1
The Relationship between the PARCC Test Scores and the College Admission Tests:
SAT/ACT/PSAT
A Research Report Submitted to Maryland State Department of Education (MSDE)
Maryland Assessment Research Center (MARC)
Executive Summary
The purpose of this research project is to explore the relationships between the PARCC
test scores and the college admission test scores. Given that the MARC team received
data with anonymous student identifiers and the sample sizes after matching students
were adequate, data analyses have been performed for the PARCC ALG01, ALG02,
ELA10 tests, SAT, ACT, and PSAT to find the relationships between the PARCC test
scores and their respective college admission test scores in the same content area. More
specifically, based on the approved proposal sent to MSDE, the following analyses have
been conducted.
1. Correlational analyses between the PARCC and SAT, ACT, and PSAT test scores;
2. Regression analyses using the PARCC test scores to predict college admission
test performance;
3. Concordance relationships established using the matched samples in which
students' PARCC test scores and college admission test scores are both available;
4. Graphical presentations of the results prepared in addition to the statistical
analysis. This perspective focused on the performance levels obtained on the
PARCC tests (levels 1 through 5);
5. A summary provided to report the findings from a series of data analyses.
Major findings are summarized below. The detailed results are reported after this
executive summary.
.
Major Findings/Talking Points
1. This study was conducted based on the matched samples for PARCC/PSAT and
SAT/ACT by subject. Only Grade 12 students’ first attempts were used in the
analyses, which usually account for more than 80% of the matched samples.
Matched sample sizes for other grades were too small to conduct statistical
analyses, which may result in reduced power to detect some significant
relationship.
2. The results of various correlational analyses and regression analyses show that
there is a moderate linear relationship between PARCC tests and college
admissions exams (SAT and ACT) for both Math and English. PARCC tests are
statistically significant predictors of SAT/ACT test scores. PARCC scores as
predictors, along with PSAT scores, explain 29% to 59% of the total variance in
the SAT/ACT outcomes. Also, the PARCC tests are moderately correlated with
similar SAT/ACT sub-test scores.
Tab D1 Attachment 1
2
3. Graphical representations show that Grade 12 students with higher performance
levels on PARCC tend to score higher on SAT/ACT. Concordance tables linking
the PARCC test scores to the SAT or ACT test scores in the same content area are
provided in the Appendix. These tables show the linkage between student
performance on PARCC tests and the college admission tests, SAT and ACT.
Though the PARCC tests and the college admission tests, SAT and ACT do not
measure exactly the same content in a similar subject area, such concordance
tables could help to see the linkage between these two types of tests: high school
graduation tests and the college admission tests.
4. Due to the limited number of administrations of PARCC tests, the matched
sample sizes between the PARCC and SAT/ACT test scores were relatively small
for the analyses carried out. Caution should be exercised when generalizing the
findings from the current study as discrepancies in the test score distributions for
the matched Grade 12 students used in the analyses and for all students in the data
received were not negligible.
3
Data Cleaning and Preparation
Based on the four datasets received, PARCC, SAT, ACT, and PSAT, data
cleaning has been conducted to prepare for data analyses. Summary tables are provided
below to provide an overview of the tests and the issues in data cleaning and preparation.
The sub-content areas for different subjects covered in the four tests are summarized in
Table 1. PARCC tests include Algebra I (ALG01) and Algebra II (ALG02), and English/
Languages Arts (ELA10). For the other three college admission tests, the subsections
differ. In general, PARCC Algebra I and II are assumed to be related to SAT, ACT, and
PSAT math while PARCC ELA10 is assumed to be related to SAT reading and Writing,
ACT Reading and English, and PSAT Verbal and Writing.
Table 1
Subjects in Each Test
Test
Subjects
PARCC
ALG01, ALG02, ELA10
SAT
Math, Reading, Writing
ACT
Math, Reading, English, Science
PSAT
Math, Verbal, Writing
The numbers of records in the data files are summarized in Table 2. Most students
are first-time takers on PARCC, SAT, and ACT, while 50% of the students have multiple
records on PSAT (259,883 out of 515,109).
