EFFECTIVENESS OF COMPUTER-BASED TRAINING FOR IMPROVING

DETECTION OF IMPROVISED EXPLOSIVE DEVICES BY SCREENERS IS
HIGHLY DEPENDENT ON THE TYPES OF IEDS USED DURING TRAINING
Diana Hardmeier
Center for Adaptive
Security Research
and Applications
(CASRA)
Zurich
Switzerland

Moritz Jaeger
Center for Adaptive
Security Research
and Applications
(CASRA)
Zurich
Switzerland

Rebekka Schibli
Center for Adaptive
Research and
Applications
(CASRA)
Zurich
Switzerland

Abstract - Several previous studies have shown that threat
detection performance of X‐ray operators can be increased
substantially if computer‐based training (CBT) is used. This
applies particularly to objects that are never or rarely
encountered at a security checkpoint as for example
Improvised Explosive Devices (IEDs). However, little is known
so far about the importance of using different types of IEDs in
CBT such as conventional, unconventional and inert IEDs.
Conventional IEDs are made with real explosive and
conventional detonators with a primary and secondary
charge. Unconventional IEDs are made with real explosive
and unconventional detonators that do not contain a primary
charge. Inert IEDs are made with explosive stimulants and
fake detonators.
Two studies conducted with 420 and 433 airport security
screeners are reported. All of them had already used an
individually adaptive CBT program for improving x‐ray image
interpretation competency during several years however
containing only conventional IEDs. In study 1, screeners were
tested with a computer‐based test containing conventional
and inert IEDs before and after CBT for several months
including both types of IEDs. In study 2 X‐ray operators were
tested with a computer‐based test containing conventional,
unconventional and inert IEDs before and after CBT for
several months including all three types of IEDs. The following
results were revealed in both studies: 1) Detection
performance in the first test was high for those types of IEDs
that X‐ray operators knew from previous CBT and 2) IEDs
that were not detected well were effectively trained using CBT
that contains these types of IEDs.
In summary, CBT can be a very effective tool for increasing
X‐ray image interpretation competency of screeners if it
contains conventional, unconventional and inert IEDs.
Index Terms - aviation security, computer‐based training,
detection performance, human factors, improvised explosive
devices, x‐ray screening.

Adrian Schwaninger
Center for Adaptive
Security Research
and Applications
(CASRA)
Zurich
Switzerland

I.

INTRODUCTION

Several terror attacks in aviation evidence the importance of a
reliable security control at checkpoints. Even though state-ofthe-art machines with automatic detection of explosive
material, liquid detection and other functionalities help in
detecting threat items, the human factor is still essential. X-ray
screening of passenger bags is one important measure to
prevent passengers bringing in threat items into the security
restricted area and on board an aircraft. So far several studies
showed that the detection of threat items in X-ray images
depends highly on training as they often look quite different
than in reality [1, 2, 3, 4, 5]. Thus, screeners not only need to
know which threat items are prohibited but also what they look
like in X-ray images of passenger bags. Studies including
threat items of the four categories guns, knives, Improvised
1
Explosive Devices (IEDs) and other were able to show that
especially the detection of the two categories IEDs and other
is rather difficult without training. Threat objects of both
categories vary enormously in shape and look quite different
in X-ray images than in reality. Additionally, IEDs are normally
neither seen in reality nor at a checkpoint. Despite the variety
of IEDs, a study by [6, 3] revealed not only large training
effects, but also large transfer effects for this category. That
is, detection performance of IEDs is much higher after several
months of training and IEDs that are visually similar to the
learned ones can be detected even if they were never seen
before. However, this transfer effect was only revealed for one
IED type, namely conventional IEDs.
Whether the training effect depends on the IED type learned
or whether a transfer effect could as well be shown for
different types of IEDs was investigated in this study.
Therefore, three different types of IEDs: conventional,
unconventional and inert IEDs were used. All screeners who
participated in this study already used an individual adaptive
training program for several years including conventional IEDs
only. To test the performance for all three IED types before

1

The category other includes threat items like electric shock devices,
chemicals, etc.

