The Effects of Reusing Written Test Items: A Study Using the Rasch Model
Table 3
Psychometric results.
Item difficulty (classical test theory)
Item difficulty and fit statistics (Rasch model)
Number of persons per analysis
Test 1–Test 4 basic analysis
First use/reuse analysis
Test 1
Test 2
Test 3
Test 4
Number of items per analysis
19
14
14
15
33
33 + 15 + 2 = 50 items
Easiness (s.e.)
I.Nb.
p
p
p
p
Easiness
Outfit
Infit
1st use
2nd use
3rd use
Published items
1
0.99
0.98
0.94
0.97
1.66
0.68
0.86
1.86 (0.28)
—
19
0.79
0.79
0.85
0.25
−1.63
1.07
1.04
−1.49 (0.12)
—
23
0.96
0.95
0.91
0.88
0.56
1.02
0.94
0.75 (0.18)
—
37
0.86
0.85
0.81
0.94
0.02
0.89
0.96
0.21 (0.16)
—
New in test 1
5
0.90
−0.11
0.80
0.97
−0.25 (0.36)
7
0.91
0.01
1.13
0.96
−0.13 (0.38)
25
0.87
0.76
−0.71
1.00
0.96
−0.55 (0.33)
−1.05 (0.36)
27
0.98
1.60
0.52
0.88
1.46 (0.72)
31
0.89
0.91
−0.11
0.61
0.81
−0.35 (0.35)
−0.01 (0.29)
32
0.92
0.14
1.16
0.96
0.00 (0.39)
33
0.94
0.99
0.95
0.76
1.00
0.32 (0.44)
2.28 (0.58)
40
0.92
0.87
−0.05
0.96
0.97
0.00 (0.39)
−0.24 (0.42)
41
0.82
0.78
−0.98
1.03
1.02
−0.87 (0.31)
−1.07 (0.23)
43
0.78
−1.01
1.11
1.01
−1.14 (0.29)
44
0.72
0.75
−1.35
0.94
0.96
−1.49 (0.27)
−1.24 (0.22)
45
0.40
0.45
0.40
−2.74
1.02
1.02
−2.95 (0.27)
−2.66 (0.21)
−2.73 (0.32)
47
0.94
0.97
1.63
0.73
0.88
0.32 (0.44)
2.90 (0.59)
48
0.65
0.64
−1.21
1.05
1.07
−1.84 (0.26)
−0.60 (0.20)
49
0.98
0.98
0.96
1.88
0.71
0.89
1.46 (0.72)
1.56 (0.52)
2.61 (0.51)
New in test 2
2
0.99
2.97
0.29
0.84
2.98 (1.00)
6
0.90
0.93
0.03
0.86
0.96
−0.08 (0.29)
0.50 (0.52)
8
0.86
−0.55
0.91
0.92
−0.52 (0.26)
30
0.87
0.62
−0.80
0.82
0.91
−0.37 (0.27)
−0.65 (0.25)
34
0.65
0.60
−1.47
1.08
1.06
−1.78 (0.21)
−0.79 (0.20)
New in test 3
4
0.91
0.22
0.51
0.72
0.28 (0.48)
15
0.85
0.98
1.30
1.13
0.90
−0.38 (41)
3.32 (0.71)
29
0.76
−1.10
1.01
1.03
−1.05 (0.36)
36
0.94
0.70
−0.49
0.87
0.93
0.76 (0.56)
−0.30 (0.21)
50
0.93
0.44
0.96
0.78
0.50 (0.52)
New in test 4
3
0.82
0.15
0.82
0.93
0.53 (0.24)
9
0.84
0.29
0.73
0.90
0.67 (0.25)
10
0.82
0.15
0.88
0.93
0.53 (0.24)
42
0.87
0.34
0.75
0.91
0.72 (0.25)
I.Nb. = Item number; = item difficulty (classical test theory); easiness = relative item difficulty (probabilistic psychometric analysis); higher values indicate easier items; s.e. = standard error; outfit/infit = fit statistics (indicating the degree of fit with the Rasch model); values around +1 are regarded as indicating a fit with the Rasch model; values close to 0 indicate an overfit (less variation than expected); values exceeding 1 indicate an underfit (more variation than expected).