Table 2
The Number of Records in the Four Datasets
PARCC
SAT
ACT
PSAT
Number of records
158077
301349
73119
842414
Total number of students (unique)
138232
296753
72755
515109
Number of students with multiple records
19775
4585
364
259883
Students with multiple records were checked to investigate the nature of the
repetition. Most of the multiple records were valid for PARCC and PSAT indicating
attempts in different test administrations. Most often, students with two records on
PARCC took ELA10 and ALG01/ALG02. However, many students with multiple
records on SAT/ACT were not really retakers. Table 3 shows such an example. This
student has two records with exactly the same scores in 2010 and 2011, but in 2011 many
other variables such as grade, gender, race, and special education were missing. It is also
observed that some records are repetitions except differences in the academic year and
the grade. Table 4 shows an example of a student with multiple records on SAT who is a
real retaker. The scores are different and there are no missing values for other variables.
4
Table 3
An Example of Two Possible Duplicated Records
No. Year SASID
Grade
Gender
Race
Reading
Score
Math
Score
Writing
Score
Combined
Score
78732
2010
9380449262
12
F
WH
520
500
440
1460
114728
2011
9380449262
NA
NA
NA
520
500
440
1460
Table 4
An Example of Actual Retakers
No. Year SASID
Grade
Gender
Race
Reading
Score
Math
Score
Writing
Score
Combined
Score
240841
2014
9363963321
12
M
MU
440
530
340
1310
240842
2014
9363963321
12
M
MU
490
530
380
1400
As the study is to explore the relationship between the PARCC test scores and the
college admission test scores, the following 20 samples were matched with respect to
different content areas. Matched sample sizes are reported in Table 5 by grade for each
matched sample. The matched sample sizes are the same for both first-time takers and
retakers. It is noted that the sample sizes for samples 3, 4, 6, 11, 13, 14, 16 to 20 in the
shaded area are not sufficient to conduct the proposed analyses. For the matched samples
with sufficient sample sizes, the numbers of retakers were checked and reported in
parentheses if there were any (see Table 5). As some matched samples such as sample 5
and 15 contain a large proportion of retakers, correlation analyses were conducted
between SAT/ACT and PARCC/PSAT for both the first attempts and the highest scores
respectively.
As the majority of the matched samples were from Grade 12, all the proposed
analyses were conducted using the matched samples for Grade 12 only. The descriptive
statistics for Grade 12 students’ PARCC test scores and all the cases in the data files
received were compared to check whether Grade 12 students in the matched samples
were representative.
The descriptive statistics for all the data and Grade 12 students on PARCC tests
are summarized in Table 6. It is noted that the means and standard deviations for Grade
12 students are consistently lower than those for the whole data. This indicates that the
Grade 12 students are not completely representative of the population. This is consistent
with the expectations that students who took high school graduation tests (in this case it is
the PARCC tests) at Grade 12 are usually relatively lower performing students. Thus it is
highlighted that caution should be exercised in generalizing the findings from this
present study based on Grade 12 data to all Maryland students taking PARCC tests.
5
Table 5
Sample Sizes for Twenty Matched Samples
No.
