Vermont students' statewide scores were competitive with those of their peers elsewhere on the computerized Smarter Balanced Assessment Consortium test, according to the most recent results.
However, gaps among some ethnic groups and students of different economic backgrounds in Vermont continue.
This year was the second administration of the online test, which is given in grades three through eight and 11 in English language arts and mathematics and is aligned to the Common Core standards.
The Agency of Education released statewide results Tuesday.
This year, more students scored proficient and above in both subjects. The standard for proficiency is considered tough, according to Education Secretary Rebecca Holcombe.
"The proficiency thresholds are intentionally ambitious and give us a baseline to track our progress moving forward," she said in a statement. "Our children from more prosperous families continue to rank near the top nationally. Our most vulnerable youth — those living in poverty, with disabilities, from marginalized populations and who speak English as a second language — continue to have test scores that are on average lower than our general population."
More than half of students scored proficient or above in English this year, meaning they met or exceeded the standards tested:
• 54 percent of Vermont students in grades three and four.
• 58 percent in fifth grade.
• 56 percent in sixth grade.
• 58 percent in seventh grade.
• 59 percent in eighth grade.
• 57 percent in 11th grade (the only year that the percentage dropped, from last year's average of 58 percent).
In mathematics, the percentages meeting or exceeding the standard were somewhat lower:
• 56 percent of Vermont students in third grade.
• 50 percent in fourth grade.
• 43 percent in fifth grade.
• 41 percent in sixth grade.
• 46 percent in seventh grade.
• 44 percent in eighth grade.
• 38 percent in 11th grade (a 1 percentage point increase over the previous year).
Among the subgroups, more students of Asian descent scored at proficient and above than the state average across subjects and grades, while blacks and American Indians (including Alaskan Natives) scored below the state average for proficiency across subjects and grades.
Notably, Vermont Hispanic students scored better than the state average for proficiency and above in English in grades six (60 percent to 56 percent) and eight (68 percent to 59 percent).
In math they didn't do as well, falling below the state average for proficiency in all grades except eighth. In that grade, they did better than the rest of their peers at meeting and exceeding the standard by an additional 11 percentage points (44 percent statewide meeting or exceeding proficiency, versus 55 percent of Hispanic students).
Low-income students did not perform well and didn't meet the state average for proficient and above in any subject or grade. Their worst performance was in 11th grade math, where 19 percent reached proficiency.
"Although we hesitate to draw major conclusions from only one measure, we know that literacy and numeracy are first steps to improved life outcomes for all students," Holcombe said. "Children can't thrive independently in life beyond school if they can't read, write and use mathematics. These results reinforce our commitment to finding strategies that will help us eliminate these gaps in achievement."
Six other states in the 15-state testing consortium have released their preliminary scores, four of which published results that are comparable: California, Connecticut, Delaware and Washington. Among those, only California tested third through eighth grades and high school, as Vermont did.
Vermont students scored higher than California's in English language arts in all grades except 11th, and they scored higher than Golden State students in every grade in mathematics.
More Vermont students scored proficient or above in English than Connecticut students in grades three, five and six. But Vermont's students had a higher proficient and above score in mathematics in every grade from three through eight. Connecticut students did not take the test in high school.
Delaware students did better than Vermont kids in grades four and five in English language arts, but Vermont performed better in grades six through eight. As for math, Vermonters did better than Delaware students in grades five through eight. Delaware students didn't take the test in high school.
Washington students scored better in grades four and five in English, and they performed better than Vermonters in every grade in math.
The other states that have published results are using last year's SBAC scores as their baseline.
Although Vermont officials said last year that the 2015 scores would become the baseline for comparisons, they have since changed their minds.
Last year, the agency's news release said: "Since the student test scores released today establish a new baseline aligned with the Common Core, they should not be compared to previous statewide test scores on different tests." It was the first year the SBAC tests were administered across Vermont, and results could not be compared to the previous year's scores from the New England Common Assessment.
Tuesday, the agency again said this year's results should be the baseline.
"Although this is the second year the Smarter Balanced Assessments have been administered in Vermont, the Agency of Education is encouraging schools to consider this year's results as the baseline for determining progress over time," the news release said.
Testing Director Michael Hock said the move is due to the test being new and also being online.
"Vermont's students and teachers had very little experience with taking tests on computer. For that reason we want to treat the 2014-15 administration as a pilot test and use those results with considerable caution," Hock said.
Haley Dover, a spokesperson for the Education Agency, added that even this year's results aren't a good place to start comparing. "Next year there will be three years of results to look back on. Students will be used to the test, and we will get a better understanding of how they are doing," Dover said.
Vermont doesn't like reporting test scores by percent proficient because it says it is not as accurate as using scale scores, but federal law requires releasing the information to the public in this way.
"We find that proficiency scores alone have negative effects in supporting school improvement and recognizing the gains students make each year," the agency said. "For this reason, Vermont is moving to an increased focus on scale scores and identifying reliable methods for calculating growth scores. Both of these endeavors require multiple years of testing and are not currently available."
The agency plans to release school-level SBAC results Sept. 19.