The Power $ Efficiency of a Data Warehouse

Does this happen in your school district?
 
The Scenario
A director of curriculum and instruction asked each school lead teacher to create a Google spreadsheet of student scores. The file needed to contain each student enrolled in the school Grade 3-8 and list last year’s reading, math, and science score information, the EVAAS Projected score, the Check-in percent correct with the subscores for CI #1 and #2, and I-Ready scores and subscores.
 
This information would be used to help identify students at-risk and to create remediation groups.
The lead teachers started the task and found that it was a very time-consuming task.
 
The Solution
One lead teacher reached out to me and asked if any of this information was in the district’s new data mart system. Yes, all of the data had been uploaded to the system and was easily put into a spreadsheet report by writing SQL code to pull the data from the tables.
Consequently, WITHIN a few hours each school had reports of all of their students with all of the required data.
Exporting the information from the database saved the lead teachers about 15 hours of work EACH!
 
Critical Question
Can you collect all of that information on your hundreds /thousands of students as efficiently as a school district with a data warehouse?
 
A hosted database is the answer. Upload spreadsheets, link the data with student number, wirte simple SQL code using a query tool and run the code and download your spreadsheet with the merged data that your administrators and teachers need.
 
A databses and attached reporting solution does not need to cost tens of thousands of dollars. It can be done for most school districts for less than $10,000. And Data Smart LLC can provide the training so that the data warehouse can be managed in-house by your school district data stewards.  

Integrating Student Performance Data – Using APEX

As I teach data management for school districts and consult on building data systems I focus on assisting school districts in the importance of integrating data into a coherent information source. As school districts administer North Carolina Check-In assessments, the reports are distributed to schools via paper reports or electronic .pdf documents. The districts also administer other assessments like the NWEA or I-Ready. Then for grades K-3, the district administers DIBELS. Each of these electronic assessments has a data portal and presents reports based on the single data source. This “siloed” data system, frequently managed by different administrators in the district, provides a myopic view of the student’s performance.

All of these data sources do not take into consideration a student’s prior performance on assessments like the End of Grade tests. My recommendation to school districts is to develop a singular portal for access all student performance in one data system so that various student scores during the year can be accessed and viewed alongside of each other. For example, it is important to know that a student on this year’s I-Ready test is testing at the 40th percentile rank during the BOY test and at the 20th percentile rank on the MOY testing, while last year the student achieved an achievement level of 4.

By integrating the student performance data in one data system, with tables of student test scores and demographic information including the student identification number, all data can be linked using the SID. This can be accomplished by creating a database and a reporting system using Oracle Application Express. For more information on how this can be accomplished visit Data Smart LLC in Greensboro, NC at www.Data-Smart.net.

Item Analysis Program: Iteman 4.5

Item Analysis Program: Iteman 4.5

Assessment Systems Corp. has released version 4.5 of the popular program Iteman, an item analysis program that I have used for since being director of testing and accountability.

The program now self-generates a report. For my testing of the program on 100 subjects on an 18 item Math 1 test using the demo version created a 30-page report. The report provides item-level analysis including p and discrimination values. The quartile plot created by the program provides insight into responses not available in other easy-to-use programs. The 46-page User Manual is very complete and provides file setup and some interpretive information.

More information is available at: https://assess.com/2021/03/31/what-is-item-analysis/

Measuring the Impact of the Pandemic on Grade 2-3 Reading Performance for 2020-21

For this brief report, the impact of being out of school and doing home-based learning for grade 3 students was examined in a rural school district. The sample size is approximately 300 students.

Measures:

  1. I-Ready beginning of the year (BOY) reading test overall percentile score converted to an NCE (normal curve equivalent) score.
  2. Beginning of Grade 3 (BOG) ELA test percentile rank score converted to NCE score.
  3. Difference scores between each student’s I-Ready BOY NCE and the corresponding BOG NCE score. By using intra-individual score differences, differences in overall average score from the two different cohort groups were not a confounding factor for this analysis.
  4. BOG percent proficient was computed using counts of level 3-5 and dividing it by the total number of scores. This was computed for each school.