College
Admission Tests
PARCC and PSAT
Tests
Grade
9
Grade
10
Grade
11
Grade
12
Total
1
SAT Math
PARCC ALG01
1
1
6
58
66
2
SAT Math
PARCC ALG02
0
11
106
1454
(2)
1571
(2)
3
SAT Math
PARCC ALG01,
PARCC ALG02
0
0
0
7
7
4
SAT Math
PARCC ALG01, PSAT
Math
0
1
2
35
38
5
SAT Math
PARCC ALG02, PSAT
Math
0
7
95
(40)
1253
(416)
1355
(456)
6
SAT Math
PARCC ALG01,
PARCC ALG02, PSAT
Math
0
0
0
5
5
7
SAT Reading
PARCC ELA10
1
7
10
76
94
8
SAT Reading
PARCC ELA10, PSAT
Verbal, PSAT Writing
1
5
6
46
58
9
SAT Writing
PARCC ELA10
1
7
10
76
(1)
94
(1)
10
SAT Writing
PARCC ELA10, PSAT
Verbal, PSAT Writing
1
5
(1)
6
(3)
46
(9)
58
(13)
11
ACT Math
PARCC ALG01
0
0
6
6
12
ACT Math
PARCC ALG02
1
8
314
(1)
323
(1)
13
ACT Math
PARCC ALG01,
PARCC ALG02
0
0
1
1
14
ACT Math
PARCC ALG01, PSAT
Math
0
0
4
4
15
ACT Math
PARCC ALG02, PSAT
Math
1
7
(2)
277
(108)
285
(110)
16
ACT Math
PARCC ALG01,
PARCC ALG02, PSAT
Math
0
0
1
1
17
ACT Reading
PARCC ELA10
1
0
9
10
18
ACT Reading
PARCC ELA10, PSAT
Verbal, PSAT Writing
1
0
8
9
19
ACT English
PARCC ELA10
1
0
9
10
20
ACT English
PARCC ELA10, PSAT
Verbal, PSAT Writing
1
0
8
9
6
Table 6
Descriptive Statistics on PARCC Tests for Student from All Grades and Grade 12
Students
Subject
Mean
SD
Min.
Max.
All
G12
All
G12
All
G12
All
G12
ALG01
734.29
714.67
32.80
25.62
650
650
850
817
ALG02
718.32
692.59
36.04
26.10
650
650
850
794
ELA10
737.79
695.15
44.95
33.40
650
650
850
807
7
Data Analysis and Results
Correlational Analyses between PARCC/PSAT and SAT/ACT
Correlational analyses were conducted to evaluate the relationships between
PARCC/PSAT and SAT/ACT using the matched samples. As samples 5 and 15 contain a
large proportion of retakers, pairwise correlations were examined for both the first
attempts and the highest scores respectively. Table 7 summarizes the correlation
coefficients with the correlations using the highest scores reported in the parenthesis. In
general, there are moderate relationships between PARCC tests and the college admission
tests, with significant correlations ranging from .52 to .75 at a significance level of 0.001.
Overall, the association between the PARCC tests and the college admission tests are
similar for the first attempts and the highest scores. The largest difference is the
correlation between the SAT Writing and the PSAT Verbal for sample 10, which
increases from .52 to .61. Note that in this case, the sample size is relatively small. The
sample size for sample 10 is only 46, and 9 of them are retakers. It is likely that using the
highest scores would have a larger impact on this kind of sample; compared to other
samples with a larger sample size and a smaller number of retakers such as sample 9. As
the majority of the correlations were very similar, the subsequent analyses were
conducted using first attempts for Grade 12 students only.
Table 7
Correlations between College Admission Tests and PARCC/PSAT Tests for First
Attempts (and Highest Scores)
No.
College
Admission
Tests
PARCC/PSAT Tests
PARCC
ALG01
PARCC
ALG02
PARCC
ELA10
PSAT
Math
PSAT
Verbal
PSAT
Writing
1 SAT Math
0.63
(0.63)
2 SAT Math
0.54
(0.54)
5 SAT Math
0.54
(0.54)
0.65
(0.67)
7 SAT Reading
0.60
(0.60)
8 SAT Reading
0.62
(0.62)
0.64
(0.64)
0.52
(0.52)
9 SAT Writing
0.74
(0.74)
10 SAT Writing
0.75
(0.75)
0.52
(0.61)
0.61
(0.66)
12 ACT Math
0.55
(0.55)
15 ACT Math
0.55
(0.55)
0.63
(0.60)
Note. All correlations in the table are significant at 0.001 level.
8
In addition, Figure 1 displays two example scatterplots showing the relationships
between the SAT subtests and the PARCC tests. Both plots show an approximately linear
pattern. Such plots are available for other pairwise relationships in the correlational
analyses and the regression analyses. Therefore, the results from both the correlational
and regression analyses conducted in the next section are warranted.
Figure 1. Scatterplots for (a) SAT Math scores vs. PARCC ALG02 scores; (b) SAT
Writing scores vs. PARCC ELA10 scores.