978-1-4244-7401-1/10/$26.00 ©2010 IEEE

1

and A’ which is a detection performance measure that takes
into account both the Hit and the False Alarm Rate [7].
Three groups of IEDs were used: trained conventional IEDs,
conventional IEDs, and inert IEDs. IEDs of the first two groups
belonged to the same IED type and thus showed both the
most common IED characteristics: organic explosives of a
quadratic shape and detonators with primary and secondary
explosive charge. Trained conventional IEDs were already
trained by the screeners before the study started whereas
new conventional IEDs were not trained before. Inert IEDs
were non-functioning IEDs visually different from conventional
IEDs amongst others in color and in the constitution of the
detonator. They were not trained by the screeners before this
study started.
In each group, 16 IEDs were used and each of the IED was
inserted into both a bag with high and low complexity. Thus,
both tests were composed of 192 images: 3 (IED groups) * 16
(objects) * 2 (bag complexity levels) * 2 (harmless images vs.
threat images).

and after training a computer based test was used before and
after Computer Based Training (CBT). Furthermore, the test
design of the study not only allowed measuring the training
and transfer effect but as well the influence of the bag
complexity level on the detection performance of all types of
IEDs used. Last, improvement in the detection for screeners
who trained most and screeners who trained least was
evaluated.

II.

STUDY 1

In the first study the two types conventional IEDs and inert
IEDs were compared before and after CBT using a computer
based X-ray screening test.
A. Methods
1) Participants and Procedure: The sample consisted of
420 airport security screeners working at one European
airport. All screeners had already several years of computer
based training with the training system X-Ray Tutor (XRT)
including conventional IEDs only. To measure the training
effect in X-ray image interpretation, all screeners conducted
an X-ray test before and after a 4 months period of recurrent
computer-based training with a special IED version of XRT.
During the training period, screeners trained on average 5.64
hours per week (SD = 5.02).
To avoid test repetition as the reason for performance
increase, two tests were developed and screeners randomly
distributed into two groups: group 1 started with test A and
finished with test B whereas group 2 conducted the tests in
the reversed order. The data of both groups are combined for
the analysis.

b) X-ray screening training: Between the two tests,
screeners conducted recurrent computer-based X-ray training
using X-Ray Tutor, a widely-used training tool that effectively
improves X-ray image interpretation competency [3, 8]. Using
XRT, screeners have to inspect X-ray images of passenger
bags created at the point of use by an adaptive algorithm that
merges threat items and passenger bags with increasingly
difficult combinations of viewpoint, superposition, and bag
complexity. The images have to be judged as OK (contains no
threat item) or NOT OK (contains a threat item). XRT has the
same appearance as the test except there is immediate
feedback and screeners have to click on the IED to show that
they correctly identified it. For details on XRT and its algorithm
please refer to [1, 9].
For this study, an XRT edition specialized on IEDs was
created with IEDs belonging to the same groups as those
used in the test: trained conventional IEDs, conventional
IEDs, and inert IEDs. In order to measure the transfer effect,
only half of the IEDs (Set A) used in the test were also
integrated in the training. The other IEDs (Set B) were only
included in the test version.

2) Material:
a)
X-ray screening test: The two X-ray image
interpretation tests (test A and test B) were developed to
measure how well airport security screeners are able to detect
IEDs in X-ray images of passenger bags. The two parallel
tests consisted of 192 X-ray images of passenger bags using
images of Smiths-Heimann Hi-Scan 6040i machines.
Whereas half of the bags where harmless bags (no threat
item included), the other half of the bags contained an IED
that had been virtually inserted into the bag by aviation
security experts.
The tests were integrated into the training system XRT and
took about 1-2 sessions of 20 minutes to complete. Images
were shown for a maximum of 15 seconds on the screen. The
task was to visually inspect the images and judge whether a
bag was OK (contains no IED) or NOT OK (contains an IED).
X-ray image interpretation competency was measured by the
Hit Rate (% of threat images correctly identified), the False
Alarm Rate (% of harmless bags falsely identified as a threat),

