High School Data Team Understanding Socioeconomic Status (SES) and Racial Gaps

1
High School Data TeamUnderstanding Socioeconomic
Status (SES) and Racial Gaps

• Co-Chairs
• Scott Summers
• Ororo T'Challa-Wakandas
• Members
• Rachel Grey
• James Howlett
• Katherine Pryde
• Kurt Wagner
• Consultant
• Sean Parker

This Demo Uses Fictional Data For Purposes of
Example The Ultimate Goal is to Answer Your
Questions with Your Data
DataTeamConsulting.Com
2
Three Questions About SES and Racial Gaps

• Are there socioeconomic and/or racial gaps in the
  percentage of general education (G.E.) classes
  taken?
• Are there SES and/or racial gaps in academic
  achievement?
• When we compare students who take the same
  percentage of G.E. classes, do the achievement
  gaps persist?

Yes.
Yes.
Yes.
DataTeamConsulting.Com
3
The First of Three Questions About SES and Racial
Gaps

• Are there SES and/or racial gaps in the
  percentage of G.E. classes taken?
• 397 students, our current juniors
• We gathered data on the course level of students
  freshman and sophomore courses.
• For each student, we considered what percentage
  of his/her leveled courses were G.E.
• We separated the data by SES and race.

DataTeamConsulting.Com
4
How Many G.E. Classes do Students Take?
397 Students
Percentage of 9th/10th Leveled Classes that were
G.E. Classes
G.E. Category
DataTeamConsulting.Com
5
Separated By SES
347 Students
50 Students
G.E. Category
DataTeamConsulting.Com
6
Separated By Race
26 Students
82 Students
15 Students
274 Students
G.E. Category
G.E. Category
DataTeamConsulting.Com
7
The Second of Three Questions About SES and
Racial Gaps

• Are there SES and/or racial gaps in academic
  achievement?
• 418 students, our current juniors
• In order to measure academic achievement, we
  created a composite.
• Unweighted Grade Point Average (GPA)
• English Language Arts MCAS (ELA MCAS)
• Mathematics MCAS (Math MCAS)
• We used histograms to explore the data.
• We separated the data by SES and race.

DataTeamConsulting.Com
8
How Do We Measure Academic Achievement? Building
A Composite
DataTeamConsulting.Com
9
How Do We Measure Academic Achievement? Building
A Composite
ELA MCAS
Math MCAS
Academic Achievement
Unweighted GPA
DataTeamConsulting.Com
10
How Do We Measure Academic Achievement? Building
A Composite
Jackie Joyner-Kersee American Heptathlete "Achiev
ement is difficult. It requires enormous effort.
Those who can work through the struggle are the
ones who are going to be successful."
DataTeamConsulting.Com
11
How Do We Measure Academic Achievement? Building
A Composite
800m Run
High Jump
Shot Put
Javelin Throw
Jackie Joyner-Kersee American Heptathlete "Achiev
ement is difficult. It requires enormous effort.
Those who can work through the struggle are the
ones who are going to be successful."
2 minutes 8.51 seconds
6 feet 4 inches
55 feet 3 inches
164 feet 5 inches
12.61 seconds
22.30 seconds
24 feet 7 inches
7291 heptathlon points
100m Hurdles
200m Sprint
Long Jump
DataTeamConsulting.Com
12
How Do We Measure Academic Achievement? Building
A Composite
800m Run
High Jump
Shot Put
Javelin Throw
Academic Achievement
100m Hurdles
200m Sprint
Long Jump
DataTeamConsulting.Com
13
How Do We Measure Academic Achievement? Building
A Composite
Shot Put
Javelin Throw
ELA MCAS
Academic Achievement
100m Hurdles
200m Sprint
Long Jump
DataTeamConsulting.Com
14
How Do We Measure Academic Achievement? Building
A Composite
ELA MCAS
Math MCAS
Academic Achievement
100m Hurdles
200m Sprint
Long Jump
DataTeamConsulting.Com
15
How Do We Measure Academic Achievement? Building
A Composite
ELA MCAS
Math MCAS
Academic Achievement
Unweighted GPA
DataTeamConsulting.Com
16
Building A Composite Why Combine Measures?