Regression Analyses using Test Scores on PARCC/PSAT to Predict SAT/ACT
Regression analysis results are summarized in Table 8 and Table 9 for samples 1,
2, 5, 7 to 10, 12, and 15 with sufficient sample sizes. Unstandardized regression
coefficients and standardized coefficients are reported in Table 8 and Table 9,
respectively. Again, the regression analyses were conducted using the matched samples
using Grade 12 students’ first attempt’s test scores. It is noted that all coefficients of
PARCC tests are statistically significant in predicting SAT/ACT test scores at a
significance level of 0.05. For example, the unstandardized coefficient of the PARCC
ALG01 is 1.99 in the regression analysis for matched sample 1, indicating that one unit
increase in the PARCC ALG01 scale score will lead to 1.99 units increase in the SAT
Math scale score after controlling for other variables’ effects. The adjusted R
2
for each
regression is also reported in the tables. As one can see, the adjusted R
2
ranges from .29
to .59, meaning that the predictors, PARCC (and PSAT) explain 29% to 59% of the total
variance in the outcome variables, SAT or ACT subtest scores.
9
Only two coefficients are not statistically significant: PSAT Writing test in
predicting the SAT Reading scores and the PSAT Verbal test in predicting the SAT
Writing scores. The potential multi-collinearity issues among predictors were examined
when more than one predictor was included (samples 5, 8, 10, and 15). There is no
serious multi-collinearity issue observed, thus all predictors were kept in the regression
equations.
Table 8
Unstandardized Regression Coefficient Estimates
No. Outcome
Predictors
2
adj
R
PARCC
ALG01
PARCC
ALG02
PARCC
ELA10
PSAT
Math
PSAT
Verbal
PSAT
Writing
1
SAT Math
1.99***
0.39
2
SAT Math
1.64***
0.29
5
SAT Math
0.94***
5.81***
0.50
7
SAT Reading
1.70***
0.35
8
SAT Reading
1.03*
5.73**
1.42
0.48
9
SAT Writing
1.92***
0.55
10
SAT Writing
1.38***
0.55
3.09*
0.59
12
ACT Math
0.06***
0.30
15
ACT Math
0.04***
0.20***
0.48
Note. *p<0.05, **p<0.01, ***p<0.001.
Table 9
Standardized Regression Coefficient Estimates
No. Outcome
Predictors
2
adj
R
PARCC
ALG01
PARCC
ALG02
PARCC
ELA10
PSAT
Math
PSAT
Verbal
PSAT
Writing
1
SAT Math
0.63***
0.39
2
SAT Math
0.54***
0.29
5
SAT Math
0.31***
0.51***
0.50
7
SAT Reading
0.60***
0.35
8
SAT Reading
0.35*
0.38**
0.10
0.48
9
SAT Writing
0.74***
0.55
10
SAT Writing
0.57***
0.04
0.27*
0.59
12
ACT Math
0.55***
0.30
15
ACT Math
0.34***
0.48***
0.48
Note. *p<0.05, **p<0.01, ***p<0.001.
The standardized regression coefficient estimates are presented in Table 9. These
standardized regression coefficients provide us with another perspective about the
relationship between the PARCC tests and the SAT/ACT test performances. For instance,
the standardized coefficient of the PARCC ALG01 is 0.63 in the regression analysis for
matched sample 1, indicating that one standard deviation increase in the PARCC ALG01
scale score will lead to 0.63 standard deviations increase in the SAT Math scale score.
10
One advantage of using standardized regression coefficients is to provide a common scale
to compare the prediction power of predictors in the same regression equation. For
example, PSAT Math has more prediction power of SAT Math compared with PARCC
ALG02 based on the sample 5 results; PARCC ELA10 and PSAT verbal have similar
prediction power of SAT Reading based on sample 8 results; PARCC ELA10 has more
prediction power of SAT Writing than PSAT Verbal and Writing as shown in sample 10.
Concordance Relationships between Test Scores on PARCC and SAT/ACT
To demonstrate the linkage of scores on different tests, an equipercentile linking
procedure was carried out for pairwise matched samples between PARCC ALG01 and
SAT Math, PARCC ALG02 and SAT Math, PARCC ELA10 and SAT Reading, PARCC
ELA10 and SAT Writing, and PARCC ALG02 and ACT Math. The common-group
design was used in linking based on the 12th graders’ first attempt scores on the tests.
