Tip:
Highlight text to annotate it
X
There are four common levels of measurement, also known as scale measurement, that we see
in statistics. These include nominal, ordinal, interval, and ratio scales. Each level of
measurement providers the user access to different types of data, some stronger than others.
It also has an impact on the types of statistical tests that can be performed, and subsequently
the conclusions that can be drawn as well.
The most basic and least powerful level of measurement is the nominal scale. With the
nominal scale question, respondents are only able to select a descriptor as a response.
As a result, data obtained through these types of questions is restricted to being classified
and counted. Examples of nominal scales include asking respondents to indicate their marital
status. Available options often include married, single, separated, divorced, and widowed.
Since these are qualitative variables we know that we're fairly limited in terms of the
types of tests that can be performed. For example, we can't average the results. So
we can really only tally the number of responses into mutually exclusive categories and provide
overall results. Another limitation of nominal scales is we can't determine the level of
intensity between responses. For example, a question asking respondents whether they
like or dislike a certain type of music fails to establish the intensity to which respondents
like or dislike a particular genre genre of music.
A more powerful level of measurement is the ordinal scale. The ordinal scale gives respondents
the opportunity to express some relative level of intensity between responses. Responses
obtained through an ordinal scale can be ranked and allow the user to determine both the median
and mode in their analysis. Restaurants and other businesses commonly ask patrons to rate
the level of service that they received during a recent visit. Available options may include
excellent, good, average, poor and inferior. So respondents can select from one of the
available options, which allows restaurants the opportunity to understand customer perceptions
with regards to service quality. But although users can determine how many respondents rated
the service quality as superior as opposed to just good, they can't determine how much
better a superior rating is over a good rating. We know that it's obviously better, but we
can't determine the level of difference between the two ratings.
The third level of measurement is the interval scale. The interval scale allows the user
to determine preference or order in responses just like the ordinal scale, but it also establishes
an absolute difference between each available option. Questions that use the interval scale
allow the user to determine not only the mode and median, but also the means and the standard
deviation of the responses. An example of an interval scale is the commonly used Likert
scale where respondents may assess a variable along a five, seven, and even ten point scale.
Take our previous example related to restaurant service quality. Using an interval scale,
we could ask respondents to rank the service they received on a seven point scale ranging
from 1, which represents not satisfied at all, to a 7, which represents extremely satisfied.
By using this type of scale, we can not only determine which responses receive a higher
rating than others, but also the level of magnitude between those responses.
The fourth and final level of measurement is the ratio scale. The ratio scale adds to
the interval scale an absolute zero. The fact that zero is an available option, and is more
importantly that it is meaningful, means that the ratio between two numbers is significant.
Although an interval scale can provide the absolute difference between each scale point,
like from extremely satisfied to somewhat satisfied, the ratio scale provides an absolute
comparison between the responses themselves. For example, a question asking respondents
how many times they have exercised during the past seven days is a ratio scale. In addition
to being able to provide a numerical value, respondents can also list zero as their response.
This zero has a great deal of significance, as it indicates they did not exercise at all
during a seven day period. Furthermore, we know that someone who exercised twice in a
seven day period, was twice as active as someone you exercised once. So in addition to being
able to determine the mean, median, mode, and standard deviation, users can also make
comparisons between responses.
Thank you for watching this video on the levels of measurement. For questions please leave
them in the comment box below and I'll do my best to get back to them in a timely fashion.
If you want more from Alanis Business Academy, you can subscribe to our channel or go to
alanisbusinessacademy.com where you'll find additional content, quizzes, and more. Thanks
for watching.