Sei sulla pagina 1di 18

Item analysis

Item analysis
The effectiveness and usefulness of any test depends upon the qualities of
the items that are included in it. The score of the test is obtained as a
result of its validity, reliability and the intercorrelation between two items.
To make the test more effective, the test constructor should study one by
one all the items which are to be included in it. This process is known as
item analysis.
In this method of item analysis, all the item of the test are studied
individually to see as to what number of persons of a group or percentage
has actually tried to respond or solve each item. Under this method, the
usefulness of the item is analysed because the quality and utility of the
test depends on those items which finally construct the test.
Item analysis is quite necessary for the selection of items, as the final test
form will be according to the objectives and subject matter of the test.
In the context of the item analysis, according to Freeman (1962), two
valuable things that should be considered while analyzing an item are
item difficulty and item validity.
According to Guilford (1954), item analysis is the best method for the
selection of items in a test finally.
According to Harper and Stevens (1948), item analysis is the most
appropriate way to bring improvement in measurement results.

ITEM DISCRIMINATION
There are two methods for this:
(a)significance of the difference
between proportion and
(b)correlational technique.

Significance of the difference


between proportion
In case of significance of difference between
proportions, the percentage or proportion of
individuals who answer the items correctly in the
high group is tested against the proportion in the
low group. If the difference is significant for a
particular item, then that item is accepted as
being the one which discriminates.
On the other hand, if the difference is nonsignificant for a particular item, then that item is
rejected as being one which does not discriminate.
The difficulty with this method is that it does not
reveal how well does each item discriminate.

Correlational technique
In case of the second method, that
is, correlational approach to item
analysis, a correlational coefficient is
computed to indicate the relationship
of the responses to the total test
score. This means how well the item
is doing and what the test itself is
doing.

ITEM DIFFICULTY (ID)


Another problem in determining difficulty level is
encountered after determining item validity.
These two independent aspects of item analysis are
interrelated because both item validity and difficulty
levels are determined simultaneously.
The number of difficulty levels equals the number of
items in a test.
Usually, the percentage of those who pass the test
item and those who fail is determined to find out the
difficulty level.
Neither a very easy item which can be responded to
by every individual of a group nor the difficult items
that cannot be responded by every individual of a
group are included in the test.

ITEM DIFFICULTY (ID)


According to Bradfield and Moredock (1957) the degree
of discrimination may be taken as an index of item
difficulty. An item which 90 percent of the group
answered correctly would be considered an easy item.
One which only 10 percent answered would be termed
very difficult.
An item that half of the class answered correctly and half
answered incorrectly is said to have 50 percent difficulty.
According to Tate (1955), item difficulty can be
calculated by finding out the proportion of the subjects
that have answered the item correctly. According to him,
difficulty value of an item is inversely proportional to the
proportion of students that have answered it correctly.

1. Descriptive statistics for the test


(subtest)
N
TR1
TR2
TR3
TR4
TR5
TR6
TR7
TR8
TR9
TR10
TR11
TR12
TR13
TR14
TR15
TR16
TR17
TR18
TR19
TR20
TR21
TR22
TR23
Valid N

Min.
296
296
296
296
296
296
296
296
296
296
296
296
296
296
296
296
296
296
296
296
296
296
296
296

Max.
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0

M (p)
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1

,95
,92
,70
,85
,56
,73
,56
,36
,58
,74
,31
,43
,25
,50
,41
,49
,41
,35
,46
,43
,15
,13
,34

St.dev.
,220
,279
,458
,353
,497
,445
,497
,481
,495
,439
,462
,495
,436
,501
,493
,501
,493
,478
,499
,496
,356
,335
,476

1. Item selection for final version of


test
1. Chose items that have mean closer
to 0.5, pick items from 0.36 to 0.75,
or from 0.3 to 0.8, where first one is
better.
2. We arrange those items from the
heaviest to hardest if the test is
about cognitive abilities.

2. Discriminativity of items
Discriminant validity indicates the possibility of distinguishing by
the general success in the test above-average underperformance
of respondents on the basis of their creations in the task
( Krkovi , 1978).
The validity of the correlation between the extent of the problem
creations and masterpieces throughout the test , and is oriented
to the homogenization of test results.
When you are talking about , discrimination test , usually states
that the correlation coefficient 0.0 to 0.2 indicates no or very
weak correlation , from 0.2 to 0.4 on a weak correlation , from 0.4
to 0.7 on high correlation , and values above 0.7 on the great
relationship between the variables (Petz , 1992) .
The average discriminant validity of the tasks of this test is riu =
0.368 , when talking about a pilot version of the test (23 items) , a
shortened version of the test has an average amount,
discrimination in riu = 0.4485 ( 12 items) , which would therefore
fall within the area of secondary links .

