close

Вход

Забыли?

вход по аккаунту

код для вставкиСкачать
MEASUREMENT
Research Methods
University of Massachusetts at Boston
©2011 William Holmes
1
WHAT IS MEASUREMENT?
• Connecting variables with
empirical data
• Assessment of presence of
characteristic (classification)
• Assessment of how much of a
characteristic is present
(amount)
2
BASIS OF MEASUREMENT
•
•
•
•
Definition of variables
Clarity of definitions
Examples of variables
Procedures for classification or
assessing amount of a
characteristic
3
WAYS OF DEFINING VARIABLES
•
•
•
•
Written definitions (Nominal)
Examples (Epistemic)
Theoretical (Nomothetic)
Procedural (Operational)
4
HOW TO CLARIFY DEFINITIONS
• Focus your definition.
• Have it refer to only one
thing.
• Identify what it is not.
• Give examples
• Provide some framework
for variable
5
PROCEDURES FOR MEASUREMENT
• Decide whether you want to measure
the presence of something, different
types of something, or amounts of
something
• Decide on the Level of Measurement
• Be sure categories are exhaustive,
exclusive, and unidimensional
6
MEASURE PRESENCE WHEN
• The variable only refers to
presence or absence of a single
characteristic, or
• The variable refers to a
dichotomy of characteristics, or
• The variable provides categories
7
MEASURE TYPES WHEN
• The variable refers to a
classification, or
• The variable refers to a discrete
list, or
• The variable refers to categories
of something, or
• The variable is a typology
8
MEASURE AMOUNT WHEN
• The variable counts something, or
• The variable refers to a length of
time, or
• The variable refers to the amount of
something, or
• The variable refers to the degree of
something
9
LEVELS OF MEASUREMENT
• Nominal/Categorical—presence of
characteristics
• Ordinal—ranks of characteristics
• Interval—equal distance between
ranks
• Ratio—equal distance and
meaningful zero point
10
ASSESING RELIABILITY
•
•
•
•
Test-Retest Procedure
Split Halves Procedure
Inter-Item Correlation Procedure
Inter-Rater Procedure
11
ASSESSING VALIDITY
•
•
•
•
Face Validity Method
Construct Method
Criterion Method
Predictive Method
12
RECOGNIZING GOOD MEASURES: 1
• Categories are exclusive,
don’t overlap
• Categories are exhaustive,
cover all possibilities
• Categories are
unidimensional, refer to
only one thing
13
RECOGNIZING GOOD MEASURES: 2
• Categories are clear, easy
to understand
• Evidence of reliability is
offered
• Evidence of validity is
offered
14
1/--страниц
Пожаловаться на содержимое документа