Due to the extremely small sample sizes for the matched groups of students taking both
tests as shown in Table 5, such concordance relationship could not be established for
PARCC ELA10 and ACT Reading, PARCC ELA10 and ACT English, and PARCC
ALG01 and ACT Math.
Equipercentile linking based on the matched samples was carried out using the
software program, Linking with Equivalent Group or the Single Group Design,
abbreviated as LEGS (Kolen & Brennan, 2004). The reported scale scores were used to
link the pairwise tests listed above. After specifying the format of the data input,
subgroup information, input data file names, smoothing values, the score range for the
old test form and the truncation choice, the program conducts equipercentile linking and
outputs the results in the window. In Appendix A, a screenshot captures the input window
for linking the PARCC ALG02 and SAT Math tests. Two smoothing values were
compared in post-linking: 0.3 and 1. The choice of using smoothing values is supported
by the results from simulation studies that the smoothed results outperform non-smoothed
method in reducing linking errors when the population test scores are in fact smooth (Cui
& Kolen, 2009; Hanson et al., 1994). The results with a smoothing value of 1 are
presented in this report due to the fact that after rounding there was little difference
between the results based on the two smoothing values. As some scale scores were not
present in the matched samples, extrapolation had to be done. A scatterplot was generated
to examine the relationship between each pair of tests linked. Based on the total variance
explained and visual inspections of the scatterplots of scores between two linked tests, a
prediction equation was developed using Excel’s best fitting line function. Using the
prediction equation, values not presented in the matched samples were extrapolated. The
extrapolation equations for all pairs of linked tests are presented in Table 10. One thing
worth of note is that when doing extrapolation for PARCC ALG02 and ACT Math based
on the prediction equation, linked ACT scores for PARCC ALG02 scores 784-799 are
actually lower than the linked ACT score for a PARRC ALG02 score of 783 yielded
from LEGS. As ACT Math scores should keep ascending as PARCC ALG02 scores
11
ascending which is a pattern based on the available data in the matched sample, linked
ACT Math scores for PARCC ALG02 scores from 784 to 799 have been changed to be
equal to the linked ACT score for a PARCC ALG02 score of 783. Table 11 is the
concordance table between PARCC ALG01 and SAT Math. Concordance tables for the
other four pairs of linked tests can be found in Appendix B.
Table 10
Prediction Equations for Linked Tests
Linked Tests
Prediction Equation
PARCC ALG01 and SAT Math
2.9927 1792.7yx= −
PARCC ALG02 and SAT Math
2.67 1474.6yx= −
PARCC ELA10 and SAT Reading
13 5.246
4*10y x
−
=
PARCC ELA10 and SAT Writing
2.5913 1484.9yx= −
PARCC ALG02 and ACT Math
42
0.433.79 1157729* 57 310 1 2.53xxy
−
+= −
Table 11
Concordance Table for PARCC ALG01
and SAT Math
Performance
Level
PARCC
ALG 01
SAT Math
1
650
200
1
651
200
1
652
200
1
653
200
1
654
200
1
655
200
1
656
200
1
657
200
1
658
200
1
659
200
1
660
200
1
661
200
1
662
200
1
663
200
1
664
200
1
665
200
1
666
200
1
667
200
1
668
210
1
669
210
1
670
210
1
671
210
1
672
210
1
673
210
1
674
210
1
675
210
1
676
210
1
677
210
1
678
210
1
679
210
1
680
210
1
681
220
1
682
220
1
683
220
1
684
220
1
685
220
1
686
230
1
687
230
1
688
230
1
689
230
1
690
240
1
691
260
1
692
260
1
693
260
1
694
260
1
695
260
1
696
260
1
697
260
1
698
270
1
699
290
12
1
700
290
2
701
290
2
702
300
2
703
300
2
704
300
2
705
300
2
706
310
2
707
310
2
708
310
2
709
320
2
710
320
2
711
320
2
712
330
2
713
330
2
714
340
2
715
350
2
716
350
2
717
350
2
718
360
2