B. Results and Discussion
In both studies the detection performance measure A’ was
calculated in order to compare the results in the computer
based test before and after computer-based training. A’ takes
the Hit Rate (H) and False Alarm Rate (F) into account. The
Hit Rate shows how many times an IED was correctly
identified in percent and the False Alarm Rate shows how
many times a harmless bag was wrongly judged as NOT OK
in percent. A’ is calculated by the following formula:

2

A’ = 0.5 + [(H–F)(1+H-F)]/[4H(1-F)], if H > F
A’ = 0.5 - [(F–H)(1+F-H)]/[4F(1-H)], if H < F
For more information about detection performance measures
and A’ please refer to [10] or [11].
1) Reliability:
To examine whether the two tests which were used to
measure detection performance are reliable and may be used
for evaluation purposes an item analysis was conducted.
Table 1 shows Cronbach Alpha and Guttman split-half
reliabilities for each test version (test A and test B) before and
after training. Reliabilities are calculated based on hit and
correct rejection rates. In order to calculate Guttman split-half
reliabilities the images were ordered according to the level of
difficulty of the bags and then assigned into two halves.
Cronbach Alpha values above .825 indicate good reliability
[12]. As well all split-half coefficients > .864 are clearly above
the limit of acceptable reliability [13]. Thus, these results
affirm a reliable measurement of detection performance
before and after training using our computer based test.

Figure 1. Detection performance A’ for each IED type before and after
training. The bars represent standard deviations. Please note that due
to security reasons detection performance values are not specified.

In addition to the independent sample t-test a two-way
analysis of covariance (ANCOVA) with the two withinparticipant factors IED group and measurement and the
covariate training hours was conducted with the following
results: There were main effects of IED group F(2,836) =
572.03, p < .001 and measurement F(1,418) = 25.70, p < .001
2
2
with an effect size of η = .58 and η = .06, respectively. The
two-way interaction of IED group and measurement was also
2
significant, F(2,836) = 6.00, p < .01 with an effect size of η =
.01. These results reveal that the improvement in detection
performance is different for the different groups of IEDs. Thus,
the detection performance of IEDs which have not been
trained before, but are very similar to the trained ones
(belonging to the same IED type) is higher than for another
IED type which is visually completely different. Furthermore,
the significantly better improvement for inert IEDs shows that
screeners are able to learn new types of IEDs very quickly.

TABLE I
RELIABILITIES
1A

1B

2A

2B

Cronbach Alpha

.845

.825

.880

.873

Guttman
split-half

.865

.867

.903

.864

N = 420

Cronbach Alpha and split-half reliabilities in the first study.
Note. 1A and 1B are the two test versions before CBT, and 2A and
2B are the ones after CBT. The same persons passed test 1A/2B
respectively 1B/2A.

2) Overall Detection Performance: As illustrated in Fig. 1,
the objects with the highest detection performance are the
trained conventional IEDs, followed by new conventional IEDs
and inert IEDs. Independent of the IED group, a higher A’
value results from computer-based training with XRT. For all
IED groups the difference between the first and the second
measurement was significant with t(838) = -4.70, p < .001 for
trained conventional IEDs, t(838) = -3.82, p < .001 for
conventional IEDs and t(838) = -5.65, p < .001 for inert IEDs
indicating training effectiveness.

3) Detection Performance of Images in Training and Test
(Set A) and in Test Only (set B): In order to evaluate whether
IEDs within the same IED group - which were not included in
the training session (set B) - have a lower detection rate,
three ANCOVAs were conducted with the two withinparticipant factors measurement and set plus the covariate
training hours (see Fig. 2).

3

To conclude the IEDs presented in the test only were overall
easier than IEDs presented in both, the test and training
session. However, the non-significant interactions between
measurement and set indicate that the transfer effect for each
IED group is quite strong.
4) Comparison of Detection Performance in Easy and
Complex Bags: A two-way ANCOVA for each IED group was
conducted in order to investigate for each group whether the
detection differs in easy and complex bags.

