CAPÍTULO III: MARCO METODOLÓGICO
3.4. Métodos, técnicas e instrumentos
3.4.3. Instrumentos
The reasoning questions are split into three types: case study specific, close transfer
scenarios and far transfer scenarios.
Case Study Specific Reasoning
The first question participants answer is related to the A&E case study. In the
case study resources such as cubicles and doctors are shared between all patients
(pooled resources). This pooling of resources helps speed up a process subject to
variation and also increases resource utilisation. A simple explanation of this is that
resources, if available, can go to where demand is waiting. Participants are asked to
predict what would happen if A&E resources were split up and dedicated to specific
emergency streams (i.e. whether performance increases or decreases) and to state
why this change in performance occurs. Answers are analysed qualitatively. Firstly,
on whether the participant predicts a decrease in performance and secondly whether
the reasoning behind the prediction is correct.
Transfer Scenarios
The transfer scenarios are divided into two groupings: close and far. All four close
transfer scenarios are set in a healthcare context. This closeness is therefore the
surface similarity of the scenario to the case study. In contrast the far transfer scenarios have less surface similarity to the case study, but are stillstructurally sim-
ilar. The far transfer scenarios are either set in call centres or a food manufacturing plant. Each scenario details a problem, provides transfer cues, lists multiple choice
answers and provides space for a qualitative answer. A scenario will look for the
transfer of one out of two concepts from the simulation experiment:
1. Recognition of the relationship between resource utilisation and the speed at
which an entity can travel through a process;
2. Recognition that eliminating variation from a process can increase the speed
at which an entity can travel through a process.
The case study contains numerous examples of these concepts in action. For
example, participants run experiments analysing the allocation of nurses to shifts
and the use of resources to learn about resource utilisation and performance. If
a degree of double-loop learning has occurred the participants should be able to
transfer these concepts beyond the case study to analogous scenarios. Table 4.5
lists the context of the scenarios and the concept that is tested within each. Refer
to Appendix A.3 for details of the scenarios.
Table 4.5: Transfer Scenarios
Scenario Context Reasoning required for transfer success S1 GP’s Surgery Process Variation linked to Performance; S2 A&E department Resource Utilisation and Performance; S3 Operating Theatre Resource Utilisation and Performance; S4 NHS walkin Centre Process Variation linked to Performance; S5 Pie Factory Resource Utilisation and Performance; S6 Police Call Centre Process Variation linked to Performance; S7 Pie Factory Process Variation linked to Performance; S8 Call Centre Resource Utilisation and Performance;
There are two ways to analyse the answers participants provide to the transfer
1. Assume that the reasoning behind the multiple choice answers is known be-
forehand.
2. For those correct answers, assume that a significant proportion of the partici-
pant’s reasoning is captured in the qualitative information provided and that
it is accurate. Interpret if the basic description of reasoning provided by the
participant is correct.
The first of these approaches is problematic as it may be that participants have
chosen the correct answer by chance. The second approach requires multiple judges;
differences in interpretation of answers must also be resolved.
Multiple judges are not available for the interpretation of the reasoning answer.
Instead the data are coded multiple times by the same judge with a time lag in
between each coding. Two codings of the reasoning answers will take place with at
least three months in-between. For acceptability these are expected to have high
reliability score - measured using Cronbach’s α and intra-class correlation (Field,
2009). In addition to minimise any judge bias (as the experimental hypothesis
is known) all participant details are hidden from view and the order of answers
randomised in each coding.
Transfer of learning is coded as a binary variable: zero for correct choice and in-
correct reasoning (transfer failure) and one for correct choice and correct reasoning
(transfer success). As an example of the coding procedure consider the informa-
tion/cues provided to a participant by scenario two:
1. A hospital is going to experience an increase in demand (amount unspecified).
2. Current average utilisation of human resources is approximately 75%.
3. Current average utilisation of cubicles is 60%.
The performance target and percentages provide the cues to recall the perfor-
mance of the simulation model. The participants are asked if they agree that more
staff should be introduced. All participants are asked to give the reasons why they
think their choice would improve the performance of the system. An example answer
provided by a participant is
‘As the workload is going to increase, it is important to increase
the number of staff. Depending on increase in workload the number of
cubicles may be necessary to increase, but is unlikely as their utilisation
is currently lower than staff utilisation.’
The second sentence provides the most detail on the participants reasoning.
The participant is thinking in terms of maximum capacity. If the resources are
working at less than 100% then they can cope. If 100% is exceeded then more are
necessary to serve them. Whilst this seems sensible it fails to transfer any learning
from the experiment. An example of the transfer we are looking for would discuss
performance levels and the relationship to utilisation: even if utilisation is increased
to 90% there will be a change in the performance of the system. Hence this answer
is coded as a zero - correct choice, but failure to transfer learning.
Confidence in Reasoning
Measurement of the confidence participants hold in their answers to the transfer
scenarios is, again, based on the scales used by (Petty et al., 2002). At the bottom
of each transfer scenario participants are asked to rate the confidence they have
in their answer on a nine-point scale (1 = not at all confident to 9 = extremely