# CST-CAHSEE Correlation - Loera by zhangyun

VIEWS: 19 PAGES: 34

• pg 1
```									California Standards Test
and CAHSEE Correlation
Use of Student Data for Targeted
Preemptive Intervention

November 1, 2006
Dr. Janis Fries-Martinez, Principal
Gerardo Loera, Assistant Principal
Polytechnic High School
Los Angeles Unified School District 2
In This Presentation You Will Learn:

   How to accurately identify 10th grade students who
need intervention for the CAHSEE before they take
it in the spring

   How to identify a target group of students to
strategically raise the percentage of students
scoring Proficient or Above on the CAHSEE in order
to meet AYP requirements

   How to predict, with significant amount of certainty,
which 10th grade students will pass and fail the
CAHSEE on their initial try
No Child Left Behind and State
Accountability Criteria - CAHSEE
   AYP and API are dependent
in large part to CAHSEE
   AYP: Annual Measurable
Objectives are measured
mainly by success of 10th
CAHSEE the 1st time
CAHSEE

   Problem: High failure rate
AYP                API
on CAHSEE

   Problem: Not enough
students scoring Proficient
or Above on CAHSEE
Guiding Questions

   How do we effectively target students for pre-
emptive intervention for success on the
CAHSEE before they initially take it?

   How do we proactively identify students that

   Do we know which students are likely to pass
the test or get a proficient score ?
What are Passing and Proficient Scores?

   A passing score on the CAHSEE is:
   350 or greater on the English Language Arts
portion of the test
AND
   350 or greater on the Mathematics portion of the
test
   A proficient score is:
   380 or greater on the English portion of the test
   380 or greater on the Mathematic portion of the
test
Some Possible Attributes Contributing To
Success On The CAHSEE?
   Demographics
   Socioeconomic status
   Learning environment
   CELDT Scores
   Language Classification
   Special Education Status
   CST Scores
   Periodic Assessments
   Etc.
Studied CST/CAHSEE Correlation at
Different High Schools
Test Groups:
Poly HS: Class of 2006
Poly HS: Class of 2007
Santee HS: Class of 2007
LAUSD Class of 2008 (37,000+ scores)
LAUSD Class of 2007 (36,000+ scores)
Polytechnic High School Class of 2007
CAHSEE – Matched Scores

Note: 800+ scores included
Source: Secondary Student Information System
South LA High School #1 - Class of 2007
CAHSEE – Matched Scores
LAUSD Class of 2008 - 9th Grade ELA CST vs.
10th Grade ELA CAHSEE – Matched Scores

450

400
CAHSEE ELA SS 05-06

350

300

Y = 189.48 + 0.5512X
R2 = 0.6672

250
250     300      350           400          450        500           550            600
CST ELA SS 04-05

Note: 37,000+ scores included
Source: School Information Branch
LAUSD Class of 2008 - 9th Grade Algebra I CST
vs. 10th Grade Math CAHSEE – Matched Scores

450
CAHSEE SS 05-06

400

350

300
y = 0.4796x + 219.73
R2 = 0.4555
250
200     250    300     350     400      450      500     550     600
CST Algebra I SS 04-05

Note: 23,282 scores included
Source: School Information Branch
LAUSD Class of 2008 - 9th Grade Geometry CST
vs. 10th Grade Math CAHSEE – Matched Scores

450
CAHSEE Math 05-06

400

350

300
y = 0.4537x + 252.82
R2 = 0.602
250
100      200           300           400           500         600
CST Geometry SS 04-05

Note: 8,410 scores included
Source: School Information Branch
LAUSD Class of 2007 - 9th Grade MATH CST vs.
10th Grade Math CAHSEE – Matched Scores

450
CAHSEE Math SS 04-05

400

350

300
y = 0.5189x + 211.05
2
R = 0.4843
250
100        200             300           400        500           600
CST Math SS 03-04

Note: 36,190 scores included
Source: School Information Branch
Math CAHSEE Predictors

Mathematics – Geometry                Mathematics – Algebra I
Model                                 Model
9th Grade CST score of 214 or         9th Grade CST score of 271 or
higher predicts a passing score of    higher on Algebra I CST predicts
350 or higher on the CAHSEE           a passing score of 350 or higher
on the CAHSEE
9th Grade CST score of 280 or
higher predicts Proficient score of   9th Grade CST score of 334 or
380 or higher on the CAHSEE           higher predicts a Proficient score
(AYP)                                 of 380 or higher on the CAHSEE
Summary of Predictors

English Language Arts
Model
Based on LAUSD Class of 2008 (37,000+ scores)

9th Grade English CST score of 292 or higher predicts a passing score
of 350 or higher on the CAHSEE

9th Grade CST score of 346 or higher predicts Proficient score of 380 or
higher on the CAHSEE (To meet AYP)
ELA – “Reliability” Matrix for predicting
Passing CAHSEE ELA score – 83.4% accurate
Data from the LAUSD School
Information Branch – Class of 2008
Predicted Results
Model: 292 ELA CST = 350 ELA CAHSEE
Pass        Fail                   Total
Actual                Pass            24071       2914                    26985
Results               Fail            3549        8212                    11761

24071  8212
 83.4% Accuracy
24071  8212  2914  3549
Therefore the model for predicting a passing score or failing score on
the English portion of the CAHSEE based on the 9th grade CST is
correct 83.4% of the time.

