SAE Studio

Small Area Estimation (SAE), Without Writing Code

SAE Studio fits Fay-Herriot small area models to survey data and area-level tables. It computes direct and model-based estimates with measures of precision, runs diagnostics on every fit, and exports tables and a methods report. All computation runs in your browser.

$149 per year. One license includes every feature.

Fay-Herriot Model Estimates

Area-level Fay-Herriot models fitted by restricted maximum likelihood (REML). Every estimate carries a mean squared error. An arcsine transformation is available for proportions.

Estimates for Small Samples

The model borrows strength across areas. Areas with small samples receive model-based estimates with quantified precision in place of unstable direct estimates.

Diagnostics on Every Run

Each run reports shrinkage, coefficients of variation, and residual diagnostics beside the results table.

Complete Export Set

The export set covers the full run: the results table, one workbook per area, the master dataset (CSV, Excel, and SPSS), a methods report that records the model, the transformation, and every setting used, and a reproducibility package with an R script that recomputes the run.

Your Data Never Leaves Your Browser

All statistics run on your machine. No survey data is uploaded, transmitted, or stored on a server. The only value sent to the Maineblock server is the license key, for verification.

  • Survey data and calculations remain in your browser
  • License verification sends the email address and the license key only
  • Payment is processed by Stripe through the embedded checkout on this page
  • Exports require a valid license
Pricing

Pricing and Checkout

One tier covers every feature. New buyers check out in the New Purchase tab. Returning buyers unlock in the Returning Buyer tab with the email address and the license key from the receipt email. The tool runs in a browser on a desktop or laptop computer.

$149 per year
SAE Studio, Annual License
$149 per year
  • No limit on areas, indicators, or projects in your own files
  • Quick mode and Full mode, with bootstrap mean squared error
  • All exports: the results table (CSV), one workbook per area (Excel), the master dataset (CSV, Excel, and SPSS), the methods report (Word), and the reproducibility package
  • Diagnostics and the publication checklist on every run
  • The subscription renews each year until canceled

All buyers purchase through the checkout on this page. No invoice or purchase order option is available.

Checkout

Enter an email address to begin. The card form appears on this page, and card details go directly to Stripe. Maineblock does not receive or store card details.

The Tool

The Tool, in Four Steps

The workflow runs in four steps, from file upload to exported tables. The built-in demo runs all four steps on demo data at no charge. A license removes the limits on your own files and enables exports.

Step 1

Upload and Profile

Load a respondent-level survey file or an area-level table. The tool profiles the file, detects the column roles, and displays coverage by area before any model is fitted.

Step 2

Configure

Confirm the indicator, the subgroups, and the covariates. Select Quick mode or Full mode, and review the feasibility preview before fitting.

Step 3

Results

The results table reports the direct estimate and the model estimate for every area, with intervals, coefficients of variation, and disclosure flags.

Step 4

Diagnostics and Exports

Review the diagnostics and the publication checklist. With a license, download the results table, the per-area workbooks, the master dataset (CSV, Excel, and SPSS), the methods report, and the reproducibility package.

What You Need, and in What Format

The tool accepts two kinds of input file. A demo file and a blank template in the accepted formats are available below.

Respondent-Level Microdata

One row per respondent, in SPSS .sav, Stata .dta, SAS .sas7bdat or .xpt, Excel .xlsx, or CSV format. The file holds the area code, the outcome variable, the survey weight, and the design variables: the strata and the primary sampling units. The tool computes the direct estimates and their variances from the file.

Area-Level Table

One row per area, in CSV or Excel format. The table holds the area code, the direct estimate, its sampling variance or standard error, and the sample size for each area. A standard error works on its own, because the tool squares it.

Two additions are optional on either route. A covariate file in CSV or Excel format, keyed by the same area codes, gives the model more to borrow from. Population control totals, keyed the same way, feed the benchmarking that runs by default.

The demo file holds direct poverty estimates for the sixteen counties of Maine, with three covariates. The demo inside the tool loads this same file. The blank template shows the area-level format, with headers only.

Results Preview

See How Your Results Will Look

The tool opens on a coverage view of your areas, then reports the direct estimate and the model estimate side by side. Both views below come from the built-in demo data: direct poverty estimates for the sixteen counties of Maine, with the model fitted to them.

Coverage: Direct Estimate by CountyPoverty rate (%)
Piscataquis19.6% · n=42
Washington18.9% · n=88
Somerset17.8% · n=112
Aroostook16.4% · n=164
Androscoggin15.2% · n=268
Penobscot15.0% · n=342
Oxford14.7% · n=128
Franklin13.9% · n=64
Waldo13.6% · n=96
Kennebec12.8% · n=288
Hancock12.1% · n=118
Knox11.4% · n=84
Lincoln10.8% · n=78
York9.4% · n=412
Sagadahoc9.2% · n=58
Cumberland7.8% · n=524
Under 10% 10–14% 14–17% 17% +

Direct Estimate Beside Model Estimate

The model borrows strength across counties, so the smallest counties move the most. Piscataquis, with forty-two responses, moves up from 19.6 percent to 20.4 percent, and its coefficient of variation falls to thirteen percent.

Demo Results, Five Counties
CountyDirect EstimateModel EstimateModel CV
Piscataquis19.6%20.4%13%
Washington18.9%18.9%11%
Somerset17.8%18.3%11%
Penobscot15.0%14.7%10%
Cumberland7.8%8.0%12%

The tool runs the full demo, with all sixteen counties, the intervals, and the diagnostics.

Open SAE Studio

How It Works

From File to Published Table in Three Steps

SAE Studio serves analysts at state agencies, nonprofits, and research firms who require small area estimates without rebuilding the method in R for each project.

01

Upload and Confirm

Upload a microdata file or an area-level table. The tool profiles the file, proposes the column roles, and flags a missing variance before any model runs.

02

Fit the Model

Select the covariates, the transformation, and Quick or Full mode. The Fay-Herriot model runs in the browser, and the file never leaves the machine.

03

Check and Export

Review the shrinkage, the flags, and the diagnostics, then export the tables and the methods report.

Free Guide

How to Run a Small Area Estimation

The free guide states the complete method the tool implements. Readers can follow the method, verify it, or apply the same steps by hand. No license or email address is required.

Read the Free Guide

Companion Product

SAE Studio Companion Workbook for Excel

The workbook implements the same Fay-Herriot calculations as Excel formulas. It is a separate product with its own price, sold on Gumroad. It is not included in the SAE Studio license.

Available soon. Sold separately on Gumroad, and not included in the SAE Studio license.

Questions

Ask a Question

Questions go to the Maineblock inbox, and the reply arrives by email.