SlideShare a Scribd company logo
1 of 75
©drtamil@gmail.com - 2016
Difficulty Index, Discrimination
Index, Reliability & Rasch
Measurement Analysis
Azmi Mohd Tamil
Universiti Kebangsaan Malaysia
©drtamil@gmail.com - 2016
Steps to assess questionnaire
• Flesch reading ease – assess readability.
• Index of difficulty – proportion of persons answering
correctly.
• Item discrimination – how well the item discriminates
between those with a high & low knowledge score.
• Reliability/Ferguson’s Sigma Discriminatory Power
• Inter-item correlation matrix
• Item-total correlations
• Cronbach’s alpha – inter-item consistency.
• Factor analysis
©drtamil@gmail.com - 2016
Topics covered in this lecture
• Topics to be covered;
– Index of difficulty
– Item discrimination
– Reliability
– Item & Person Matching
(Rasch Measurement Analysis).
https://ppukm.org/2015/04/02/
calculating-omr-indexes/
https://ppukm.org/2015/04/16/matching-the-right-questions-to-the-
right-students-rasch-model-for-measurement/
©drtamil@gmail.com - 2016
Convert text file to Excel
• It is difficult to enter the answers for each
questions into SPSS if there were a lot of
questions and a large number of students.
• So instead we make use of OMR machines and
scan their answers into a text file.
• Then convert the text file into Excel.
• http://www.palmx.org/mambo/content/view/
173/45/
©drtamil@gmail.com - 2016
Sample of A Raw Text File
©drtamil@gmail.com - 2016
Sample of A Raw Text File
Matric.
No.
Answers
©drtamil@gmail.com - 2016
Convert txt into Excel
©drtamil@gmail.com - 2016
Import into SPSS
©drtamil@gmail.com - 2016
Convert into Correct=1,Wrong=0
Converted Excel file available from http://drtamil.me/2016/03/02/fk6193-practical-diff-
disc-index/ for use in the coming exercises.
©drtamil@gmail.com - 2016
Index of Difficulty
• D = students with correct answer x 100
total students
©drtamil@gmail.com - 2016
PPUKM uses <30 (Difficult), 30 to 70 (Okay), >70 (Easy)
cut-off points.
©drtamil@gmail.com - 2016
Discrimination Index
• R = (H – L)
27% of Total
• H = number of correct answers from top 27% of
students
• L = number of corrects answers from bottom 27%
of students
• 27% out of 22 = 6 students.
©drtamil@gmail.com - 2016
Interpretation of Discrimination Index
• PPUKM uses; negative & 0.00 (non-disc), 0.01
to 0.15 (poor), 0.16 to 0.25 (marginal), 0.26 to
0.35 (good),>0.35 (Excellent),
©drtamil@gmail.com - 2016
Using the earlier Excel file,
we can calculate the D & R
• Example for D
– =SUM(B2:B23)/COUNT(B2:B23)*100 for the first
question
• Example for R
– Sort the total marks from largest to smallest
– =((SUM(B2:B7))-(SUM(B18:B23)))/6 for the first
question
©drtamil@gmail.com - 2016
©drtamil@gmail.com - 2016
Select, Copy, New Sheet,
Paste Special, Transpose
Click on Transpose
©drtamil@gmail.com - 2016
Import into SPSS & Analyse
Item 27, difficult and indiscriminate,
need to review
©drtamil@gmail.com - 2016
Reliability - Kuder and Richardson
Formula 20
The test checks the internal consistency of
measurements with dichotomous choices. A
correct question scores 1 and an incorrect
question scores 0. The test statistic is
©drtamil@gmail.com - 2016
• k = number of questions
• pj = number of people in the sample who answered
question j correctly
• qj = number of people in the sample who didn’t answer
question j correctly
• σ2 = variance of the total scores of all the people taking
the test = VARP(R1) where R1 = array containing the
total scores of all the people taking the test.
• Values range from 0 to 1. A high value indicates
reliability, while too high a value (in excess of .90)
indicates a homogeneous test.
©drtamil@gmail.com - 2016
From Our Table
k = 30, Sum of pj & qj = 4.7521, σ2 = 31.827
ρKR20 = (30/29)*(1-(4.7521/31.827))
= 0.88
High reliability, almost homogeneous.
©drtamil@gmail.com - 2016
From Our Table
σ2 = 31.827, ρKR20 = 0.88
S.E.M. = Standard deviation * SQRT (1 – Reliability)
= SQRT(31.827) * SQRT (1 – 0.88)
= 1.95.
©drtamil@gmail.com - 2016
Conclusion
• Item 27 need to be reviewed due to being
both indiscriminate (R=0.00) and difficult
(D=32).
• Average DI is Easy at 76.5.
• Average RI is Good at 0.36.
• ρKR20 is very reliable at 0.87.
• SEM is very small at 1.95.
• Overall a good set of examination questions.
©drtamil@gmail.com - 2016
Were the questions too easy?
Brief Introduction to
Rasch Measurement Analysis
©drtamil@gmail.com - 2016
Rasch Measurement Analysis
• Earlier we learnt about Difficulty and Discrimination
Index.
• In Rasch, Item Measure of Difficulty (Di) is similar to
Difficulty Index (D). Instead of Discrimination Index
(R), Rasch has Persons’ Measure of Ability (Bi).
• A high Discrimination Index (R) question is able to
discriminate the good from poor students.
• With Rasch, the higher is a person’s ability (Bi) the
more likely he is able to answer a difficult question.
Therefore an indiscriminate item is detected when a
high-ability person cannot answer an easy item or a
low-ability person can answer a difficult item.
• In Rasch, they use logit instead of rate or %.
©drtamil@gmail.com - 2016
Difficulty & Ability Measures
• Item Measure of Difficulty (Di)
– Logit (number of wrong answers/number of
correct answers)
– Measured according to the items/questions.
• Persons’ Measure of Ability (Bi)
– Logit (number of correct answers/number of
wrong answers)
– Measured according to persons.
©drtamil@gmail.com - 2016
e.g. If you have 100 students, with a difficult question maybe 99 will get it
wrong and only 1 get it right. Odds of 99/1 is a logit of 4.5951.
A moderate difficulty question, maybe 50 will get it wrong and 50 will get
it right. Odds of 50/50 is a logit of 0.
A slightly easier question, maybe 25 will get it wrong and 75 will get it
right. Odds of 25/75 is a logit = -1.0986.
“Measurement is defined as the assignment of numerals to
objects or events according to rules.”
(“On the Theory of Scales of Measurement”; S.S. Stevens, 1946)
Rasch Model ‘logit’ scale for Di
25
75
e-4.6
-4.6
75
25
50
50
99
1
1
99
e0 e4.6
0 4.6-1.1 1.1
exp
logit
Now, we already have a SCALE with a unit termed ‘logit’ for Di.
-2.0
12
88
37
63
-0.5
63
37
88
12
0.5 2.0
e-1.1 e1.1
©drtamil@gmail.com - 2016
e.g. If you have 100 questions, with a good student maybe she will get 99
questions right and only get 1 wrong. Odds of 99/1 is a logit of 4.5951.
A moderately able student, maybe she will get 50 questions right and the
other 50 wrong. Odds of 50/50 is a logit of 0.
A weak student, maybe he will get 25 questions right and the other 75
wrong. Odds of 25/75 is a logit = -1.0986.
“Numerals can be assigned under different rules leads to
different kind of scales & different kinds of measurement.”
(“On the Theory of Scales of Measurement”; S.S. Stevens, 1946)
Rasch Model ‘logit’ scale for Bi
25
75
e-4.6
-4.6
75
25
50
50
99
1
1
99
e0 e4.6
0 4.6-1.1 1.1
exp
logit
Now, we already have a SCALE with a unit termed ‘logit’ for Bi.
-2.0
12
88
37
63
-0.5
63
37
88
12
0.5 2.0
e-1.1 e1.1
©drtamil@gmail.com - 2016
Measures of Item Difficulty (Di)
& Person’s Ability (Bi)
Please look closely at the Difficulty Index (D) and Measures of Item Difficulty (Di).
They are inversely related.
©drtamil@gmail.com - 2016
Import SPSS data into Winsteps
1. Open Winsteps
2. Click on Excel/RSST
3. Click on SPSS button
4. Click on Select SPSS file
5. Your SPSS file will be
imported into “SPSS
Processing for Winsteps”
window.
6. Now copy and paste the
identifying data and item
data.
©drtamil@gmail.com - 2016
1. Cut the matric. no. and
paste under “Person
Label Variables”.
2. Then cut all question
items then paste under
“Item Response
Variables”
3. Click the “Construct
Winsteps file” button .
©drtamil@gmail.com - 2016
The end
product, a
Winsteps
file, which is
really a text
file.
©drtamil@gmail.com - 2016
Winsteps
file after
convert
©drtamil@gmail.com - 2016
Open the file in Winsteps & press
Enter twice.
The squared sum residuals
of the entire matrix.
Iteration is done until this
value is as close as
possible to 0.
©drtamil@gmail.com - 2016
Fit Statistics – How well was it measured?
©drtamil@gmail.com - 2016
©drtamil@gmail.com - 2016
©drtamil@gmail.com - 2016
The Manual & Winsteps
Measures Differ!
item
Manual
Manual
As though JMLE sets item 7 measure as 0, as anchor to other items/persons measures.
Winsteps
©drtamil@gmail.com - 2016
Manual Measures and Winsteps (Rasch
Software) Measures are not the same. Why?
• Due to Joint Maximum Likelihood Estimation (JMLE)
• Winsteps implements three methods of estimating Rasch
parameters from ordered qualitative observations: JMLE
and PROX. Estimates of the Rasch measures are obtained
by iterating through the data. Initially all unanchored
parameter estimates (measures) are set to zero. Then the
PROX method is employed to obtain rough estimates. Each
iteration through the data improves the PROX estimates
until they are usefully good. Then those PROX estimates are
the initial estimates for JMLE which fine-tunes them, again
by iterating through the data, in order to obtain the final
JMLE estimates. The iterative process ceases when the
convergence criteria are met.
• Confused?
©drtamil@gmail.com - 2016
Items Difficulty Measure – similar but not
the same due to different reference point.
©drtamil@gmail.com - 2016
Persons’ Ability Measures (Bi) - similar but not
the same due to different reference point.
©drtamil@gmail.com - 2016
JMLE adjustment
• So it is as though the JMLE sets one item (item 7)
as the anchor reference point, then all other
items/persons measures are adjusted
accordingly.
• So the differences between all the measures are
still the same, as shown in the scatter diagram.
• So the manual measures and Winsteps measures
are similar. r = 1. Just the reference point is
changed or adjusted.
• Of course the real calculation is much more
complicated.
©drtamil@gmail.com - 2016
JMLE adjustment
• First they get the average of all the Item Measures of
Difficulty (Di).
• Then minus the value of the average from all the Item
Measure of Difficulty (Di);
i.e. (-1.2) - (-1.41) = 0.2
