You are on page 1of 3

Introduction

The context of the study is to identify which is a better predictor of AP United States History
Exam scores, the students weighted GPA or SAT scores (the one based on the 1600 grading
scale). In this situation, the APUSH score will be the response variables and the explanatory
variables are the weighted GPA and the SAT score. A preliminary hypothesis our group
developed was that the GPA is the better predictor of the APUSH score.

Data Collection
The data was obtained from students of the 2015-2016 APUSH class. We inquired for their
APUSH exam score, their weighted GPA, and their SAT score. The results are listed in the table
below.
Weighted GPA

SAT Score

APUSH Exam Score

4.26

1090

3.50

1100

4.07

1210

4.00

1290

4.28

1150

3.91

1250

3.70

1130

4.45

1300

4.13

1190

3.60

1090

4.43

1440

3.20

990

3.90

950

3.80

1080

4.42

1200

3.60

1060

3.30

1090

Graphs
Graph one has a positive associated relationship and its form is linear. The strength of the plot is
moderately strong. Graph two also has a positive associated relationship with a linear form, but
what differs from Graph one is the strength. The strength of Graph two is weak because the
points are scattered and doesnt form a clear linear shape.

Numerical Summaries
GPA:

r=0.52

The value of r shows that the relationship between GPA and APUSH score is moderately
strong
r 2=0.27

2
The value of r shows that the data takes into account of 27% of variation in the value

of APUSH scores that is accounted for by the least-squares regression line of APUSH
scores on GPA.
^y =2.46+1.23 x with a=-2.46 and b=1.23

The least-squares regression line shows that for every one unit of increase of GPA the
APUSH score will increase by 1.23 points. Also, it shows that with a GPA score of 0 the
APUSH score will be -2.46 which is not plausible because there cant be a negative score
value.
s=0.82

The value of s determines that the average error in predicting the APUSH score is 0.82
points using the least-squares regression line.
SAT:

r=0.35

The value of r shows that the relationship between SAT and APUSH score is weak.
r 2=0.12

2
The value of r shows that the data takes into account of 12% of variation in the value

of APUSH scores that is accounted for by the least-squares regression line of APUSH
scores on SAT scores.
^y =0.72+0.003 x with a=-0.72 and b=0.003

The least-squares regression line shows that for every one unit increase of SAT scores the
APUSH score will increase by 0.003 points. Also, it shows that with a SAT score of 0 the

APUSH score will be -0.72 which isnt plausible because there cant be negative score
values.
s=0.99

The value of s determines that the average error in predicting the APUS score is 0.99
points using the least-squares regression line.

Residual
Residual Plot 1 shows no clear pattern which means that the least-squares regression line for the
relationship between GPA and APUSH score fits the data well. Residual Plot 2 also shows no
clear pattern meaning that the least-squares regression line for the relationship between SAT and
APUSH score fits the data well.
Conclusion
From the study conducted, it can be concluded that GPA was a better predictor than SAT scores
when it came to passing the APUSH exam. As GPA had a higher correlation of 0.52 while SAT
only had 0.35 meaning that there was a stronger relationship between GPA and APUSH score
than the relationship between SAT and APUSH score. The average error in SAT scores was 0.99
while GPA had a lower rate at 0.82 which means that SAT scores estimated had a higher error
rate than GPA errors. The potential limitations of the model is that the APUSH score only ranges
from 1-5, GPA ranges from 0.00-5.00, and SAT score ranges from 200-1600. This gives
inaccurate scores for GPA and SAT at zero by giving us negative numbers which isnt possible
on the APUSH exam. Using our model, it wouldnt be appropriate to estimate a student's APUSH
score using SAT scores prior to March 2016 which was out of 2400 or an unweighted GPA,CSU
GPA, or UC GPA.

You might also like