Professional Documents
Culture Documents
Week 3
Week 3
A variable is an aspect of a testing condition that can change or take on different characteristics with
different conditions.
Types of Variables
Measurement
The ability to state laws quantitatively means two things are true:
What is Measurement?
- The process of assigning numbers to events or objects according to rules. The properties
of the events are represented by the properties of the number system. An example
would be assigning numbers to people according to a rule.
1. Nominal Scales
- A measure that divides objects or events into categories according to their similarities or
differences.
- The rule is that objects or events of the same kind get the same number and objects or
events of a different kind get a different number.
2. Ordinal Scales
- A measure that both assigns objects or events a name and arranges them in order of
their magnitude.
3. Interval Scales
- A measure in which the differences between numbers are meaningful; includes both
nominal and ordinal information.
4. Ratio Scales
- A measure having a meaningful zero point as well as all of the nominal, ordinal, and
interval properties. (should have a true zero point)
- Nominal scale gives information only about whether two events are the same or
different.
- Ordinal scale does the same, but gives us a ranking on a certain variable.
- Interval scale conveys both nominal and ordinal scale information, and allows us to make
quantitative statements about the differences between events.
- Ratio scale contains all the information of the other three scales, as well as conveying
information about ratios of magnitudes.
- The ways we can alter the assignment of numbers to individual events without distorting
the scale are called permissible transformations.
-
-
-
-
Measurement and Statistics
The scale on which a variable is measured determines the type of statistics that can appropriately be
performed on the data. Many hold that it is not appropriate to use the usual parametric statistics the
data are measured on an interval or ratio scale.
- Reliability – the property of consistency of a measurement that gives the same result on
different occasions.
- Validity – the property of measurement that tests what it is supposed to test.
- A measurement should be reliable before it can be valid.
Validity of Measurement
- Variability and Error – The task of research is to find relationships between independent
and dependent variables, to find how the dependent variable changes with changes in
the independent variable. So variability in the dependent variable is good when it is
associated with changes in the independent variable. A certain bad variability is called
error variance, or random error (variability in the dependent variable that is not
associated with the independent variable).
- Types of Validity of Measurements:
1. Construct Validity – a test that the measurements actually measure the constructs
they are designed to measure, but no others.
o Several ways to determine whether a test yields data that have construct
validity:
a. The test should actually measure whatever theoretical construct it tests.
b. The test should measure what it intends to measure but not measure
theoretically unrelated constructs.
c. The test should prove useful in predicting results related tot the theorical
concept it is measuring.
2. Face Validity – an idea that a test should appear superficially to test what it is
supposed to test. (a test should appear to any person to be a test of what it is
supposed to test).
3. Content Validity – the idea that a test should sample the range of behaviour
represented by the theoretical concept being tested.
4. Criterion Validity – the idea that a test should correlate with other measures of the
same theoretical construct.
1. Test-retest Reliability – the degree to which the same test score would be obtained on
another occasion.
2. Internal Consistency – the degree to which the various items on a test are measures of the
same thing. Split-half reliability is determined when the items on a test are divided into two
sets as if they were two separate tests. Then the scores on the two halves are correlated to
see how closely the various individuals’ scores agree on the two halves. If the test is a good
test, it will have a high split-half correlation.