire units, e.g., acknowledgment of assign-
ment, etc.
3
Data collection occurred during six consecutive evenings between
2200 and 0600 as shown at the bottom of Figure 1. Data was collected
for the expert team on the morning of the seventh day between 0600 and
0800. Each team served for a two-hour period each evening. Each period
consisted of two one-hour sessions. (Simulation equipment constraints
precluded simulated raid tapes from being over one hour in length.)
USAADS prepared scenarios from which twe raids were scripted and two
raid tapes were developed for use in the Q-73's simulation mode of opera-
tion. To assure comparable group data, all video viewed by operators on
their radar screens was generated from the tapes rather than from live
radar signals. The simulation package also included the capability for
periodic recordings of the status of all actions taken by the operators
and the times of these actions. In addition, the system periodically
recorded the status of all fire units under control of the Battalion Q-73.
During each one-hour session, a team would be exposed to three different
assaults or waves of varying size. As soon as an aircraft track entered
the system and three simulated radar sweeps verified its presence, the
track became available to the ID operatnr seated at the left console. All
tracks entered with an unknown identity. Simulated messages into the Q-73
resulted in change of status symbols replacing unknown symbols on the
operator's screen. When the ID operator observed this special symbol, he
"hooked" the track by placing his cursor over the syinbol and pressing the
appropriate function switches to reveal whether the track was friendly or
hostile.
14
.1's•1
~~~~~~ . ... j . .. ., . . . .
N
2 34 4 !4
NoV e
es. YYes
8 8 182 Yes8 ' 8 ii8
Yes Yo No
3 3 1Ye No Yes./IO
3 2
.1L
4: '-I4
No !a - ; Yes. . F-.. o
When a track symbol changed to hostile, it became the duty of the
weapons assignmen~t officer to select a fire unit to engage the track.
The Q-73 system has limited software to provide automatic assignment of
fire units, but it was the purpose of the presevit experiment to ass.,ss
human performance capabilities in the manual weapon assignment mode.
For each two-hour session, there were either four, six, or eight
fire units under control of the Q-73 (see Figure 1). The manning config-
uration consisted of having the tactical director behind the console
operators or seated at the weapons assignment console. Thus, each
combination of fire unit number and manning configuration resulted in
six different conditions constituting the six experimental se-sions.
I6
€
6
'I _____
I ~¶r - ~ - -
IF ___________________________
III. RESULTS AND DISCUSSIONS
The analyzed data consisted of 383 sets of reaction times corres-
ponding to complete actions taken by AN/TSQ-73 operators on hostile
aircraft during days 1 through 6. Thirty-three additional sets of times
were obtained from the expert group and were used in non-parametric
analysis of operator training and familiarization. Each set was composed
of three values for a given hostile aircraft: (1) the total time to identify
an aircraft and assign a firing unit; (2) the proportion of the time required
by the ID operator; and (3) the proportion of the time required by the
weapons assignment operator. As an estimate of terminal system effective-
ness, percentages of successfully engaged aircraft were tabidlated for
days 5 and 6. A breakdown of these percentages into detailed categories
of actions, e.g., number of friendly aircraft incorrectly engaged, was
precluded due to extraneoL factors such as uncontrolled instructions to
the ope~rators, system problems, and errors on the radar track tapes. It
was still possible, however, to estimate overall performance against
7 hostile aircraft. These percentages are reported in Table 1.
This -fable shows the percentage effectiveness of the operators. It
-is based on ratio of the average number of completed identification/
;~ weapon-assignment sequences to the average number of completed identifica-tions in a wave. Effectiveness remains at about 70% for low- and medium-
I~ size waves. It drops to about 49% for high-wave sizes. In this experiment
Shigh-wave size was 25 or more aircraft. It is possible that these
estimates may be conservatively low. Effectively completed sequences
7
may have been restricted by uncontrolled instructions issued during the
experiment to reduce extraneous noise on the operators' screens, such as
bogus track identifiers created by rapidly turning aircraft.
