Constructed Response Is it worth the effort? Up Front Information Mark Vanacore Technology Coordinator Albion Central School District Formerly Teacher of Physics, Earth Science, Chemistry, Mathematics, Computer Programming, Cisco Networking Disclosure I am 100% in favor of … NYS State Regents Exams Common Core Testing Computer Based Testing Making some changes in the format of State Summative Assessments There are things I wonder about… Free Response Questions that aren’t Math Questions that are allowed sometimes but not others, on both exams Science DBQ Good Bad ??? Vague Acceptables Not at all Limited to Science Exams … Things I like … Unexpected New Questions… Formative vs Summative Assessments Goals : Formative - provide feedback to student and teacher concerning the student’s ability to respond to questions and problems. Summative - A check to determine the level of competency concerning the covered content. Confusion about the Purpose of State Exams State exams are about determining competency on the covered material. Question: Can we determine if a student understands something with a multiple choice question? Some Vocab Anchors Anatomy of a Multiple Choice Question A root question An answer to the root question Several Distractors Vocab Anchors continued Anatomy of a Constructed Response Question Some Information - Data, Tables, Passage A Root Question A Blank bit of Paper A Scoring Guide with the intensions of the test writer. Some Testing Vocabulary Reliability Test - Retest - same results over time? Parallel Forms - Form A vs Form B? Inter-rater - Do all teacher grade the same? *Internal Consistency ? Checks within test? Validity Does it test what it says it tests? Concerning State Science Tests In General, State Science Exams ... Pass all Reliability measures Pass all Validity Tests Methodology The data was pulled from copies of the scanable student forms. The total points accumulated in each part of the exam The raw parts were converted in %’s for comparison Methodology Exams Looked at took place in June 2014 Earth Science * Living Environment * Chemistry Physics Integrated Algebra * Common Core Algebra * Geometry Trig Data Concerning the 2014 Earth Science Exams Sample Size 96 Students Range of Scaled Scores 44 - 98 32 Mastery 85+ 21 College Ready 75-84 24 Passing 65-74 14 Gandolfed 0-64 Comparison of MC to Overall Score Correlation between the MC % Correct and the overall Scale Score. Correlation Coefficient .9596 Very High Correlation - Score well in the MC Score Well on the Test Correlation Graph Correlation between CR and Overall Score Correlation between % of CR Points received and overall score Correlation Coefficient .9351 Still pretty High Score well on the CR score well on the test Correlation Graph Correlation between Practical and Total Score Points Received on Practical compare to Scale Score Correlation Coefficient .6311 Much Lower but expected, the Practical is a series of skills taught throughout the year which kids learn well, range of scores 8-16 with an average of 14/16 points. What more can we learn? MC to Score .9596 CR to Score .9351 Either score is good predictor of overall performance. But My lower kids need those Constructed response questions… Do they Really ??? I Subtracted the CR percent from MC Percent example %MC-%CR = Some Number Mastery Students Averaged -4.22 College Ready Averaged +1.52 Passing Averaged +3.96 Gandolfed Averaged +4.86 Take Away ... The Test as is could be modified to eliminate the Constructed response from the overall score without significantly altering the outcome of the test. A simple “scantron” form would be sufficient to determine the range of student level Mastery-5 College Ready-4 Passing-3 Nearly Passing-2 Not Passing-1 The Danger … The impression that free response is not ever needed. Free Response is critical in Formative Assessment It is the most practical way to determine if the student can complete complex processes to get at an answer. Getting a little more out of MC MC with Explanations What is the correct answer and how do you know? Of the Distractors which one is the most correct and Why? How could you change the root of the question to make one of the distractors correct? Turning CR into MC Ask Students to turn a Lab Question or CR from an old exam into a MC Question. Getting student to think like test writers may help improve their scores. Other Tests - Living Environment Sample Size 59 Mastery 19 College 19 Passing 14 Gandolfed 7 Correl MC to Score .9592 Correl CR to Score .9110 Integrated Algebra Scores Sample Size 132 Mastery 23 College 58 Passing 31 Gandolfed 20 Correl MC to Score .9106 Correl CR to Score .8033 Scatter Plot MC vs Scale Score Scatter Plot CR vs Scale Score Some Confusing Data Diff MC - CR Mastery +24 College +34 Passing +39 Gandolfed +31 It was a few points in science what happened? Scatter Plot MC vs CR Integrated Algebra A change in Focus and A Trap Revealed Math teachers focused on MC questions teaching tricks and tips to maximize the chance of selecting correct response. Including but not limited to back substitution of distractors to find the answers. Why? In Algebra you only need about 35 correct responses in part 1 to Pass the test. Common Core Algebra Results Sample Size 117 Mastery 1 College 6 Passing 69 Gandolfed 41 Correl MC to Score .8361 Correl CR to Score .7713 Diff MC to CR Master -6 (1 Student) College +19 Passing +21 Gandolfed +25 There are significant difference in expectation between the Integrated Algebra test and the Common Core Algebra Test. Common core question are more about understanding and application rather than plug and check methodologies. Summary Myth Busted We do not really need elaborate Constructed response questions to determine if kids have met the criteria for being awarded a passing grade and awarded Regents Credit. Well designed MC questions can get the information we need, and we have plenty of really good questions across the content. Summary Myth Confirmed We still need free response question to ensure that our kids have the skill needed to select the correct multiple choice answer. Understanding and being able to apply knowledge is still important regardless of the format of the summative assessment. Raw Data Slides www.albionk12.org/forum/presentations/ Spread Sheets in Excel Slide Show in PowerPoint 2010 Credits Statistician: Marge Innovera Chairman, Math Dept. Horatio Algebra Chairman, Staff Physics Dept. Victor Analysis Chairman, Underemployment Study Group Art Majors Chief Accountant Candace B. Rittenoff Child Care Provider A. Hugh Nokitov Complaint Line Operator Xavier Breath Downsizing Consultant Candace Guy Special Thanks to My Teachers for Copying the answer sheets so I could collect the data.