Figure 2: Detection performance A’, with standard deviation bars, of
images in tests only and in tests and training sessions.

The results of the ANCOVAs for each IED group were as
follows:
Trained conventional IEDs: significant main effects of
2
measurement with η = .07, F(1, 418) = 29.50, p < .001 and
2
set with η = .56, F(1,418) = 527.38, p < .001, non-significant
2
interaction with η = .00, F(1, 418) = .87, p = .35 showing that
the detection performance of images which were in the test
and of those which were in both, the test and training session
were significantly different. Also the difference in detection
performance before and after CBT was significant. However,
the interaction between measurement and set was not
significant indicating that screeners’ improvement in the
detection performance was not higher for images also shown
in the training.
Conventional IEDs: significant main effects of measurement
2
2
with η =.03, F(1, 418) = 11.77, p < .01, and set with η = .32,
F (1, 418) = 192.34, p < .001, non-significant interaction with
2
η = .00, F (1, 418) = .91, p = .34. These results are very
similar to those obtained from trained conventional IEDs,
revealing that there is again no significant interaction.
2
Inert IEDs: significant main effects of measurement with η =
2
.05, F(1, 418) = 20.65, p < .001 and set with η = .05, F (1,
2
418) = 108.97, p < .001, non-significant interaction with η =
.00, F (1, 418) = .48, p =.49 indicating that also for inert IEDs
the improvement in detection performance did not differ for
images which were only in the tests and those which were
both in the test and training sessions.

Figure 3: Comparison of detection performance A’ of easy and difficult
bags. The bars again represent the standard deviations.

As illustrated in Fig. 3, the detection performance of IEDs in
easy bags was higher for all three IED groups. For trained
conventional IEDs there were significant main effects of
2
measurement η = .03, F(1,418) = 13.58, p < .05 and
2
complexity η =.22, F(1, 418) = 114.89, p < .001, but no
2
significant interaction η = .00, F(1, 418) = .33, p = .57. The
results for conventional and inert IEDs were similar. There
2
were significant main effects of measurement η = .02, F (1,
2
418) = 7.12, p < .01 respectively η = .03, F(1,418) = 12.80, p
2
< .001 and complexity η = .15, F(1,418) = 75.56, p < .001
2
respectively η = .01, F(1,418) = 5.73, the interactions were
2
again not significant with η = .00, F(1, 418) = 1.84, p = .18
2
respectively η = .00, F(1,418) = .63, p = .43. These results
reveal that the improvement in detecting IEDs in complex and
easy bags between the first and second measurement were
comparable for all three IED groups even though inert IEDs

4

III. STUDY 2

were expected to be detected worse in complex bags
because of their missing features.
5) Effect of Training Hours: In this part it was investigated
whether frequent training affects the detection performance.
To generate two groups with a considerable difference
regarding the amount of training hours, the 24 airport security
screeners with most training hours were compared against
the 24 screeners with least training hours. For the analysis
one outlier with more than 60 hours of training was deleted.

In the second study the detection performance measure A’ in
a computer based test for the three types of IEDs, namely
conventional, unconventional and inert IEDs was calculated
again before and after CBT.
A. Methods
1) Participants and Procedure: The sample consisted of
433 airport security screeners working at the same European
airport. Again,screeners conducted an X-ray test before and
after a 2.5 months period of recurrent computer-based X-ray
training with XRT.
Equally as in study 1 two tests were created and the
screeners were randomly distributed into two groups with a
reversed test order. The screeners trained on average 2.37
hours per week (sd=2.03). Again results of both groups are
analyzed together.
2) Material:
a) X-ray screening test:
Two new X-ray image interpretation tests were developed to
measure how well airport security screeners can detect
conventional, unconventional and inert IEDs in X-ray images
of passenger bags.
Both tests consisted of 132 X-ray images of passenger bags
using images of Smiths-Heimann Hi-Scan 6046i machines
with 50% of the bags containing an IED.
Three IED types were chosen. Conventional IEDs (30 objects)
and inert IEDs (6 objects) corresponded to the two IED types
used in study 1. The objects of conventional IEDs were new
and therefore not trained in study 1. The inert IED objects
were the same as in study 1 and therefore already trained.
The new type unconventional IEDs (30 exemplars) featured
IEDs with the following characteristics: inorganic and metal
explosives, detonators without primary and secondary
explosive charge and various non-quadratic shapes of
explosives.

