IRT or Rasch–what can and cannot be done? · 2019-06-27 · 2019-06-25 1 [email protected]...
Transcript of IRT or Rasch–what can and cannot be done? · 2019-06-27 · 2019-06-25 1 [email protected]...
2019-06-25
1
[email protected] 2019-06-12 PM-Health
IRT or Rasch – what can and cannot be done?
L R Pendrill1, and the EMPIR NeuroMet 15HLT04 consortium
1RI.SE Research Institutes in Sweden, Metrology, Eklandagatan 86, 41261 Göteborg (SE), phone:+46 767 88 54 44, mailto:[email protected]
WORKSHOP
[email protected] 2019-06-12 PM-Health
35 arthritis patients have been through rehabilitation therapy. Their admission to therapy and discharge from therapy measures are to be compared. They have been rated on the 13 mobility items of the Functional Independence Measure (FIMTM).
Each item has seven levels.
At admission, the patients could not perform at the higher levels. At discharge, all patients had surpassed the lower levels (Data courtesy of C.V. Granger & B. Hamilton, ADS).
EXAM12
1 0% Independent2 25% Independent3 50% Independent4 75% Independent5 Supervision6 Device7 Independent
2019-06-25
2
[email protected] 2019-06-12 PM-Health
[email protected] 2019-06-12 PM-Health W P Fisher Jr. 1999
)Difficulty(d
2019-06-25
3
[email protected] 2019-06-12 PM-Health
Score, x1 0% Independent2 25% Independent3 50% Independent4 75% Independent5 Supervision6 Device7 Independent
[email protected] 2019-06-12 PM-Health
2019-06-25
5
[email protected] 2019-06-12 PM-Health
[email protected] 2019-06-12 PM-Health
Most able
Leastable
2019-06-25
6
[email protected] 2019-06-12 PM-Health
DifficultEasy
[email protected] 2019-06-12 PM-Health
Most able
Leastable
Most difficult
Leastdifficult
! "#$%$&'
( )$**$+,%&'k = 2
2019-06-25
7
[email protected] 2019-06-12 PM-Health
[email protected] 2019-06-12 PM-Health
!"# =1&'( )*+,-
./0"+,#CTT mean of
response x
2(!"#) =1
&'( ) &'( − 1 )*+,-
./0"+,# − !"#6u(CTT mean of
response x)
k = 1
Score, x1 0% Independent2 25% Independent3 50% Independent4 75% Independent5 Supervision6 Device7 Independent
2019-06-25
8
[email protected] 2019-06-12 PM-Health
!"# =1&'( )*+,-
./0"+,#CTT mean of
response x
Difficult
Easy
k = 1
[email protected] 2019-06-12 PM-Health
!",$,% ='∑)*+, -./01,)
∑%2341 '∑)*+, -./01,)
Score residual, y 5",$ = 6",$ − 8",$
8",$ = 9%23
41: ; !",$,%Expected mean, E, of
response x
Graphs: Test Characteristic Curve
6",$
8",$
5",$
2019-06-25
9
[email protected] 2019-06-12 PM-Health
!"#!$_&'()_*+ =2 . / 0 −
/"23 −
23
55 = 6
7.89:;<
=/
Molton: “Both statistics … require a lot of meditation and incense. For useful explanations, refer to: http://www.rasch.org/rmt/rmt162f.htm”.
0 = >?6 = @ABC
9:;
DA6
EA,?,G =H∑JKL
M NO<PQ,J
∑GBRSQ H∑JKL
M NO<PQ,J
Score residual, y DA,? = TA,? − UA,?
UA,? = @GBR
SQ
V . EA,?,GExpected mean, E, of
response x
[email protected] 2019-06-12 PM-Healthk = 2
Physical disability
2019-06-25
10
[email protected] 2019-06-12 PM-Health
[email protected] 2019-06-12 PM-Health
Physical disability
2019-06-25
11
[email protected] 2019-06-12 PM-Health
!. #$%&'# #()*+,'(-./01 = 7 − !. #$%&'# #()*+,'(
Physical disability
[email protected] 2019-06-12 PM-Health22
Acquiescence
!"#$$%"" ='()*
1 + '()*
”Agreement regardless of item content∆. = / 0 .; / < 0
2019-06-25
12
[email protected] 2019-06-12 PM-Health
Difficult
Easy
Difficulty
Physical disability
[email protected] 2019-06-12 PM-Health
DifficultEasy
Physical disability
2019-06-25
13
[email protected] 2019-06-12 PM-Healthk = 2
Physical disability
[email protected] 2019-06-12 PM-Health
Fit residual
!"#$$%"" ='()*
1 + '()*
- =.!"#$$%""./
='01 ()*
1 + ' ()* 0−
' ()*
1 + ' ()*
34,6 = 74,6 − 84,6
Response Sensitivity
9:;<=_?@=A_BC =
2 1EF −
EGHI −
23
K
L′6,*0= L6 −
.!"#$$%""./
1 ./0
= L60 − 2 1
.!"#$$%""./
1 ./ 1 L6 +.!"#$$%""./
1 ./0
Changedstimulus
F = L60 = N
4OP
QRS
34,60
2019-06-25
14
[email protected] 2019-06-12 PM-Health
Acquiesence, τ < 0Importancerescaling, τ > 0
[email protected] 2019-06-12 PM-Health W P Fisher Jr. 1999
)Difficulty(d
2019-06-25
15
[email protected] 2019-06-12 PM-Health
Metrologicalreferences
Difficulty
Mass
Tasks
)(Difficultyd
Physical disability
[email protected] 2019-06-12 PM-Health
Acknowledgments
Work performed as part of 15HLT04 NeuroMet project, European Metrology Programme for Innovation & Research (Horizon 2020), jointly funded by EMRP participating countries within EURAMET (www.euramet.org) and European Union.
Thanks are due, particularly, to: • Members of the EMPIR NeuroMet consortium • Jeanette Melin, RISE Metrology Göteborg Sweden• Stefan Cano, Modus Outcomes, Boston (USA) & Stotford UK• William P. Fisher, Jr., Ph.D., Research Associate, BEAR Center, Graduate
School of Education, University of California, Berkeley, CA (USA) & Principal, LivingCapitalMetrics Consulting
2019-06-25
16
NeuroMET: Innovative measurements for improved diagnosis and management of neurodegenerative diseases
Acknowledgments
This project has received funding from the EMPIR programme co-financed by the Participating States and from the European Union’s Horizon 2020 research and innovation programme
NeuroMET : www.lgcgroup.com/EMPIR-neuromet
[email protected] 2019-06-12 PM-Health
EURAMET’s Research Programmes