This is a Confusion Matrix as described in the work of Kohavi and Provost, 1998
ELA – “Reliability” Matrix for predicting Passing
CAHSEE ELA Proficient or Above score – 84.6% accurate

Data from the LAUSD School
Information Branch – Class of 2008
Predicted Results
Model: 346 ELA CST = 380 ELA CAHSEE
Prof/Above Below Prof
Actual                 Prof/Above 11288         2135
Results                Below Prof 3846          21477
11288  21477
 84.6%
11288  21477  2135  3846
Therefore this model is correct 84.6% of the time.

This is a Confusion Matrix as described in the work of Kohavi and Provost, 1998
Algebra I - “Reliability” Matrix for predicting
Passing CAHSEE Math Passing score – 72.9% accurate
Data from the LAUSD School
Information Branch – Class of 2008
Predicted Results
Model: 271 Alg. I CST = 350 MATH CAHSEE
Pass        Fail
Actual                 Pass               11231       2930
Results                Fail               3383        5737
11231  5737
 72.9%
11231  5737  2930  3383
Therefore the model for predicting a passing or failing score on the
Mathematics portion of the CAHSEE based on the 9th grade CST is
correct 72.9% of the time.

This is a Confusion Matrix as described in the work of Kohavi and Provost, 1998
Algebra I – “Reliability” Matrix for predicting
Passing CAHSEE Math Proficient or Above score – 85.1%
accurate
Data from the LAUSD School
Information Branch – Class of 2008
Predicted Results
Model: 334 Alg. I CST = 380 MATH CAHSEE
Prof/Above Below Prof
Actual                    Prof/Above 2654          2742
Results                   Below Prof 726           17159
2654  17159
 85.1%
2654  17159  2742  726

Therefore the model for predicting a score on the Mathematics portion
of the CAHSEE based on the 9th grade CST is correct 85.1% of the
time.

This is a Confusion Matrix as described in the work of Kohavi and Provost, 1998
Geometry - “Reliability” Matrix for predicting
Passing CAHSEE Math Passing score – 88.7% accurate
Data from the LAUSD School
Information Branch – Class of 2008
Predicted Results
Model: 214 Geom. CST = 350 MATH CAHSEE
Pass        Fail
Actual               Pass                7374        68
Results              Fail                886         82
7374  82
 88.7%
7374  82  68  886
Therefore the model for predicting a passing score or failing score on
the Mathematics portion of the CAHSEE based on the 9th grade
Geometry CST is correct 88.7% of the time.

This is a Confusion Matrix as described in the work of Kohavi and Provost, 1998
Geometry – “Reliability” Matrix for predicting
Passing CAHSEE Math Proficient or Above score 82.1% accurate

Data from the LAUSD School
Information Branch – Class of 2008
Predicted Results
Model: 334 Geom. I CST = 380 MATH CAHSEE
Prof/Above Below Prof
Actual                  Prof/Above 4539              721
Results                 Fail             781         2369

4539  2369
 82.1%
4539  2369  721  781
Therefore the model for predicting a Proficient or Above score on the
Mathematics portion of the CAHSEE based on the 9th grade Geometry
CST is correct 82.1% of the time.

This is a Confusion Matrix as described in the work of Kohavi and Provost, 1998
Math (ALL) - “Reliability” Matrix for predicting
Passing score on CAHSEE Math – 72.7% accurate
Data from the LAUSD School
Information Branch – Class of 2007
Predicted Results
Model: 268 Math CST = 350 MATH CAHSEE
Pass        Fail
Actual                Pass              19070       3634
Results               Fail              6237        7249
19070  7249
 72.7%
19070  7249  3634  6237
Therefore the model for predicting a passing score or failing score on
the Mathematics portion of the CAHSEE based on the 9th grade Math
CST is correct 72.7% of the time.