A detailed investigation of operator reaction times added insight into
the variables influencing overall effectiveness. By examining operators'
switch actions on the AN/TSQ-73 console, it was possible to accurately
determine ID and weapon assignment times. These times were analyzedin relation to the experimental variables of manning configuration,
number of fire units, number of aircraft in an attacking wave, training
group, day of test, and raid-tape scenarios. To maximally explore the
relationships ri-esent in the data, times were subjected to multiple linear
regression analysis using an IBM 360-65 computer. The data were tested
against 90 different models corresponding to plausible psychological
outcomes. Multiple F-tests were then used to choose the most parsimonious
models with the greatest degree of explained variance.1
F The following conclusions were supported by reaction time data. The
average total time to identify and successfully engage a hostile aircraft
was 208 seconds. (Start thime was estimated at about 24 seconds after the
system first contacted the radar track.) The average time to correctly
It should be noted that the degrees of freedom used in these Fstatistics represent model comparisons and are not always directly relatedto the number of experimental variables as is the case in the usual ANOVATable. The reader is referred to the appendix for reference detailingthis procedure.
r 9
identify a friendly aircraft was 57 seconds. The iD and weapons assign-
ment (WA) operators did not always respond in the same manner to the
experimental variables. Manning configuration had a significant effect on
UA reaction times (F(l, 31 8 )'9.49 P'ý.01). When the officer implemented
decisions through the console his mean reaction time was 91.6 seconds
versus 131.4 seconds when he served as tactical director. This wds an
average improvement of 39.8 seconds, or about 30%.
Number of fire units also effected WA time (F2 ,380)=4.14 P<.05).
Reaction times were approximately equal for f.our or six fire units
(X4=ll6 seconds versus X6=114 seconds). Eight fire units reduced the
average WA times to Y8=69 seconds. Caution should be used in interpreting
this result however. The percentage of explained variation due to fire
units was low 012=.0213 where M2 is squared multiple correlation
coefficient of the fire unit variable) which means that although the
effect was statistically significant, number of fire units contributed
little to overall prediction of reaction time. Increased numbers of
fire units may improve overall effectiveness as well, but due to a lack of
usable performance data, this question could not be answered.
The most powerful predictive variable was wave size. The most
elficient predictive model was a linear function of ID (F( 1 ,38 1)h7.22 Pc.O01)
4 ).and a cubic function of WA (F2 , 3 80 )=8.06 Pv.001). Least squares fitting
of alternate models show ID is best fit by the linear function:
RTID60. 3 + 2.05W
where W is the numberof aircraft in the attpacking wave. Actual mean
.1 10
A
'4'P W
-. OMI II9
values for ID times are plotted in Figure 2 as a function of wave size.
For WA the linear trend seen in ID times was not evident, F:gure 3 shows
mean reaction times already rising rapidly at eight aircraft, suggesting
that the WA operator was already overloaded. Latencies reach a peak
of 176 seconds per aircraft at 16 aircraft and recede to a steady state
level of 100 seconds thereafter. Though dropping from the highest value,
terminal WA times are still well over a minute longer per aircraft than
at a wave size of eight. This decrease coincides with a loss in overall
effectiveness and may be due to the forced adoption of faster but less
optimal WA strategies for assigning hostilc targets to the most appropriate
fire units. Comparisons between possible wave models-showed the WA data
was best fit by a cubic equation:RTwA= 3.23w-.012w3 -29.37
The cubic fit was a significantly better predictor than either a linear
or a second order equation (F(1 ,380)=4.29P<.05).
An overall best equation for estimating total reaction time is the
sum of the optimal ID and WA equations:
RTTOT= 30. 9 3+15 .2 8w-.O1 2w3.
The mean RTs for the total data are shown in Figure 4 along with the
predictive curve generated by the above formula.
IAnalysis of the variables of days and raid tapes generated new
research questions. Times from two raid tapes showed no difference for
ID (F( 1 ,381 )=2.18 PY.20) but a significant difference for
WA (F(I,3I9.05 P.05). This could imply that the WA task is
' jI
more sensitive to the attacht configuration than -the ID task, since
the raid tapes generated different numbers of artificial tracks on the
radar screens depending on the aircraft maneuvers utilized in the raid.
Reaction time data show a significant effect over days on ID and WA
times (FID(5,377)= 2 .59 P<.25, FWA(5,3 7 7)=5. 6 4 P<.OOl). Tinves over
days were best fit by simple linear function3 which are plotted in Figures
5 and 6.
RTID=14 9 .6 6 -1 3 .14d
RTwA= 3 2 .97+19.87d
where d is the day's number d c {l, 2, 3, 4, 5, 6}. The most interesting
fact about these data is that ID and WA functions had opposite slopes.
The negative slope of the ID function means RT decreased over days for the
ID operator. The positive slope for WA showed just the opposite effect.
