Tears or Tiers for Data Analysis?

Years ago, I worked for/with a dream team, and we were tasked with developing our district's data-based decision-making framework from scratch. After reviewing many districts, I landed on a model of data presentation from SCRED in MN. Leadership in SCRED at the time was generous with time and sharing of information.

We were using NWEA-MAP as our universal screener and we wanted a way to give teams an efficient way to see where we wanted students to be (proficient) and if there were scores that should prompt further review for possible intervention discussions (below 25th percentile- I/we guesstimated based on a loose quote from Fletcher et. al. about reading difficulty beginning below a standard score of 90). The work is still online, and I smile each time I can still see it (link).

We created a document called the Tiers of Performance. This table was based on the work of many districts and states (SCRED, Oregon RTI, Florida MTSS, Derry Township School District, PA). We tailored our table to include the fields we thought teams needed, such as proficiency targets, warning triggers, expected growth, and talented and gifted indicator scores.

Even though much was "borrowed", creation took a long time, was tedious, and easy to make errors from copying and pasting numbers. "Wait, did I copy from the reading or math table? Was that the right grade? What number did I just transpose?" It took forever... But we wanted a framework for reliable, quick, efficient, and consistent data analysis for our school teams. Feedback loops were critical, and I would draft, share, take suggestions, redo, share, and repeat. We landed on a final version that was the backbone of our data-based decision-making process in our MTSS framework.

I'm now in another state and district, and the district is moving from one screener to MAP. Ironically, the 10th and 25th percentiles have both been used to identify where required reading intervention is needed. The table holds up!

Disclaimer: Scores are more precise than percentiles. However, most educators can find percentiles on score reports with greater ease.

• Color Code Workaround: My recommendation is to color-code performance bands and if possible, conditionally format or build in color-coded scores for each performance band (limited, proficient, etc.). Then, scores are listed (better precision), and educators see the student's performance using the color-efficient and reliable.

Targets: Many screening assessments conduct studies of data to provide cut scores where students should score to indicate they are on track to score in the proficient range on the year-end state test. MAP does this for many states. For a better description, check this out (link). Ideally, we want all students to reach proficiency and have the best odds of success for whatever path they choose after graduation. The targets are based on scores estimated to be solid indicators that if students score at or above that target in the season of the assessment, they are on track.

For K and 1, MAP does not have linking data. Originally, I worked backward from the 2nd-grade targets using growth data and smoothed out the scores to be realistic. For this one, I was lazy and asked NotebookLM to figure it out for me. See below, I was walking my dog.

Growth: Even if students score below the target, knowing how much growth they achieved is critical. If a student is below the target or even the 25th percentile but they are exceeding growth estimates, that may indicate they can achieve proficiency in the future. On the other hand, if they are below proficiency and growth is far lower than the average growth, that's another problem. MAP has individualized growth targets for each student, but the table has average growth for the sake of team discussion and quick reference.

Dog Walking: I walk my aging herding mix dog three times a day. She now sniffs everything, and often, and I am happy to let her do that (she's not the only aging one here). Last week, as I was waiting and making sure she was not into anything too gross, I thought back on the Tiers of Performance and how it would be useful to at least me, since we would be using MAP next year (I need more podcasts, I know). I pulled out my phone, searched the old version of the Tiers, pulled up the MAP Linking Study for Ohio, and the newest norms from MAP. I then asked it to use the Tiers and create an updated document from the other sources. This is between walking and sniffing using NotebookLM. Here's my notebook if anyone wants to see (link). I asked it to create an Excel or Sheets version, which is not possible. It offered text I could copy/paste, and that worked.

Final Version: Here is the final, updated version that I continue to refine (LINK).

Disclaimer: I have repeatedly asked NotebookLM to review and check the table for accuracy and consistency. I am on summer break and plan to hand check, but it is nice out, and my dog will need to go out soon. AI makes mistakes, I know, but so do I. Use with caution and review this.

Benefits of AI for this work?

The idea of redoing the Tiers of Performance and all the copying/pasting was something I was not really up for at this point. I saw a tedious project that would take a while.