• The average from all this new Di s would be equal to 0.
©drtamil@gmail.com - 2016
Recalculate the Probability of
Answering Correctly
• Recalculate using the old Item Measure of Ability
(Bi) and the new Item Measure of Difficulty (Di).
e (βn – δi )
P(Ɵ) =
1 + e (βn – δi )
©drtamil@gmail.com - 2016
SUM UP VARIANCE OF EXPECTED VALUES
• Calculate the variance of expected values for each cell;
P*(1-P)
• Sum up the variances according to rows and columns.
• Negative sums in blue column act as denominator to tweak
persons ability logit to maximise fit.
• Negative sums in yellow row act as denominator to tweak
items difficulty logit to maximise fit.
©drtamil@gmail.com - 2016
New Bi & new adjusted Di
• Tweak the Persons Measure of Ability (Bi) based on the sum of residuals (Observed – Expected) and sum
of variances of expected values;
New Bi = Old Bi – (Sum of Variances/Sum of Residuals)
• Tweak the Item Measure of Difficulty (Di) based on negative sum of residuals (Observed – Expected) and
sum of variances of expected values;
New Di = Old adj. Di – (Sum of Variances/Negative Sum of Residuals)
• Get the average of all the new Item Measures of Difficulty (Di). Then get the new adj. Di by deducting the
value of the average from all the new Item Measure of Difficulty (Di);
i.e. (-0.046) - (-0.211) = 0.165 ~ 0.17
• Keep doing the iteration until the squared sum residuals of the entire matrix is as close as possible to 0.
• For this dataset, 4 iterations was required before that was achieved.
©drtamil@gmail.com - 2016
Need More Analysis?
©drtamil@gmail.com - 2016
Checking the Fit
Bubble Plot
©drtamil@gmail.com - 2016
Plots – Bubble Chart
©drtamil@gmail.com - 2016
Bubble Plot
Okay Erratic
Too good
to be true
©drtamil@gmail.com - 2016
Bubble Plot
- Items Measure is the Y-axis.
- Model S.E. is the diameter of the circle
(reduced to 60%).
- InFit Z-STD is the X-axis
- We expect difficult items can be
answered by the more able persons and
easy items could be answered by all.
- Item 27 is considered erratic, although it
is difficult, both able and weak person
cannot answer it.
Under the previous exercise, for item 27; the
Difficulty Index was D=32 and Discrimination
Index was R=0.00. Difficult and yet unable to
discriminate. So it was already detected as a
problem question in earlier analysis.
©drtamil@gmail.com - 2016
Bubble Plot
Okay Erratic
Too good
to be true
27; 2.7,2.89, 0.56
14; -0.8,1.21, 0.53
23; -1.3, 2.82, 0.57
8; 0.0, -3.40, 1.85
Lowest item
Largest error of 1.85,
so largest bubble.
20; 1.3, 1.21, 0.53
11; 1.4, 0.34, 0.56
17;0.2,-2.12,1.06
24; 0.2,-2.12,1.06
26; 1.7, 0.74, 0.58
28;0.5,-0.35,0.625;-1.0,-0.35,0.62
12;-1.1,-0.35,0.62
19;-1.0,0.24,0.59
Item No.; t Infit Zstd, Item
Measures, S.E.
Graph showing how it is
plotted using the Infit Zstd
and Item Measures.
Bubble Size is 60% of the
Model S.E.
Zstd for 27 larger than 2.0
therefore an erratic item.
Should be checked.
©drtamil@gmail.com - 2016
Bubble Plot
Okay Erratic
Too good
to be true
27;D32,R0.00
14;D59,R0.83
23;D32,R0.83
8;D100,R0.00
Item 8 has high Difficulty
Index & yet poor
Discrimination Index.
20;D59,R0.33
11;D73,R0.33
17;D96,R0.17
24;D96,R0.17
26;D64,R0.33
28;D82,R0.335;D82,R0.67
12;D82, R0.67
19;D73,R0.83
Item No.; Difficulty Index,
Discrimination Index.
Graph showing the
relationship of Difficulty
Index with Item Measures
and Discrimination Index
with Infit Zstd.
Item 27 has low Difficulty
Index and yet poor
Discrimination Index.
©drtamil@gmail.com - 2016
Wright Map
©drtamil@gmail.com - 2016
Person Item Map
Mean for Persons
Mean for Items
Need tougher questions
to test A, B, C D, E, F & G.
Questions too easy.
Not testing anybody.
1. Poor Students;
n=4 (8%)
2.Good students;
n=18 (82%)
On target. Between
the mean + 1sd.
13/30 = 43%
©drtamil@gmail.com - 2016
Item Measures
©drtamil@gmail.com - 2016
Scan the InFit
Zstd for values
larger than 2.0.
For item 27; the
InFit Zstd is 2.7,
therefore larger
than 2.0. As
stated earlier,
such items are
considered
erratic and
should be
removed or
changed.
©drtamil@gmail.com - 2016VERY DIFFICULT
= +2.89 logit
N=21, score=7
avg.=0.33,
Many got it
wrong.
BOTH y,z BREACHED
ITEM NEED REVIEW
Large +Z due to inconsistency
in response. e.g. poorly able
person can answer a difficult
question.
-2 < Z < +20.5 < y < 1.5 0.32 < x < 0.8
LOW PT. MEASURE
CORELATION . SOME
POOR STUDENTS CAN
ANSWER ITEM
CORRECTLY WHILE
GOOD STUDENTS GOT
IT WRONG
©drtamil@gmail.com - 2016
LOW PT. MEASURE
CORELATION . SOME
POOR STUDENTS CAN
ANSWER ITEM
CORRECTLY WHILST
GOOD STUDENTS GOT
IT WRONG
0.32 < x < 0.8
EXTREMELY EASY
=-3.40 logit
N=22, score=22
ave.=1, all correct
©drtamil@gmail.com - 2016
Person Measures
©drtamil@gmail.com - 2016
Scan the InFit
Zstd for values
larger than 2.0.
For Person T & U;
the InFit Zstd is
larger than 2.0.
Such erratic
performance
could be due to
them getting
some very easy
questions wrong.
©drtamil@gmail.com - 2016
Expected Score ICC
The InFit Zstd of item 27
is 2.7, therefore larger
than 2.0.
So we will check the ICC
of item 27, why it is
erratic.
For comparison we will
also look at ICC of item
23, with the Infit Zstd of
-1.3.
©drtamil@gmail.com - 2016
ICC
Item
27,
Di= 2.9
R= 0.0
For a difficult
question, a less able
person shouldn’t be
able to answer.
e (βn – δi )
P(Ɵ) =
1 + e (βn – δi )
where;
e= Euler’s Number, 2.7183
βn= Person’s ability measure
δi= item difficulty measure
But a more able
person should be
able to answer.
But the blue line
is not following
the red line. So
the ability is not
consistent.
©drtamil@gmail.com - 2016
ICC
Item
23,
Di= 2.8
R=0.8
For a difficult
question, a less able
person shouldn’t be
able to answer.
e (βn – δi )
P(Ɵ) =
1 + e (βn – δi )
where;
e= Euler’s Number, 2.7183
βn= Person’s ability measure
δi= item difficulty measure
But a more able
person should be
able to answer.
Here the blue
line is following
the red line. So
the ability is
consistent with
the difficulty.
©drtamil@gmail.com - 2016
0.66; ‘Poor’ item instrument
reliability in measuring student
learning ability. Poor targeting.
Summary
Statistics (All)
+ve Person mean
μ = 1.75 logit
P[Ɵ]LOi=e(1.75+0.11)/(1+e(1.75+0.11))
= 6.4237/7.4237
= 0.865 (easy to pass)
G=1.90
Separation index=(4G+1)/3=2.9
‘Good’ Person separation into
3 groups. So can have 3 grades!
0.78 ‘Fair’ person reliability
Cronbach-α :0.87 Good
reliability assessment of student
learning. Same value as ρKR20
calculated manually.
Presence of one extreme
person, “Mrs A” and one
extreme item, “Item 27”,
causes the InFit & OutFit
MNSQ & Zstd not calculated.
e (βn – δi )
P(Ɵ) =
1 + e (βn – δi )
where;
e= Euler’s Number, 2.7183
βn= Person’s ability measure
δi= item difficulty measure
©drtamil@gmail.com - 2016
How ρKR20 was calculated earlier.
k = 30, Sum of pj & qj = 4.7521, σ2 = 31.827
ρKR20 = (30/29)*(1-(4.7521/31.827))
= 0.88
High reliability, almost homogeneous.
©drtamil@gmail.com - 2016
0.65; ‘Poor’ item
measurement reliability in
measuring student learning
ability. Poor targeting.
Summary
Statistics
(minus1)
+ve Person mean
μ = 1.58 logit
P[Ɵ] LOi= e(1.58-0)/(1+e(1.58-0))
= 4.855/5.855
= 0.83 (easy to pass)
InFit Zstd close to 0 & MNSQ
close to 1, so data fit the
model.
G=1.97
Separation index=(4G+1)/3=2.96
‘Good’ Person separation of 3
groups. So can have 3 grades!
0.80 ‘Fair’ person reliability
Cronbach-α not available after
exclusion of extreme person
data.
Interpretation of Person & item measurement reliability;
-<0.67 is Poor
- 0.67 – 0.80 Fair
- 0.81 – 0.90 Good
- 0.91 – 0.94 Very Good
e (βn – δi )
P(Ɵ) =
1 + e (βn – δi )
where;
e= Euler’s Number, 2.7183
βn= Person’s ability measure
δi= item difficulty measure
©drtamil@gmail.com - 2016
Guttman Scalogram
©drtamil@gmail.com - 2016
Guttman
Scalogram
SMART Student-01
POORStudent-22
EASY ITEMS DIFFICULT ITEMS
RESPONSE SORTED: EASY TO TOUGH
©drtamil@gmail.com - 2016
Guttman
Scalogram
Theorem 1. Persons who are more able / more developed
have a greater likelihood of correctly answer all the items /
able to complete a given task.
Theorem 2. Easier items / task are more likely to be
answered correctly by all persons.
CARELESS
PREDICT=1
PREDICT=0
GUESS
©drtamil@gmail.com - 2016
Guttman
Scalogram
as similarly
arranged
as
descending
Difficulty
Index.
©drtamil@gmail.com - 2016
Item Misfit Response String
©drtamil@gmail.com - 2016
Item Misfit Response String
Most misfit item:
Exceed MNSQ Limit:
0.5 < y < 1.5
High Rating Response Zone 1.
Item in red circles for the
respective Persons were under
rated
Low Rating Response 0.
Item in blue circles for
the respective Persons
were over rated
©drtamil@gmail.com - 2016
Conclusion
• Item 27 needs to be relooked at. Needs better
rephrasing of the question. Or better choice of
answers
• Rasch came out with almost the same kind of
conclusion as our earlier analysis.
• Rasch also identifies gaps in questions selection,
showing us that there were too many easy
questions, and absence of difficult questions that
could test those with higher ability. Something
the conventional analysis couldn’t show clearly.
©drtamil@gmail.com - 2016
Conclusion (continued)
• Among the figures that were similarly generated
between both methods were;
– Difficulty Index D versus Item Measure of Difficulty Di
(inverse relationship)
– ρKR20 of 0.88 (manual) and 0.87 (Rasch).
– Both method detected item 27 as an erratic item. Of
course Rasch proved it with multitudes of figures and
data.
– With the conventional method, it requires experience
to realise items with low difficulty index and non-
discriminatory discrimination index as problem items.
©drtamil@gmail.com - 2016
Conclusion (continued)
• Most importantly, Rasch detected the following;
– ‘Poor’ item instrument reliability of 0.66; in
measuring student learning ability. Poor targeting.
– μ Bi= 1.75 logit; μ Di=-0.11
P[Ɵ]LOi=e(1.75+0.11)/(1+e(1.75+0.11)) = 0.865
– Based on the above, questions were too easy.
– Separation index of 2.9 indicating ‘Good’ Person
separation into 3 groups. So can have 3 grades!
Excellent, Pass, Fail? Instead of the usual
A, A-, B+, B, B-, C+, C, C-, D & E?

