learning from the national evaluation of i3: challenges...
TRANSCRIPT
Learning from the
National Evaluation of
i3: Challenges,
responses, and future
plans
i3 Learning Community Meeting
September 11-12, 2014
pg
What We Have Learned
A grant program/agency can incentivize high-quality evaluations
Across the first 3 cohorts, the majority of the studies have the potential to meet NEi3 standards with or without reservations
– 89% of all grants have potential to meet standards
– 83% of development grants using RCTs, QED (twice as many QEDs)
– 100% of validation and scale-up grants using RCTs or QEDs (about half and half)
National Evaluation of i3 1
pg
What We Have Learned
At the same time, this emphasis and the success of helping evaluators design strong studies does not eliminate risks that can undermine what can be learned
– Low power: 17 development grants (as designed), 3 validation grants (as implemented)
– Serious confounds: 6 development grants, 2 validation grants, 1 scale-up grant (1 of multiple studies)
– Only studying part of the full i3-supported intervention or sample: 2 development grants, 2 validation grants, 1 scale-up grant
– Dropping down in rigor of design: 6 development grants proposed QEDs but ended up implementing promise evaluations
National Evaluation of i3 2
pg
What We Have Learned
Institutionalizing rewards for prior evidence has increased attention on/efforts to implement strong studies
– OII and other federal agencies are increasingly focused on evidence—as rationale for funding and as a result of funding
– OII/IES are increasingly clear about the evidence requirements for winning another i3 grant—bar has been set higher
In earlier meeting, this link to evidence cited as one definition for the “innovation” in Investing in Innovation
National Evaluation of i3 3
pg
What We Have Learned
A tiered evidence program such as i3 faces issues
about appropriate design standards for lower level
grants
– Many of the development grants are developing,
investigating innovative approaches
– Guidance on rigorous designs may be misaligned with stage
of intervention
• Focus on implementation may be more aligned?
• How to weigh innovation versus need for believable evidence
National Evaluation of i3 4
pg
What We Can Learn
Effectiveness of each individual
intervention/approach at improving student
achievement
– Which interventions have been shown to be
“effective” and “promising”
This is the smallest ‘grain size’
5 National Evaluation of i3
pg
What Else Can We Learn
Can we learn something more systematic about
educational interventions, can we go beyond “one-off”
evaluations
Is there a way to group interventions and summarize
findings within groups
– WWC has moved away from topic area summaries and
towards intervention reports
– Dangers of summarizing effects of groups of i3 interventions
6 National Evaluation of i3
pg
What Else We Can Learn
Group the i3 interventions by
–Absolute Priority?
–Outcome domains?
– Intervention components?
7 National Evaluation of i3
pg
Group by broad policy area/goal
National Evaluation of i3 8
Author, Affiliation, i3 role;
FY2010-2013 i3 Grants by Absolute Priority
pg
Group by intended outcome?
National Evaluation of i3 9
Author, Affiliation, i3 role;
80
43
18
7
9
8
7
5
4
5
5
3
2
4
4
4
4
4
4
11
4
6
3
1
1
2
1
1
1
1
1
1
0 20 40 60 80 100
ELA achievement
Math achievement
Science achievement
Social studies achievement
Classroom/school climate
General Achievement
Attitudes / beliefs / self-efficacy / motivation
Attendance
Understanding of content / pedagogical content knowledge
Instructional quality or effectiveness
College/career readiness
High school graduation
Retention
Engagement / attentiveness / involvement
Use and expertise of assessment, monitoring, measures of achievement
Collaboration among students/teachers/admin/staff
Quality of leadership
Language/Literacy development
Cognitive development/ Executive functioning Meet NEi3 Criteria With or Without Reservations
Provide Evidence of Promise
Will not meet NEi3 criteria for grant type
*nearly all grants are conducting multiple contrasts, some across multiple domains
pg
Group by approach to affecting change?
National Evaluation of i3 10
Author, Affiliation, i3 role;
78%
57%
41%
27%
24%
18%
18%
16%
16%
8%
8%
6%
6%
4%
2%
Educator / School Leader Professional Development
Curriculum / Instructional materials / Instructional tools / Coursework
Support from Admin / leadership
Learning communities (PLCs, Online networks, etc)
Coaching / Mentoring of Educator / School Leader
Data Collection and analysis; monitoring and/or reporting of data
Selection, recruitment, hiring, and/or placement of Ss/Ts/Others
College / Career Readiness
Parental/community involvement, engagement, support, and training
System of student assessment
Teacher Assessment/Evaluation
Schedule / Calendar
Student Mentoring
Student Case Management
Environmental / Physical structure
0% 20% 40% 60% 80% 100%
*Many grants utilize multiple key components
pg
Revisiting Original Goal of Learning Community
Trying to identify site-level contextual factors that help
us understand variation in effects across sites
– Site-level variation can be studied only in evaluations where each site has
an effect (i.e., each site has treatment and comparison units)
Grantees selected for this learning community based on having
evaluation designs that would allow for studying site-level
variation
The field has not moved far in measuring common contextual
factors
National Evaluation of i3 11