The interpretation of this finding is difficult. One possible explanation
could be that the WA task is more fatiguing to the operator and increased
latency due to fatigue obscured normal improvement resulting from practice
effects. If this is the case, it implies that WA operator fatigue during
combat could markedly increase response times. To more fully investigate
repetition effects, an additional analysis was conducted using the 33 time
, sets from the expert group. Since the expert group participated only
during one special session of the experiment but were still highly trained,
y Vit was felt that their performance could reflect ID and WA training effects
without experiment fatigue. Expert times were paired with mean times from
the experimental groups on days 5 and 6 on the basis of aircraft simulation
numbers coded into the raid tapes.
12
"j .-
Due to the low number of expert scores and the possible violation
of normality assumptions, a non-parametric Wilcoxen matched pairs signed-
ranks test was chosen. The results showed no significant difference for
ID times (Z=1.03 p<.15), but faster weapons assignment times for the
expert group (Z=3.39, pc.Ol). Mean WA time for experts was 62.6 versus
82.0 seconds for days 5 and 6. This difference could still represent only
greater familiarity with the Q-73 operating system. Such a possibility
receives support from the experts greater use of system switch optionsj as well as superior ability to handle extraneous verbal "squawk" evidenced
on tape recordings made of the sessions. However, it could also be a result
of fatigue differences. Additional research could provide the answer by
testing the issue directly.
t 13
IV. CONCLUSIONS
Tentative conclusions may be drawn for some of the original research
questions. First, TEWA in the semi-automatic mode is not instantaneous.
Average time for system contact, identification, and weapons assignment in
the AN/TSQ-73 system was 3.47 minutes. Second, manning configuration in
which the officer implements decisions directly through the WA console
significantly improves assignment times. Third, wave size has its greatest
effect on the WA operator, although ID times do increase linearly as the
number of aircraft increase. Fourth, WA is overloading at eight aircraft
and reaction times increase rapidly. Increased numbers of fire units do
improve WA times but only if eight or more are available. Finally, oper-
ator fatigue could be a consideration in WA under heavy task demands.
L I REFERENCES
Ward, Joe H. and Jenninqs, Earl, Introduction to Linear Models, 1973,
Prentice-Hall, Inc.
i
14
APPENDIX
The techniques used in this report to determine predictive formulas
involved a multiple linear regression program modified by Jorgensen to
permit rapid generation of model vectors. A discus;ion of the statistical
logic, model construction, and general programing methods is available
in "An Introduction to Linear Models" (Ward & Jennings, 1973).
Briefly the technique involves solving for parameters in a theoretical
equation previously defined by the experimenter to reflect specific
psychological'assumptions. This is done through a least squares fit of
the equation parameters to raw or transformed data. The program then
generates the standard beta weights and regression constant for that solution.
The predictive power of the equation is specified by M2 , the squared multiple
correlation coefficient. M2 is mathematically equivalent to the percent
of explained variation. Through the use of alternate equations, F ratios
can be set up to compare various assumptions in terms of their explained
variation. Analysis of variance and factorial analysis of covariance are
two of the many possible model formulations which could be utilized within
the regression framework.!n
Usually the most efficient strategy for finding optimal moaels
consists of postulating a general all-encompassing equation which takes
-into account a linearly independent set of predictor variables based on
conditions used in the experiment. The M for this model serves as a
maximum against which the predictive power of other simpler models is
06. 21 p' .c-W'. i.- .... 3¶*iSS
judged. The simplest possible predictive model is the mean of the
distributions. An F ratio formed between the general model and the
mean model tests the usual null hypothesis for equality of means. By
using the same technique with more complex models, F tests can be set up
comparing various models to the general model and to each other. This
is analogous to analysis of covariance. The pattern of significant F
ratios as well as the M2 efficiency of the models quickly points out the
most efficient and parsimonious equations for a given set of assumptions
and predictive power.
For example, in this experiment the three general models were all
predictive equations taking into account raid tapes, low, medium, or
high wave sizes, days, manning configuration, groups and fire units.
A different general equation was generated for total, ID, and weapons
assignment times. In this analysis four types of models were condidered:
orthogonal component models for each variable of interest (these are
equivalent to a one-way analysis of variances for each variable), linear
models (equivalent to simple linear regression fits), second and third
order polynomial models (which investigated parabolic curves) and mixed
models, such as a linear model going second order at a specific point.
