Transcript document

Does Attending a Private
School Cause Children to
Have Higher Achievement?
John Q. Scholar
Center for Basic Research
Harvard University
Graduate School of Education
April, 2008
Private Schools & Student Achievement
John Q. Scholar, HGSE, April, 2008
Slide 1
My Research Questions:
1. Does receipt of a private school
tuition voucher lead families to
place their children in private,
rather than public, school ?
2. Does attending a private school
cause children to achieve at
higher levels than children who
attend public schools?
Private Schools & Student Achievement
John Q. Scholar, HGSE, April, 2008
Slide 2
Where Was My Study Conducted?
• In New York City, among low-income families
who had children about to enter kindergarten.
• Etc.
• Etc.
Private Schools & Student Achievement
John Q. Scholar, HGSE, April, 2008
Slide 3
Who Is In My Sample?
• 1449 families drawn randomly from the
population of 12,042 low-income
families who volunteered to participate
in the study.
– Summary stats on families?
– Can only generalize results to
population of volunteers.
• Each family provided one target child,
for a total sample of 1449 children.
– Summary stats on children?
– Statistical power analysis indicates that
this sample size provides moderate to
high power (.80) to detect small effect
sizes (.10 of a standard deviation) at
typical levels of Type I error.
Private Schools & Student Achievement
John Q. Scholar, HGSE, April, 2008
Slide 4
What Special Procedures Did I Employ:
• Conducted A Randomized Experiment:
– Families were randomized to either a Treatment
or a Control group.
• Treatment families received a private-school tuition
voucher,
• Control families did not.
– Families chose to enroll the target child in either a
private or a public school, making up any
difference in fees or expenses themselves.
– All target children were monitored through the
end of their kindergarten year.
• Administered Standardized Tests of Reading
Achievement “pre” and “post”:
– Pretest: at “baseline,” before the beginning of
kindergarten.
– Post-test: at the end of the kindergarten year.
Private Schools & Student Achievement
John Q. Scholar, HGSE, April, 2008
Slide 5
What Are My Measures:
• Outcome Variable:
– READ1: Nationally normed reading
0
Public
.05
achievement, measured at the end of
Kindergarten (expressed in national
percentiles).
0
indicating whethe the child attended a
private school during kindergarten
(1=Private; 0=Public).
Mean = 24.19
1
Private
.05
Reading Achievement (%ile)
• Question Predictor:
– PRIVATE: Dichotomous variable
• Instrument:
– VOUCHER: Dichotomous variable
• Covariate:
– READ0: Nationally normed reading
0
indicating whether the child’s family
received a private school tuition voucher
(1=Voucher; 0=No Voucher).
Mean = 25.12
0
50
100
read1
Graphs by Revised 1st fu indicator for at
achievement, measured at the beginning
of Kindergarten(expressed in national
percentiles).
Private Schools & Student Achievement
John Q. Scholar, HGSE, April, 2008
Slide 6
What Data-Analyses Did I Conduct:
Addressed my research questions by instrumental variable estimation
(IVE), using a Two-Stage Least-Squares (2SLS) approach:
RQ #1: Does receipt of a private school tuition voucher lead families to
place their children in private school?
 1st stage of 2SLS/IV analysis.
 Linear probability model, full sample.
PRIVATEi   0   1VOUCHERi   2 READ0i   i
RQ #2: Does attending a private school cause children to achieve
at higher levels than children who attend public schools?
 2nd stage of 2SLS/IV analysis.
 Multiple linear regression model, full sample.
READ1i  0  1PRIVˆATEi   2 READ0i   i
Private Schools & Student Achievement
John Q. Scholar, HGSE, April, 2008
Slide 7
Vouchers Impact Enrollment in Private School
RQ #1: Does receipt of a private school tuition voucher lead families
to place their children in private school?
Outcome = PRIVATE
Parameter
Estimate
(standard
error)
Intercept, 0
.0447**
(.0154)
VOUCHER, 1
.7710***
(.0169)
READ0, 2
.0004
(.0004)
2
.1021
R2
.59
Interpretation, as usual … being
awarded a tuition voucher
increases the fitted probability that
a family will enroll its child in
private school by 77 percentage
points … etc.
