Comparing Results for the Computer-based Assessments

Aggregate results can be viewed at and downloaded from the public Test Results for California’s Assessments website as well as from CERS (for properly credentialed users).

When making comparisons across years within a given grade level and content area, it is important to understand that even when the number of students is the same, the group’s composition from year to year may be quite different if student mobility (transiency) is high. Such comparisons are actually comparisons of different groups of students with different traits taking different assessments. Generally, there will be more variance in scores from year to year, when small numbers of students are tested.

Smarter Balanced Summative Assessments

Because of the vertical scaling of the Smarter Balanced assessments, scale scores for an assessment may be compared to scale scores for the same student or groups of students in different years for the same content area, as well as between specific grade levels and content areas. This allows users to say that achievement for a given content area and grade level was higher or lower one year as compared with another. Scale scores for the Smarter Balanced assessments may be compared across grade levels since the scales are vertically aligned across grade levels. Refer to the CDE Assessments - Average Scale Score Change web page to download spreadsheets that report the average change in scale score points from one grade level to the next by administration year. These reports show how scale scores changed among all students statewide in prior years. Note that this web page is updated with data from the just-tested year in the fall following the test administration.

Scores for the Smarter Balanced Summative Assessments PPT forms are linear but have the same scale as the computer-based assessments.

In addition to the scores for the current administration, results for comparison are available for past administrations. Refer to the CERS User Guide for more information about reviewing data for previous years.

Comparing ELA and Mathematics Scale Scores and Achievement Levels for Groups

An example of how group-level scale scores for 20XX–YY may be compared to the 20YY–ZZ scale scores for the same content area and grade level is shown in table 1 using Smarter Balanced for ELA scores. In this table, hypothetical average scale scores for ELA are compared between the consecutive years of 20XX–YY and 20YY–ZZ for the students in a particular school. In addition to comparisons for all students, similar grade level–by–grade level comparisons of scale scores may be made for different student groups of interest. Finally, an average scale score is not provided for all students across grades in the same school in table 1 because each grade level has its own scale score range.

It is important to note that the difference in the average scale scores from two years may be a result of different student groups.

Table 1. Hypothetical Example of Using the Smarter Balanced for ELA to Measure Progress by Comparing Average Scale Scores
Grade	20XX–YY No. of Students	20XX–YY Mean Scale Score	20YY–ZZ No. of Students	20YY–ZZ Mean Scale Score	Difference
Grade 5	120	2440.0	111	2451.3	11.3
Grade 6	100	2510.0	124	2510.3	0.3
Grade 7	90	2590.0	102	2593.2	3.2

Table 2 uses Smarter Balanced for Mathematics scores to provide a second hypothetical example of how group-level Smarter Balanced results may be compared. In this example, the percentage of students scoring at Standard Met or Standard Exceeded in mathematics is compared between 20XX–YY and 20YY–ZZ across grade levels for the same school.

Table 2. Hypothetical Example of Using the Smarter Balanced for Mathematics to Measure Progress by Comparing Percentages of Students at Standard Met or Standard Exceeded
Grade	20XX–YY No. of Students	20XX–YY % Standard Met or Standard Exceeded	20YY–ZZ No. of Students	20YY–ZZ % Standard Met or Standard Exceeded	Difference
Grade 5	120	31%	111	35%	4%
Grade 6	100	33%	124	33%	0%
Grade 7	90	29%	102	31%	2%
All Grades	310	31%	337	33%	2%

Comparisons between 20XX–YY and 20YY–ZZ in table 2 indicate the same trends as indicated by table 1: a slightly higher percentage of students in grades five and seven scored at Standard Met or Standard Exceeded and the same percentage of grade six students scored at Standard Met or Standard Exceeded.

Note that different from table 1, table 2 provides a comparison of overall results for the entire school that is not provided in table 1. Because Standard Met or Standard Exceeded in mathematics is a standards-based classification, the percentage of students meeting or exceeding standards can be calculated for the entire school across grade levels by dividing the total number of students meeting or exceeding standards in the school from all grade levels by the total number of students, instead of by averaging the percentages from different grade levels. The resulting school-level averages may be compared from year to year.