The following are examples of a general 12-element orthogonal model for
[ .total times, a model based on six wave sizes, a linear model, and a
polynomial model:
22
a7i P 77
7I
Time (total) " aoXI a1S2 Ja2 13 " ". l7
Time (total) -aoYI+aIY 2+a2Y3Y4+a4Y5 +a5Y6
Time (total)
Tite (total) U+aW
2 3Time U+aWI~aW +a W
where X1... X12 are colrmn vectors of O's and I's partitioning total data
times into orthogonal sets as a function of such variables as days, t *ves
or raid tapes.
Y...Y 6 are column vectors of O's or l's partitioning total times
into sets for wave size one through 6, respectively.
U is a unit vector of l's.. W is a column vector containing the
numbers one through 6 corresponding to the size of the wave.
W is the direct product of W with itself.
a is a beta weight for the regression.
Though many models were considered and fit, only the most parsimonious
and experimentaly meaningful are reported in this paper. The equations
derived represent best fits for the available data. Their generalizability
assumes the generalizability of the experimental situation to the combat
usage of the AN/TSQ-73.
23
I,i
S •y 23
' -
A l. - . ... -::_ .. "-- --
•-- !- ,--• ,• : ..... ... .• .. ... ....
DISTRIBUTION
I US ARM'! CINCPAL bjPPUHT bVUOP PELRSON~4tL DIVISION
1 HODA ATTN: pMAC
I TAGITAGCEN AT IN: UAAGj-EO
I HQi, TC*ATA ATTN: AICAT-UP-U
e t-0j)A RESLARCS AND SIUDILti 01C
1 MILITARY UCCUPATI'JNAL U t.LOPMENf DIV UAPC-SPO-0 RM 852C HOFFMAN BLDG I
4 OASU (mRA ANDI)I NOL)A ATTN: UAMO-RUR
1 HQ TCATA TECHNICAL LIbHAmY
I HOI)A CHILF9 HUMANJ kLSUURM-LS UEVLLOPMENT DIV
1 USHALJCOO SICI HQ0A ATTN: LJAMI-TSI
I USA AVIATION SYSTEMSi COMO ATTN: URSAV-ZDR
I USA CORADCOM ATTN: AMbLLPA-WmII USA A~kAUCOM ATTN: ATFE-LO-A.I HEAU)UUARTLkSt US 14AR!NL. LORPS ATTNt COUE MPI-a8
r- US ARMY EUROPE A1ND SEVENTH ARMY
1 IST INFANTRY DIVISION ANU eTo RILEY ATTN$ AAFZN-DIT-T
IUSA INTELLIGENCE. AND St.CUNITY CUMMAND ATTN: IAQPS-7NG
eHO TRAIOC TECtiNICAL LIIIRAHY
I NAVAL TRAINING3 t.WJIPMLr t..tN AIlN: TtCHNICAL LldRARY
I MILIlAPY OCCUPATIONAL UEVt.LOIPMEiNI lV ATTN: UAP(.-MSP-S, Rm u52C9 HOFFMAN BLDG I
1 MILITARY UCCUPA1IIUNAL UEVLLOPMFNT DIV ATTN: UAPC.MSPO0, Am 8,.,C* HOFFMAN 8LUG I
1 MILIIAPY OCCUPATIONAL ULVLLOPMFNI DIV ATTN: OAPC-MSP-T, RH 8b2CO HOFFMAN BLDG I
I USikFACFA CH1E9 DWGANIZA]IONAL 0L~CIIvENESS 11RANCH
I 8TH- INFANTRY DIVISION
I 8400A TANK FORCES MANAt3LALNT OFC
1 NAVAL AIR SYSTE.M COMMAND ATTN: AIN-413F (CUR JAMES H. ASHBURN)
I 1 CSOPS (DIST 4) ATTN: DAMO-WUT + UAMO-HQC
1 1230 USAHLOM XMhEHVL CENTtR
1 FT. Ht.NJAI41N HAHRISON9 11N 46kib
I DI14EC1)RATEF OF AHN40R AVIAIION A11N: ATS13-AAU
J ~I AVIA~iIN UlVISIUvN ATTN: A1ZK-jIyI-AVN 1UHIFS)
I USA 7ýO&CEh COMMAND AFIN - OLI'lhTY C OF S FOR INTELLIGENCE
1 USA IORCES COMMAND AFUP - ULP0Y~ CHIEF OF STAFF FOR OPERAYIONS
I US ARMY Alp UE~EkiSL
1 L)INECTOHATE OF (HAINING ATTIN AILU-T
I DIRECTOHAIIE OF CU-1bAT UtL~tLOP#4INTS ATTN: ATZA.D
I HUUARCOM MARINE CORPS LIAISON UtIC
LSI DEPARTMENT OF* )HE AIRY US ARMY INTELLIGENCE * SECURITY COMMAND
1 US ARMY AUENCY F,) AVIATION SA0LY ATTN: LIORARIAN, BLDGi 4905
I USA MIsSILE MATERIEL HLAUINES!S COMMANU ATTN: ORSMI-NTN
1 ARTADS ATTN: URCPM-TUS-TuI USA FORCES COMMAND
k PM TRA)E' ATTN: URCPM-INU-NE Wk.~ CRuNSIOLM)
I US MILITARY D1STRICT Oý WA5111N('lON oFC OF EUUAL OPPORTUNITY
1 NAVAL C'IVILIAN PERSONNEL L-UMU SOUTHERN FLU DIV
2U ARI LIAISONq L2FýICEI IJSACDEC ATTN: AIEC-t.X-L MUMAN tACIORS
I USAFAI3OS/TAC SLNIuR ARmY ADVISOR?
