Title: The Impact of Literacy Coaching on Teachers
1The Impact of Literacy Coaching on Teachers
Value-Added to Student Learning in Literacy
Collaborative
- Gina Biancarosa, University of Oregon
- Anthony S. Bryk, The Carnegie Foundation for the
Advancement of Teaching - Allison Atteberry, Stanford University
- Heather Hough, Stanford University
- Institute of Education Sciences
- Annual Conference
- June 2010
2Key Features of Literacy Collaborative
- Comprehensive school reform program designed to
improve elementary childrens reading, writing,
and language skills primarily through
school-based coaching - Used in over 700 elementary schools in 200
districts across 26 states - Intensive professional development of coaches
(selected from school faculty) - Trained over one year (Lesley University and the
Ohio State University) - Ongoing support from local and national network
- Coachs role and duties
- Half-time teaching, half-time coaching
- In-school professional development courses
- One-on-one coaching sessions
3Main Research Questions
- Does Literacy Collaborative improve the
value-added to student literacy learning? - Can Literacy Collaborative effects be attributed
to coaching, either directly or indirectly?
4Student Data
- Value-added analyses focused on grades exposed to
LC professional development (K-2) - Sample 8576 children, 341 teachers, and 17
coaches in 17 public schools across 8 states in
the Eastern U.S. - Children tested in fall and spring for 4 years to
measure change over time in students literacy
learning using - Dynamic Indicators of Basic Early Literacy Skills
(DIBELS) - Terra Nova in spring
Low Income 46.0
Race/Ethnicity African-American Latino Other White 15.5 5.8 7.2 70.6
Limited English Proficiency 4.0
5Accelerated Longitudinal Cohort Design 6 cohorts
studied over 4 years
Year of Study Year of Study Year of Study Year of Study Year of Study Year of Study Year of Study Year of Study
First Year First Year Second Year Second Year Third Year Third Year Fourth Year Fourth Year
Fall Spring Fall Spring Fall Spring Fall Spring
K C C D D E E F F
1 B B C C D D E E
2 A A B B C C D D
Grade
Training year
Year 1 of implementation
Year 2 of implementation
Year 3 of implementation
6Our early literacy scale
Mean at 2nd grade end Mastery of component
skills Reads 90 wpm Answers 2/3 of 1st grade
comprehension questions correctly, 1/3 of 2nd
grade questions correctly
4
- Equal differences on scale imply equal
differences on the trait measured at any level - Reported in logits (which describe the
probability of a student with a given ability
level getting a particular item right or wrong) - But what do they mean given the particular
assessments used?
Mean at 1st grade end 2nd grade entry Accurate
(not fast) PA Reads 50-60 wpm Answers 1/3 of 1st
grade comprehension questions correctly
3
Mean at K end 1st grade entry Accurate and fast
letter recognition Good initial sound PA Little
evidence of decoding
2
Mean at K entry Names about 30 letters in a
minute Very low phonemic awareness (PA)
1
7Value-added Hierarchical Cross-classified Effects
Modeling
- Four Levels time (students x teachers)
school - Repeated measures on students (level 1)
- Students (level 2) who cross Teachers (level 3)
over time - All nested within Schools (level 4)
- The analysis model can be conceptualized as a
joining of 2 separate multi-level models - One two-level model for individual growth in
achievement over time, and - A second two-level model which represents the
value-added that each teacher in a school
contributes to student learning in that school in
a particular year.