More Related Content

What's hot

What's hot (20)

Performance-based Assessment
Performance-based AssessmentPerformance-based Assessment
Performance-based Assessment
 
Item analysis
Item analysis Item analysis
Item analysis
 
Topic 8b Item Analysis
Topic 8b Item AnalysisTopic 8b Item Analysis
Topic 8b Item Analysis
 
Item Analysis
Item AnalysisItem Analysis
Item Analysis
 
Item analysis report
Item analysis report Item analysis report
Item analysis report
 
Item Analysis
Item AnalysisItem Analysis
Item Analysis
 
Standard Scores
Standard ScoresStandard Scores
Standard Scores
 
Item Response Theory in Constructing Measures
Item Response Theory in Constructing MeasuresItem Response Theory in Constructing Measures
Item Response Theory in Constructing Measures
 
Affective Assessment
Affective AssessmentAffective Assessment
Affective Assessment
 
Item analysis
Item analysisItem analysis
Item analysis
 
Analyzing and using test item data
Analyzing and using test item dataAnalyzing and using test item data
Analyzing and using test item data
 
Item analysis
Item analysisItem analysis
Item analysis
 
Reliability
ReliabilityReliability
Reliability
 
Item analysis2
Item analysis2Item analysis2
Item analysis2
 
Principles of Test Construction 1
Principles of Test Construction 1Principles of Test Construction 1
Principles of Test Construction 1
 
Distracter Analysis - Index of Effectiveness
Distracter Analysis - Index of EffectivenessDistracter Analysis - Index of Effectiveness
Distracter Analysis - Index of Effectiveness
 
Normal curve
Normal curveNormal curve
Normal curve
 
58519966 review-of-principles-of-high-quality-assessment
58519966 review-of-principles-of-high-quality-assessment58519966 review-of-principles-of-high-quality-assessment
58519966 review-of-principles-of-high-quality-assessment
 
Z score presnetation
Z score presnetationZ score presnetation
Z score presnetation
 
Item analysis ppt
Item analysis pptItem analysis ppt
Item analysis ppt
 

Viewers also liked

MOOC for Public Health Physicians: Innovative Method in Ensuring Continuous C...
MOOC for Public Health Physicians: Innovative Method in Ensuring Continuous C...MOOC for Public Health Physicians: Innovative Method in Ensuring Continuous C...
MOOC for Public Health Physicians: Innovative Method in Ensuring Continuous C...Azmi Mohd Tamil
 
Kaedah penyelidikan action method
Kaedah penyelidikan action methodKaedah penyelidikan action method
Kaedah penyelidikan action methodfathira90
 
How to get your courses listed on UKM MOOC (OpenLearning)
How to get your courses listed on UKM MOOC (OpenLearning)How to get your courses listed on UKM MOOC (OpenLearning)
How to get your courses listed on UKM MOOC (OpenLearning)Azmi Mohd Tamil
 
Pendekatan Penyelidikan Kuantitatif
Pendekatan Penyelidikan KuantitatifPendekatan Penyelidikan Kuantitatif
Pendekatan Penyelidikan KuantitatifIta Kamis
 
How to run Pearson's Chi-square for SPSS
How to run Pearson's Chi-square for SPSSHow to run Pearson's Chi-square for SPSS
How to run Pearson's Chi-square for SPSSAzmi Mohd Tamil
 
How to draw Scatter plot on SPSS
How to draw Scatter plot on SPSSHow to draw Scatter plot on SPSS
How to draw Scatter plot on SPSSAzmi Mohd Tamil
 
How to run Simple Linear Regression on SPSS
How to run Simple Linear Regression on SPSSHow to run Simple Linear Regression on SPSS
How to run Simple Linear Regression on SPSSAzmi Mohd Tamil
 
How to run ANOVA on SPSS
How to run ANOVA on SPSSHow to run ANOVA on SPSS
How to run ANOVA on SPSSAzmi Mohd Tamil
 
How to run Student's t-test on SPSS
How to run Student's t-test on SPSSHow to run Student's t-test on SPSS
How to run Student's t-test on SPSSAzmi Mohd Tamil
 
Taklimat pentingnya ifolio untuk kenaikan pangkat pensyarah UKM
Taklimat pentingnya ifolio untuk kenaikan pangkat pensyarah UKMTaklimat pentingnya ifolio untuk kenaikan pangkat pensyarah UKM
Taklimat pentingnya ifolio untuk kenaikan pangkat pensyarah UKMAzmi Mohd Tamil
 
Running Pearson's Correlation on SPSS
Running Pearson's Correlation on SPSSRunning Pearson's Correlation on SPSS
Running Pearson's Correlation on SPSSAzmi Mohd Tamil
 
Introduction to spss: define variables
Introduction to spss: define variablesIntroduction to spss: define variables
Introduction to spss: define variablesAzmi Mohd Tamil
 

Viewers also liked (14)

MOOC for Public Health Physicians: Innovative Method in Ensuring Continuous C...
MOOC for Public Health Physicians: Innovative Method in Ensuring Continuous C...MOOC for Public Health Physicians: Innovative Method in Ensuring Continuous C...
MOOC for Public Health Physicians: Innovative Method in Ensuring Continuous C...
 
Kaedah penyelidikan action method
Kaedah penyelidikan action methodKaedah penyelidikan action method
Kaedah penyelidikan action method
 
How to get your courses listed on UKM MOOC (OpenLearning)
How to get your courses listed on UKM MOOC (OpenLearning)How to get your courses listed on UKM MOOC (OpenLearning)
How to get your courses listed on UKM MOOC (OpenLearning)
 
Pendekatan Penyelidikan Kuantitatif
Pendekatan Penyelidikan KuantitatifPendekatan Penyelidikan Kuantitatif
Pendekatan Penyelidikan Kuantitatif
 
Borang kaji selidik
Borang kaji selidikBorang kaji selidik
Borang kaji selidik
 
How to run Pearson's Chi-square for SPSS
How to run Pearson's Chi-square for SPSSHow to run Pearson's Chi-square for SPSS
How to run Pearson's Chi-square for SPSS
 
How to draw Scatter plot on SPSS
How to draw Scatter plot on SPSSHow to draw Scatter plot on SPSS
How to draw Scatter plot on SPSS
 
How to run Simple Linear Regression on SPSS
How to run Simple Linear Regression on SPSSHow to run Simple Linear Regression on SPSS
How to run Simple Linear Regression on SPSS
 
How to run ANOVA on SPSS
How to run ANOVA on SPSSHow to run ANOVA on SPSS
How to run ANOVA on SPSS
 
How to run Student's t-test on SPSS
How to run Student's t-test on SPSSHow to run Student's t-test on SPSS
How to run Student's t-test on SPSS
 
Taklimat pentingnya ifolio untuk kenaikan pangkat pensyarah UKM
Taklimat pentingnya ifolio untuk kenaikan pangkat pensyarah UKMTaklimat pentingnya ifolio untuk kenaikan pangkat pensyarah UKM
Taklimat pentingnya ifolio untuk kenaikan pangkat pensyarah UKM
 
Running Pearson's Correlation on SPSS
Running Pearson's Correlation on SPSSRunning Pearson's Correlation on SPSS
Running Pearson's Correlation on SPSS
 
Introduction to spss: define variables
Introduction to spss: define variablesIntroduction to spss: define variables
Introduction to spss: define variables
 
Borang soal selidik
Borang soal selidikBorang soal selidik
Borang soal selidik
 

Similar to Difficulty Index, Discrimination Index, Reliability and Rasch Measurement Analysis

Introduction to Structural Equation Modeling
Introduction to Structural Equation ModelingIntroduction to Structural Equation Modeling
Introduction to Structural Equation ModelingAzmi Mohd Tamil
 
Chapter 6 data analysis iec11
Chapter 6 data analysis iec11Chapter 6 data analysis iec11
Chapter 6 data analysis iec11Ho Cao Viet
 
Top schools in noida
Top schools in noidaTop schools in noida
Top schools in noidaEdhole.com
 
Statistical Learning and Model Selection module 2.pptx
Statistical Learning and Model Selection module 2.pptxStatistical Learning and Model Selection module 2.pptx
Statistical Learning and Model Selection module 2.pptxnagarajan740445
 