Private Schools & Student Achievement
John Q. Scholar, HGSE, April, 2008
Slide 8
Vouchers Impact Enrollment in Private School
RQ #1: Does receipt of a private school tuition voucher lead families
to place their children in private school?
700
600
500
400
Fre
300
200
100
1
0
The fitted odds that a family
awarded a tuition voucher will
enroll its child in private school
are 82.9 times the odds that a
family without a voucher will do
the same.*
* Setting Pretest reading achievement to the sample average of 22.54
0
Priva
0
te
Vou
che
r
1
Private Schools & Student Achievement
John Q. Scholar, HGSE, April, 2008
Slide 9
Private & Public Schools Equally Effective
RQ #2: Does attending a private school cause children to achieve at
higher levels than children who attend public schools?
Outcome = READ1
Parameter
Estimate
(standard
error)
Intercept, β0
10.963***
(.866)
PRIVATE, β1
.9061
(1.185)
READ0, β2
.5875***
(.0211)
2
299.3
R2
.35
Interpretation, as usual … cannot reject the null
hypothesis that the effect of private school is
zero, but the associated coefficient is positive,
suggesting that enrolling in private school for
the kindergarten year causes a child to score
less than one percentile point higher on a
nationally-normed test of reading achievement.
Private Schools & Student Achievement
John Q. Scholar, HGSE, April, 2008
Slide 10
Private & Public Schools Equally Effective
RQ #2: Does attending a private school cause children to achieve at
higher levels than children who attend public schools?
Interpretive Visuals
Displaying RQ#2 Findings
Private Schools & Student Achievement
John Q. Scholar, HGSE, April, 2008
Slide 11
Big Picture:
 Vouchers Affect Enrollment. The probability
that a child will enroll in a private school, for
their kindergarten year, is 77 % points
greater, if the family is provided with a tuition
voucher (p<.001).
 Schools in Both Sectors Equally Effective. A
child who has attended private school for
kindergarten scores less than 1 percentile
point higher on a nationally-normed test of
reading, at the end of the year (p>.40).
Private Schools & Student Achievement
John Q. Scholar, HGSE, April, 2008
Slide 12
What Does This Mean for Kids and
Families?
 Need to change state and federal
educational policy by …
 We do not need to buy shares in voucher
printing companies.
 Need to watch “The Bachelor: London
Calling.”
 Etc.
Private Schools & Student Achievement
John Q. Scholar, HGSE, April, 2008
Slide 13
Where Do We Go Next?
In the future, we need to:
• Distinguish the impact of vouchers on private school enrollment for important subgroups
of children:
– Race/ethnicity?
– Gender?
• Examine larger samples of children to buttress the statistical power:
– Must ensure there is sufficient statistical power to detect effects in important subgroups.
• Extend the study longitudinally, beyond the kindergarten year, through High School and
into the workforce:
– Examine long term impact of both vouchers and private school on subsequent outcomes.
• Focus on other kinds of academic and non-academic outcomes:
–
–
–
–
Mathematics, science knowledge, languages.
Measures of later school success (grade repetition, dropout, graduation, college entry).
Social and affective outcomes (self-confidence, satisfaction, reliability, character).
Economic outcomes (future labor force participation, wages). other than reading , and
non-academic outcome
• Replicate the study in other major cities and states.
• Join the French Foreign Legion.
Private Schools & Student Achievement
John Q. Scholar, HGSE, April, 2008
Slide 14
Include a Bibliography
(pay due homage -- Peterson,
Krueger and Angrist may be in
the audience!)
Private Schools & Student Achievement
John Q. Scholar, HGSE, April, 2008
Slide 15
Appendices:
Include a data appendix.
Describe approaches you used to deal with
missing data, and other analytic problems.
Provide ancillary descriptive statistics on the full
sample and important subgroups.
Provide additional tables containing the results of
fitting alternative model specifications, sensitivity
analyses, etc.
Include ancillary slides that anticipate potential
audience questions.