This is a Confusion Matrix as described in the work of Kohavi and Provost, 1998
Math (ALL) – “Reliability” Matrix for predicting
Passing CAHSEE Math Proficient or Above score 81.5% accurate

Data from the LAUSD School
Information Branch – Class of 2008
Predicted Results
Model: 326 Math CST = 380 MATH CAHSEE
Prof/Above Below Prof
Actual                  Prof/Above 5652           5127
Results                 Below Prof 1564           23974
5652  23847
 81.5%
5652  23847  5127  1565

Therefore the model for predicting a Proficient or Above score on the
Mathematics portion of the CAHSEE based on the 9th grade Math CST
(regardless of which CST test was taken such as Algebra I, Geometry,
General Math etc.) is correct 81.5% of the time.

This is a Confusion Matrix as described in the work of Kohavi and Provost, 1998
Summary of Predictors

English Language Arts                           Mathematics
Model                                           Model
Based on LAUSD Class of 2008 (37,000+ scores)   Based on LAUSD Class of 2007 (36,190 scores)

9th Grade English CST score of                  9th Grade CST score in Math of
292 or higher predicts a passing                268 or higher predicts a passing
score of 350 or higher on the                   score of 350 or higher on the
CAHSEE                                          CAHSEE

9th Grade CST score of 346 or                   9th Grade CST score in Math of
higher predicts Proficient score of             326 or higher predicts a 380 or
380 or higher on the CAHSEE                     higher on the CAHSEE (AYP)
(AYP)
Deciding Which Students To Target –
Possible Action Plans For Applying the ELA Model

   For Program Improvement Schools:
   A Review of the 2006 AYP Progress Report
needs to be made in order determine which
subgroups, if any, fell short of meeting the Annual
Measurable Objectives (AMO)
   Recall that the AMOs are directly associated with the
percentages of 10th grade students that score Proficient
or Above on the ELA and Math portions of the CAHSEE
Deciding Which Students To Target –
Possible Action Plans For Applying the ELA Model

   For Program Improvement Schools (cont.):
   Once a subgroup has been identified, an SSIS
extract can be performed to create a list of
students needing pre-intervention
   The extract should produce a list of 10th grade students
that scored less than a 346 on their English 9th Grade
CST. This list should be sorted in decreasing order by
their English 9th Grade CST scaled score.
Deciding Which Students To Target –
Possible Action Plans For Applying the Math Model

   For Program Improvement Schools (cont.):
   Once a subgroup has been identified, an SSIS
extract can be performed to create of list of
students needing pre-intervention
   The extract should produce a list of 10th grade students
that scored less than a 326 on their Math 10th Grade
CST. This list should be sorted in decreasing order by
their Math 9th Grade CST scaled score.
Deciding Which Students To Target –
Possible Action Plans For Applying the ELA Model

   For Program Improvement Schools (cont.):
   The resulting lists are prioritized lists of students
that will benefit from pre-intervention and would
most likely not score a Proficient or Above on the
Deciding Which Students To Target –
Possible Action Plans For Applying the Predictive Model

   How to identify 10th grade students that will
probably fail one or both parts of the
CAHSEE?
   Perform an SSIS extract for 10th Grade students
that scored less than a 268 on their 9th Grade
Math CST
   Perform an SSIS extract for 10th grade students
that scored less than a 292 on their 9th Grade
English CST
Deciding Which Students To Target –
Technical Assistance
   Cynthia Lim in the LAUSD School Information
Branch has offered to provide schools with
technical support on identifying students at-
risk of not passing the CAHSEE or not getting
a Proficient or Above score.
Cynthia.lim@lausd.net
CAHSEE Preemptive Intervention
Case Study – Polytechnic High School 2006

   Master schedule permitted for approximately
170 seats for CAHSEE preemptive
intervention in Spring of 2006
   Selected a window of scores predicted to be
300-370
   Based on student performance this resulted
in a target range as follows:
For ELA: CSTs of 212 - 328
For Math: CSTs of 164 - 308
CAHSEE Preemptive Intervention
Case Study – Polytechnic High School 2006 - Results

   On average, non prepped 10th Graders
matched the predicted values.
   Math predicted mean score: 347.5
   Math actual mean score:       348.5
   ELA predicted mean score: 353.0
   ELA actual mean score:        352.2
CAHSEE Preemptive Intervention
Case Study – Polytechnic High School 2006 - Results
ELA          Math
# In preemptive          130          166
preparation
# Predicted to pass      108          118
# Actual did pass        120          143
Increase                 +12          +25
% Increase               11.1%        +21.2
# Predicted proficient   48           16
# Actual Proficient      61           67
Increase                 +13          +51
% Increase               27.1%        319%
CAHSEE Preemptive Intervention
Case Study – Polytechnic High School 2006 - Results

 CAHSEE Preemptive intervention
works!
 Preparation was a cost effective
means of positively impacting the
study group.

```
To top