Statistics for analytics
Statistics for analyticsStatistics for analytics
Statistics for analyticsdeepika4721
 
QUANTITATIVE-DATA.pptx
QUANTITATIVE-DATA.pptxQUANTITATIVE-DATA.pptx
QUANTITATIVE-DATA.pptxViaFortuna
 
Stepwise Selection Choosing the Optimal Model .ppt
Stepwise Selection  Choosing the Optimal Model .pptStepwise Selection  Choosing the Optimal Model .ppt
Stepwise Selection Choosing the Optimal Model .pptneelamsanjeevkumar
 
Dive into the Data
Dive into the DataDive into the Data
Dive into the Datadr_jp_ebejer
 
Simple rules for building robust machine learning models
Simple rules for building robust machine learning modelsSimple rules for building robust machine learning models
Simple rules for building robust machine learning modelsKyriakos Chatzidimitriou
 
3Measurements of health and disease_MCTD.pdf
3Measurements of health and disease_MCTD.pdf3Measurements of health and disease_MCTD.pdf
3Measurements of health and disease_MCTD.pdfAmanuelDina
 
Introduction to Data Analysis With R and R Studio
Introduction to Data Analysis With R and R StudioIntroduction to Data Analysis With R and R Studio
Introduction to Data Analysis With R and R StudioAzmi Mohd Tamil
 
Statistical Learning and Model Selection (1).pptx
Statistical Learning and Model Selection (1).pptxStatistical Learning and Model Selection (1).pptx
Statistical Learning and Model Selection (1).pptxrajalakshmi5921
 
A data science observatory based on RAMP - rapid analytics and model prototyping
A data science observatory based on RAMP - rapid analytics and model prototypingA data science observatory based on RAMP - rapid analytics and model prototyping
A data science observatory based on RAMP - rapid analytics and model prototypingAkin Osman Kazakci
 
Basics of Stats (2).pptx
Basics of Stats (2).pptxBasics of Stats (2).pptx
Basics of Stats (2).pptxmadihamaqbool6
 
Thompson (2009) -_classical_item_analysis_with_citas
Thompson (2009) -_classical_item_analysis_with_citasThompson (2009) -_classical_item_analysis_with_citas
Thompson (2009) -_classical_item_analysis_with_citasMelanio Florino
 
The 8 Step Data Mining Process
The 8 Step Data Mining ProcessThe 8 Step Data Mining Process
The 8 Step Data Mining ProcessMarc Berman
 
Systems Modelling and Qualitative Data
Systems Modelling and Qualitative Data Systems Modelling and Qualitative Data
Systems Modelling and Qualitative Data mikeyearworth
 
CHAPTER 11 LOGISTIC REGRESSION.pptx
CHAPTER 11 LOGISTIC REGRESSION.pptxCHAPTER 11 LOGISTIC REGRESSION.pptx
CHAPTER 11 LOGISTIC REGRESSION.pptxUmaDeviAnanth
 

Similar to Difficulty Index, Discrimination Index, Reliability and Rasch Measurement Analysis (20)

Introduction to Structural Equation Modeling
Introduction to Structural Equation ModelingIntroduction to Structural Equation Modeling
Introduction to Structural Equation Modeling
 
Chapter 6 data analysis iec11
Chapter 6 data analysis iec11Chapter 6 data analysis iec11
Chapter 6 data analysis iec11
 
Top schools in noida
Top schools in noidaTop schools in noida
Top schools in noida
 
Statistical Learning and Model Selection module 2.pptx
Statistical Learning and Model Selection module 2.pptxStatistical Learning and Model Selection module 2.pptx
Statistical Learning and Model Selection module 2.pptx
 
Statistics for analytics
Statistics for analyticsStatistics for analytics
Statistics for analytics
 
Credit scoring i financial sector
Credit scoring i financial  sector Credit scoring i financial  sector
Credit scoring i financial sector
 
QUANTITATIVE-DATA.pptx
QUANTITATIVE-DATA.pptxQUANTITATIVE-DATA.pptx
QUANTITATIVE-DATA.pptx
 
Stepwise Selection Choosing the Optimal Model .ppt
Stepwise Selection  Choosing the Optimal Model .pptStepwise Selection  Choosing the Optimal Model .ppt
Stepwise Selection Choosing the Optimal Model .ppt
 
Dive into the Data
Dive into the DataDive into the Data
Dive into the Data
 
Simple rules for building robust machine learning models
Simple rules for building robust machine learning modelsSimple rules for building robust machine learning models
Simple rules for building robust machine learning models
 
3Measurements of health and disease_MCTD.pdf
3Measurements of health and disease_MCTD.pdf3Measurements of health and disease_MCTD.pdf
3Measurements of health and disease_MCTD.pdf
 
Introduction to Data Analysis With R and R Studio
Introduction to Data Analysis With R and R StudioIntroduction to Data Analysis With R and R Studio
Introduction to Data Analysis With R and R Studio
 
Statistical Learning and Model Selection (1).pptx
Statistical Learning and Model Selection (1).pptxStatistical Learning and Model Selection (1).pptx
Statistical Learning and Model Selection (1).pptx
 
Kaggle KDD Cup Report
Kaggle KDD Cup ReportKaggle KDD Cup Report
Kaggle KDD Cup Report
 
A data science observatory based on RAMP - rapid analytics and model prototyping
A data science observatory based on RAMP - rapid analytics and model prototypingA data science observatory based on RAMP - rapid analytics and model prototyping
A data science observatory based on RAMP - rapid analytics and model prototyping
 
Basics of Stats (2).pptx
Basics of Stats (2).pptxBasics of Stats (2).pptx
Basics of Stats (2).pptx
 
Thompson (2009) -_classical_item_analysis_with_citas
Thompson (2009) -_classical_item_analysis_with_citasThompson (2009) -_classical_item_analysis_with_citas
Thompson (2009) -_classical_item_analysis_with_citas
 
The 8 Step Data Mining Process
The 8 Step Data Mining ProcessThe 8 Step Data Mining Process
The 8 Step Data Mining Process
 
Systems Modelling and Qualitative Data
Systems Modelling and Qualitative Data Systems Modelling and Qualitative Data
Systems Modelling and Qualitative Data
 
CHAPTER 11 LOGISTIC REGRESSION.pptx
CHAPTER 11 LOGISTIC REGRESSION.pptxCHAPTER 11 LOGISTIC REGRESSION.pptx
CHAPTER 11 LOGISTIC REGRESSION.pptx
 

More from Azmi Mohd Tamil

Hybrid setup - How to conduct simultaneous face-to-face and online presentati...
Hybrid setup - How to conduct simultaneous face-to-face and online presentati...Hybrid setup - How to conduct simultaneous face-to-face and online presentati...
Hybrid setup - How to conduct simultaneous face-to-face and online presentati...Azmi Mohd Tamil
 
Audiovisual and technicalities from preparation to retrieval how to enhance m...
Audiovisual and technicalities from preparation to retrieval how to enhance m...Audiovisual and technicalities from preparation to retrieval how to enhance m...
Audiovisual and technicalities from preparation to retrieval how to enhance m...Azmi Mohd Tamil
 
Broadcast quality online teaching at zero budget
Broadcast quality online teaching at zero budgetBroadcast quality online teaching at zero budget
Broadcast quality online teaching at zero budgetAzmi Mohd Tamil
 
Video for Teaching & Learning: OBS
Video for Teaching & Learning: OBSVideo for Teaching & Learning: OBS
Video for Teaching & Learning: OBSAzmi Mohd Tamil
 
Bengkel 21-12-2020 - Etika atas Talian & Alat Minima
Bengkel 21-12-2020 - Etika atas Talian & Alat MinimaBengkel 21-12-2020 - Etika atas Talian & Alat Minima
Bengkel 21-12-2020 - Etika atas Talian & Alat MinimaAzmi Mohd Tamil
 
GIS & History of Mapping in Malaya (lecture notes circa 2009)
GIS & History of Mapping in Malaya (lecture notes circa 2009)GIS & History of Mapping in Malaya (lecture notes circa 2009)
GIS & History of Mapping in Malaya (lecture notes circa 2009)Azmi Mohd Tamil
 
Blended e-learning in UKMFolio
Blended e-learning in UKMFolioBlended e-learning in UKMFolio
Blended e-learning in UKMFolioAzmi Mohd Tamil
 
How to Compute & Recode SPSS Data
How to Compute & Recode SPSS DataHow to Compute & Recode SPSS Data
How to Compute & Recode SPSS DataAzmi Mohd Tamil
 
Hack#38 - How to Stream Zoom to Facebook & YouTube Without Using An Encoder o...
Hack#38 - How to Stream Zoom to Facebook & YouTube Without Using An Encoder o...Hack#38 - How to Stream Zoom to Facebook & YouTube Without Using An Encoder o...
Hack#38 - How to Stream Zoom to Facebook & YouTube Without Using An Encoder o...Azmi Mohd Tamil
 
Hack#37 - How to simultaneously live stream to 4 sites using a single hardwar...
Hack#37 - How to simultaneously live stream to 4 sites using a single hardwar...Hack#37 - How to simultaneously live stream to 4 sites using a single hardwar...
Hack#37 - How to simultaneously live stream to 4 sites using a single hardwar...Azmi Mohd Tamil
 
Cochran Mantel Haenszel Test with Breslow-Day Test & Quadratic Equation
Cochran Mantel Haenszel Test with Breslow-Day Test & Quadratic EquationCochran Mantel Haenszel Test with Breslow-Day Test & Quadratic Equation
Cochran Mantel Haenszel Test with Breslow-Day Test & Quadratic EquationAzmi Mohd Tamil
 
New Emerging And Reemerging Infections circa 2006
New Emerging And Reemerging Infections circa 2006New Emerging And Reemerging Infections circa 2006
New Emerging And Reemerging Infections circa 2006Azmi Mohd Tamil
 
Hacks#36 -Raspberry Pi 4 Mini Computer
Hacks#36 -Raspberry Pi 4 Mini ComputerHacks#36 -Raspberry Pi 4 Mini Computer
Hacks#36 -Raspberry Pi 4 Mini ComputerAzmi Mohd Tamil
 
Hack#35 How to FB Live using a Video Encoder
Hack#35 How to FB Live using a Video EncoderHack#35 How to FB Live using a Video Encoder
Hack#35 How to FB Live using a Video EncoderAzmi Mohd Tamil
 
Hack#34 - Online Teaching with Microsoft Teams
Hack#34 - Online Teaching with Microsoft TeamsHack#34 - Online Teaching with Microsoft Teams
Hack#34 - Online Teaching with Microsoft TeamsAzmi Mohd Tamil
 
Skype for Business for UKM
Skype for Business for UKM Skype for Business for UKM
Skype for Business for UKM Azmi Mohd Tamil
 
Safe computing (circa 2004)
Safe computing (circa 2004)Safe computing (circa 2004)
Safe computing (circa 2004)Azmi Mohd Tamil
 
Introduction to 20 Classroom Hacks For Education 4.0 (updated)
Introduction to 20 Classroom Hacks For Education 4.0 (updated)Introduction to 20 Classroom Hacks For Education 4.0 (updated)
Introduction to 20 Classroom Hacks For Education 4.0 (updated)Azmi Mohd Tamil
 

More from Azmi Mohd Tamil (20)

Hybrid setup - How to conduct simultaneous face-to-face and online presentati...
Hybrid setup - How to conduct simultaneous face-to-face and online presentati...Hybrid setup - How to conduct simultaneous face-to-face and online presentati...
Hybrid setup - How to conduct simultaneous face-to-face and online presentati...
 