1. Descriptive statistics for the test


(subtest)
N
TR1
TR2
TR3
TR4
TR5
TR6
TR7
TR8
TR9
TR10
TR11
TR12
TR13
TR14
TR15
TR16
TR17
TR18
TR19
TR20
TR21
TR22
TR23
Valid N

Min.
296
296
296
296
296
296
296
296
296
296
296
296
296
296
296
296
296
296
296
296
296
296
296
296

Max.
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0

M (p)
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1

,95
,92
,70
,85
,56
,73
,56
,36
,58
,74
,31
,43
,25
,50
,41
,49
,41
,35
,46
,43
,15
,13
,34

St.dev.
,220
,279
,458
,353
,497
,445
,497
,481
,495
,439
,462
,495
,436
,501
,493
,501
,493
,478
,499
,496
,356
,335
,476

riu
0,249**
0,177**
0,405**
0,351**
0,458**
0,410**
0,478**
0,372**
0,533**
0,451**
0,390**
0,456**
0,209**
0,525**
0,374**
0,506**
0,349**
0,209*
0,348**
0,469**
0,193*
0,118*
0,435**

Ways of testing discriminant


validity of items
Fajgelj (2005.), cites two possible
ways of determining, discrimination
test items:
1. item correlation with the total score
(item-total correlations) and
2. correlation with external standard or
external criterion.

Fajgelj (2005) also states and two indices,


1. the index of validity of items and
2. reliability index of items.
Ad 1. Index validity of items (iriC) is the product of the
standard deviation of item score and correlation of
items with some external criterion,
Ad 2. the index of reliability of items (iriy) is a product
of standard deviations and correlations of items with
the total score of the test.
. Cohen and Swerdlik (1999th, according Fajgelj,
2005). Indicate that the choosing the items
that have a high index of validity maximizes
the criterion validity of the test, the dial of
items that have a high level of reliability
maximizes the internal consistency test.

1. the index of reliability of items iriy


riu

iriy

St.dev.
,95 ,220

0,249**

0,055

0
0

1
1

,92 ,279
,70 ,458

0,177**
0,405**

0,049
0,185

296
296

0
0

1
1

,85 ,353
,56 ,497

0,351**
0,458**

0,124
0,228

TR6
TR7

296
296

0
0

1
1

,73 ,445
,56 ,497

0,410**
0,478**

0,182
0,238

TR8
TR9

296
296

0
0

1
1

,36 ,481
,58 ,495

0,372**
0,533**

0,179
0,264

TR10
TR11

296
296

0
0

1
1

,74 ,439
,31 ,462

0,451**
0,390**

0,198
0,180

TR12
TR13

296
296

0
0

1
1

,43 ,495
,25 ,436

0,456**
0,209**

0,226
0,091

TR14
TR15

296
296

0
0

1
1

,50 ,501
,41 ,493

0,525**
0,374**

0,263
0,184

TR16
TR17

296
296

0
0

1
1

,49 ,501
,41 ,493

0,506**
0,349**

0,254
0,172

TR18
TR19

296
296

0
0

1
1

,35 ,478
,46 ,499

0,209*
0,348**

0,100
0,174

TR20
TR21

296
296

0
0

1
1

,43 ,496
,15 ,356

0,469**
0,193*

0,233
0,069

TR22
TR23

296
296

0
0

1
1

,13 ,335
,34 ,476

0,118*
0,435**

0,040
0,207

Min.

Max.

TR1

296

TR2
TR3

296
296

TR4
TR5

Valid N

296

M (p)

2. Index validity of items iriC


N

Min.

Max.