719
360
2
720
360
2
721
360
2
722
360
2
723
360
2
724
370
2
725
380
3
726
380
3
727
380
3
728
390
3
729
400
3
730
400
3
731
400
3
732
400
3
733
400
3
734
400
3
735
400
3
736
410
3
737
420
3
738
420
3
739
450
3
740
460
3
741
460
3
742
460
3
743
460
3
744
460
3
745
460
3
746
460
3
747
460
3
748
460
3
749
460
3
750
460
4
751
460
4
752
460
4
753
460
4
754
460
4
755
460
4
756
470
4
757
470
4
758
470
4
759
480
4
760
480
4
761
480
4
762
490
4
763
550
4
764
550
4
765
550
4
766
550
4
767
550
4
768
550
4
769
550
4
770
560
4
771
560
4
772
560
4
773
560
4
774
560
4
775
560
4
776
560
4
777
560
4
778
560
4
779
560
4
780
570
4
781
570
4
782
570
4
783
570
4
784
570
4
785
570
13
4
786
570
4
787
570
4
788
570
4
789
570
4
790
570
4
791
570
4
792
580
4
793
580
4
794
580
4
795
580
4
796
590
4
797
590
4
798
590
4
799
590
4
800
590
4
801
590
4
802
590
4
803
590
4
804
590
4
805
590
5
806
600
5
807
600
5
808
600
5
809
600
5
810
600
5
811
600
5
812
600
5
813
600
5
814
600
5
815
600
5
816
610
5
817
610
5
818
660
5
819
660
5
820
660
5
821
660
5
822
670
5
823
670
5
824
670
5
825
680
5
826
680
5
827
680
5
828
690
5
829
690
5
830
690
5
831
690
5
832
700
5
833
700
5
834
700
5
835
710
5
836
710
5
837
710
5
838
720
5
839
720
5
840
720
5
841
720
5
842
730
5
843
730
5
844
730
5
845
740
5
846
740
5
847
740
5
848
750
5
849
750
5
850
750
14
Graphical Presentations of the Relationships between the Performance Levels on
PARCC Tests and SAT/ACT Scores
In addition to the exploration of the relationships between the PARCC test scores
and the college admission test scores, the performance levels on the PARCC tests were
also examined in relation with the SAT and ACT test scores. Again given the adequacy
of the matched sample sizes, five analyses were conducted to examine the relationships
between the PARCC performance levels and the college admission test scores for Grade
12 students’ first attempts. The sample sizes at each PARCC performance level in the
matched samples are reported in Table 12. There are five performance levels on the
PARCC tests with 5 indicating the highest performance level. Generally speaking, the
sample size decreases as the performance level increases. This pattern indicates that the
majority of students in the matched samples fall in low performance levels, while fewer
students fall in high performance levels. Table 5 summarizes the number of students in
each PARCC performance level for each of the matched samples with columns 2 to 6
corresponding to the matched samples 1, 2, 12, 7, and 9 5 respectively.
Table 12
Matched Sample Size at Each PARCC Performance Level
Performance
Level
SAT Math
& ALG01
SAT Math
& ALG02
ACT Math
& ALG02
SAT Reading
& ELA10
SAT Writing
& ELA10
1
15
859
185
37
37
2
21
388
86
17
17
3
16
155
30
13
13
4
5
52
13
8
8
5
1
0
0
1
1
Total
58
1454
314
76
76
The mean scale scores at each performance level are reported for the five matched
samples in Table 13. There is a clear pattern that the mean scale score on SAT/ACT for
all five samples monotonically increases as the performance level increases from 1 to 5.
While Table 13 compares the mean score locations across performance levels, the
probability density plots displayed in Figure 2 provide a more detailed picture about how
the distributions of SAT/ACT test scores change across PARCC performance levels. As
the sample size for performance level 5 was too small to fit a density plot, only the plots
for performance levels 1 to 4 were generated. As displayed in Figure 2, the center of each
score distribution of SAT/ACT gradually shifts to the right as the performance level
increases from 1 to 4. The dispersion of the distributions also vary across samples and
performance levels. Generally speaking, students with higher performance levels on the
PARCC tests tend to have higher scores on SAT/ACT.
