0% found this document useful (0 votes)
22 views

Week 4 - Reliability

The document discusses different types of reliability estimates used in psychometrics including test-retest reliability, parallel forms reliability, internal consistency reliability, and inter-scorer reliability. It also discusses factors that can impact reliability such as test characteristics, sampling, and error sources.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
22 views

Week 4 - Reliability

The document discusses different types of reliability estimates used in psychometrics including test-retest reliability, parallel forms reliability, internal consistency reliability, and inter-scorer reliability. It also discusses factors that can impact reliability such as test characteristics, sampling, and error sources.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 8

Psychometric Reliability →

Reliability Coefficient
SOURCES OF ERROR VARIANCE
Item Sampling or Content Sampling
The Concept of Reliability


Error content sampled
the way the content is sampled
𝑿=𝑻+𝑬
Where

Variance (𝜎 2 ) Test Administration

𝝈𝟐 = 𝝈𝟐𝒕𝒉 + 𝝈𝟐𝒆 → Testtaker Variables


Where 𝝈 𝟐

𝝈𝟐𝒕𝒉 𝝈𝟐𝒆

Reliability → Examiner-Related Variables

→  

Test Scoring and Interpretation

→ Scorer (or Rater)

Measurement Error

Sampling Error
Random Error


Methodological Error

Systematic Error


Reliability Estimates Parallel Forms

TEST-RETEST RELIABILITY ESTIMATES


Test-Retest Method or Time Sampling →


Test-Retest Reliability
→ Parallel Forms Reliability


Alternate Forms


→ Alternate Forms Reliability

Coefficient of Stability



→ Carryover Effects ➢

→ Practice Effects ➢

INTERNAL CONSISTENCY ESTIMATE OF RELIABILITY

Internal Consistency Estimate of Reliability or Estimate of


PARALLEL-FORMS & ALTERNATE-FORMS RELIABILITY
Inter-Item Consistency
ESTIMATES

Coefficient of Equivalence

Inter-Item Consistency



equivalent forms reliability
➢ 𝟐𝒓
𝒄𝒐𝒓𝒓𝒆𝒄𝒕𝒆𝒅 𝒓 =
𝟏+𝒓

→ Homogeneity Where 𝒄𝒐𝒓𝒓𝒆𝒄𝒕𝒆𝒅 𝒓 =

𝒓=


𝒓 =. 𝟕𝟖
𝟐(. 𝟕𝟖) 𝟏. 𝟓𝟔
→ Heterogeneity 𝒄𝒐𝒓𝒓𝒆𝒄𝒕𝒆𝒅 𝒓 = = =. 𝟖𝟕𝟔
𝟏+. 𝟕𝟖 𝟏. 𝟕𝟖

➢ →

𝜶
Cronbach’s Coefficient Alpha (𝜶)

Split-Half Method


→ →

→ →

Odd-Even System

𝟐 [𝝈𝟐𝒙 − (𝝈𝟐𝒚𝟏 𝝈𝟐𝒚𝟐 )]


𝜶=
𝝈𝟐𝒙
Where 𝜶 =
𝝈𝟐𝒙 =
𝝈𝟐𝒚𝟏 𝝈𝟐𝒚𝟐 =

➢  
Kuder–Richardson Formula 20 (KR-20)

Spearman-Brown Formula


➢ ➢

𝒌 ∑ 𝒑𝒒
𝒓𝑲𝑹𝟐𝟎 = ( ) (𝟏 − ) ➢
𝒌−𝟏 𝝈𝟐
Where 𝒓𝑲𝑹𝟐𝟎 ;𝒌
𝟐
;𝝈 ;