Audiovisual and technicalities from preparation to retrieval how to enhance m...
Audiovisual and technicalities from preparation to retrieval how to enhance m...Audiovisual and technicalities from preparation to retrieval how to enhance m...
Audiovisual and technicalities from preparation to retrieval how to enhance m...
 
Broadcast quality online teaching at zero budget
Broadcast quality online teaching at zero budgetBroadcast quality online teaching at zero budget
Broadcast quality online teaching at zero budget
 
Video for Teaching & Learning: OBS
Video for Teaching & Learning: OBSVideo for Teaching & Learning: OBS
Video for Teaching & Learning: OBS
 
Bengkel 21-12-2020 - Etika atas Talian & Alat Minima
Bengkel 21-12-2020 - Etika atas Talian & Alat MinimaBengkel 21-12-2020 - Etika atas Talian & Alat Minima
Bengkel 21-12-2020 - Etika atas Talian & Alat Minima
 
GIS & History of Mapping in Malaya (lecture notes circa 2009)
GIS & History of Mapping in Malaya (lecture notes circa 2009)GIS & History of Mapping in Malaya (lecture notes circa 2009)
GIS & History of Mapping in Malaya (lecture notes circa 2009)
 
Blended e-learning in UKMFolio
Blended e-learning in UKMFolioBlended e-learning in UKMFolio
Blended e-learning in UKMFolio
 
How to Compute & Recode SPSS Data
How to Compute & Recode SPSS DataHow to Compute & Recode SPSS Data
How to Compute & Recode SPSS Data
 
Hack#38 - How to Stream Zoom to Facebook & YouTube Without Using An Encoder o...
Hack#38 - How to Stream Zoom to Facebook & YouTube Without Using An Encoder o...Hack#38 - How to Stream Zoom to Facebook & YouTube Without Using An Encoder o...
Hack#38 - How to Stream Zoom to Facebook & YouTube Without Using An Encoder o...
 
Hack#37 - How to simultaneously live stream to 4 sites using a single hardwar...
Hack#37 - How to simultaneously live stream to 4 sites using a single hardwar...Hack#37 - How to simultaneously live stream to 4 sites using a single hardwar...
Hack#37 - How to simultaneously live stream to 4 sites using a single hardwar...
 
Cochran Mantel Haenszel Test with Breslow-Day Test & Quadratic Equation
Cochran Mantel Haenszel Test with Breslow-Day Test & Quadratic EquationCochran Mantel Haenszel Test with Breslow-Day Test & Quadratic Equation
Cochran Mantel Haenszel Test with Breslow-Day Test & Quadratic Equation
 
New Emerging And Reemerging Infections circa 2006
New Emerging And Reemerging Infections circa 2006New Emerging And Reemerging Infections circa 2006
New Emerging And Reemerging Infections circa 2006
 
Hacks#36 -Raspberry Pi 4 Mini Computer
Hacks#36 -Raspberry Pi 4 Mini ComputerHacks#36 -Raspberry Pi 4 Mini Computer
Hacks#36 -Raspberry Pi 4 Mini Computer
 
Hack#35 How to FB Live using a Video Encoder
Hack#35 How to FB Live using a Video EncoderHack#35 How to FB Live using a Video Encoder
Hack#35 How to FB Live using a Video Encoder
 
Hack#34 - Online Teaching with Microsoft Teams
Hack#34 - Online Teaching with Microsoft TeamsHack#34 - Online Teaching with Microsoft Teams
Hack#34 - Online Teaching with Microsoft Teams
 
Hack#33 How To FB-Live
Hack#33 How To FB-LiveHack#33 How To FB-Live
Hack#33 How To FB-Live
 
Skype for Business for UKM
Skype for Business for UKM Skype for Business for UKM
Skype for Business for UKM
 
Safe computing (circa 2004)
Safe computing (circa 2004)Safe computing (circa 2004)
Safe computing (circa 2004)
 
Introduction to 20 Classroom Hacks For Education 4.0 (updated)
Introduction to 20 Classroom Hacks For Education 4.0 (updated)Introduction to 20 Classroom Hacks For Education 4.0 (updated)
Introduction to 20 Classroom Hacks For Education 4.0 (updated)
 
HUKM Patient Portal
HUKM Patient PortalHUKM Patient Portal
HUKM Patient Portal
 

Recently uploaded

See the 2,456 pharmacies on the National E-Pharmacy Platform
See the 2,456 pharmacies on the National E-Pharmacy PlatformSee the 2,456 pharmacies on the National E-Pharmacy Platform
See the 2,456 pharmacies on the National E-Pharmacy PlatformKweku Zurek
 
Call Girls Hebbal Just Call 7001305949 Top Class Call Girl Service Available
Call Girls Hebbal Just Call 7001305949 Top Class Call Girl Service AvailableCall Girls Hebbal Just Call 7001305949 Top Class Call Girl Service Available
Call Girls Hebbal Just Call 7001305949 Top Class Call Girl Service Availablenarwatsonia7
 
Call Girls Service in Bommanahalli - 7001305949 with real photos and phone nu...
Call Girls Service in Bommanahalli - 7001305949 with real photos and phone nu...Call Girls Service in Bommanahalli - 7001305949 with real photos and phone nu...
Call Girls Service in Bommanahalli - 7001305949 with real photos and phone nu...narwatsonia7
 
High Profile Call Girls Jaipur Vani 8445551418 Independent Escort Service Jaipur
High Profile Call Girls Jaipur Vani 8445551418 Independent Escort Service JaipurHigh Profile Call Girls Jaipur Vani 8445551418 Independent Escort Service Jaipur
High Profile Call Girls Jaipur Vani 8445551418 Independent Escort Service Jaipurparulsinha
 
Bangalore Call Girls Marathahalli 📞 9907093804 High Profile Service 100% Safe
Bangalore Call Girls Marathahalli 📞 9907093804 High Profile Service 100% SafeBangalore Call Girls Marathahalli 📞 9907093804 High Profile Service 100% Safe
Bangalore Call Girls Marathahalli 📞 9907093804 High Profile Service 100% Safenarwatsonia7
 
Call Girls Kanakapura Road Just Call 7001305949 Top Class Call Girl Service A...
Call Girls Kanakapura Road Just Call 7001305949 Top Class Call Girl Service A...Call Girls Kanakapura Road Just Call 7001305949 Top Class Call Girl Service A...
Call Girls Kanakapura Road Just Call 7001305949 Top Class Call Girl Service A...narwatsonia7
 
Call Girls Service Chennai Jiya 7001305949 Independent Escort Service Chennai
Call Girls Service Chennai Jiya 7001305949 Independent Escort Service ChennaiCall Girls Service Chennai Jiya 7001305949 Independent Escort Service Chennai
Call Girls Service Chennai Jiya 7001305949 Independent Escort Service ChennaiNehru place Escorts
 
Aspirin presentation slides by Dr. Rewas Ali
Aspirin presentation slides by Dr. Rewas AliAspirin presentation slides by Dr. Rewas Ali
Aspirin presentation slides by Dr. Rewas AliRewAs ALI
 
call girls in Connaught Place DELHI 🔝 >༒9540349809 🔝 genuine Escort Service ...
call girls in Connaught Place  DELHI 🔝 >༒9540349809 🔝 genuine Escort Service ...call girls in Connaught Place  DELHI 🔝 >༒9540349809 🔝 genuine Escort Service ...
call girls in Connaught Place DELHI 🔝 >༒9540349809 🔝 genuine Escort Service ...saminamagar
 
Book Call Girls in Kasavanahalli - 7001305949 with real photos and phone numbers
Book Call Girls in Kasavanahalli - 7001305949 with real photos and phone numbersBook Call Girls in Kasavanahalli - 7001305949 with real photos and phone numbers
Book Call Girls in Kasavanahalli - 7001305949 with real photos and phone numbersnarwatsonia7
 
Mumbai Call Girls Service 9910780858 Real Russian Girls Looking Models
Mumbai Call Girls Service 9910780858 Real Russian Girls Looking ModelsMumbai Call Girls Service 9910780858 Real Russian Girls Looking Models
Mumbai Call Girls Service 9910780858 Real Russian Girls Looking Modelssonalikaur4
 
Low Rate Call Girls Mumbai Suman 9910780858 Independent Escort Service Mumbai
Low Rate Call Girls Mumbai Suman 9910780858 Independent Escort Service MumbaiLow Rate Call Girls Mumbai Suman 9910780858 Independent Escort Service Mumbai
Low Rate Call Girls Mumbai Suman 9910780858 Independent Escort Service Mumbaisonalikaur4
 
Artifacts in Nuclear Medicine with Identifying and resolving artifacts.
Artifacts in Nuclear Medicine with Identifying and resolving artifacts.Artifacts in Nuclear Medicine with Identifying and resolving artifacts.
Artifacts in Nuclear Medicine with Identifying and resolving artifacts.MiadAlsulami
 
Sonagachi Call Girls Services 9907093804 @24x7 High Class Babes Here Call Now
Sonagachi Call Girls Services 9907093804 @24x7 High Class Babes Here Call NowSonagachi Call Girls Services 9907093804 @24x7 High Class Babes Here Call Now
Sonagachi Call Girls Services 9907093804 @24x7 High Class Babes Here Call NowRiya Pathan
 