NotebookLM allowed me to control my sources, ask questions, reframe, redo, and check and recheck my work. It also allows me to share what I did and provide step-by-step directions if anyone else finds this useful. I spent more time on this blog than the Tiers.

AI-Generated Blog with Directions:

I also asked NotebookLM to write up a blog post about what I did to generate the updated Tiers of Performance. Here's the AI version, no edits.

The process used in this notebook to build the 2025 Ohio Performance Tiers represents a systematic integration of national descriptive norms and state-specific prescriptive benchmarks. By synthesizing data from multiple technical reports, we transformed raw RIT scores into an actionable instructional framework.

Below are the steps taken during this process and a guide for how a K-12 practitioner can build a similar table from scratch using any universal screening assessment.

The Steps Used in This Notebook

1. Framework Selection: We began by using an existing "Tiers of Performance" model (originally from a 2018 Wisconsin study) to establish the layout: identifying intervention thresholds, on-track proficiency, and high-achievement targets [1, 3, conversation history].

2. State Alignment Update: We replaced outdated data with the 2025 Ohio MAP Growth Linking Study, pulling exact "Proficient" cut scores to ensure the table correctly predicted success on the Ohio State Tests (OST).

3. National Norm Integration: We cross-referenced state scores with the 2025 NWEA Technical Manual to add national percentile context. This allowed us to include the 10th and 25th percentiles for intervention and the 95th percentile for talent development.

4. Predictive Modeling for K–1: Since state linking studies typically begin at Grade 3 (with projections for Grade 2), we used a "working backward" methodology. By taking the Grade 2 Spring proficient target and subtracting the 2025 mean growth norms, we mathematically derived targets for Kindergarten and Grade 1 [172, 233, conversation history].

5. Secondary Readiness Integration: We incorporated data from the 2025 Ohio EOC Study to add specific targets for Algebra 1 readiness and Graduation Competency requirements, ensuring the table served middle and high school practitioners.

6. Accuracy Verification: Every RIT score was verified against the raw data tables in the source appendices to ensure that the Math and Reading scores were not transposed and that all calculations for K–1 were precise [106, 108, conversation history].

How to Create Your Own Table from Scratch

To build a unified performance table for your district using MAP Growth or other screeners, follow these six linear steps:

Step 1: Gather Your Three Essential Documents

• National Technical Manual: Provides the descriptive norms (how students typically perform).

• State Linking Study: Provides the prescriptive benchmarks (the minimum score needed to pass state tests).

• Course-Specific/EOC Reports: Essential for secondary schools to identify targets for high-stakes courses like Algebra 1.

Step 2: Define Your Performance Tiers

Decide which benchmarks your district uses to trigger support:

• Intervention: Commonly, the 10th percentile (Tier 3) and 25th percentile (Tier 2).

• Proficiency: The state-aligned "Proficient" cut score.

• High Achievement: The 95th percentile for "Talent Development" or gifted identification.

Step 3: Extract RIT Scores for Tested Grades

Open your state linking study and locate the Summary of MAP Growth RIT Cut Scores (usually Table E.1). Record the Fall, Winter, and Spring RIT scores for every grade that takes the state assessment.

Step 4: "Work Backward" for Kindergarten and Grade 1

If your linking study lacks K–1 data, you can calculate it using the following formula: $RIT_{PreviousTerm} = RIT_{Spring} - g$

• Spring Cut: Start with the Grade 2 Spring Proficient RIT.

• g: Subtract the "Mean Growth" for the preceding grade found in the Growth Norms section of the Technical Manual (e.g., Table C.1 for Math).

Step 5: Add Growth Expectations

To help teachers evaluate student progress, add columns for Mean Seasonal Growth. Locate the "Mean Growth" tables in your technical manual (Appendix C) and pull the values for the Fall-to-Winter and Fall-to-Spring intervals for each grade.

Step 6: Format for Practitioners

Organize the data into a single, color-coded spreadsheet.

• Place intervention targets on the left (red/orange).

• Place the state proficiency target in the center (green).

• Place high-achievement targets on the right (blue).

• Final Touch: Add a "Special Notes" section for graduation requirements or course-readiness targets to make it a one-stop shop for your administrators.

I have to get the dog out.