8Value-added effects by year (prior to adding
coaching as predictor)
Average student learning growth in an academic
year (1.02 logits)
Year 1 Year 2 Year 3
Average value-added (overall) .164 .280 .327
Performance improvement 16 28 32
Effect size .22 .37 .43
School 95 plausible value-added range .23 .28 .37
Teacher 95 plausible value-added range .51 .71 .91
9Explaining variability in value-added effects
- Tested models with cumulative number of coaching
sessions per year (derived from coach logs) - Per teacher
- Averaged across teachers at school-level
- Also tested a variety of controls thought to
influence teachers openness to, participation
in, and selection for coaching - Prior use of reform literacy practices
- Role conception
- School commitment
- New to school
10Summary of findings
- Coaching at the teacher level significant
- Coaching at the school level not significant
- Teacher expertise of implementation not
significant - Only one teacher characteristic significant (role
conception), but only in one year
11Conditional value-added effects
Year 1 Year 2 Year 3
Average value-added for teacher receiving NO coaching 0.26 0.17 0.14ns
Role conception -.01ns .04 .01ns
Teacher expertise 0.02ns -0.03ns 0.03ns
Value-added per coaching session (cumulative) -.026 .012 .012
12Comparing Coaching Value-added to Unconditional
Mean Value-Added
Year 1 Year 2 Year 3
Value-added per coaching session (cumulative) -.026 .012 .012
Effect size per session -0.03 0.02 0.02
Mean cumulative coaching sessions 2.60 8.96 15.70
Mean coaching value-added -0.07 0.09 0.19
Unconditional mean value-added .164 .280 .327
Proportion accounted for by coaching NA 0.32 0.57
13Variability in Coaching between Schools
17
14Variability of Coaching between and within
Schools A Tale of Two Schools
- Staff size 14
- Value-added
- Y1 ? Y2 ? Y3
- 0.07 ? 0.22 ? 0.25
- Starts below average and improves
- Variability between teachers decreases from Y1 to
Y3
- Staff size 14
- Value-added
- Y1 ? Y2 ? Y3
- 0.17 ? 0.13 ? 0.01
- Starts at average and deteriorates
- Variability between teachers increases from Y1 to
Y3
How can we make sense of what happened in these
two schools?
15School 10 Coaching Sessions Accumulated per
Teacher
16School 16 Coaching Sessions Accumulated per
Teacher
17Network Analysis
- Conducted by Allison Atteberry Tony Bryk
- Pre- and post-teacher surveys asked who talked to
about instruction and student problems in
literacy (up to 7) and how often
Black box coach
Lines and arrows reported ties
Arrow weight reported frequency of consultation
Shape color grade level Shape size PD dosage
18School 10Riverside 2005
19School 16Tyson William Elem 2005
20School 10Riverside 2008
21School 16Tyson William Elem 2008
22Summary of findings
- Evidence that the mechanism for improved
value-added shifts from over time - Year 1 Coaching has no value-added
- Year 2 Coaching begins to add to value-added for
student learning - Year 3 Coaching becomes the primary mechanism
for value-added to student learning - Cumulative coaching explains differences in
teacher value-added effects, but not school
effects
23Implications
- Coaching largely mediates teachers value-added
to student learning - Not in Year 1, but in Year 2 and especially 3
- Selection effects?
- Dosage effects?
- Developmental/expertise effect for teachers?
- Developmental/expertise effect for coaches?
24Future Steps
- Exploration of the role of coach
- Expertise and its development
- Coach decision-making, especially what influences
whom coaches target - Exploration of the influence of school context
- Teacher influence in school decisions
- Principal leadership and supportiveness
- Trust
25Limitations
- Limited sample, especially at school level,
limits ability to explore contextual mechanisms - Professional development for coaches is more
intense than in most other models - Coaching embedded in a school-wide reform model
that included - Professional development courses
- Detailed literacy instruction framework
26Thank you!
27Variability in school value-added, year 1
Average student gain per academic year
High value-added schools
Low value-added schools
Year 1 mean effect (.16)
No effect
28Variability in school value-added, year 2
Average student gain per academic year
Year 2 mean effect (.28)
Year 1 mean effect (.16)
No effect
29Variability in school value-added, year 3
Average student gain per academic year
Year 3 mean effect (.33)
Year 2 mean effect (.28)
Year 1 mean effect (.16)
No effect
30Variability in teacher value-added within 2
schools
Average student gain per academic year
Year 3 mean effect (.33)
Year 2 mean effect (.28)
Year 1 mean effect (.16)
No effect