While these examples have made comparisons across only one year, it is important for program evaluation that results be compared across a number of years to verify that the trend is stable. The same sort of table could be used to compare year-to-year results for any test group.

Comparing Results for the Science Assessments

Comparisons of CAST or CAA for Science results should only be made within the same grade level; that is, it is acceptable to compare the grade five CAST in the previous test administration year to the grade five CAST in the current test administration year. However, no direct comparisons should be made between grade levels and between subjects; for example, results for the CAASPP Smarter Balanced for ELA (grade five) should not be compared with results for the CAST (grade eight).

Two types of comparisons are possible:

comparing the average scale score within a grade level; or
comparing the percent of students scoring at each achievement level within a grade level.

Comparing Science Scale Scores and Achievement Levels for Groups

Because CAST is not vertically scaled, the scores between different assessments are not comparable. For example, the scale scores from grade five cannot be compared to those from grade eight or high school to measure growth. However, the scores between grades ten, eleven, and twelve are comparable because students in different high school grade levels are taking the same high school assessment.

In addition to comparisons for all students from the same school, similar cross-year comparisons of scale scores may be made for different student groups of interest. Note that an average scale score is not provided for all students across grades in the same school in table 3 because the scores from different grade-level or grade-band assessments are not comparable.

An example of how group-level scale scores for 20XX–YY may be compared to the 20YY–ZZ scale scores for the same content area and grade level is shown in table 3 using CAST scores. In this table, hypothetical average scale scores for CAST are compared between the consecutive years of 20XX–YY and 20YY–ZZ for the students in a particular school.

Table 3. Hypothetical Example of Using the CAST to Compare Average Scale Scores across Years
Grade	20XX–YY No. of Students	20XX–YY Mean Scale Score	20YY–ZZ No. of Students	20YY–ZZ Mean Scale Score	Difference
Grade 5	100	198.0	111	199.2	1.2
Grade 8	80	400.5	95	401.1	0.6
Grade 10	10	598.5	12	597.0	−1.5
Grade 11	100	601.2	105	600.8	−0.4
Grade 12	25	600.0	28	601.2	1.2
High School	135	600.8	145	600.6	−0.2

Table 4 uses CAST scores to provide a second hypothetical example of how group-level CAST results may be compared. In this example, the percentage of students scoring at Standard Met or Standard Exceeded in CAST is compared between 20XX–YY and 20YY–ZZ by grade levels for the same school.

Table 4. Hypothetical Example of Using the CAST to Compare Percentages of Students at Standard Met or Standard Exceeded across Years
Grade	20XX–YY No. of Students	20XX–YY % Standard Met or Standard Exceeded	20YY–ZZ No. of Students	20YY–ZZ % Standard Met or Standard Exceeded	Difference
Grade 5	400	27%	360	28%	1%
Grade 8	80	28%	95	28%	0%
Grade 10	10	31%	12	30%	−1%
Grade 11	100	30%	105	30%	0%
Grade 12	25	29%	28	28%	−1%
High School	135	30%	145	30%	0%
All Grades	615	28%	600	28%	0%

Note that table 4 provides a comparison of overall results for the entire school that is not provided in table 3. Because Standard Met or Standard Exceeded in CAST is a standards-based classification, the percentage of students meeting or exceeding standards can be calculated for the entire school across grades levels by dividing the total number of students meeting or exceeding standards in the school from all grades by the total number of students, instead of by averaging the percentages from different grade levels. The resulting school-level percentages may be compared from year to year.

Comparing Results for the California Alternate Assessments for ELA and Mathematics

Comparisons of CAA results should only be made within the same content area and grade level; that is, compare grade four ELA in the previous test administration year to grade four ELA in the current test administration year, or grade eight mathematics in the previous test administration year to grade eight mathematics in the current test administration year. No direct comparisons should be made between grade levels and between subjects; for example, results for the CAA for Mathematics (grade seven) should not be compared with results for the CAA for Mathematics (grade eight), and results for the CAA for Mathematics (grade eleven) should not be compared with the result of the CAA for ELA (grade eleven).