Etc. …
Private Schools & Student Achievement
John Q. Scholar, HGSE, April, 2008
Slide 16
Overview of Main Points:
The Four p’s …
Private Schools & Student Achievement
John Q. Scholar, HGSE, April, 2008
Slide 17
Prepare Your Presentation
• Establish A Sensible Slide Format That Is Common Across All Slides:
–
–
–
Use footers and headers to maintain your title, authorship, affiliation, date.
Make sure each slide has an explicit title that explains what’s on that slide.
Don’t put too much on any one slide.
• Provide An Attention-Grabbing Opening:
–
–
Slide #1:
Slide #2:
State title, author, affiliation, date, mention your personal interests.
State your research questions, discuss the literature and abiding issues.
• Document The Background And Context Of Your Research:
–
–
–
–
–
–
Slide #3:
Slide #4:
Slide #5:
Slide #6:
Slide #7:
Slide #8:
Identify and describe your site.
Identify and describe your dataset.
Describe your sample, include selected descriptive statistics/plots, if possible.
Describe any important procedures used in the design or data collection.
Define your measures –outcome, question predictors, instruments, covariates.
Overview your data-analyses, by question, identifying critical parameters.
• Present Your Main Findings, By Research Question:
–
–
Slide #9a: Prepare decent, sparse, well-annotated summary tables, by RQ.
Slide #9b: Create interpretive visuals to accompany answer to each RQ, if possible.
• Provide A Memorable Closing:
–
–
–
Slide #10: Provide a “big picture” summary in words, directly addressing the RQs.
Slide #11: Comment on the “real-life” benefits or consequences of your work.
Slide #12: Outline the implications of your work for the future.
• Prepare Appendices and Supplementary Slides (optional)
• Prepare A Handout For Your Audience, Including a Bibliography (optional, but a good idea).
Private Schools & Student Achievement
John Q. Scholar, HGSE, April, 2008
Slide 18
Practice Your Presentation
• Review Your Visuals Carefully:
– Examine their clarity, clutter and relevance.
– Decide, in advance, which slide(s) you’ll skip, if time runs out.
– Identify some additional slides you’ll comment on if time allows.
• Prepare To Present:
–
–
–
–
Plan every word you intend to say.
Write your script out in a large font, formatted to emulate spoken phrasing.
Use cautious animation to stage the presentation of each slide.
Avoid unnecessary animation (just like me?).
• Rehearse Your Talk:
– Videotape a practice run and review your performance with remote control & coach.
– Obtain feedback from your coach on:
•
•
•
•
•
The strength of your opening & closing.
Your logical flow.
The credibility and clarity of evidence presentation.
The clarity and comprehensibility of your key points.
Your many distracting verbal mannerisms and physical ticks.
• Check the physical environment in which you’ll present, in advance:
–
–
–
–
–
Check that you have everything you need (including a pointer).
Check the audience sightlines in advance, moving or removing seats as needed.
Fix your computer so that it doesn’t go to sleep while you’re being introduced.
Make sure that the projected image can be read from the back of the room.
Establish a reasonable surface to place (hide) your notes in plain sight.
Private Schools & Student Achievement
John Q. Scholar, HGSE, April, 2008
Slide 19
Be Positive During Your Presentation
• Be Positive About Your Talk:
– Project a sense that you really value your material.
– Do NOT read a script, but give your talk using your slides as “prompts.”
• Be Positive About Yourself:
– Be confident, yet relaxed.
– Don’t be afraid to acknowledge a mistake, but don’t grovel.
• Build Rapport With Your Audience:
–
–
–
–
Be yourself, be sincere.
Establish and maintain eye-contact with your audience.
Connect your message to what you think their experiences are.
Don’t adopt a “defensive” position -- move out towards, and into, the
audience if that is possible.
– If you don’t want questions until the end, say so!
– If you don’t know the answer, don’t bluster – say “I don’t know the
answer to that right not, but I will certainly look into it!”
• Hold the Audience’s Attention:
– Don’t stand rigidly, gripping the podium with white knuckles.
– Have good decisive body animation.
– Don’t be afraid to direct audience questions back to other audience
members (or to the rest of the panel!).
Private Schools & Student Achievement
John Q. Scholar, HGSE, April, 2008
Slide 20