assessment

6. Student Assessment: The teacher maintains accurate records of student completion of assignments and student progress. The teacher also plans for performance assessment and provides students with feedback to improve progress.

APPR Planning Sheet 2011-2012

Evidence
I have given students a Benchmark pretest for Science 8. They will retake the same test at the end of the year. The students will also be taking 10 week assessments to track progress. Test and data analysis are done on Castle Learning.

Summary of Progress
I began this process during the summer of 2011 by writing the Benchmark test. Each quarter test is a variation on the Benchmark test, but testing only up to the information covered in class. The final exam was given completely online and was exactly the same as the Benchmark pretest. Questions were randomized for each student and ELL and IEP students were able to have questions read aloud using Castle Learning software.

Reflection
There are a few key things that enable these assessments to be useful and not be a burden on teacher or student. To begin with, the test are all online. Everything else of value from these tests comes from this feature. Item analysis is immediate. Changes to instruction can begin the NEXT DAY based on results. This is the case for all my tests as well. I can do remedial instruction if I find students tested poorly on a specific topic.

Another key aspect of these tests is seeing who will test well on the final exam and may need extra help. This is different than seeing what items tested poorly which is student independent. Generally, the same students that failed or didn't show for the quarterly tests, failed or didn't show for the final exam. I have yet to connect my testing to state testing since they are not identical measures. I do expect they would be similar in many respects though.

I also make the tests low impact. The Benchmark pretest is only graded as to whether it was completed or not. They get full credit regardless of how they do. I find at the beginning of the year especially that students want to do well on tests. Quarter tests and the final exam are curved so that only students who guess randomly risk failing. I'm more interested in the data than failing students.

The questions for next year include whether to keep the tests at the change of quarters or place them elsewhere in the school year. I'd also like to follow student progress better through tests. I wish Castle Learning had some better statistics features.

Evidence
Establishment of New Assessments
The following new assessments were added this year:

Pretest or Benchmark Test
Quarterly Assessment 1
Midterm: Quarterly Assessment 2
Quarterly Assessment 3
Final Exam (same as Benchmark Test)
Castle
Learning View of Assessments

Final Exam Administered Online
The Final Exam was successfully administered online.

Pre-Post Score Comparison

Pre	Post
37	62
30	63
41	68
38	63
38	76
32	56
40	63
33	40
40	76
22	35
51	78
32	56
41	41
32	46
33	73
33	60
40	54
48	81
43	73
37	56
40	76
29	57
40	68
14	27
19	25
48	73
32	65
29	43
35	70
46	59
60	81
33	35
40	73
52	92
44	59
35	68
32	57
46	63
35	54
46	68
29	67
41	57
41	59
41	68
41	76
56	81
22	32
57	67
38	49
38	63
29	52
29	67
30	40

A paired t-test was used to compare means.

Paired t test results

P value and statistical significance:
  The two-tailed P value is less than 0.0001
  By conventional criteria, this difference is considered to be extremely statistically significant.

Confidence interval:
  The mean of PreTest minus PostTest equals -23.08
  95% confidence interval of this difference: From -25.91 to -20.24

Intermediate values used in calculations:
  t = 16.3529
  df = 52
  standard error of difference = 1.411

Review your data:

Group	PreTest	PostTest
Mean	37.51	60.58
SD	9.14	14.76
SEM	1.26	2.03
N	53	53

Based on the results I can confidently say the Science 8 curriculum boosted test scores by an average of 23 points. Note, only those with a pre and post test scores have been included in order to do the paired t-test. I don't suspect this greatly changes the results. Only one student scored the same on both pre and post tests and no student scored worse on the post test than on the pretest. I will also admit that I have not determined the validity or reliability of the Benchmark test. Though the test is exactly the same I doubt students remember the questions from Septmember, but it is something to keep in mind.

Besides the change in means between the two tests; I also find the the standard deviations (SD) interesting. The smaller SD on the pretest shows that students all began at a similar level of ignorance. The larger SD on the posttest shows a disparity in knowledge: some learned more than others. There was a larger difference between high and low scores.

Back to Main Page