Figure 4: Detection performance A’ with standard deviations bars for
the 24 persons that trained most and the 24 persons that trained
least.

As shown in Fig. 4, the 24 screeners that trained most have a
higher improvement in detection performance for all three IED
groups, and they improved the most for inert IEDs. However,
the analysis of variance (ANOVA) with the two withinparticipant factors measurement and training hours indicates
2
no significant interactions of all three types of IEDs with η =
2
.04, F (1, 23) = 1.03, p = .32 for trained conventional IEDs, η
2
= .04, F (1, 23) = 1.02, p = .32 for conventional IEDs, and η =
.15, F (1, 23) = 3.89, p = .06 for inert IEDs. However, the pvalue for inert IEDs which is just above the significance level
of .05 indicates that by trend airport security screeners who
trained more also show a higher increase in their detection
performance.

b) X-ray screening training: Again, an XRT edition
specialized on IEDs was created with IEDs belonging to the
same IED types as those used in the test. This edition was
adjusted as follows: images were now shown in SmithsHeimann Hi-Scan 6046i quality and the amount of
implemented IEDs was largely increased. The IEDs in the
training belonged to the same types as in the tests:
Conventional IEDs, unconventional IEDs, and inert IEDs. In
contrast to study 1, all IEDs in the tests were also used in
training thus no measurement of the transfer effect within the
same category was conducted in this second study.

5

B. Results
Once more the psychophysical
measure A’ was used for analysis.

detection

performance

1) Reliability:
Cronbach Alpha and split-half reliability coefficients are listed
in table 2, again separately for both test versions. All reliability
coefficients show reliable measurements with
Cronbach Alpha coefficients >.805 and split-half coefficients
>.803.
TABLE II
RELIABLITIES
N = 433
Cronbach Alpha
Guttman
split-half

1A

1B

2A

2B

.817

.805

.811

.825
Figure 5: Detection Performance A’ for each IED type before and after
training, n = 433. All three types showed significant effects (p < .005).

.806

.817

.803

.848
The ANCOVA with the two within-participant factors IED type
and measurement plus the covariate training hours gave the
following results: there were significant main effects of IED
2
type η = .36, F(2,862) = 246.90, p < .001 and measurement
2
η = .05, F(1.431) = 23.56, p < .001 and a significant
2
interaction of IED type and measurement η =.04, F(2,862) =
16.46, p < .001. These results are also illustrated in Figure 5,
which visualizes that the influence of training is different for
unconventional IEDs than for the other two IED types.
Because inert IEDs had now been trained for several months,
the detection performance of this IED type was much higher
compared to the detection performance of unconventional
IEDs. It was also much higher in comparison to the detection
performance of this type revealed in the first study.

Cronbach Alpha and split-half reliabilities of the second study.
Note. 1A and 1B are the two tests before CBT, and 2A and 2B are the
ones after CBT.

2) Overall Detection Performance:
As in study 1, airport security screeners showed a better
result after several months of training. This improvement in
detection performance was significant for all three IED types
with t(864) = -4.92, p < .001 for conventional, t(864) = -10.60,
p < .001 for unconventional and t(864) = -2.47, p < .05 for
inert (see Fig. 5).

