Database Reference
In-Depth Information
weight fluctuated each time, and nothing changed about you between weigh-
ings (that is, you're not munching on Twinkies after each round!), you would
conclude that the scale is
unreliable
.
Validity
, however, refers to whether the measure accurately reflects the object
or phenomenon that you are trying to measure.7 Your scale may tell you each
time that you are 185 pounds; however, in reality you may be pushing 225. In
this case, your scale would not be
valid
.
In Figure 4-22, Target A shows a reliable measure that is not valid—notice how
the blue dots consistently arrive at a similar point, but they are not accurate
and miss the bull's eye. Target B shows a valid measure in that the blue dots,
on average, will arrive at a measure that is at or near the bull's eye; however,
the accuracy is poor, making it unreliable. Target C, however, lacks accuracy
and precision—the dots do not produce a valid result on average, nor do they
consistently arrive at the same point. Finally, Target D shows a valid and reli-
able measure, one that arrives at a precise and consistent location, and one
centered on the bull's eye.
A
Reliable
Not Valid
B
Valid
Not Reliable
C
Neither Reliable
Nor Valid
D
Both Reliable
and Valid
Figure 4-22
Understanding reliability and validity
Source: Juice
In assessing the reliability and validity of the data, it is important to consider
a few threats to each. For reliability, think about the instrument of measure-
ment. Is it a person? A computer? A tool or machine? Or is it a process that
might include each of these? Although there are a number of formal statistical
tests to measure reliability, these questions can be an important first step in
questioning a data's reliability. In addition, processes that repeatedly measure
the object or phenomenon of interest can also help determine the data's reli-
ability. For example, you can never find out if your scale is unreliable based
on a single measurement. Stepping onto the scale repeatedly in a short time
will quickly help you determine if it produces a reliable measure of weight.
The same is true for other forms of measurement, especially those that involve
Search WWH ::
Custom Search