Level Of Measurement

Instructional Measurement

1. Research Methods I
Kinds of Data and
Levels of Measurement

2. Review
The Scientific Method – Conducting a
Study
1.Understanding Nature of Problem
Literature Review
Research Question - Hypothesis
2.Test Hypothesis empirically
Deciding on measurements
Data collection
Data analysis
Interpret results and draw conclusions

3. Variables and Measurements
Variables:
Characteristics that can take on different values for
different members of a group
Independent variables
Dependent variables
Construct:
Hypothetical concepts that describe and explain
behavior (e.g. self-esteem)
operational definition of construct
Measurements:
“Assignment of numbers to aspects of objects,
persons or events.”

4. Kinds of Data
Levels of Measurement
Qualitative / Discrete Data
Separate, indivisible categories (e.g. male, female)
Nominal (categorical)
(Ordinal)
Quantitative / Continuous Data
Infinite number of possible values that fall between
two observed values
Interval
Ratio

5. Nominal Level of Measurement
Data in a set of categories that have
different names
It is arbitrary; no logical ordering
Has to do with names
E.g. gender, race, religion, kind of profession
N-category nominal scales
Dichotomies (gender)
Five category (ethnicity: African-American,
Caucasian, Asian, Native American, Hispanic)

6. Ordinal Level of Measurement
Ranked in terms of magnitude
Distances between variables or exact
amount of variables does not have to be
known
In papers grouped ordinal data is often
used (how many people are in each
category

7. Ordinal Level of Measurement –
Likert Scales
Item pool concerning referent in question
Level of agreement to each statement
Average responses to get final score
Logical sequence (order)
May be treated as continuous variables in
analyses even though they are actually
ordinal

8. Interval Level of Measurement
Ordered categories that are all intervals of
exactly the same size
For interval data zero is an arbitrary point
Does not mean the absence of measured
characteristic
Arithmetic operations can be performed with
interval data
There are some limitations

9. Ratio Level of Measurement
Ratio data is like interval data, except
the origin of the scale represents the
absence of the characteristic
measured
Examples of ratio level measurements
are

10. Is our measure valid?
Definition:
Validity describes how well as measure
actually assesses what you want it to
Decide how to measure variables
Describes soundness and appropriateness
of a measure for purpose of study

11. Validity
Content validity: Does measure cover all
different domains of the concept?
Face validity: How is measure viewed by others
as covering the concept?
Sampling-content validity: everything covered?
Criterion validity: How well do measures of
convenience assess criterion of interest
Construct validity: Does the measure assess
underlying theoretical construct?

12. Reliability
To which extent do two sets of measurements of
the same characteristic on the same people
duplicate each other
A reliable measure is free of measurement error
Test-retest reliability (same people, different time)
Inter-rater agreement (same people, same time)
Internal-consistency (consistency of answers across
items)
Problem with measurement error and reliability -
variability