finish the The SWOT Analysis Activity(1) (1)
xiabuantjngo1
measurement.pptxMeasurement
Unit 4
There are several related topics in this unit…
Defining Measurement
Levels of Measurement
Specific Measurement Techniques
General Measurement Concerns
Defining Measurement Measurement
Measurement is the process of determining the existence,
characteristics, size, and/or quantity of some variable through
systematic recording and organizing of observations.
That is, the specific scheme for assigning numbers or symbols to
designate characteristics of a variable.
Measurement is the difference between saying…
“the box is heavy”
and “the box weighs 80 lbs”.
Levels of Measurement Measurement
There are four levels of measurement…
Nominal
Ordinal
Interval
Ratio
Nominal Level of Measurement
Nominal level of measurement is a measurement scale that classifies a variable into
qualitatively different categories. It does not take into account amounts, but rather only
differences in type.
We use the nominal level of measurement therefore simply to classify or categorize
objects or observations according to certain characteristics.
For example, we may differentiate people as male or female or as management or
nonmanagement employees. We may differentiate conversations as online or face-to-
face. We may differentiate people’s political affiliations as Democrat or Republican.
We may differentiate cultures as Eastern or Western.
All of these classify according to some categorization by type
male/female categorizes by biological sex
management/nonmanagement categorizes by organizational role
online/face-to-face categorizes by location of the interaction
Democrat/Republican categorizes by political affiliation
Eastern/Western categorizes by features of culture
Nominal Level of Measurement
In order for nominal level of measurement to be effective the
categories used must be:
mutually exclusive
exhaustive
equivalent
Mutually exclusive refers to the idea that the categories are
independent and separate from one another. It is a fancy way of
saying that they should not overlap. So, for example, we may
classify cars as compact, mid-size, large, and economical. We
would have trouble with this because economical likely will
overlap with compact. So these categories are not mutually
exclusive.
Nominal Level of Measurement
Exhaustive means that the categories exhaust all of the possible categories that should
be used. We do not always have every possible category represented, but can make our
nominal level of measurement exhaustive by adding an “other” category, which will
capture observations that do not fit into any of the existing categories.
So, for example, compact, mid-size, and large cars may not be an exhaustive list, but by
adding an “other” category it will be. Someone who drives a van will choose “other”.
Finally, equivalent means the categories are measuring or getting at the same idea.
With regard to cars our example from earlier works here too. Economical speaks to
how fuel efficient a car is, which is related to the vehicle’s size, but is not the same as a
reference to the vehicle’s size (e.g., compact, mid-size, large). Thus, not only do we
have a problem with the categories being mutually exclusive, but also with them not
being equivalent.
Ordinal Level of Measurement
Ordinal level of measurement classifies a variable into nominal categories, but also rank orders
those categories along some “greater than” and “less than” dimension. Thus, it serves to provide a
rank ordering of observations relative to one another. It does not however assume that each of
the ranked observations are equidistant with regard to how much greater or lesser they are from
one another.
For example, in a horse race the first horse could win be 3 lengths, the second by a nose. The
distance between the first and second horse (3 lengths) is considerably greater than the distance
between the second and third horse (a nose).
Similarly, the number one ranked college football team may be much better than then number two
team, which is only marginally better than the third team.
Thus, the ordinal level of measurement provides an overall ranking of objects relative to one
another, but does not give us the magnitude of those differences.
Ordinal level of measurement also would include:
a list of consumers’ top rated video games
a ranking of the most gas efficient vehicles
a music critic’s list of the top 10 albums of all time
Interval Level of Measurement
Interval level of measurement categorizes a variable by rank ordering along some
dimension that has assumed equivalence (i.e., distances between points on the scale are
assumed to be equidistant from one another).
Thus, the interval level of measurement specifies the magnitude of differences among
observations by using a numerical scaling that has equidistant points. A classic example
is temperature expressed as either Fahrenheit or Celsius. In both cases each degree of
temperature represents the same amount of increase or decrease in temperature as any
other degree.
Additionally, interval level of measurement contains an “arbitrary zero”, meaning that
the zero is just another point on the continuum and not an absolute. So, zero degrees
Fahrenheit or Celsius both represent very cold temperatures, but not the point at which
temperature ceases to exist.
Interval level of measurement often appears in traditional forms of scaling such as 1
through 5 with values assigned that range from strongly disagree to strongly agree.
Ratio Level of Measurement
Ratio level of measurement carries all of the characteristics of the interval level of measurement
(use of numbers to indicate a rank ordering with assumed equivalence), but adds an absolute zero.