Two types of comparisons are possible:

comparing the average scale score within a grade level; or
comparing the percent of students scoring at each achievement level within a grade level.

Comparing Results for the California Spanish Assessment

CSA results for the previous administrations are available. Comparisons of CSA results should only be made within the same grade level; that is, it is acceptable to compare the grade five CSA in the previous test administration year to the grade five CSA in the current test administration year. However, comparing scale scores from different grade levels for the CSA is not appropriate, because the curricula are different across grade levels and the scale scores are not vertically linked between grade levels.

Comparing Results for the Summative ELPAC

An individual can make comparisons within the same grade level, grade span, and overall score across years. Because the ELPAC results are scaled vertically, scale scores for an assessment at one grade level may be compared to scale scores at another grade level or grade span; this allows for the comparison of the same student’s performance over time, as well as comparison of student groups at different grade levels or grade spans. Thus, it can be said that proficiency for a given grade level or grade span was higher or lower one year as compared with another. However, caution should be taken when comparing scale scores from different grade levels or grade spans, especially nonadjacent grade levels, because the curricula are different across grades. Comparing scores obtained in different composite scores (i.e., oral language scores and written language scores) should be avoided because the results are not on the same scale.

Two types of comparisons are possible:

comparing the average scale score within a grade level; or
comparing the percent of students scoring at each performance level.

An example of how group-level scale scores for 20XX–YY may be compared to the 20YY–ZZ scale scores for the same grade level is shown in table 5 using ELPAC overall scores. In this table, hypothetical average scale scores for the ELPAC are compared between the consecutive years of 20XX–YY and 20YY–ZZ for the students in a particular school. In addition to comparisons for all students, similar grade level–by–grade level comparisons of scale scores may be made for different student groups of interest. It is important to note that the difference in the average scale scores from two years may be a result of different student groups. An overall average scale score is not provided for students across all grades levels in the same school in table 5 because each grade level or grade span has its own scale score range.

Table 5. Hypothetical Example of Using the ELPAC to Measure Progress by Comparing Average Scale Scores
Grade	20XX–YY No. of Students	20XX–YY Mean Scale Score	20YY–ZZ No. of Students	20YY–ZZ Mean Scale Score	Difference
Grade 5	120	1520	111	1518	−2
Grade 6	100	1525	124	1524	−1
Grade 7	90	1535	102	1537	2

Table 6 uses ELPAC scores to provide a second hypothetical example of how group-level ELPAC results may be compared. In this example, the percentage of students scoring at Level 4 (Well Developed) is compared between 20XX–YY and 20YY–ZZ across grade levels for the same school.

Table 6. Hypothetical Example of Using the ELPAC to Measure Progress by Comparing Percentages of Students at Level 4 (Well Developed)
Grade	20XX–YY No. of Students	20XX–YY % Reaching Level 4	20YY–ZZ No. of Students	20YY–ZZ % Reaching Level 4	Difference
Grade 5	120	20%	111	19%	−1%
Grade 6	100	18%	124	16%	−2%
Grade 7	90	20%	102	22%	2%
All Grades	310	19%	337	19%	0%

Note that table 6 provides a comparison of overall results for the entire school that is not provided in table 5. Because performance levels are standards-based classifications, the percentage of students reaching level 4 can be calculated for the entire school across grade levels by dividing the total number of students reaching level 4 in the school from all grade levels by the total number of students, instead of by averaging the percentages from different grade levels. The resulting school-level percentages may be compared from year to year.

Comparing Results for the Summative Alternate ELPAC

Comparisons of Summative Alternate ELPAC results should only be made within the same grade level; that is, it is acceptable to compare the grade two Summative Alternate ELPAC results in the previous test administration year to the grade two Summative Alternate ELPAC results in the current test administration year.

Two types of comparisons are possible:

comparing the average scale score within a grade level; or
comparing the percent of students scoring at each performance level within a grade level.