Analisis Soal Ujian Tengah Semester Matematika Kelas VII Tingkat SMP Negeri di Kota Padang Tahun Ajaran 2016/2017

Abstract

The purpose of this research is to know the quality mid term of math class VII SMP Negeri in Padang City academic year 2016/2017 when viewed from aspects of logical validity, empirical validity, reliability, difficulty, differentiation and distractor problems. This research is a descriptive research with two approaches that is qualitative approach and quantitative approach. Qualitative approach is used to find out the logical validity of the problem, while the quantitative approach is used to determine the empirical validity, reliability, difficulty level, distinguishing power and distractor problems. The results of this study indicate that the quality of items in terms of the analysis of logical validity about mid term of math class VII SMP Negeri in Padang City academic year 2016/2017 there are 2 invalid questions, 7 questions have medium validity and 31 questions with high validity. The quality of empirical validity about mid term of math class VII SMP Negeri in Padang City academic year 2016/2017 is 0.49 with interpretation of empirical validity is moderate. Quality of reliability about mid term of math class VII SMP Negeri in Padang City academic year 2016/2017 is 0.63 with the interpretation of reliability is quite enough. Levels of difficulty about mid term of math class VII SMP Negeri in padang at study year 2016/2017 of 40 problems there are: 6 problems are relatively easy, 30 questions are classified as moderate, 4 the problem is quite difficult. The different power of mathematics mid term of math class VII SMP Negeri in Padang City academic year 2016/2017 from the 40 item is classified into; 14 items of the test have low power different, 13 items of the test have middle power different, 4 items test have high power different, and 9 items of the test have very good power different. From the distractor of mathematics mid term of math class VII SMP Negeri in Padang City academic year 2016/2017 can be as much 30% distraktor can't yet function and needs to be revised, because is still qualified bad and very bad, meanwhile 70% another distraktor can function.

Keywords: validity, reliability, levels of difficulty, different power, distractor.