曝光臺 注意防騙 網曝天貓店富美金盛家居專營店坑蒙拐騙欺詐消費者

analysis shows that duration was a more appropriate measure. Something that can
happen with inexperienced observers is that they leave data analysis until
observations are complete, only to discover that the question remains unanswered.
Once preliminary observations are complete, developing a taxonomy is the first step
in systematic observational research (it may even be the entire purpose.) Defining the
behaviours is an essential aspect, but the detail depends on the specific question. The
objective may be a comparison of two or more communication architectures with
respect to qualitative and quantitative aspects of behaviour. Alternatively, it may be
to develop a taxonomy of what exists to assist in the design of an appropriate
decision support aid.
Several ways of describing behaviour may be used in a taxonomy. The physical
description of behaviour may be “molecular”. This includes minute detail of muscle
or skeletal action. Alternatively, it can be “molar”. A molar description of “walk”
DSTO-TR-1034
9
might be a slow, quadrupedal locomotion. Behaviour can also be described in terms
of its consequences. That is, it is defined in terms of effects. For example, “approach”
has the effect of decreasing the distance between two subjects, regardless of how the
action was performed.
A behaviour taxonomy might be restricted to discrete categories of behaviour, or if
detailed sequences are not of interest, some tasks may be recorded as single units (for
example, “issue command”). This compares to recording detail of the actions
involved in issuing the command. The level we record depends on the problem at
hand. Functionally organised taxonomies do have advantages. After experience
develops, observers find that it makes sense to group specific behaviours into higher
order functions, or into factors. This can be done during data analysis, and if there is
enough data, factor analysis can be performed.
3.6 Exhaustive and Mutually Exclusive Recording Categories
For purposes of data recording and analysis, it can be an advantage to define
categories that are both exhaustive and mutually exclusive. Exhaustive means that
the subject is always observed as doing something, even if that something is
“inactive”. Mutually exclusive means that the subject is never recorded as doing
more than one thing simultaneously. For example, they can be “travelling” or
“approaching”, but not both. In “approach”, they are performing both actions, so
recording “travel” is redundant. In such a situation, though, we would still want
data for “travel” if “approach” were not occurring. Consequently, the recording
system needs rules for establishing priorities, such as “approach” over “travel”.
It should be noted that during an observational block, more than one set of mutually
exclusive and exhaustive categories could be included. For example, you can score
one behaviour, one location, and one relation simultaneously. By using clearly
defined collection methods, different behaviours can be analysed separately and
together to determine relationships.
3.7 Recording Behaviour
There are two kinds of events that activate the observer to record a score: a change in
behaviour or the passage of time. A behaviour change scoring system usually
involves recording the onset of a new behaviour, but may also include recording the
termination of the current behaviour. Alternatively, the transition time between two
behaviours may be recorded. Behaviour change scoring is usually associated with
continuous sampling methods. However, for some behaviour, the transition from
one bout to another can be ambiguous. In such cases, the behaviour taxonomy
should include defining events that signal when a new behaviour should be recorded
(eg. a certain number of seconds of inactivity). The second approach to recording
behaviour involves time sampling. At the end of a predetermined time interval, the
observer scores either the behaviour occurring at that point (scan, instantaneous or
point sampling), or scores the occurrence or non-occurrence of each behaviour in the
interval (one-zero sampling). The techniques for recording behaviour are described
in detail in the following section.
DSTO-TR-1034
10
3.8 Sampling Methods
Sampling methods are used to make estimates about the entire population based on
a set of that population. It is just about impossible to observe the entire thing we
seek to understand. Over the years, certain methods of sampling behaviour have
been developed to reduce the possibility of bias. Even though a project might have
　
中國航空網 www.k6050.com
航空翻譯 www.aviation.cn
本文鏈接地址：航空資料29(64)