An absolute zero exists when zero represents the absence of value or no observable amount of the
variable of interest. So, for example, money would be ratio because you could report not having
any money. In this case, saying you have zero money would mean you do not possess any money
at all.
To best understand the difference between ratio and interval levels of measurement we need to
simply ask if there is a zero used in the scaling and if so is it an absolute zero (representing none
of the variable of interest) or an arbitrary zero (just another point on the continuum).
So, for example, a researcher may ask people to report about the percentage of time they spend
“chatting on line” using a scale from 0 to 100. In this case, 0 percent would equate to not spending
any time doing this (an absolute zero). Or a researcher may ask people to rate how likely they are
to meet people online with a scale that ranges from 0 to 10 with 0 being very unlikely and 10
being very likely. In this case the zero is not absolute because it represents “very unlikely” instead
of no chance at all. The first example above involves ratio level of measurement, the second
interval level of measurement.
Exercise Level of Measurement
Let’s try an exercise to help you better understand the levels of measurement. What is the level of
measurement implicit in the following examples?
1. Ranking of college football teams
2. Money in English Pounds
3. 5-point Likert type scale that measures employee commitment that ranges from (0) not very committed to highly committed (5)
4. Temperature in Fahrenheit
5. The amount of miles a long-haul trucker has driven in the past week
6. Makes of mountain bikes
7. Top 10 consumer-rated sport utility vehicles
8. 14-point Likert type scale that measures parental satisfaction with television programming that ranges from very dissatisfied (-7) to very satisfied (+7)
9. A 5-point Likert type scale that measures how often people have visited foreign countries that ranges from (0) never to (5) very often
10. The types of computers students use to complete assignments
Exercise Level of Measurement
1. Ranking of college football teams
This involves a ranking, must be…
Ordinal level of measurement.
2. Money in English Pounds
Can you have no money in English pounds? Yes, then you have an absolute zero. Also, equidistant points, one pence is the same difference from 2 pence as 2 pence is from 3 pence.
Correct answer: Ratio level of measurement.
3. 5-point Likert-type scale that measures employee commitment that ranges from (0) not very committed to (5) highly committed.
Is this zero absolute or arbitrary? It has a value assigned to it that is something other than “none” and therefore it is an arbitrary zero and the correct answer is…
Interval level of measurement.
Exercise Level of Measurement
4. Temperature in Fahrenheit
There’s a zero here, but what type of zero is it? Not an absolute zero as it simply represents another point on the scaling, which makes this…?
Interval level of measurement.
5. The amount of miles a long-haul trucker has driven in the past week
An absolute zero? Yes, the trucker could have driven 0 or no miles. Also miles are
equidistant from one another right?
Thus, Ratio level of measurement is the correct answer here.
6. Makes of mountain bikes
Not ranked, just differentiated by type. That would be…
Nominal level of measurement.
Exercise Level of Measurement
7. Top 10 consumer-rated sport utility vehicles.
Now we have something that is ranked, given a greater than/less than characteristic, but not assumed to be equivalent. So this is…
Ordinal level of measurement.
8. 14-point Likert type scale that measures parental satisfaction with television programming that ranges from very dissatisfied (-7) to very satisfied (+7)
Equidistant points, a zero, but what type of zero? Absolute or not? 0 here seems to simply be a point on the continuum, it is not the absence of satisfaction is it? Thus, not an absolute zero and the appropriate level of measurement is…
Interval level of measurement.
Exercise Level of Measurement
9. A 5-point Likert type scale that measures how often people have visited foreign countries that ranges from (0) never to (5) very often
Absolute zero? 0 here means you have not visited foreign countries, absence of the value right? Therefore this is…
Ratio level of measurement.
10. The types of computers students use to complete assignments
Simply differentiated by? Categories, right?
Nominal level of measurement.
Specific Measurement Techniques Measurement
There are several measurement techniques, the most
prevalent and popular include…
Semantic Differential
Likert
Likert-Type
Semantic Differential Measurement
The semantic differential measurement technique involves providing people with
carefully chosen words that are polar opposites of one another and having them choose
a point between those opposite ideas.
For example, the Source Credibility Scale requires respondents to choose between the
“anchors” below when rating the degree to which they find a given speaker to be
credible:
reliable ____ ____ ____ ____ ____ unreliable
uniformed ____ ____ ____ ____ ____ informed
unqualified ____ ____ ____ ____ ____ qualified
poised ____ ____ ____ ____ ____ nervous
dishonest ____ ____ ____ ____ ____ honest
Semantic differential measurement only works well when in fact people perceive the
anchors provided to be polar opposites. For example, happy/sad would work well,
while happy/edgy wouldn’t as these are not regularly seen as polar opposites and
therefore would confuse respondents.