College Call Girls Pune Mira 9907093804 Short 1500 Night 6000 Best call girls...
College Call Girls Pune Mira 9907093804 Short 1500 Night 6000 Best call girls...College Call Girls Pune Mira 9907093804 Short 1500 Night 6000 Best call girls...
College Call Girls Pune Mira 9907093804 Short 1500 Night 6000 Best call girls...Miss joya
 
Call Girls Jp Nagar Just Call 7001305949 Top Class Call Girl Service Available
Call Girls Jp Nagar Just Call 7001305949 Top Class Call Girl Service AvailableCall Girls Jp Nagar Just Call 7001305949 Top Class Call Girl Service Available
Call Girls Jp Nagar Just Call 7001305949 Top Class Call Girl Service Availablenarwatsonia7
 
Call Girls Service Nandiambakkam | 7001305949 At Low Cost Cash Payment Booking
Call Girls Service Nandiambakkam | 7001305949 At Low Cost Cash Payment BookingCall Girls Service Nandiambakkam | 7001305949 At Low Cost Cash Payment Booking
Call Girls Service Nandiambakkam | 7001305949 At Low Cost Cash Payment BookingNehru place Escorts
 
Call Girls Hsr Layout Just Call 7001305949 Top Class Call Girl Service Available
Call Girls Hsr Layout Just Call 7001305949 Top Class Call Girl Service AvailableCall Girls Hsr Layout Just Call 7001305949 Top Class Call Girl Service Available
Call Girls Hsr Layout Just Call 7001305949 Top Class Call Girl Service Availablenarwatsonia7
 
Low Rate Call Girls Pune Esha 9907093804 Short 1500 Night 6000 Best call girl...
Low Rate Call Girls Pune Esha 9907093804 Short 1500 Night 6000 Best call girl...Low Rate Call Girls Pune Esha 9907093804 Short 1500 Night 6000 Best call girl...
Low Rate Call Girls Pune Esha 9907093804 Short 1500 Night 6000 Best call girl...Miss joya
 

Recently uploaded (20)

See the 2,456 pharmacies on the National E-Pharmacy Platform
See the 2,456 pharmacies on the National E-Pharmacy PlatformSee the 2,456 pharmacies on the National E-Pharmacy Platform
See the 2,456 pharmacies on the National E-Pharmacy Platform
 
Call Girls Hebbal Just Call 7001305949 Top Class Call Girl Service Available
Call Girls Hebbal Just Call 7001305949 Top Class Call Girl Service AvailableCall Girls Hebbal Just Call 7001305949 Top Class Call Girl Service Available
Call Girls Hebbal Just Call 7001305949 Top Class Call Girl Service Available
 
Escort Service Call Girls In Sarita Vihar,, 99530°56974 Delhi NCR
Escort Service Call Girls In Sarita Vihar,, 99530°56974 Delhi NCREscort Service Call Girls In Sarita Vihar,, 99530°56974 Delhi NCR
Escort Service Call Girls In Sarita Vihar,, 99530°56974 Delhi NCR
 
Call Girls Service in Bommanahalli - 7001305949 with real photos and phone nu...
Call Girls Service in Bommanahalli - 7001305949 with real photos and phone nu...Call Girls Service in Bommanahalli - 7001305949 with real photos and phone nu...
Call Girls Service in Bommanahalli - 7001305949 with real photos and phone nu...
 
High Profile Call Girls Jaipur Vani 8445551418 Independent Escort Service Jaipur
High Profile Call Girls Jaipur Vani 8445551418 Independent Escort Service JaipurHigh Profile Call Girls Jaipur Vani 8445551418 Independent Escort Service Jaipur
High Profile Call Girls Jaipur Vani 8445551418 Independent Escort Service Jaipur
 
Bangalore Call Girls Marathahalli 📞 9907093804 High Profile Service 100% Safe
Bangalore Call Girls Marathahalli 📞 9907093804 High Profile Service 100% SafeBangalore Call Girls Marathahalli 📞 9907093804 High Profile Service 100% Safe
Bangalore Call Girls Marathahalli 📞 9907093804 High Profile Service 100% Safe
 
Call Girls Kanakapura Road Just Call 7001305949 Top Class Call Girl Service A...
Call Girls Kanakapura Road Just Call 7001305949 Top Class Call Girl Service A...Call Girls Kanakapura Road Just Call 7001305949 Top Class Call Girl Service A...
Call Girls Kanakapura Road Just Call 7001305949 Top Class Call Girl Service A...
 
Call Girls Service Chennai Jiya 7001305949 Independent Escort Service Chennai
Call Girls Service Chennai Jiya 7001305949 Independent Escort Service ChennaiCall Girls Service Chennai Jiya 7001305949 Independent Escort Service Chennai
Call Girls Service Chennai Jiya 7001305949 Independent Escort Service Chennai
 
Aspirin presentation slides by Dr. Rewas Ali
Aspirin presentation slides by Dr. Rewas AliAspirin presentation slides by Dr. Rewas Ali
Aspirin presentation slides by Dr. Rewas Ali
 
call girls in Connaught Place DELHI 🔝 >༒9540349809 🔝 genuine Escort Service ...
call girls in Connaught Place  DELHI 🔝 >༒9540349809 🔝 genuine Escort Service ...call girls in Connaught Place  DELHI 🔝 >༒9540349809 🔝 genuine Escort Service ...
call girls in Connaught Place DELHI 🔝 >༒9540349809 🔝 genuine Escort Service ...
 
Book Call Girls in Kasavanahalli - 7001305949 with real photos and phone numbers
Book Call Girls in Kasavanahalli - 7001305949 with real photos and phone numbersBook Call Girls in Kasavanahalli - 7001305949 with real photos and phone numbers
Book Call Girls in Kasavanahalli - 7001305949 with real photos and phone numbers
 
Mumbai Call Girls Service 9910780858 Real Russian Girls Looking Models
Mumbai Call Girls Service 9910780858 Real Russian Girls Looking ModelsMumbai Call Girls Service 9910780858 Real Russian Girls Looking Models
Mumbai Call Girls Service 9910780858 Real Russian Girls Looking Models
 
Low Rate Call Girls Mumbai Suman 9910780858 Independent Escort Service Mumbai
Low Rate Call Girls Mumbai Suman 9910780858 Independent Escort Service MumbaiLow Rate Call Girls Mumbai Suman 9910780858 Independent Escort Service Mumbai
Low Rate Call Girls Mumbai Suman 9910780858 Independent Escort Service Mumbai
 
Artifacts in Nuclear Medicine with Identifying and resolving artifacts.
Artifacts in Nuclear Medicine with Identifying and resolving artifacts.Artifacts in Nuclear Medicine with Identifying and resolving artifacts.
Artifacts in Nuclear Medicine with Identifying and resolving artifacts.
 
Sonagachi Call Girls Services 9907093804 @24x7 High Class Babes Here Call Now
Sonagachi Call Girls Services 9907093804 @24x7 High Class Babes Here Call NowSonagachi Call Girls Services 9907093804 @24x7 High Class Babes Here Call Now
Sonagachi Call Girls Services 9907093804 @24x7 High Class Babes Here Call Now
 
College Call Girls Pune Mira 9907093804 Short 1500 Night 6000 Best call girls...
College Call Girls Pune Mira 9907093804 Short 1500 Night 6000 Best call girls...College Call Girls Pune Mira 9907093804 Short 1500 Night 6000 Best call girls...
College Call Girls Pune Mira 9907093804 Short 1500 Night 6000 Best call girls...
 
Call Girls Jp Nagar Just Call 7001305949 Top Class Call Girl Service Available
Call Girls Jp Nagar Just Call 7001305949 Top Class Call Girl Service AvailableCall Girls Jp Nagar Just Call 7001305949 Top Class Call Girl Service Available
Call Girls Jp Nagar Just Call 7001305949 Top Class Call Girl Service Available
 
Call Girls Service Nandiambakkam | 7001305949 At Low Cost Cash Payment Booking
Call Girls Service Nandiambakkam | 7001305949 At Low Cost Cash Payment BookingCall Girls Service Nandiambakkam | 7001305949 At Low Cost Cash Payment Booking
Call Girls Service Nandiambakkam | 7001305949 At Low Cost Cash Payment Booking
 
Call Girls Hsr Layout Just Call 7001305949 Top Class Call Girl Service Available
Call Girls Hsr Layout Just Call 7001305949 Top Class Call Girl Service AvailableCall Girls Hsr Layout Just Call 7001305949 Top Class Call Girl Service Available
Call Girls Hsr Layout Just Call 7001305949 Top Class Call Girl Service Available
 
Low Rate Call Girls Pune Esha 9907093804 Short 1500 Night 6000 Best call girl...
Low Rate Call Girls Pune Esha 9907093804 Short 1500 Night 6000 Best call girl...Low Rate Call Girls Pune Esha 9907093804 Short 1500 Night 6000 Best call girl...
Low Rate Call Girls Pune Esha 9907093804 Short 1500 Night 6000 Best call girl...
 