I INTEH-UN1V SFMINAR ON ARMLU FORCES # touc
1 USA ELECTRONIC PIRDVING uUHUNU ATTN: STEEP-mT-S
I I (OASU) (R AND U) UEPUTY FOR SCILNLE ANO TECHNOLUGY
jOFC OF NAVAL RESEARHC AIIN: COUL 455 (UP. MARTIN A* TOLCOTT)
I AFHRL/7TI AFHRL/AS
e AIR FORCE HUMAN HESOUHCtS LAO jTTN: Pt.
I FEOEIRAI AVIATION ADMINISTkýATIUN CENTRAL R~bION LIBRARY, ACE-66
1 6510 AmWL/8H
1 6510 AmRqL/HEINVLpERSONNEL H ANDU uCENTERI 6OMMAND AND SUPPORT SYSTEMS
25
PRECEDWII'G PAG9 BW.A1d s
I NAVY PFNSUNNEL R AND U) CL14ILH ATTN: (COt)E .1081 kOMUNO 0. THOMASI NAVY PFI4SONNEL W AND U CENTER uIRLCTOR OF PROGRAMS
I NAVY PtRSUNNLL H AND U CEN~TER ATTN$ UN. d. RIMLANOII iUS ARMY AVN ENbiINEEHINb. FLIGHT ACTIVITY ATTNI SAVTE-Tfle OFC Of NAVAL RESEARHC~ PERSONNEL AND TRAININGI RESEARCH PROGRAMSI OFC OF NAVAL R~hANCH ASS]. uIRLCTOR PERS # TRAINING RSCH PROGSI OFC OF NAVAL Nk.SLANCH PRUJf.CT utFICER94 ENVIRONMENTAL PHYSIOLOGY1 NAVAL AEROSPACE MEDICAL Rblvi LAW ATT-41 (COOL L51l.I dUREAU OF NAVAL PERSUNNtL SCjt:NTIýIC ADVISOR (PEAS-OR)I'NAVAL AEROSPACE MEDICAL NHt.H LAU AEROSPACE PSYCHOLOGY DEPARTMENT1 USA TRADOC SYSTEMS ANALYSIS ACTIVIIY ATTN% ATAA-TCAI HEADQUARTERS, COAST bUARU CHIF'I, PSYCHOLO61CAL RSCH 8PI USA~ RESEARCH AND TLCNNULUtlY LA4 ATTN$ QAVOL-ASI USA ENGINEER TOPOGHAPHIC LAWS ATTN: O1L-GSLI USA ENGINEER TUPU0 NRAPHIL LAW$ ATTN: bTINFO CENTCRI USA LNGINEER TUOI~SkAPHIL LABS AITN:'ETL-TD)-S1 USA 14UHILITY EWiU12MtNT H AND u LUMOU ATTN: ORUmE-TQ1 NIGHT VISION LA4I ATTN: UmbEL-NV-bUD
I USA TRAINING dUAHO1 OSA HUMAN EN(3INELRINb LAW1 US HLL/USAAVNC AT(N: URAVIL-FN (UN. HOFMANN)I USA MATERIEL SYSTEMS ANALYSIS ACTIVITY ATTN: 0RXSV-M1 LISA RESEANCH OýC AnN: k. W. MUHTHLAW4I NAFEC HUMAN EN.I1NLEEINU UwANCHI HAtTTLLE.-COLUMWUS LAUOOAIUWiLS TALTICAL TECHNICAL 01CI USA ARCTIC TEST CEN ATTN: AMSyt-PL-TSI UJSA ARCTIC TEST CEN AITN9 STEAL-L-ML1 USA CONCEPTS ANALYSIS Ak7LY ATTN: MOCA-W6,I USA CONCEPTS A14ALYSIS A(3CY ATTN- MOCA-JFI HU WRATH DIV OF 4ILUHUPbYLHIAlWYI USACACflA ATTN9 ATLLCA-CICI USACACflA ATTN: ATLLCA-Ci-MI USACACIlA ATTN: ATZLCA-CI-AI USACACInA AT TN: ATZLCA-CAI USA LI.FCTNONIC VARFAH4 LAW LHjItt INTELLIGENCE "ATER DEVEL SUPP OFFI USA IHOPIC TEST CENTER ATTN: TtLHNICAL LIBRARYI USA HSCH bEVEL + S1ANL)AR&JLA upt U.K.I AFFDL/FGR (CUIC)A USA NATICKS RESLAHCH AND ULVLLOPMLNI COMMAND CHIEF$ BEHAV SCIENCES DIV, FOOD SCI LAB1 0ASOl E A~fl E (L AND LS) MILITARY ASST FOR TNG *PERS TECHNOLI TRAJANA ATTN: SAJS-OHIHWOIA ATTN: UASG-?tb (Lr(. XICHMRu L. HANT7ELL)
I NAVAL AIR SYSTLMb COMMAND ATTN: AIN-53131 ECOM ATTN: AMbeL-CT-0I USACUEcEC TE.CHNICAL Ll"AINAYI USAAHL LIWNANYI HUMAN FRESOURCES, NSCH O~lb tHUMI4NU) LIdNARYI S~vILLI. RESEARCH CORPORATAONI US-^ FHAUOC SYSTEMS ANALYSIS ACTIVITY ATTN: ATAA-SL (TECH LIBRARY)i UNLFORMEu St"RVICLS U1411 OF THt. HLALTH SCI ULPARtMENT OF PSYCHIATRYI I USA CUmPUTER SYSTEI4S COMMAND ATTN: COMMAND TECHNICAL LIBRARYIH UMA4A WESOURCEb HSCH UNG (flUMNIRO)
I HUAiRRO WESTERN LldkRAH'I EUSTIS OIHECTONAIE, USAAMNuL TECHNICAL LIBRARYI RAID CORPORATION ATTN: LIbRANy.-MAITLAND1 RAND) CokFURATIUNj j&TN: LioRA~y UI F E;ENAt AVIA71UN AUMINISTWATION AITNt CAM! LIWdRARV ACC-4401INAFEC 1.8HNARY, ANA-6*#
1 L3RONINbt.R L18RARY ATTN: AIZP-PS-L UL~)b 1313I CENTER FOR NAVAL ANALYbib1 NAVAL iiEALTH RbCH CEN LAWRANYI NAVAL FLLCTRONICS LAU A11I4 HFSLANCH LIARARY
26
I NAVAL. Ft~NONNEL hANI) U Ct.N L1t5HAKY ATTN: CODE 9D201L
I AIR FORCE HUMAN~ HES0UHLk.S LAO AIJN: TSi
I AIR IýiRCE HUMAN kESOuNCES LAb i-LYINO TRAININL6 DIVISIONI HQ, FT. HUACHUCA ATTN: ItCH REF DIV
1 USA ACADEMY OF HELALTH1 bCILtNCt.S bTLMS'IN LIBRARY (DOCUMENTS)
I SCH-OOL OF SYSTEM4S ANU LO~.jIbT1cs ATTN& AFIT/LSCM
I USANMNCnC TECHNICAL LlbHAHY
I DEPIARTmENT OF IHL NAVY THAININU ANALYSIS AND EVALUATION OP
I NATIONAL LENTEN F31o HEALIM STATISTICS ATTN: he LUPUY (PSYCHOLOGICAL ADVISOJR) -
I USMA nOj'T OF bEF1AVIUHAL t)Ll ANU LLADEHSHIPI US NAVY CNET bUI'O0RI HL!3LARCH LIbNAWY
I OLD) UDMINION UNIVERSITY PLRFO)PdANCL ASSESSMENT LABORATORY
I USA COMMAND ANU t;ENEMAL bIAFi- COLLk(EGE ATTN; LIBRARY
X~ USA TRANSPOR1AlIU'N SCHOOL USA INANSP IECm INFO ANO 145CM CEN
I NASA Hu ATTN: COOL. kTt.-b (L)R. MLLVIN MONTEMk.NLOI
I NMROC PROGRAM mA'4AGL.H FON HUMAN~ FtRF3RMANCL
I NAVAL MLUICAL H A~4L U COMMANU 144)
