Lecture 1 EPSY 625

1
Lecture 1EPSY 625
2
MEASUREMENT THEORY TIMELINE
Chinese Greek European Stat- Measure- Exams s
ystem univ. exam istics meant
2000BCE 400BCE 800AD 1800AD 1900AD 2000AD
3
1000 BC Chinese civil service exam

• Democratic in principle anyone could sit for the
  exam
• limitation opportunity to learn access to
  schooling low for peasants
• many modern issues
• reliability test-retest, interrater agreement
• validity content, face, predictive
• test formats written vs. oral

4
GREEK-ROMAN EDUCATION SYSTEMS

• Greek emphasis on oral examination
• Roman emulation of Greek systems (used Greek
  slaves as teachers for private tutoring)
• Use of papyrus for written work
• Carryover to European university systems created
  about 800 AD- 1800
• Oral examinations for degrees- paper was very
  expensive
• Arabic university systems somewhat like European
  systems in 1000-1500AD

5
1800 Error of Measurement - astronomy gambling

• Discovery of Normal distribution
• scoring agreement among astronomers
  disagreements as poor work vs. individual
  differences
• gambling with dice or cards statistics and
  probability were invented- gambling noblemen
  wanted an edge

6
1850 Wundt's lab - physiological measurement

• Measurement of human performance and perception
• jnd just noticeable difference in perception

7
WRITTEN EXAMINATIONS 1850

• Cheap paper and lead pencil invented in 1840s
• Written essay advocated by American educators
  such as Horace Mann
• Reduction in emphasis on oral presentation,
  recitation
• Essay examination becomes important

8
1880 Galton's anthropometry human physical
measurement

• Measurement of human physiognomy
• notion of relationships among variables
• correlation and causation
• ratio scaling led to Pearsons correlation
• nominal scaling led to chi square association

9
1890 reaction time, "mental test"

• J.M. Cattell - child intelligence
• Binet - Simon - individual testing
• developmental perspective
• deviation at an age compared to cross-age
  development
• factor analysis - Spearman concept of construct
  underlying observed measurement

10
1905 educational psychology - Thorndike

• learning and assessment
• First educational psychology text- 1906
• 1914 text topics
• The Original Nature of Man
• The Psychology of Learning
• Individual Differences and Their Causes
• sex and race
• ancestry (genetics) or family etc.

11
1920s Group testing

• IQ testing from WW I
• need to classify men efficiently for military
  duties
• need to assess large numbers group testing
• attitude assessment - Thurstone - Chave
• Fisher ANOVA (beginning of experimental design)

12
1930s Theory development

• Thurstone and Likert assessment of affect,
  methodology
• KR-20, Rulon, etc. internal consistency
• Spearman-Brown prediction of test characteristics

13
1940s More theory

• Hoyts reliability using ANOVA
• Guttman scaling of attitude
• Decision theory and WWII

14
1950s

• Computer-based analysis
• Cronbachs alpha- pulling theory together
• construct validity - Cronbach, Meehl
• multitrait-multimethod validity - MacCorquodale

15
1960s Computers and statistics

• latent trait theory- Rasch
• formalized true score theory - Lord Novick
• covariance structure analysis - Jorekog
• multidimensional scaling - Torgeson
• latent partition analysis - Wiley

16
1970s

• generalizability theory- Cronbach
• criterion-referenced testing (CRT)
• bias in testing theory, differential item
  function (DIF)
• 2- and 3- parameter latent trait models - Lord
• statewide assessments minimum competency

17
1980s multitrait linear logistic model - Embretson

• Multitrait linear logistic model- Embretson
• Item-individual interaction cognitive
  function task demand item
  function (Mislevy, Willson)
• Development of testing standards

18
1990s Web-based assessment

• Automated response web testing, computer
  adaptive testing
• AI systems
• widespread use of measurement model in
  quantitative theory (Structural equation
  modeling)
• accountability systems based on testing

19
2000s HIGH STAKES TESTING AND RESEARCH

• No Child Left Behind
• Multilevel modeling (HLM)
• Federal testing requirements for US States
• Great demands on testing industry to develop many
  tests