Procedure:
For the fall 2019 and the 2020 (current year) the BOG scores were collected and matched into a single table with the previous year I-Ready BOY data score.

  1. The joined table included columns for the year, school, SID, I-Ready tier, NCE BOG, and NCE I-Ready scores.
  2. A difference score was computed for each student by subtracting the BOY I-Ready NCE score from the BOG NCE score. For example:
    2020 BOG NCE = 50 and the BOY I-Ready NCE score = 45  difference = +5
  3. For each school the average of the differences was computed. 
  4. NCE score differences were disaggregated by school and by reading tier (Tier 1, Tier 2, and At-risk for Tier 3).

Findings: 

  1. The average I-Ready to BOG NCE difference for the 2019-20 school year for the district was 6.5 with a range of 1.0 to 9.7.
  2. The average I-Ready to BOG NCE difference for the 2020-21 school year for the district was -19.8 with a range of -17.8 to -24.2. 
  3. Differences by Tier for the two different school years were as follows:
    1. 2019 Tier 1   5         Tier 2  6.7        At-Risk for Tier 3  10.1
    2. 2020 Tier 1   -26.8      Tier 2  -20.7     At-Risk for Tier 3  -9.9
  4. The 2019 BOG district proficiency was 24.1% and this is a consistent score for the last three years. The 2020 district proficiency was 11.5%.

Conclusion:
The impact of being out of school for the spring of 2020 due to the pandemic is measurable for grade 3 students using the methodology described above. Furthermore, the results suggest that there is a negative impact on reading performance for the student in this district, and the impact cuts across all reading tier levels.

Actions:
 Present this data to the schools and make a school roster of students for each school so that the school leaders and teachers can identify the students who experienced the greatest negative impact so that interventions can be provided.

Use the I-Ready diagnostic data to determine if there is a pattern of weak areas across the most negatively impacted students. For each student determine specific areas had the greatest negative impact and provide targeted individual intervention.  

Policy Impact:
While growth for grade 3 students can be determined by comparing BOG to EOG NCE scores, Percent proficient for grade 3 is quite likely to be lower than the 2018 EOY percent proficient. This is likely to be the case for ELA for grades 3-8. Accountability using the current targets will be problematic.

Test Score Analytics Using APEX

What is the value for a school district to have all of its student performance data in one database?

The answer is quick and accurate multi-year analytics. After the initial report is created, each year is uploaded and populates all of the reports.

Example

Here is an example from a school district, that is looking forward to this year’s Beginning of Grade 3 test (BOG) data and making comparisons to other test data. Using a brief SQL statement the mean scores and standard deviations can be computed and reported for each school for each year in the APEX data system.

Then the distribution of scores in the form of a histogram for each of the years can be viewed. Using a simple drop-down filter, the year and the individual school can be overlaid and viewed .

The question is then what is the relationship between the BOG scores and the i_Ready reading scores when both are converted to Normal Curve Equivalent (NCE) scores? Here is the sample SQL code for that report:

select
    BOG.SCH_CODE as SCH_CODE,
    count(BOG.BOG_NCE),
    round(CORR(BOG.BOG_NCE, I_READY_RD.IR_RD_NCE), 3)
 from I_READY_RD I_READY_RD,     BOG BOG
   where I_READY_RD.SID=BOG.SID and BOG.YEAR = 2020
      and I_READY_RD.TEST_DATE = ‘BOY’
        group by BOG.SCH_CODE
      order by BOG.SCH_CODE

The next question is: What does this data look like as a scatter plot. In the scatter plot, an entire district can be shown for a testing year or the scatter plot can be made to show one school, and then the other data color will show al of the remaining schools. In conclusion, when a district has the BOG and I_Ready (or I_Station) data for this year, it will be better able to understand the significance of the information and perhaps understand the impact on being out of school for months has had on its students.