I USA ADMINLEN TLC.INICAL Nt.SEAWCh ONANCH LIBRARY
e HUDA IISA MEL) HSCI1 AND Ut.VLL CoMMAND
I USA FIFID ARTY HU) ATTN: A1LR.I-iuIJUT (IQLSTWOM)
I NAT LLFANLNUHOUSL 1-ON MENIAL HEALTH 14FO PA~RKLAWN BLDG
I U o)F T F XAS CEN Fam CUMMU141CAIIUN NSCrII INSTITIUTE FOR Uk.FENSL A1NALYSES
I USA TRAINING SUpPPJRi CLNILN U1-VLL SYSTEmS TNG * DEVICES DIRECTORATE
1 AFHRL TE.CHNOLUUY 'JFC (H)I PURDUE UNIV ULPT OF PSYLHULOIJILAL SCIENCES
1 USA mUHIL1TY EUUIPMENT R AND U LUMMANIO ATTN: ORUME-ZG
I HU, JSA MO~W ATTN: ANPE-OE
I D)A US ARMY HETKgIIiIN'i DLIL REitt-AHCH # EVALUATION DIR
I ULSAF SCHOOL OF AERUSlPACt NELCILI'iE AEikumLUICAL LIBRARY (TSK-4)
I US MILTTAHY ACAuL4Y LIBRANY
I USA INIELLIGENCE CEN ANU tSCM ATTN*. SCHOOL LIbRWARY
IUSA INIELLIGLNILL CLN ANO SPLH 11EtPT OF (2ROUNI) SENSORS1 MAWINE CORPS IN'S11IUTL
i NAVAL SAFETY CENTER ATTN: HU6LHI A. ALKOV PH.U
1 USAAVNC AND) I-T HJCKLH ATTN: A~liUES$
I US ANMY AVN TN(3 LIbRARY ATTN: i..LIE LIbRARIAN
I USAAVNC ATTNS AIIUL)
I US MILTIARY ACAOL41 Ui-C UI* MILITARY LEADERSHIP
1 US MILITARY ACAI~kAY UILtLIOR Di- INSTITUTIONAL RSCR
I USA AIP Ut.FENSL SCtIOUL AI'NS AlSACU..MS
I klSAAUS-L~tNRAyDO)CUMtNI S
I IIUSA AIR DEFENSE bZ)MNL) Al IN: FILLS REPuSITORYI USA INFANIRY IHUAki AIINZ ATlu-lU-TS-ri
I iJSA INTkLLIGLNCF- CEN A14u SLH1 LLJUCATIJNAL ADVISOR
I USA ORDNANCE CEN ANU SCM ATINI ATbL-TtM-C
I USA ARMOR SCHOOL ATTN: AIl "-DpT-
1 USA AW:OUA CLNTER UINECIUKATL t0 F cOmtiAT iEVELOPMENTS
I NAVAL POStGRAUUAIE ý,CH ATTN: Uul-tY KNOX LIbRARY (CODE '1424)
*i USA IRANSPONIRAlIIU SCHOOL UEPI)IY ASS[. COMMANDANT EDUCA. TLCHNOLOGY
I 1ISA bI(NAL, SCHUUL ANDU it. knORuoN AIT49 ATZH-k.T
I USA ARmUH SCHd- DL. VAL b4AwCHq UIHLCTJRATL Of INSTRUCTION
I CHIEF (IF NAvAL ILJJCATIUN AND I-jt ATTu: RAYdUHN A. WILLIAMS (COuE 333)
I USASIGS% STAFF AND FACULTY ULV AND INJ DIV
I HQ ATC/XpTI TRAI'Nil, SYbIEMS oEVLLOP~ENT
I USAISL) ATTN: AlblE-UT-t
I US ARMY AHMOR 5CtiJUL UIHLLTtJRATL OF TR~jidIN5
I USA IJUAR.TERMA:lL" SCHOOL LDIREBURATt UF IRAININ6U EVELOPMENTS
i US COAST OL-AHD ACAUEMY ATTN: CAUtl C3UNSELLO (DICK SLIMAK)
I USA THANSPORTATIUIJ SCHOOL UDIR:LtOR OF IRAININO
