Inaugural report · September 22, 2026
This launch edition analyzes the evidence available on September 22. September is still in progress; the completed-month edition is scheduled for October 5, 2026.
The September view is current through September 22, 2026. OpenAlex returned 28 candidate title matches for “STEM education” with publication dates from September 1–22, compared with 27 in the same dates of 2025. The “science education” query returned 47 candidates versus 66. These are records currently indexed, not a settled measure of research growth: recent work has had less time to enter the database. U.S. search interest is also available for partial September, while complete-month and annual comparisons provide context below.
These signals describe different things. Together they help identify questions for further research; they do not establish whether the public feels more or less positively about STEM education. No validated sentiment score is available in this launch edition.
Source & method updatesSeptember 22–24, 2026 · The original launch findings are preserved.
The evidence at a glance
| Measure | Observation period | Reading | Comparison |
|---|---|---|---|
| STEM education research candidates · current | September 1–22, 2026 | 28 | 27 during September 1–22, 2025; both queried September 22, 2026 |
| Science education research candidates · current | September 1–22, 2026 | 47 | 66 during September 1–22, 2025; both queried September 22, 2026 |
| U.S. search interest · partial September | September 2026, through September 22 | STEM education: 7; science education: 34 | Provisional 0–100 values; not compared with full months |
| STEM education research candidates | 2025 | 871 | 868 in 2024; +0.3% |
| Science education research candidates | 2025 | 1,415 | 1,329 in 2024; +6.5% |
| U.S. STEM education search interest | August 2026 | 9 | 11 in July; 8 in August 2025 |
| U.S. science education search interest | August 2026 | 35 | 50 in July; 31 in August 2025 |
| STEM education book phrase frequency | 2022 | 0.330 per million ngrams | 0.268 in 2021, same corpus |
Research figures are unvalidated OpenAlex English title-query matches. Search figures use a shared Google Trends export and its relative 0–100 scale. Book figures use the newer saved English corpus, with case-sensitive matching and no smoothing. The readings above are not directly comparable across measures.
September research activity: a current, provisional reading
At this edition’s September 22 cutoff, the Index opened with monthly Research activity. The matching September 1–22 windows contain 28 versus 27 STEM education candidates and 47 versus 66 science education candidates in 2026 and 2025, respectively. The elapsed calendar windows match, but the maturity of indexing does not. We therefore show the counts without presenting percentage changes as an estimate of underlying research growth.
The monthly chart spans September 2025 to September 2026. Earlier points cover complete months; the final point is month to date. It must not be read as a full-month decline. The collector retains the exact date-filtered queries and raw responses, and rejects missing or failed responses rather than treating them as zero.
Annual context: a higher level, followed by a pause in growth
The preserved OpenAlex snapshot contains 871 candidate title matches for “STEM education” in 2025, compared with 868 in 2024: an increase of three, or approximately 0.3%. Relative to 598 candidates in 2023, the 2025 count is 45.7% higher. Within this query-based series, the sharp increase occurred before 2025; the latest complete year mostly maintained that higher level.
The “science education” comparison series increased from 1,329 candidates in 2024 to 1,415 in 2025, or 6.5%. The two phrases can retrieve overlapping works and represent different terminology. Their counts should not be added together or treated as market shares.
These are discovery counts, not a census of verified studies. Relevance, duplicate versions and publication dates remain under review, and later indexing can revise earlier years. The project has not yet validated an education-research denominator. Consequently, a higher count cannot by itself establish a larger share of education research, improved research quality or greater public approval. The incomplete 2026 count is excluded from year-to-year growth comparisons.
Search interest: softer than July, higher than a year earlier
In the saved U.S. joint export, “STEM education” moved from 11 in July to 9 in August 2026, a decline of two index points. “Science education” moved from 50 to 35, a decline of 15 points. Compared with August 2025, however, both August 2026 readings were higher: 9 versus 8 and 35 versus 31, respectively.
The choice of comparison therefore matters. The latest monthly movement points downward, while the same-month annual comparison points upward. Neither comparison identifies the cause. Seasonality, news coverage and changes in search behavior are possibilities to investigate, not established explanations in this report.
Google explains that Trends uses sampled, normalized search data. Its index is not a count of people, an absolute search-volume total or a public-opinion poll. Small movements require caution. These comparisons use values within one preserved export; separately normalized downloads must not simply be appended. The September 22 export also includes partial September readings of 7 for STEM education and 34 for science education. These appear in the explorer, explicitly marked partial, but are excluded from the complete-month comparisons above.
Book attention: historical context with a clear endpoint
In the newer saved English book corpus, “STEM education” appears approximately 0.330 times per million ngrams in 2022, compared with 0.268 in 2021—an increase of about 23.1% using the unrounded values. This describes phrase frequency in the corpus. It does not measure how many people read those books, what they thought, or activity in September 2026.
The book series ends in 2022. The explorer also retains a separate legacy corpus ending in 2019. Google’s Ngram documentation describes how corpus revisions can change historical values; this project therefore keeps the two versions separate rather than joining them into an apparently continuous series.
What this means—and what to watch next
For educators, researchers and organizations, the immediate value is a common starting point for examining evidence. The candidate research series suggests sustained activity at a higher recent level, while search interest shows why one short-term movement should not be mistaken for a settled trend. The older book series helps place terminology in historical context.
Priorities for the next edition are to incorporate a complete September search observation when a supported export is available, identify revisions to research metadata, and advance relevance, duplicate and date review. The current media sentiment pipeline reports its own source coverage and scoring status; it does not claim a representative public-opinion reading. If a source update is unavailable, its limitation will be stated rather than filled with an invented value.
Next scheduled edition: October 5, 2026. That report will cover the completed September reporting month, while identifying the actual observation period of each source. This launch report records the evidence available through September 22, including the current-month update.
Sources and reproducibility
Calculations use the project’s preserved September 22 snapshots, including fresh date-filtered monthly queries and matching prior-year windows: OpenAlex English title queries for “STEM education” and “science education,” the joint U.S. Google Trends export beginning in January 2004, and the newer Google Books English Ngram series beginning in 1990. Source responses, query settings, input hashes and arithmetic are retained with this edition. Annual percentage changes use the prior value as the denominator and are rounded to one decimal place.
Explore the data and download observations · Read the methodology and limitations · Browse the Index report collection