M (p)
St.dev.
,95 ,220

ric

iriC

iriy

0,137*

0,030

0,055

TR1

296

TR2
TR3

296
296

0
0

1
1

,92 ,279
,70 ,458

0,081
0,067

0,023
0,031

0,049
0,185

TR4
TR5

296
296

0
0

1
1

,85 ,353
,56 ,497

0,136*
0,129*

0,048
0,064

0,124
0,228

TR6
TR7

296
296

0
0

1
1

,73 ,445
,56 ,497

0,083
0,059

0,037
0,029

0,182
0,238

TR8
TR9

296
296

0
0

1
1

,36 ,481
,58 ,495

0,049
0,093

0,024
0,046

0,179
0,264

TR10
TR11

296
296

0
0

1
1

,74 ,439
,31 ,462

0,132*
-0,023

0,058
-0,001

0,198
0,180

TR12
TR13

296
296

0
0

1
1

,43 ,495
,25 ,436

0,108
0,001

0,053
0,000

0,226
0,091

TR14
TR15

296
296

0
0

1
1

,50 ,501
,41 ,493

0,146*
0,120*

0,073
0,059

0,263
0,184

TR16
TR17

296
296

0
0

1
1

,49 ,501
,41 ,493

0,172**
0,051

0,086
0,025

0,254
0,172

TR18
TR19

296
296

0
0

1
1

,35 ,478
,46 ,499

0,011
0,176*

0,005
0,088

0,100
0,174

TR20

296
296

0
0

1
1

,43 ,496
,15 ,356

0,219**
0,065

0,108
0,023

0,233
0,069

296

,13 ,335

0,048

0,016

0,040

296

,34 ,476

0,200**

0,095

0,207

TR21
TR22
TR23
Valid N

296

What are appropriate values


Guilford (1968th, according Fajgelj
2005) lists of acceptable values
range from 0.30 to 0.80 when
talking about the item-total
correlations, and the size of the
required coefficient varies from
author to author.

What are appropriate values


Nunnally and Bernstein summarize the discussion of the item
discrimination and their choice in the final form of the test as
follows:
1. "If the discrimination coefficient is calculated as the linear
correlation coefficient must be used correction" according to
the expression:
rjY Y j

r j (Y j )

j Y 2 j Y rjY
2

rjY linear correlation j-th of items with a total score;


j standard deviation j-th item;
Y standard deviation of the total score;
rj(Y-j) corrected item-total correlation.

2. "Reasonable lower limit discrimination coefficient is 0.30"

1. Descriptive statistics for the test


(subtest)
riu

rj(Y-j)

ric

iriC

iriy

0,249**

0,194

0,137*

0,030

0,055

0,177**
0,405**

0,106
0,292

0,081
0,067

0,023
0,031

0,049
0,185

,85 ,353
,56 ,497

0,351**
0,458**

0,267
0,347

0,136*
0,129*

0,048
0,064

0,124
0,228

1
1

,73 ,445
,56 ,497

0,410**
0,478**

0,308
0,369

0,083
0,059

0,037
0,029

0,182
0,238

0
0

1
1

,36 ,481
,58 ,495

0,372**
0,533**

0,258
0,430

0,049
0,093

0,024
0,046

0,179
0,264

296
296

0
0

1
1

,74 ,439
,31 ,462

0,451**
0,390**

0,354
0,282

0,132*
-0,023

0,058
-0,001

0,198
0,180

TR12
TR13

296
296

0
0

1
1

,43 ,495
,25 ,436

0,456**
0,209**

0,347
0,098

0,108
0,001

0,053
0,000

0,226
0,091

TR14
TR15

296
296

0
0

1
1

,50 ,501
,41 ,493

0,525**
0,374**

0,421
0,257

0,146*
0,120*

0,073
0,059

0,263
0,184

TR16
TR17

296
296

0
0

1
1

,49 ,501
,41 ,493

0,506**
0,349**

0,399
0,229

0,172**
0,051

0,086
0,025

0,254
0,172

TR18
TR19

296
296

0
0

1
1

,35 ,478
,46 ,499

0,209*
0,348**

0,087
0,229

0,011
0,176*

0,005
0,088

0,100
0,174

TR20
TR21

296
296

0
0

1
1

,43 ,496
,15 ,356

0,469**
0,193*

0,358
0,103

0,219**
0,065

0,108
0,023

0,233
0,069

TR22
TR23

296
296

0
0

1
1

,13 ,335
,34 ,476

0,118*
0,435**

0,039
0,328

0,048
0,200**

0,016
0,095

0,040
0,207

Min.

Max.

M (p)
St.dev.
,95 ,220

TR1

296

TR2
TR3

296
296

0
0

1
1

,92 ,279
,70 ,458

TR4
TR5

296
296

0
0

1
1

TR6
TR7

296
296

0
0

TR8
TR9

296
296

TR10
TR11

Valid N

296

Potrebbero piacerti anche