• Measurement Reliability
• Statistical Power
• We Teach To Students, Not To Tests
• Schoolwide Picture, Schoolwide Vision

DataTeamConsulting.Com
17
A Histogram of Academic Achievement
418 Students
Number of Students
Number of Students
Zero is the average (i.e., mean) for all 418
students.
Negative is below average.
Positive is above average.
Academic Achievement Composite
DataTeamConsulting.Com
18
A Histogram of Academic Achievement
418 Students
3.11 GPA (m) 250 Prof. ELA 268
Adv. Math
2.31 GPA (f) 242 Prof. ELA 228 NI Math
0.00 Composite
3.81 GPA (m) 262 Adv. ELA 274 Adv.
Math
-1.95 Composite
0.98 Composite
3.08 GPA (m) 260 Adv. ELA 264 Adv. Math
3.28 GPA (f) 248 Prof. ELA 268
Adv. Math
Number of Students
0.30 Composite
0.00 Composite
Academic Achievement Composite
DataTeamConsulting.Com
19
A Histogram of Academic Achievement
50 Line i.e., Median
418 Students
Mean
Number of Students
Academic Achievement Composite
DataTeamConsulting.Com
20
A Histogram of Academic Achievement
418 Students
Number of Students
Academic Achievement Composite
DataTeamConsulting.Com
21
Academic Achievement Separated by SES
364 Students
Mean 0.12
54 Students
Number of Students
Mean -0.82
Academic Achievement Composite
DataTeamConsulting.Com
22
Academic Achievement Separated By Race
Examine the shapes of these histograms.
86 Students Mean 0.27
29 Students Mean -1.25
18 Students Mean -0.85
285 Students Mean 0.10
Academic Achievement Composite
DataTeamConsulting.Com
23
The Third of Three Questions About SES and Racial
Gaps

• When we compare students who take the same
  percentage of G.E. classes, do the achievement
  gaps persist?
• 395 students, our current juniors
• We first saw that percentage of G.E. classes is
  correlated with SES and race.
• We just saw that academic achievement is
  correlated with SES and race.
• We will see that percentage of G.E. classes is
  correlated with academic achievement (using
  histograms and a scatterplot).
• We will disentangle the correlations using a
  statistical model with statistical controls.

DataTeamConsulting.Com
24
Academic Achievement Separated By Percentage of
G.E.
295 Students Took No G.E. Courses Mean 0.41
No
G.E.
44 Students Took Some G.E. Courses Mean -0.58
Some
G.E.
24 Students Took Half G.E. Courses Mean -0.87
Half
G.E.
32 Students Took Mostly G.E. Courses Mean -1.92
Mostly
G.E.
We can look at the same information without
grouping students into four categories (None,
Some, Half, Most). We will use a scatterplot to
do so.
Academic Achievement Composite
DataTeamConsulting.Com
25
A Scatterplot of Academic Achievement vs. G.E.
Percentage
3.11 GPA (m) 250 Prof. ELA 268
Adv. Math 0.0 Achievement 0 G.E.
3.11 GPA (m) 250 Prof. ELA 268
Adv. Math
2.31 GPA (f) 242 Prof. ELA 228 NI
Math -2.0 Achievement 57 G.E.
2.31 GPA (f) 242 Prof. ELA 228 NI Math
3.81 GPA (m) 262 Adv. ELA 274 Adv. Math
3.81 GPA (m) 262 Adv. ELA 274 Adv.
Math 1.0 Achievement 0 G.E.
3.08 GPA (m) 260 Adv. ELA 264 Adv. Math
3.08 GPA (m) 260 Adv. ELA 264 Adv.
Math 0.2 Achievement 9 G.E.
3.28 GPA (f) 248 Prof. ELA 268
Adv. Math
3.28 GPA (f) 248 Prof. ELA 268
Adv. Math 0.0 Achievement 0 G.E.
Percentage of 9th/10th Leveled Classes that were
G.E. Classes
DataTeamConsulting.Com
26
Trend Curves For Differing Subgroups
Estimated trend curves for students who receive
neither SPED services nor LEP services.
White, Non-Free Lunch
White, Free Lunch
African American, Non-Free Lunch
African American, Free Lunch
Percentage of 9th/10th Leveled Classes that were
G.E. Classes
DataTeamConsulting.Com
27
Trend Curves For Differing Subgroups
Estimated trend curves for students who receive
neither SPED services nor LEP services.
White, Non-Free Lunch
White, Free Lunch
African American, Non-Free Lunch
African American, Free Lunch
Percentage of 9th/10th Leveled Classes that were
G.E. Classes
DataTeamConsulting.Com
28
Three Questions About SES and Racial Gaps

• Are there socioeconomic and/or racial gaps in the
  percentage of general education (G.E.) classes
  taken?
• Are there SES and/or racial gaps in academic
  achievement?
• When we compare students who take the same
  percentage of G.E. classes, do the achievement
  gaps persist?

Yes.
Yes.
Yes.
DataTeamConsulting.Com
29
Where do we go from here?

• Feedback from you
• Principals working group on achievement
• A renewed focus on achievement as part of our
  school improvement goals

DataTeamConsulting.Com
30
Notes Not For The General Audience
GLH testing tells us to keep RACE and
FREELUNCHEVER. Zero-skew transformation helps to
meet normality assumption. Robust standard errors
address heteroskedasticity. No interactions were
stat sig, but we want to keep an eye on them as
our sample size (and statistical power) increases.
DataTeamConsulting.Com