6

3) Effect of Training Hours: Again, the results are similar
as in study 1 (see Fig. 6).

years are detected the best, whereas the detection
performance of new IED types was much lower in the
beginning. However, this detection performance was
increased substantially after training with XRT. Especially
results for inert, but also for unconventional IEDs indicate that
screeners are able to improve their visual knowledge within
short time. These results are also consistent with former
studies showing that training increases detection performance
for all kind of threat items within short time [8, 3, 6, 5],
Further, a transfer effect for similar looking IEDs within the
same IED type was found. Conventional IEDs which had been
trained for several years by the screeners were detected best.
Whereas new conventional IEDs similar in shape to the
trained ones had a slightly lower A’ value. This transfer effect
was also found within each IED type as there was no
significant interaction between measurement and set. That is,
IEDs which were presented in both the test and training
condition were not detected better after training than IEDs
only presented in the test condition. These results are also
consistent with previous studies which showed that screeners
are able to detect visually similar items also without training of
these specific items [3, 6]. However, unconventional and inert
IEDs, two visually completely different types of IEDs have a
significantly lower detection performance before training.
Thus, learning how conventional IEDs look like does not help
to recognize other IED types such as unconventional and inert
IEDs in X-ray images.
Furthermore, the improvement in the detection performance
of IEDs in bags with different complexity levels reveals that
with training all types of IEDs can be detected quite well
independent of the complexity level of the bag, as well as inert
IEDs which have no concise features.
Although there is no significant effect between screeners who
trained the most and screeners who trained the least, there
was a clear trend that more training between the two test
sessions leads to better detection performance especially for
new IED types, namely unconventional and inert IEDs. This
affirms that the most marked improvement in detection
performance is for items which have not been trained before
and that the training libraries need not only to be very broad,
but also need to be constantly updated and adapted with new
threat items.
In short, results from this study evidence the importance of a
variety of different IED types in X-ray screening training
systems. For this study three types of IEDs were defined:
conventional, unconventional and inert IEDs. Further studies
should investigate whether other types and how many types
of IEDs in total should be included in the training session in
order to guarantee both, effectiveness and efficiency of the
training system.

Figure 6: Detection performance A’ (with standard deviation bars) for
the 24 persons that trained most and the 24 persons that trained
least.

Although the interactions of the two-way ANOVA were not
2
significant η = .11, F (1, 23) = 2.71, p = .11 for conventional
2
IEDs, η = .07, F (1, 23) = 1.67, p = .21 for unconventional
2
IEDs and for inert IEDs η = .00, F (1, 23) = .02, p = .90
Figure 6 indicates that screeners who trained the most have a
slightly higher improvement in the detection performance than
those who trained the least. Furthermore, the biggest
improvement was found for the types of IEDs which never had
been trained before.

IV. SUMMARY AND CONCLUSIONS
The aim of this study was to investigate the importance of
having different types of IEDs in a computer based x-ray
screening training. This was tested using three types of IEDs,
namely conventional, unconventional and inert IEDs. In both
studies screeners conducted a test before and after several
months of computer based IED training with XRT.
The results of both studies indicate that the detection
performance of the three types of IEDs differs significantly.
Conventional IEDs which were already trained for several

7

V.

performance,” Psychological Bulletin, vol. 102 (3), pp
439-442, 1987.
[12] D. George and B. Watt, SPSS for Windows step by step:
A simple guide and reference, Boston: Allyn and Bacon,
2003.
[13] L. Crocker and J. Algina, Introduction to classical and
modern test theory, Orlando: Holt, Rinehard and
Winston, 1986.

ACKNOWLEDGEMENTS

We are thankful to Zurich Airport, Switzerland and the security
companies at Zurich Airport supporting this research study.