15
Table 13
Mean Scale Scores at Each PARCC Performance Level
Performance
Level
SAT Math
& ALG01
SAT Math
& ALG02
ACT Math
& ALG02
SAT Reading
& ELA10
SAT Writing
& ELA10
1
288.67
348.92
15.46
313.51
278.38
2
325.24
402.50
16.86
318.82
338.24
3
364.38
460.32
19.20
428.46
390.77
4
464.00
509.23
23.15
472.50
475.00
5
610.00
NA
NA
640.00
580.00
Figure 2. Probability density plots of SAT/ACT scores by PARCC performance level.
Summary
A series of statistical analyses have been conducted to explore the relationship
between the PARCC tests (ALG01, ALG02, and ELA10) and the college admission tests
(SAT, ACT, and PSAT). Data cleaning and exploratory analyses were carried out for
each of the four tests to understand the first-time takers and retakers. Descriptive statistics
on the PARCC test scores for all the data received and Grade 12 students were
summarized and compared. Correlational and regression analyses were conducted to
examine the relationship between the scores on the PARCC and the SAT/ACT tests using
the Grade 12 first-time takers. Equipercentile linking based on the matched samples for
different pairwise PARCC and SAT/ACT tests was used to establish the concordance
table for scores on PARCC and SAT/ACT. Last, the relationships between SAT/ACT and
the PARCC performance levels were investigated as well.
Although data were available for retakers with the highest scores, only first
attempts were included in the analysis for the following reasons. First, the correlations
16
between PARCC and SAT/ACT test scores were similar for the first attempts and the
highest scores. It was expected that the findings would be similar, had we used the
highest scores instead. Second, there are some issues in the data with the highest scores.
For example, both PARCC ALG02 and PSAT Math were included as predictors for
regression analysis for matched sample 5. Some students have the highest scores on
ALG02 in the first attempt but the highest score on the PSAT Math in the second attempt.
It would be difficult to decide whether to use the highest score on each test, the attempt
with the highest ALG02 score and that with the highest PSAT Math score, or the attempt
with the highest combined score. Third, from a diagnostic point of view, using the first
attempt data gives teachers the most information to develop remedial instruction to help
students improve and prepare for the college admission tests.
Major Findings
1. This study is conducted based on the matched samples for PARCC/PSAT and
SAT/ACT by subject. Only Grade 12 students’ first attempts were used in the
analyses, which usually account for more than 80% of the matched samples.
Matched sample sizes for other grades were too small to conduct statistical
analysis, which may result in reduced power to detect some significant
relationship.
2. The results of various correlational analyses and regression analyses show that
there is a moderate relationship between PARCC tests and college admissions
exams (SAT and ACT) for both Math and English. PARCC tests are statistically
significant predictors of SAT/ACT test scores. PARCC scores as predictors, along
with PSAT scores, explain 29% to 59% of the total variance in the SAT/ACT
outcomes. The PARCC tests are moderately correlated with similar SAT/ACT
sub-test scores.
3. Graphical representations show that Grade 12 students with higher performance
levels on PARCC tend to score higher on SAT/ACT. Concordance tables linking
the PARCC test scores to the SAT or ACT test scores in the same content area are
provided. These tables show the linkage between student performance on the
PARCC tests and the college admission tests, SAT and ACT. Though the PARCC
tests and the college admission tests, SAT and ACT do not measure exactly the
same content in a similar subject area, such concordance tables could help to
show the linkage between these two types of tests: high school graduation tests
and the college admission tests.
Limitations
Due to the limited number of administrations of the PARCC tests, the matched
sample sizes between the PARCC and SAT/ACT test scores were relatively small for the
analyses carried out. Caution should be exercised when generalizing the findings from
the current study based on the matched Grade 12 samples to all Maryland students. This
is due to the un-negligible discrepancies in the test score distributions for the matched
Grade 12 students used in the analyses and for all students in the data received.
17
References
Cui, Z., & Kolen, M. J. (2009). Evaluation of Two New Smoothing Methods in Equating:
The Cubic B‐ Spline Presmoothing Method and the Direct Presmoothing
Method. Journal of Educational Measurement, 46(2), 135-158.
