SW 5050 Overview of Class 3

1
SW 5050Overview of Class 34

• Intro to Measurement
• Reliability Validity
• Levels of Measurement
• Continue Lab Assignment 1

2
Measurement Defined

• Measurement is the process of describing abstract
  concepts in terms of real world indicators by the
  assignment of numbers or symbols according to
  rules.

3
Measurement

• Measurement is essentially the process of
  operationalizing concepts.
• It involves moving from concepts to variables.

4
Reductionism

• THEORY
• CONCEPTS
• HYPOTHESIS
• VARIABLES

5
Real World Indicators

• An indicator is an observation that is assumed to
  be evidence of the attributes or properties of
  some phenomenon.
• The more abstract a concept, the more indicators
  that are necessary to operationally define the
  concept into a variable.

6
Item vs. Index

• Item A single indicator of a variable.
• Index A series or number of items needed to
  operationalize a variable.

7
Example Complex Concept

• Depression An emotional condition characterized
  by decrease in functional activity (nominal
  definition).
• Indicators
• Excessive Sleeping
• Crying
• Loss of weight
• Loss of appetite

8
Example Complex Concept

• Alcohol Dependence
• Number of drinks per day.
• Time missed from work.
• Daily activities involving alcohol consumption.
• Number of DUIs.

9
Reliability and Validity

• are really about determining whether our real
  world indicators capture the complex concept we
  are trying to measure.

10
RELIABILITY

• Refers to the consistency of the measure.
• If I measure over and over, will I get the same
  results?

11
VALIDITY

• The accuracy of the measure.
• Am I measuring what I think I am measuring?
• Not to be confused with internal or external
  validity.

12
RELIABILITY The Inside Story

• If I measure over and over, will I get the same
  results?
• Applies to both complex and non-complex
  variables. (Attendance?)
• Two Principles
• Stability and Equivalency.

13
Establishing Stability Equivalency
14
TEST-RETEST

• Most commonly used method.
• Application of instrument at two points in time.
• Two week interval recommended.
• Scores generated from two tests should be
  comparable (correlated).
• Correlation of .70-.80 necessary.

15
PARALLEL FORMS

• Two separate but equal (parallel) instruments are
  developed.
• Instruments are combined and administered jointly
  to the same group.
• Results are correlated. (.70-80)
• Feasibility?

16
SPLIT-HALF

• Best approach for addressing equivalency.
• Responses randomly divided into congruent halves.
• Scores from each half are totaled and correlated.
• Cronbach Alpha

17
ValidityThe Inside Story

• Golden Rule You cannot have validity without
  reliability. You can have reliability without
  validity.
• The ultimate validity of a measure can never be
  proven.

• Types of Validity Tests
• Content Validity
• Criterion Validity
• Construct Validity

18
CONTENT VALIDITY

• A.K.A. Face Validity
• Assessment of logical relationship between the
  concept and the proposed measure.
• Weakest form of validity.
• Expert participation Bolstering face validity.

19
CRITERION VALIDITY

• Show a correlation between our measurement device
  and some other criterion.
• Key Find solid criterion for purpose of
  comparision.

• Two Types
• Concurrent Validity
• Predictive Validity

20
CONCURRENT VALIDITY

• Most common approach.
• Comparison to existing (well established)
  instrument.
• Minimum correlation of .50 is needed for
  establishing concurrent validity.

21
Predictive Validity

• An instrument is used to predict some future
  state of affairs or future event.
• Quality of the measure (its validity) is
  determined by how well it predicts the event.
• SAT/GRE??

22
CONSTRUCT VALIDITY

• Most complex form of validity.
• Involves relating a measure to an overall
  theoretical framework.
• Domains underpinning the complex concept must
  be clearly delineated.

23
Additional Issues

• Use of existing instruments.
• Reliable and Valid for whom?
• Creation of instrument??

24
Level of Measurement

• Nominal
• Ordinal
• Interval
• Ratio

25
Nominal Level Variables

• Observations classified into mutually exclusive
  categories of kind or type.
• Examples
• Gender 1) Female 2) Male
• Ethnicity 1) Anglo 2) African American 3)
  Latino 4) Asian American
• Political Affiliation 1) Republican
  2) Democrat 3) Independent

26
Ordinal Level Variables

• Mutually exclusive categories of kind or type.
• Rank-ordered/hierarchy.
• Examples
• SES 1) Upper 2) Middle 3) Lower
• Education 1) College 2) Secondary 3)
  Primary

27
Interval Level Variables

• Mutually Exclusive Categories
• Rank Order/Hierarchy
• Equal Spacing Between Categories
• Examples
• Temperature
• IQ
• Likert Scale

28
Ratio Level Variables

• Mutually Exclusive Categories
• Rank Order
• Equal Spacing
• True Zero
• Example
• Income total monthly salary