Difficulty Index, Discrimination Index, Reliability and Rasch Measurement Analysis

  • 1. ©drtamil@gmail.com - 2016 Difficulty Index, Discrimination Index, Reliability & Rasch Measurement Analysis Azmi Mohd Tamil Universiti Kebangsaan Malaysia
  • 2. ©drtamil@gmail.com - 2016 Steps to assess questionnaire • Flesch reading ease – assess readability. • Index of difficulty – proportion of persons answering correctly. • Item discrimination – how well the item discriminates between those with a high & low knowledge score. • Reliability/Ferguson’s Sigma Discriminatory Power • Inter-item correlation matrix • Item-total correlations • Cronbach’s alpha – inter-item consistency. • Factor analysis
  • 3. ©drtamil@gmail.com - 2016 Topics covered in this lecture • Topics to be covered; – Index of difficulty – Item discrimination – Reliability – Item & Person Matching (Rasch Measurement Analysis). https://ppukm.org/2015/04/02/ calculating-omr-indexes/ https://ppukm.org/2015/04/16/matching-the-right-questions-to-the- right-students-rasch-model-for-measurement/
  • 4. ©drtamil@gmail.com - 2016 Convert text file to Excel • It is difficult to enter the answers for each questions into SPSS if there were a lot of questions and a large number of students. • So instead we make use of OMR machines and scan their answers into a text file. • Then convert the text file into Excel. • http://www.palmx.org/mambo/content/view/ 173/45/
  • 6. ©drtamil@gmail.com - 2016 Sample of A Raw Text File Matric. No. Answers
  • 9. ©drtamil@gmail.com - 2016 Convert into Correct=1,Wrong=0 Converted Excel file available from http://drtamil.me/2016/03/02/fk6193-practical-diff- disc-index/ for use in the coming exercises.
  • 10. ©drtamil@gmail.com - 2016 Index of Difficulty • D = students with correct answer x 100 total students
  • 11. ©drtamil@gmail.com - 2016 PPUKM uses <30 (Difficult), 30 to 70 (Okay), >70 (Easy) cut-off points.
  • 12. ©drtamil@gmail.com - 2016 Discrimination Index • R = (H – L) 27% of Total • H = number of correct answers from top 27% of students • L = number of corrects answers from bottom 27% of students • 27% out of 22 = 6 students.
  • 13. ©drtamil@gmail.com - 2016 Interpretation of Discrimination Index • PPUKM uses; negative & 0.00 (non-disc), 0.01 to 0.15 (poor), 0.16 to 0.25 (marginal), 0.26 to 0.35 (good),>0.35 (Excellent),
  • 14. ©drtamil@gmail.com - 2016 Using the earlier Excel file, we can calculate the D & R • Example for D – =SUM(B2:B23)/COUNT(B2:B23)*100 for the first question • Example for R – Sort the total marks from largest to smallest – =((SUM(B2:B7))-(SUM(B18:B23)))/6 for the first question
  • 16. ©drtamil@gmail.com - 2016 Select, Copy, New Sheet, Paste Special, Transpose Click on Transpose
  • 17. ©drtamil@gmail.com - 2016 Import into SPSS & Analyse Item 27, difficult and indiscriminate, need to review
  • 18. ©drtamil@gmail.com - 2016 Reliability - Kuder and Richardson Formula 20 The test checks the internal consistency of measurements with dichotomous choices. A correct question scores 1 and an incorrect question scores 0. The test statistic is
  • 19. ©drtamil@gmail.com - 2016 • k = number of questions • pj = number of people in the sample who answered question j correctly • qj = number of people in the sample who didn’t answer question j correctly • σ2 = variance of the total scores of all the people taking the test = VARP(R1) where R1 = array containing the total scores of all the people taking the test. • Values range from 0 to 1. A high value indicates reliability, while too high a value (in excess of .90) indicates a homogeneous test.
  • 20. ©drtamil@gmail.com - 2016 From Our Table k = 30, Sum of pj & qj = 4.7521, σ2 = 31.827 ρKR20 = (30/29)*(1-(4.7521/31.827)) = 0.88 High reliability, almost homogeneous.
  • 21. ©drtamil@gmail.com - 2016 From Our Table σ2 = 31.827, ρKR20 = 0.88 S.E.M. = Standard deviation * SQRT (1 – Reliability) = SQRT(31.827) * SQRT (1 – 0.88) = 1.95.
  • 22. ©drtamil@gmail.com - 2016 Conclusion • Item 27 need to be reviewed due to being both indiscriminate (R=0.00) and difficult (D=32). • Average DI is Easy at 76.5. • Average RI is Good at 0.36. • ρKR20 is very reliable at 0.87. • SEM is very small at 1.95. • Overall a good set of examination questions.
  • 23. ©drtamil@gmail.com - 2016 Were the questions too easy? Brief Introduction to Rasch Measurement Analysis
  • 24. ©drtamil@gmail.com - 2016 Rasch Measurement Analysis • Earlier we learnt about Difficulty and Discrimination Index. • In Rasch, Item Measure of Difficulty (Di) is similar to Difficulty Index (D). Instead of Discrimination Index (R), Rasch has Persons’ Measure of Ability (Bi). • A high Discrimination Index (R) question is able to discriminate the good from poor students. • With Rasch, the higher is a person’s ability (Bi) the more likely he is able to answer a difficult question. Therefore an indiscriminate item is detected when a high-ability person cannot answer an easy item or a low-ability person can answer a difficult item. • In Rasch, they use logit instead of rate or %.
  • 25. ©drtamil@gmail.com - 2016 Difficulty & Ability Measures • Item Measure of Difficulty (Di) – Logit (number of wrong answers/number of correct answers) – Measured according to the items/questions. • Persons’ Measure of Ability (Bi) – Logit (number of correct answers/number of wrong answers) – Measured according to persons.
  • 26. ©drtamil@gmail.com - 2016 e.g. If you have 100 students, with a difficult question maybe 99 will get it wrong and only 1 get it right. Odds of 99/1 is a logit of 4.5951. A moderate difficulty question, maybe 50 will get it wrong and 50 will get it right. Odds of 50/50 is a logit of 0. A slightly easier question, maybe 25 will get it wrong and 75 will get it right. Odds of 25/75 is a logit = -1.0986. “Measurement is defined as the assignment of numerals to objects or events according to rules.” (“On the Theory of Scales of Measurement”; S.S. Stevens, 1946) Rasch Model ‘logit’ scale for Di 25 75 e-4.6 -4.6 75 25 50 50 99 1 1 99 e0 e4.6 0 4.6-1.1 1.1 exp logit Now, we already have a SCALE with a unit termed ‘logit’ for Di. -2.0 12 88 37 63 -0.5 63 37 88 12 0.5 2.0 e-1.1 e1.1
  • 27. ©drtamil@gmail.com - 2016 e.g. If you have 100 questions, with a good student maybe she will get 99 questions right and only get 1 wrong. Odds of 99/1 is a logit of 4.5951. A moderately able student, maybe she will get 50 questions right and the other 50 wrong. Odds of 50/50 is a logit of 0. A weak student, maybe he will get 25 questions right and the other 75 wrong. Odds of 25/75 is a logit = -1.0986. “Numerals can be assigned under different rules leads to different kind of scales & different kinds of measurement.” (“On the Theory of Scales of Measurement”; S.S. Stevens, 1946) Rasch Model ‘logit’ scale for Bi 25 75 e-4.6 -4.6 75 25 50 50 99 1 1 99 e0 e4.6 0 4.6-1.1 1.1 exp logit Now, we already have a SCALE with a unit termed ‘logit’ for Bi. -2.0 12 88 37 63 -0.5 63 37 88 12 0.5 2.0 e-1.1 e1.1
  • 28. ©drtamil@gmail.com - 2016 Measures of Item Difficulty (Di) & Person’s Ability (Bi) Please look closely at the Difficulty Index (D) and Measures of Item Difficulty (Di). They are inversely related.
  • 29. ©drtamil@gmail.com - 2016 Import SPSS data into Winsteps 1. Open Winsteps 2. Click on Excel/RSST 3. Click on SPSS button 4. Click on Select SPSS file 5. Your SPSS file will be imported into “SPSS Processing for Winsteps” window. 6. Now copy and paste the identifying data and item data.
  • 30. ©drtamil@gmail.com - 2016 1. Cut the matric. no. and paste under “Person Label Variables”. 2. Then cut all question items then paste under “Item Response Variables” 3. Click the “Construct Winsteps file” button .
  • 31. ©drtamil@gmail.com - 2016 The end product, a Winsteps file, which is really a text file.
  • 33. ©drtamil@gmail.com - 2016 Open the file in Winsteps & press Enter twice. The squared sum residuals of the entire matrix. Iteration is done until this value is as close as possible to 0.
  • 34. ©drtamil@gmail.com - 2016 Fit Statistics – How well was it measured?
  • 37. ©drtamil@gmail.com - 2016 The Manual & Winsteps Measures Differ! item Manual Manual As though JMLE sets item 7 measure as 0, as anchor to other items/persons measures. Winsteps
  • 38. ©drtamil@gmail.com - 2016 Manual Measures and Winsteps (Rasch Software) Measures are not the same. Why? • Due to Joint Maximum Likelihood Estimation (JMLE) • Winsteps implements three methods of estimating Rasch parameters from ordered qualitative observations: JMLE and PROX. Estimates of the Rasch measures are obtained by iterating through the data. Initially all unanchored parameter estimates (measures) are set to zero. Then the PROX method is employed to obtain rough estimates. Each iteration through the data improves the PROX estimates until they are usefully good. Then those PROX estimates are the initial estimates for JMLE which fine-tunes them, again by iterating through the data, in order to obtain the final JMLE estimates. The iterative process ceases when the convergence criteria are met. • Confused?
  • 39. ©drtamil@gmail.com - 2016 Items Difficulty Measure – similar but not the same due to different reference point.
  • 40. ©drtamil@gmail.com - 2016 Persons’ Ability Measures (Bi) - similar but not the same due to different reference point.
  • 41. ©drtamil@gmail.com - 2016 JMLE adjustment • So it is as though the JMLE sets one item (item 7) as the anchor reference point, then all other items/persons measures are adjusted accordingly. • So the differences between all the measures are still the same, as shown in the scatter diagram. • So the manual measures and Winsteps measures are similar. r = 1. Just the reference point is changed or adjusted. • Of course the real calculation is much more complicated.
  • 42. ©drtamil@gmail.com - 2016 JMLE adjustment • First they get the average of all the Item Measures of Difficulty (Di). • Then minus the value of the average from all the Item Measure of Difficulty (Di); i.e. (-1.2) - (-1.41) = 0.2 • The average from all this new Di s would be equal to 0.