Hanson, B. A. (1994). A Comparison of Presmoothing and Postsmoothing Methods in
Equipercentile Equating. ACT Research Report Series 94-4.
Kolen, M. J., & Brennan, R. L. (2004). Test equating, scaling, and linking (pp. 201-205).
New York: Springer.
18
Appendix A
LEGS example input window for linking PARCC AlG02 and SAT Math
Relationship Between PARCC
and College Admission Test
Scores
Maryland State Board of Education
Assessment Update
February 23, 2016
Purpose
To share with the State Board the results
of a research study examining the
relationship between PARCC test scores
and college admission test scores.
Background
The State Board requested information on
the relationship between PARCC test
scores and college admission test, SAT
and ACT scores.
The Maryland Assessment Research Center
(MARC) conducted a research study
examining those relationships.
Methods
This study was conducted based on matched
samples for PARCC/PSAT and SAT and
PARCC/PSAT and ACT, by subject.
Grade 12 students’ first attempts at college
admission tests were used in the analyses.
Accounted for more than 80% of the matched
samples.
*At the time of the research study, there was only one administration of the PARCC
tests; therefore, the matched sample sizes between the PARCC and SAT/ACT test scores
were relatively small. See the full report for the limitations of the study.
Matched Samples for Algebra
College
Admission
Tests
PARCC and PSAT
Tests
Total
SAT PARCC ALG01 58
SAT PARCC ALG02 1454
SAT PARCC ALG02, PSAT 1253
ACT PARCC ALG02 314
ACT PARCC ALG02, PSAT 277
Matched Samples for English
College
Admission Tests
PARCC and PSAT
Tests
Total
SAT Reading PARCC ELA10 76
SAT Reading PARCC ELA10 PSAT
Verbal and Writing
46
SAT Writing PARCC ELA10 76
SAT Writing PARCC ELA10 PSAT
Verbal and Writing
46
Results
Results indicate that there is a moderate
relationship between PARCC test scores
and the college admissions test scores.
PARCC test scores are statistically
significant predictors of SAT and ACT test
scores.
Findings for PARCC and SAT
PARCC
Performance
Level
PARCC
Threshold
Score
Correlated
SAT Score
Subject Alg II ELA 10 Math Reading Writing
Level 1 650 650 210 200 200
Level 2 700 700 400 340 340
Level 3 725 725 480 400 380
Level 4 750 750 530 490 460
Level 5 809 795 690 630 570
Findings for PARCC and ACT
PARCC
Performance
Level
PARCC Threshold
Score
Correlated
ACT Score
Subject Alg II ELA 10 Math
*
Reading/
Writing
Level 1 650 650 12
Level 2 700 700 17
Level 3 725 725 20
Level 4 750 750 23
Level 5 809 795 32
*Given the small sample sizes for the matched groups of students taking both tests,
such concordance relationship could not be established.
MD Proposed College and Career
Readiness (CCRD) Minimum Scores
Content PARCC SAT ACT
English
Language
Arts
(ELA)
Score of 750 or
greater on English
10 and English 11
Score of 500 or
greater on Evidence-
based Reading &
Writing (EBRW)
Section
Score of 21 or
greater on
Composite/
Average of
English Test &
Reading Test
scores
MATH Algebra II
score of 750 or
greater
Score of 500 or
greater Mathematics
Section
Score of 21 or
greater on
Mathematics Test
Study Findings Summary
PARCC
Performance
Level
PARCC Threshold
Score
Correlated SAT Score
Correlated
ACT Score
Subject Alg II ELA 10 Math Reading Writing Math
Level 1 650 650 210 200 200 12
Level 2 700 700 400 340 340 17
Level 3 725 725 480 400 380 20
Level 4 750 750 530 490 460 23
Level 5 809 795 690 630 570 32
MD minimum College and Career Determination scores: PARCC-750;
SAT-500; ACT-21
Summary
Despite limited information, the results
support Maryland’s proposed College and
Career Ready determination scores.
Subsequent research studies will be
conducted when additional administrations of
PARCC data is available.