Likert Measurement
Likert scaling is a type of interval measurement that asks
respondents to indicate the degree to which they agree/disagree in
response to a set of items using a 5-point scale that ranges from
“strongly agree” to “strongly disagree”.
For example, the following items use Likert scaling.
I am satisfied with the communication climate where I work.
I communicate well with members of my family.
1 2 3 4 5
strongly disagree neutral agree strongly
disagree agree
Likert-Type Measurement
Likert-type measurement or scaling is an adaptation of Likert measurement/scaling. The adaptation can occur
in one of two ways.
1. We can use different anchors.
We may use anchors that reflect how often someone does something.
For example, the item “I text my friends” may provide the following anchors…
1 = infrequently 4 = frequently
2 = sometimes 5 = always
3 = regularly
2. We can use a different range of numerical indicators.
So, for example, we may use a 7-point or a 3-point scale instead of the traditional 5 points.
While Likert can only ever occur at the interval level of measurement because it does not include an absolute
zero, Likert-type allows us to use ratio level of measurement when necessary. For example, we may be
interested in how many foreign friends someone communicates with on a regular basis. We would want an
absolute zero for measuring this variable because it is quite possible that someone will not communicate with
any foreign friends.
General Measurement Concerns Measurement
There are several general measurement concerns
including…
Dimensionality of the Concept
Types of Questions
Types of Observation
Dimensionality of the Concept Measurement
Concepts of interest, those that we want to study, can be either
unidimensional or multidimensional.
Unidimensional Concepts
A concept that can be measured with multiple items that all add together to form a composite or overall score. So, even though there are multiple items being used the concept only ends up being represented and discussed as a single idea.
The Intercultural Willingness to Communicate Scale, for example, is a 6-item scale that measures a single concept—the degree to which we are comfortable interacting with people who we perceive to be different culturally from us.
Dimensionality of the Concept Measurement
Multidimensional Concepts
A concept measured by a series of scale items that assess more than one factor or idea. The factors are relatively independent of one another so that people’s scores on a measure of one factor are not necessarily related to their scores on another factor.
For instance, the Organizational Dissent Scale measures how often people express dissent to…
their supervisors (upward dissent)
their coworkers (lateral dissent)
to family and friends outside of work (displaced dissent).
The measurement of employee dissent, then, is multidimensional as the
concept occurs in different forms or along different dimensions.
Types of Questions Measurement
The types of questions we ask relate to both the level of measurement and the
type of data or observations we will collect.
Directive/Closed-Ended Questions
Provide respondents with a set of choices or responses, thereby limiting the way in which they respond. Directive in the sense that they direct respondents by providing particular response choices, closed-ended in the sense that respondents do not have the opportunity to answer outside of the choices provided.
Include:
All Likert, Likert-type, and semantic differential measures
Items like: What is your sex (check one): ____ male ____ female
Usually produce quantitative data/observations.
Types of Questions Measurement
Nondirectional/Open-Ended Questions
Questions that neither direct nor provide responses to participants. These questions do not have a clear set of responses. Rather, we phrase them so that participants can report about the phenomenon of interest in their own words.
Allows respondents to be as descriptive as they would like to be.
For example,
“Think of a time when you felt it was necessary to go around your boss to express a concern you had about a workplace issue. As best you can please describe what you did in this situation?”
Produces qualitative data.
Types of Observation Measurement
Finally, we may choose to “measure” an idea by observing it in some form.
Observation can be either direct or indirect.
Direct Observation
Involves watching people as they engage in actual communication behavior. That is, examining communication as it unfolds between people in real time.
So, for example, we could go to court and observe how people respond to questions on the witness stand. Or we could ask people to come to a lab and have them interact with someone of the opposite political affiliation to see how they engage in political conversations.
In both cases, whether it be in a natural setting or one that we have created as the researcher, we are conducting direct observation as we’re watching people interact in real time.
Types of Observation Measurement
Indirect Observation
This type of observation involves examining “artifacts” (i.e., something left over) from communication interactions. We commonly refer to these as communication “texts”. They need not be “texts” in that traditional sense, something written down in a book, although they certainly can be.
Texts also include any communicative medium that produces an artifact, such as:
videotapes audiotapes
web pages emails
minutes from meetings company memos
company profiles/reports mission statements