  • 43. ©drtamil@gmail.com - 2016 Recalculate the Probability of Answering Correctly • Recalculate using the old Item Measure of Ability (Bi) and the new Item Measure of Difficulty (Di). e (βn – δi ) P(Ɵ) = 1 + e (βn – δi )
  • 44. ©drtamil@gmail.com - 2016 SUM UP VARIANCE OF EXPECTED VALUES • Calculate the variance of expected values for each cell; P*(1-P) • Sum up the variances according to rows and columns. • Negative sums in blue column act as denominator to tweak persons ability logit to maximise fit. • Negative sums in yellow row act as denominator to tweak items difficulty logit to maximise fit.
  • 45. ©drtamil@gmail.com - 2016 New Bi & new adjusted Di • Tweak the Persons Measure of Ability (Bi) based on the sum of residuals (Observed – Expected) and sum of variances of expected values; New Bi = Old Bi – (Sum of Variances/Sum of Residuals) • Tweak the Item Measure of Difficulty (Di) based on negative sum of residuals (Observed – Expected) and sum of variances of expected values; New Di = Old adj. Di – (Sum of Variances/Negative Sum of Residuals) • Get the average of all the new Item Measures of Difficulty (Di). Then get the new adj. Di by deducting the value of the average from all the new Item Measure of Difficulty (Di); i.e. (-0.046) - (-0.211) = 0.165 ~ 0.17 • Keep doing the iteration until the squared sum residuals of the entire matrix is as close as possible to 0. • For this dataset, 4 iterations was required before that was achieved.
  • 47. ©drtamil@gmail.com - 2016 Checking the Fit Bubble Plot
  • 49. ©drtamil@gmail.com - 2016 Bubble Plot Okay Erratic Too good to be true
  • 50. ©drtamil@gmail.com - 2016 Bubble Plot - Items Measure is the Y-axis. - Model S.E. is the diameter of the circle (reduced to 60%). - InFit Z-STD is the X-axis - We expect difficult items can be answered by the more able persons and easy items could be answered by all. - Item 27 is considered erratic, although it is difficult, both able and weak person cannot answer it. Under the previous exercise, for item 27; the Difficulty Index was D=32 and Discrimination Index was R=0.00. Difficult and yet unable to discriminate. So it was already detected as a problem question in earlier analysis.
  • 51. ©drtamil@gmail.com - 2016 Bubble Plot Okay Erratic Too good to be true 27; 2.7,2.89, 0.56 14; -0.8,1.21, 0.53 23; -1.3, 2.82, 0.57 8; 0.0, -3.40, 1.85 Lowest item Largest error of 1.85, so largest bubble. 20; 1.3, 1.21, 0.53 11; 1.4, 0.34, 0.56 17;0.2,-2.12,1.06 24; 0.2,-2.12,1.06 26; 1.7, 0.74, 0.58 28;0.5,-0.35,0.625;-1.0,-0.35,0.62 12;-1.1,-0.35,0.62 19;-1.0,0.24,0.59 Item No.; t Infit Zstd, Item Measures, S.E. Graph showing how it is plotted using the Infit Zstd and Item Measures. Bubble Size is 60% of the Model S.E. Zstd for 27 larger than 2.0 therefore an erratic item. Should be checked.
  • 52. ©drtamil@gmail.com - 2016 Bubble Plot Okay Erratic Too good to be true 27;D32,R0.00 14;D59,R0.83 23;D32,R0.83 8;D100,R0.00 Item 8 has high Difficulty Index & yet poor Discrimination Index. 20;D59,R0.33 11;D73,R0.33 17;D96,R0.17 24;D96,R0.17 26;D64,R0.33 28;D82,R0.335;D82,R0.67 12;D82, R0.67 19;D73,R0.83 Item No.; Difficulty Index, Discrimination Index. Graph showing the relationship of Difficulty Index with Item Measures and Discrimination Index with Infit Zstd. Item 27 has low Difficulty Index and yet poor Discrimination Index.
  • 54. ©drtamil@gmail.com - 2016 Person Item Map Mean for Persons Mean for Items Need tougher questions to test A, B, C D, E, F & G. Questions too easy. Not testing anybody. 1. Poor Students; n=4 (8%) 2.Good students; n=18 (82%) On target. Between the mean + 1sd. 13/30 = 43%
  • 56. ©drtamil@gmail.com - 2016 Scan the InFit Zstd for values larger than 2.0. For item 27; the InFit Zstd is 2.7, therefore larger than 2.0. As stated earlier, such items are considered erratic and should be removed or changed.
  • 57. ©drtamil@gmail.com - 2016VERY DIFFICULT = +2.89 logit N=21, score=7 avg.=0.33, Many got it wrong. BOTH y,z BREACHED ITEM NEED REVIEW Large +Z due to inconsistency in response. e.g. poorly able person can answer a difficult question. -2 < Z < +20.5 < y < 1.5 0.32 < x < 0.8 LOW PT. MEASURE CORELATION . SOME POOR STUDENTS CAN ANSWER ITEM CORRECTLY WHILE GOOD STUDENTS GOT IT WRONG
  • 58. ©drtamil@gmail.com - 2016 LOW PT. MEASURE CORELATION . SOME POOR STUDENTS CAN ANSWER ITEM CORRECTLY WHILST GOOD STUDENTS GOT IT WRONG 0.32 < x < 0.8 EXTREMELY EASY =-3.40 logit N=22, score=22 ave.=1, all correct
  • 60. ©drtamil@gmail.com - 2016 Scan the InFit Zstd for values larger than 2.0. For Person T & U; the InFit Zstd is larger than 2.0. Such erratic performance could be due to them getting some very easy questions wrong.
  • 61. ©drtamil@gmail.com - 2016 Expected Score ICC The InFit Zstd of item 27 is 2.7, therefore larger than 2.0. So we will check the ICC of item 27, why it is erratic. For comparison we will also look at ICC of item 23, with the Infit Zstd of -1.3.
  • 62. ©drtamil@gmail.com - 2016 ICC Item 27, Di= 2.9 R= 0.0 For a difficult question, a less able person shouldn’t be able to answer. e (βn – δi ) P(Ɵ) = 1 + e (βn – δi ) where; e= Euler’s Number, 2.7183 βn= Person’s ability measure δi= item difficulty measure But a more able person should be able to answer. But the blue line is not following the red line. So the ability is not consistent.
  • 63. ©drtamil@gmail.com - 2016 ICC Item 23, Di= 2.8 R=0.8 For a difficult question, a less able person shouldn’t be able to answer. e (βn – δi ) P(Ɵ) = 1 + e (βn – δi ) where; e= Euler’s Number, 2.7183 βn= Person’s ability measure δi= item difficulty measure But a more able person should be able to answer. Here the blue line is following the red line. So the ability is consistent with the difficulty.
  • 64. ©drtamil@gmail.com - 2016 0.66; ‘Poor’ item instrument reliability in measuring student learning ability. Poor targeting. Summary Statistics (All) +ve Person mean μ = 1.75 logit P[Ɵ]LOi=e(1.75+0.11)/(1+e(1.75+0.11)) = 6.4237/7.4237 = 0.865 (easy to pass) G=1.90 Separation index=(4G+1)/3=2.9 ‘Good’ Person separation into 3 groups. So can have 3 grades! 0.78 ‘Fair’ person reliability Cronbach-α :0.87 Good reliability assessment of student learning. Same value as ρKR20 calculated manually. Presence of one extreme person, “Mrs A” and one extreme item, “Item 27”, causes the InFit & OutFit MNSQ & Zstd not calculated. e (βn – δi ) P(Ɵ) = 1 + e (βn – δi ) where; e= Euler’s Number, 2.7183 βn= Person’s ability measure δi= item difficulty measure
  • 65. ©drtamil@gmail.com - 2016 How ρKR20 was calculated earlier. k = 30, Sum of pj & qj = 4.7521, σ2 = 31.827 ρKR20 = (30/29)*(1-(4.7521/31.827)) = 0.88 High reliability, almost homogeneous.
  • 66. ©drtamil@gmail.com - 2016 0.65; ‘Poor’ item measurement reliability in measuring student learning ability. Poor targeting. Summary Statistics (minus1) +ve Person mean μ = 1.58 logit P[Ɵ] LOi= e(1.58-0)/(1+e(1.58-0)) = 4.855/5.855 = 0.83 (easy to pass) InFit Zstd close to 0 & MNSQ close to 1, so data fit the model. G=1.97 Separation index=(4G+1)/3=2.96 ‘Good’ Person separation of 3 groups. So can have 3 grades! 0.80 ‘Fair’ person reliability Cronbach-α not available after exclusion of extreme person data. Interpretation of Person & item measurement reliability; -<0.67 is Poor - 0.67 – 0.80 Fair - 0.81 – 0.90 Good - 0.91 – 0.94 Very Good e (βn – δi ) P(Ɵ) = 1 + e (βn – δi ) where; e= Euler’s Number, 2.7183 βn= Person’s ability measure δi= item difficulty measure
  • 68. ©drtamil@gmail.com - 2016 Guttman Scalogram SMART Student-01 POORStudent-22 EASY ITEMS DIFFICULT ITEMS RESPONSE SORTED: EASY TO TOUGH
  • 69. ©drtamil@gmail.com - 2016 Guttman Scalogram Theorem 1. Persons who are more able / more developed have a greater likelihood of correctly answer all the items / able to complete a given task. Theorem 2. Easier items / task are more likely to be answered correctly by all persons. CARELESS PREDICT=1 PREDICT=0 GUESS
  • 70. ©drtamil@gmail.com - 2016 Guttman Scalogram as similarly arranged as descending Difficulty Index.
  • 71. ©drtamil@gmail.com - 2016 Item Misfit Response String
  • 72. ©drtamil@gmail.com - 2016 Item Misfit Response String Most misfit item: Exceed MNSQ Limit: 0.5 < y < 1.5 High Rating Response Zone 1. Item in red circles for the respective Persons were under rated Low Rating Response 0. Item in blue circles for the respective Persons were over rated
  • 73. ©drtamil@gmail.com - 2016 Conclusion • Item 27 needs to be relooked at. Needs better rephrasing of the question. Or better choice of answers • Rasch came out with almost the same kind of conclusion as our earlier analysis. • Rasch also identifies gaps in questions selection, showing us that there were too many easy questions, and absence of difficult questions that could test those with higher ability. Something the conventional analysis couldn’t show clearly.
  • 74. ©drtamil@gmail.com - 2016 Conclusion (continued) • Among the figures that were similarly generated between both methods were; – Difficulty Index D versus Item Measure of Difficulty Di (inverse relationship) – ρKR20 of 0.88 (manual) and 0.87 (Rasch). – Both method detected item 27 as an erratic item. Of course Rasch proved it with multitudes of figures and data. – With the conventional method, it requires experience to realise items with low difficulty index and non- discriminatory discrimination index as problem items.
  • 75. ©drtamil@gmail.com - 2016 Conclusion (continued) • Most importantly, Rasch detected the following; – ‘Poor’ item instrument reliability of 0.66; in measuring student learning ability. Poor targeting. – μ Bi= 1.75 logit; μ Di=-0.11 P[Ɵ]LOi=e(1.75+0.11)/(1+e(1.75+0.11)) = 0.865 – Based on the above, questions were too easy. – Separation index of 2.9 indicating ‘Good’ Person separation into 3 groups. So can have 3 grades! Excellent, Pass, Fail? Instead of the usual A, A-, B+, B, B-, C+, C, C-, D & E?