MEASUREMENT

1
MEASUREMENT

• ASSIGNING NUMBERS TO
• CHARACTERISTICS OF PEOPLE
• ACCORDING TO A SET OF RULES

2
How satisfied are you with?1. Very dissatisfied
2. Dissatisfied3. Satisfied 4. Very satisfied

• LIVING SITUATION?
• WITH YOUR COURSES THIS SEMESTER?
• RELATIONSHIPS WITH YOUR FAMILY?
• WITH THE WAY YOU LOOK?

• WITH YOUR MONEY SITUATION?
• WITH THE NUMBER OF FRIENDS YOU HAVE?
  
• WITH THE AMOUNT OF FREE TIME YOU HAVE?

3
Measurement

• Involves the process of operationalizing a
  variable and measuring it

4
Level of Measurement

• Numbers that you assign to attributes of variable
  have certain properties which determines what
  mathematical manipulations are permissible.

5
LEVELS OF MEASUREMENT

• Nominal
• use numbers to name categories
• have no meaning other than a shorthand for
  letting you know which category person assigned
  to.

6
Gender 1. Male 2. FemaleReligion 1.
Catholic 2. Protestant 3. OtherYour student
identification number.
7
Ordinal

• 1. Use numbers to name categories
• 2. Use numbers to order categories

8
Year in Program

• 1. MSW first year
• 2. MSW second year

9
How much do you like the course?

• 1. Not at all
• 2. A little
• 3. Some
• 4. A lot

10
Interval

• 1. Use numbers to name categories
• 2. Use numbers to order individuals
• 3. Use numbers to indicate equivalent distance
  between intervals

11
IQ Scores

• Difference in intelligence between two people
  with IQs of 100 and 110 is same as difference in
  intelligence between two people
• with IQs of 90 and 100

12
Interval Level

• Temperature Degree Fahrenheit
• 32 degrees to 42 degrees
• same distance as
• 72 degrees to 82 degrees

13
Ratio

• 1. Use numbers to name categories
• 2. Use numbers to order categories
• 3. Use numbers to indicate intervals between
  categories is equivalent
• 4. Use numbers to indicate ratio between
• scores.

14
Ratio Scale

• Have a meaningful zero point.
• Absence of the trait

15
Ratio Level Variables

• age
• education
• number of group sessions attended
• number of hospitalizations

16
Score on Measure/ScaleConsist of Two Parts

• True Score
• Error-- that part of score that does not reflect
  the underlying variable that we are trying to
  measure

17
Two types of Error

• Systematic
• Random

18
Test score true ability systematic error
random error

• 85 89 -5 1
• 7587 -2 -10

19
SYSTEMATIC ERROR

• Errors in the measuring instrument that are
  repeatable and consistent from one setting or
  test taking situation to the next

20
Sources of Systematic Error

• Way the question is asked (questionnaire bias)
• Who is asking question (interviewer bias)
• Individual who is answering question (response
  sets)
• Yea say - Respond randomly
• Central tendency - Lie
• Social desirability

21
Systematic ErrorDue to Way Ask Question

• Do you support the killing of babies?
• YES
• NO

22
Do you support a womans right to choose?Yes NO
23
Systematic Error Due to Who Is Asking Question

• Interviewer bias

24
Systematic Error due to Dynamics of Those
Providing the Data

• Acquiescent Response set agree/disagree
  regardless of content of statement
• Central tendency
• Test Anxiety
• Social Desirability
• responses that indicate or imply approval of
  actions that are generally considered good.

25
Social Desirability

• Tendency of people to say or do things that will
  make them or their reference group look good.

26
Social Desirability

• I have never intensely disliked someone.
• I sometimes feel resentful when I dont get my
  way.
• I sometimes try to get even, rather than forgive
  and forget.
• I never resent being asked to return a favor.
• I have almost never felt the urge to tell people
  off.

27
Random Error

• Those errors that are neither consistent or
  patterned.
• They are random. Unpredictable from one test
  taking situation to the next.

28
Random Errors

• Fluctuations in Mood
• Read items carelessly.
• Guessing -- dont have required knowledge
• Telephone rings during exam momentarily lose
  your concentration.

29
Florida Vote

• Error
• Original Bush 1800 or so odd votes
• Various recounts 900, 600, 400
• Systematic or random error?

