Manuel Barrera, PhD
Associate Professor, Metropolitan State
Research Associate, National Center on
Educational Outcomes, University of Minnesota
First presented at Free Minds, Free People
Conference, Chicago, Illinois
July 11, 2013
Three Points
• Historical results of state accountability testing with
specific case of MN
• Convolution of Summative testing to make formative
• The case for a return to classroom-based assessment,
especially for English language learners and students with
Historical Results of State Testing
• For students of color, ELLs, and students with disabilities:
• Chronic Failure Across Years, Across Tests, Across States
• Persistent Academic Disparity
• $Millions spent to demonstrate the obvious: Children of
color, English language learners and students with
disabilities do worse than their White peers
• For White students:
• Despite doing better than their non-White peers, demonstrate
persistently low performance than their “advantages” would seem
to dictate, especially in mathematics and science
• Charter Schools are the Worst
MN Reading Results by Major Groups 2008
(publically accessible data)
10th grade Passing Rates on MN Reading 2005-2008
(publically accessible data)
MN Charter School 10th grade Reading Passing Rates
(publically accessible data)
Summative Testing as Formative
• Successive Administrations’ Policy:
• State tests can be used to help determine educational
improvements based on “data-based decision-making”
• Teachers can use state tests as a way to determine how to teach
• Teacher “merit” can be determined by progress on state tests and
improvement based on improved test results
Basics of Standardized Tests
• Designed to provide stable results over time: all 5th
graders should perform, on average, at a 5th grade level
when they are 5th graders
• “Average” is the observed Mean of scores when a test is
given, which means exactly ½ of all students tested lower
and ½ tested higher: a score within 1 standard deviation
from the mean on either side represents 64% of all
scores. “Normal” is considered to be within 2 SD of the
• An average score can “bleed” into other average scores
across grades
Basics II
• The average score is not actually the “passing score”
(e.g., “partially met” vs. “met” standard)
State tests are explained based on varying and often
opposed “derived” scores (e.g., a percentile rank, a
percent score, a standard “z” or “T” score, or a stanine)
None of these score is actually used to determine
“passing”, which is often a taskforce of experts’ or more
aptly, a political, decision known as a “cut score”
In short, these scores are Summative (How did they do?)
not Formative (what explains the results?).
Hence, none of these interpreted scores provides relevant
information for guiding instruction nor determining
program changes
Classroom-based Assessment
• Analyze what students know now and determine what
they need to know to learn what I am going to teach.
Teach that
What did I teach and how did I teach it?
Informs what I should test and how to test it?
Measure progress on what was taught and the skills
needed to learn it.
Provide individual student progress and teach students
how to monitor their own progress
Improved Classroom Assessment and
Instruction Cost Money
• What you need
• More Teachers
• More resources—including technology and teacher support
• Fewer students per class
• More student support including community supports
• More professional development
• Better leadership development (as opposed to better
It All Costs LESS than Current Policies
• It All Costs LESS Than Spending on Tests That Tell You
NOTHING about the problems and then sending students
into the “prison-industrial complex”
• It Costs Less than funding the worst kind of schooling,
charter schooling
• It Cost Less to Make Less Violent Individuals in a Less
Violent Society Than to Produce Wars where violent
individuals can play out their fantasies
Language Policy is a Social Policy
• Testing policy of learners who we already know will not do
well on an English test is social violence
• It produces violence in retaliation
• Supporting different languages of learners and
communities creates the best potential for creating
multicultural integration and human interactions
• No More Useless and Inappropriate Tests; More Teaching
that facilitates learning

History of An(cad)emic Progress of English Language