𝒑 ;𝒒
∑ 𝒑𝒒
𝑝𝑞


𝟏
(𝒓𝟏𝟏 + 𝒓𝟐𝟐 ) − 𝒓𝟏𝟐
𝒓=𝟐
𝟏 − 𝒓𝟏𝟐
Coefficient Alpha Where 𝒓𝟏𝟏 = 𝒓𝟐𝟐 =
𝒓𝟐𝟐 =

𝒌 ∑ 𝝈𝟐𝒊
𝒓𝜶 = ( ) (𝟏 − )
𝒌−𝟏 𝝈𝟐
𝟏
(. 𝟗𝟎+. 𝟕𝟎) − . 𝟕𝟎 . 𝟏𝟎
Where 𝒓𝜶 𝒌 𝒓=𝟐 = = . 𝟑𝟑
𝟏− . 𝟕𝟎 . 𝟑𝟎
𝝈𝟐𝒊 ∑ 𝝈𝟐𝒊
𝟐
𝝈
MEASURES OF INTER-SCORER RELIABILITY

➢ Inter-Scorer Reliability


Coefficient of Inter-Scorer Reliability

➢ 


➢ Factor Analysis Using & Interpreting a Coefficient


of Reliability

THE PURPOSE OF THE RELIABILITY COEFFICIENT


Type
Average Proportional Distance (APD) Uses
Number of Testing Sessions
Source of Error Variance
Type Static Characteristic

Uses →

Number of Testing Sessions


Source of Error Variance The range of test scores is or is not restricted.

Type Restriction of Range/Variance

Uses

Number of Testing Sessions Inflation of Range/Variance


Source of Error Variance

Type
The test is a speed or a power test.
Uses
Power Test

Number of Testing Sessions


Speed Test
Source of Error Variance

THE NATURE OF THE TEST


➢ →

The test items are homogeneous or heterogeneous in →


nature.
Homogenous Test Items
The test is or is not criterion-referenced.
→ Criterion-Referenced Test

Heterogenous Test Items →

The characteristic, ability, or trait being measured is



presumed to be dynamic or static.
Dynamic Characteristic


THE TRUE SCORE MODEL OF MEASUREMENT AND →
ALTERNATIVES TO IT 𝝈𝒎𝒆𝒂𝒔

Classical Test Theory (CTT)

→ true score model of


measurement –


𝑿 = 𝑻 + 𝑬 𝑜𝑟 𝑿 − 𝑻 = 𝑬
Where 𝑿= 𝑻= 𝑬=

→ →



Domain Sampling Model



STANDARD ERROR OF MEASUREMENT

Standard Error of Measurement (SEM or 𝝈𝒎𝒆𝒂𝒔 )


Item Response Theory


→ standard error of a score
→ 

→ latent-trait theory

→ →

𝝈𝒎𝒆𝒂𝒔 = 𝝈√𝟏 − 𝒓𝒙𝒙

Where 𝝈𝒎𝒆𝒂𝒔 = 𝝈=
→ 𝒓𝒙𝒙 =

Given that


𝝈𝒎𝒆𝒂𝒔 = 𝟏𝟎√𝟏−. 𝟖𝟒 = 𝟒


→ Difficulty

→ Discrimination


Reliability & Individual Scores

𝝈𝒅𝒊𝒇𝒇 = √𝝈𝟐𝒎𝒆𝒂𝒔𝟏 + 𝝈𝟐𝒎𝒆𝒂𝒔𝟐

Where 𝝈𝒅𝒊𝒇𝒇 =
→ 𝝈𝟐𝒎𝒆𝒂𝒔𝟏 =
𝝈𝟐𝒎𝒆𝒂𝒔𝟐 =

→ confidence →
interval

𝝈𝒅𝒊𝒇𝒇 = 𝝈√𝟐 − 𝒓𝟏 − 𝒓𝟐
𝝈𝒅𝒊𝒇𝒇 =
75 ± 1.96𝜎𝑚𝑒𝑎𝑠 = 75 ± 1.96(2.37) = 75 ± 4.645 ≈ 5 𝒓𝟏 =
𝒓𝟐 = 𝝈=

STANDARD ERROR OF THE DIFFERENCE BETWEEN TWO


SCORES →
Standard Error of the Difference (SED)


𝝈𝒅𝒊𝒇𝒇 = 𝟏𝟒√𝟐 − . 𝟗𝟐 − . 𝟗𝟐 = 𝟏𝟒√. 𝟏𝟔 = 𝟓. 𝟔


You might also like