30
How to evaluate the Goodness or Quality of a Scale

• Self-esteem
• Marital satisfaction

31
Two Criteria1. Reliability2. Validity
32
Reliability

• Refers to a measures ability to yield consistent
  results each time it is applied.

33
Reliability

• An assessment of how much random error in
  measuring instrument

34
ReliabilityTwo Principles

• Equivalence -- all the items that make up a
  measuring instrument should be measuring the same
  thing and thus responses should be consistent
  from one item to the next.
• Stability -- a reliable measure should not change
  from one application to the next.

35
Reliability -- Equivalence
36
Reliability -- Equivalence
37
Optimism Scale

• 1. Expect best to happen
• 2. Will go wrong for me
• 3.Look on bright side
• 4. Optimistic about future

• 5. Hardly expect things to go my way
• 6. Never work out way want them
• 7. Every cloud has a silver lining
• 8. Rarely count on good things happening

38
Stabilility

• A reliable measure should not change form one
  application to the next

39
Weight Take One Day Apart
40
Weight Taken One Day Apart
41
Types of Reliability

• Internal Consistency (Equivalence)
• Idea that items in your instrument are all
  measures of the same variable (construct).
• If items are all measuring the same construct,
  they (the items) should be highly correlated with
  one another.

42
Cronbachs alpha

• Measure of the internal consistency of the items.
• Cronbachs alpha tells you how much random error
  you have.

43

• If Cronbachs alpha is .90, means that 10 of
  score is random error.
• If Cronbachs alpha is .40, means that 60 of
  score is random error.
• As a general rule, want Cronbachs alpha to be at
  least .70 and higher.

44
Other Types of Internal Consistency Reliability

• Split Half
• Parallel forms

45
Split halfOptimism Scale
46
Test-Retest Reliability (Stability)

• Administer a test at one point in time and then
  re-administer it at a second point in time.
• Assuming nothing in your life has changed which
  would influence test scores, the scores on the
  two test-taking occasions should be close to one
  another.

47
Reliability of Raters

• Interrater Agreement
• or
• Interrater Reliability

48
(No Transcript)
49
Percent Agreement

• Percent agreed/total of subjects rated X 100
• 8/10 X 100 80

50
Kappa Coefficient

• A coefficient of agreement.
• Takes into account chance agreement.

51
Kappa always equal to or lessthan percent
agreement.
52
Validity

• Is our scale measuring what we think it is
  measuring?

53
Depression

• During the last week, how any days have you felt
• sad
• blue
• tearful
• no interest in things

54
Four Types of Validity

• Face
• Content
• Criterion
• Construct

55
Face Validity

• Do the items look like the variable/construct
  that you say you are measuring?

56
Vocabulary Test

• 5 8
• 20/4
• X 519.
• Log X10.

57
Research Methods Test

• List four symptoms of major depressive disorder.
• How is a schizophrenia different from a
  schizoaffective disorder?

58

• The degree to which my spouse and I share common
  interests.
• The amount of consideration shown by my spouse.
• How my spouse reacts when I share feelings.
• The frequency of sexual or other physically
  intimate relations with my spouse.

59
Content Validity

• Are the items a fair representation of the
  population of items that could be used to measure
  the construct.

60
Midterm Exam

• 50 Questions
• Exam to cover all material up through
  reliability.
• All 50 questions are on reliability

61
Satisfaction with School

• Relationships with professors
• Access to schools
• Courses
• Field placement

• Access to schools resources
• Advising

62
Criterion Related Validity

• Does the measure (that you are trying to
  validate) correlate with some outside CRITERION
  that is known or believed to measure this
  construct/variable?

63
SAT and Success in College

• Does SAT scores predict
• success in college?
• Criterion success in college

64
Does scale for depression correlate with
psychiatric diagnosis of depression?

• Criterion psychiatric diagnosis of depression

65
Criterion Validity

• The criterion may be a
• well-validated scale.

66
Marital Satisfaction Questionnaire for Older
Persons correlate with the Locke-Wallace

• Criterion Locke-Wallace

67
Test Bias

• Test does better job of predicting success for
  one demographic/cultural group than another.

68
Construct Validity

• To what extent does the measuring instrument act
  the way it is suppose to act.

69
Construct Validity

• One of hypothesis testing.
• To the extent hypotheses are supported, have
  evidence for construct validity.
• Construct validity is evaluated by the
  accumulation of evidence.