2.1. The relationship between characteristic
and variable
2.1.1. Ensuring a valid relationship between
a characteristic and a variable
For measurements on a variable to be valid, they
must be:
correct, and
relevant to the issue being studied.
Let’s consider an example. In 1996, the Australian Bureau
of Statistics (ABS) conducted a study to assess the literacy skills
of Australians. The ABS decided that a measure of literacy would
be obtained by asking participants to complete some simple literacy-related
tasks. These consisted of "using a bus timetable, working out
the cost of discounted items 'on sale', and following a manufacturer's
warranty instructions for a refrigerator."[2]
The survey was designed to assess 'prose literacy', 'document literacy'
and 'quantitative literacy', as shown in the table below.
The ABS did not attempt to measure every aspect of the characteristic
‘literacy’. They clearly identified the sub-characteristics
of literacy in which they were interested, and then defined the
aspects that they were attempting to measure. The survey indicated
that over 40 percent of Australians would be expected to experience
some difficulty making sense of printed materials.
SCENARIO
Upon reading the ABS report, a reporter for a newspaper
commented that the study showed that over 40 percent
of Australians could not read or write. Explain if this
was an accurate or inaccurate interpretation of the
data.
Are the variables selected valid measures of the characteristic?
Did the variables identified by the ABS represent valid
measures of the characteristic? According to the ABS this 'task-based'
method for assessing literacy is valid because it has been developed
and tested in other countries and then modified for the Australian
context. It was also trialled in Australia before being used more
widely. Such statements serve to support their argument for the
validity of the variables that they used to measure aspects of literacy.
If the characteristic is well understood, then it is usually easier
to identify a variable that is valid. For example, if you wanted
to measure height, then you could use an instrument like a measuring
tape and measure people's height in centimetres. Most people would
accept your variable as a valid measure of height. The accuracy
of your measure would, of course, depend on the accuracy of your
measuring tape and on how it was used to measure heights.