VI. REFERENCES
[1]

A. Schwaninger, “Increasing efficiency in airport security
screening,“ in Proceedings of AVSEC World 2004,
November 3-5, Vancouver, B.C., Canada, 2004.
[2] A. Schwaninge, D. Hardmeier and F. Hofer, “Airport
security screeners visual abilities & visual knowledge
measurement,” IEEE Aerospace and Electronic
Systems, vol. 20(6), pp 29-35, 2005.
[3] S. M. Koller, D. Hardmeier, S. Michel, and A.
Schwaninge,r, “Investigating training, transfer, and
viewpoint effects resulting from recurrent CBT of x-ray
image interpretation,“ Journal of Transportation Security,
vol 1(2), pp 81-106, 2008.
[4] J. D. Smith, J. S. Redford, D. A. Washburn, L. A.
Taglialatela, “Specific-Token Effects in Screening Tasks:
Possible Implications for Aviation Security”, Journal of
Experimental Psychology: Learning, Memory, and
Cognition, vol. 31(6), pp 1171-1185, 2005.
[5] J. S. McCarley, A. F. Kramer, C. D. Wickens, E. D.
Vidoni and W. R. Boot, “Visual skills in airport
screening,“ Psychol Science, vol. 15(5), pp 302–306,
2004.
[6] A. Schwaninger and F. Hofer,.”Evaluation of CBT for
increasing threat detection performance in X-ray
screening,” in: K. Morgan and M. J. Spector, The
Internet Society 2004, Advances in Learning, Commerce
and Security, pp 147-156, Wessex: WIT Press, 2004.
[7] I. Pollack, D. A. Norman, “A non-parametric analysis of
recognition experiments,” Psychonomic Science, vol 1,
pp 125-126, 1964.
[8] A. Schwaninger and A. W. J. Wales, ”One year later:
how screener performance improves in X-ray luggage
search with computer-based training,” in Proceedings of
the Ergonomics Society Annual Conference, 2009, pp
381-389.
[9] A. Schwaninger, S. Michel, A. Bolfing, ”A statistical
approach for image difficulty estimation in x-ray
screening using image measurements,” in Proceedings
of the 4th Symposium on Applied Perception in Graphics
and Visualization, ACM Press, New York, USA, 2007, pp
123-130.
[10] J. B. Grier, ”Nonparametric indexes for sensitivity and
bias: Computing formulas,” Psychological Bulletin, vol.75
(6), pp 424-429, 1971.
[11] D. Aaronson and B. Watt, Extensions of Grier’s
computational formulas for A’ and B’’ to below-chance

VII. VITA
Dr. Diana Hardmeier is the General Manager of CASRA since
2009. Prior to that she worked nearly three years at Zurich
State Police Airport Division and was responsible for the
development, deployment and supervision of quality control
measures for airport security control. This allowed her
combining both the operational and the theoretical
perspectives which she acquired at the University of Zurich.
During her employment at the University of Zurich, Dr.
Hardmeier was project manager of testing and certification
projects in Switzerland, Belgium and Germany. She received
her doctoral degree in 2008.
Moritz Jäger is in his master studies in the Psychological
Department of the University of Zurich. He has practical
experience in aviation security x-ray screening and in the field
of computer-based training and testing of x-ray screeners.
Rebekka Schibli graduated from Zurich University in 2009
with a Bachelor of Natural Science degree. She works for
CASRA since 2010.
Prof. Dr. Adrian Schwaninger lectures at the University of
Zurich since 1999 and at the University of Applied Sciences
Northwestern Switzerland since 2008. He is the head of the
Center for Adaptive Security Research and Applications
(www.casra.ch) in Zurich and the head of the Institute
Humans in complex Systems (MikS) at the School of Applied
Psychology, University of Applied Sciences Northwestern
Switzerland (www.fhnw.ch/miks). His areas of expertise are
aviation security, human factors, scientifically based software
development, applied cognitive psychology, and humanmachine interaction. Prof. Schwaninger is a member of the
ECAC Training Task Force, the ECAC Technical Task Force,
the ICAO Working Group on Training, and he leads the ECAC
Technical Task Force TIP Study Group. Prof. Schwaninger is
recognized as a leading authority of aviation security. He has
more than 70 publications and more than 150 invited
presentations. In 1999 he received the Young Researcher
Award in Psychology. In 2003 he received the ASI
International Award of Excellence in Aviation Security:
Enhancement of Human Factors.

8