I USA CHAP)LAIN CENTERN AND bLHOUL AU;N: L18WARIAN
I LISA LNFANIkY SCMUDL LIORAHAY ATII MISS15 L.IVINOSTONj
27
FI USA INFANTRY SCtiUDL ATNN; ATSqI-VI US ARMY INFANTHY SCHOOL. ATTN$ AFSti-CUI USA INFANTRY SCHUOL ATTN$ AT51-UtJII 1J5A INiANTRY SL.MU3L ATTN; ATSH4LvIUSA MILITARY POLICkL SCHUUL/TkAjNIN1, CLNTER ATTN$ ATZN-PTS
I11lSA MILITARY PULICk SCH1UOL/TRAININ(, CLNTVN' UIA, C0OMBAT UEYLLOPMENTI USA MILITARY PULICE bCHUOL,'THAININbi Cr.NTFR ULM* TRAINING DL.VELOPMENT1 USA Mu ITARY POLICt '3CMOOL/IRAININ1, CENTER ATTNt ATZN-ACLI USA INSrITOTL. OF AI)MINIbT8ATIQ-\I ATTN$ HESIOLNT TRAINING MNAhGEMENT
1 USA FIFLU ARTILLL.RY SCHOOL ýOjkklb StvET[ LIBRARYI USA INTITUTE Ui AUMtNISI"A;IUN ALAU04IC LIUI4ARYI USA WAI4 CULLEOL ATTN: LidU.y[I USA LN(AN~kW SCtiUOL LIUNANY ANU LtAN'41N(, REWDURUES CENTER4I USA A4MOW SCHOUL (USARMS) A17N: L1BRARY1 OHtANIIATLONAL LfFtZCIIV4LNL!.S 11413 CtN + SCM ATTN: L1IOPARIAN1 US ARMY INTELLliv4(C. CLNILH + SCHOOL ATTN: ATSI-ToI 1U~S ARMY INTLLLIuL'4tL CL~i~tr( * !bLHOOL ATTN: ATSI.'Rm-mI US ARMY INTELL1INCE CEIMI~ + lioiuuL AITNt ATSI-ToLDI US AHMY IN~TELL~iU~4Ct C.Idt~it + SCHOOL ATTN: AISI-CO'CS-CI US ,ýRMY INTELLI~jtNCE CL.NILH + SL.)OUL ATTN: ATSI-DT-SF-IM
4 RRTISH LMAASSY jkIfIbH UJtkLNCt. STAFF2CA14AUIAN JOINT S1AIF
I CULS (w) LItHkAHYI FRENCH MILITARY AIIACHL.I AUSTRIAN LMtiASbY MILITARY ANU AIR ATTACHE.3 CANAUIAN UFIýLNLt LIAISON bIAtIý ATTN: COOUNSELLOR, UEFENCE R AND D
1 ROYAL NtmtHFmLANUi Er~dAbSY MILIIAH'1 AIIACHEI CANALfiAN PORCEb IIASI CUHNWALLIS A17N: PERSONNEL SELkLCTIONe CANADIaN tORCE4 PEkSUNNLL APPL RSCh UNITI WARY PFRSUNNEL Rk.SLAmCh LbIAbLlbhMtN7I AWA4Y PFHýbUN NtL RLSLARCH LbIAdLISHMtNT ARI SCIENTIFIC coOORDINATION OFFICEI NLrMI.HLANW1S EMOA.'SY UýILL U1 Itit AlIR ATTACHE
6 LIHRARy UF CUNIjRLSS LXCHAN(iL LANU bIFn UIVI IJE~kNSF TLCHNILAL INH.JHMAIION CL1I' ATTN: UTIC-IC
15Sý LIBRAHY OF* CON~3HLSS UNIT UO(.UmLNIS EJIPEDITIN4 PROJECT1 EDITOR. Rt ANU U MA1,iAL'NE ATTN: UNCut-LNIUS (30VFI4NMENT VRL4~TING UH'. LI.HMAHY, PUbLIC UUCU14LNTS DEPARTMENTIUS GOVFHNMINT PH147NGI1 Utt. LIHNARY A4U STATurORY, L119 DIV (SLI)
1 THF ARmY LlbRAHY
28
top related