v Validity v Reliability v Objectivity v Comprehensiveness

ลกษณะขอสอบ ทด v Validity v Reliability v Objectivity v Comprehensiveness v Practicability v v Good level of difficulty Good discrimination power

การวเคราะหขอสอบท งฉบบ m Mean, SD m Median m Mode m Range : Max - Min m Standard error of measurement ( SEM ) m Reliability ( rtt : KR-21, KR- 20 ) m Difficulty index ( p( m Discrimination power ( r ) ¡ MPL = S )AI ) n

Reliability

Validity v. s Reliability not Valid, not Reliable not Valid, Reliable

Reliability ( rtt ) K-R 21 K-R 20 = n x (n-x) 1 n. Sx 2 n-1 = n piqi 1 S x 2 n-1 n S 21 1 Cronbach’ s coefficient - = S 2 x n-1

Pretest Analysis The task should be done by the teachers to evaluate the test to be used in term of : Ø Validity Ø Difficulty factor (DF) / Acceptability index (AI) Ø Minimum pass index (AI)

Modified Angoff Method กลมทคาบเสนควรจะตอบขอสอบแตละขอไดอยางนอ ยรอยละเทาใดของคะแนนเตม 1. 00 Question 1 2 3 4 5 6 7 8 9 10 Expected passing score = 60% Probability for Correct Answer. 75. 60. 50. 40. 75. 50. 40 Sum 6. 00

Item analysis base on student, at one point in the time, reflecting some characteristics of the item

87 83 81 80 80 79 79 78 76 76 75 73 73. 72. 71. . . 47 46 46 45 45 44 42 42 42 41 39 39 38 37 33 31 ����� ��� H 27% H 33% H 50% L 33% L 27% �����

Item # 41 High 13 ( n = 53 ) Low ( n = 53 ) = = P H+L 2 n r H–L n การคำนวณคา r p และ A B C D ** E 3 4 3 3 40 1 9 8 29 0. 04 0. 11 0. 10 0. 65 0. 05 - 0. 12 - 0. 09 0. 20 6 0. 08 - 0. 06

Item Discrimination Index Point BIserial Correlation( rpbs ) What does the point biserial measure? The relationship between two variables, a categorical variable (i. e. , the student either answered the test item correctly or incorrectly), with a continuous variable(i. e. , percent score on the examination. ( How do you interpret a point biserial correlation? As with all correlation indices, the point biserial ranges from – 1. 00 to +1. 00. A positive point biserial tells us that those scoring higher on the exam were more likely to have better item “discriminates” between high-scoring and low-scoring students. ( www. som. tulane. edu.

To calculate the rpbi for each item, use the following formula: rpbi = Mp–Mq St pq rpbi = point-biserial correlation coefficient Mp = whole-test mean for students answering item correctly (i. e. , those coded as 1 s) Mq = whole-test mean for students answering item incorrectly (i. e. , those coded as 0 s) St = p= q= standard deviation for whole test proportion of students answering correctly (i. e. , those coded as 1 s) proportion of students answering incorrectly (i. e. , those coded as 0 s)

Uses of Indices : Review of Question Difficulty Index (p) = R / T x 100 (20)30 40 Difficult 50 60 Recommended Acceptable 70(80) Easy

Uses of Indices : Review of Question Discrimination Index (r) = H-L x 100 N Þ 0 0. 05 0. 15 0. 25 0. 35 (0. 20) Discard (most likely) Revise Good Excellent