MSSS-20

A look at the current normative dataset for the Megalophobia Severity Screening Scale The Megalophobia Severity Screening Scale (MSSS-20) is a 20-item self-report measure designed to screen for megalophobia — discomfort or fear related to very large objects, structures, or spaces. Since data collection opened on June 11, 2026, the dataset has grown to 359 participant responses from 37 countries, collected over roughly two months (through July 14, 2026). The data is collected from intractive version megalophobia test.

Anthropic's overall assessment classifies the dataset as "Strong," scoring 70 out of 100 on a composite maturity index that weighs sample size, geographic spread, and internal consistency.

Who's in the Sample

The 359 respondents skew young: 37 participants are under 18, and 29 fall in the 18–24 range, with a meaningful share (25) preferring not to disclose age. Gender representation is close to balanced between female (59) and male (57) respondents, with 8 selecting "prefer not to say."

Geographically, the sample leans heavily toward a handful of countries. India contributes the largest share at 29.2% of geo-tagged responses (105 participants), followed by the United States (15.3%, 55 participants) and Germany (9.5%, 34 participants). The remaining 34 countries each contribute smaller slices, from a few percent down to single respondents in places like Nigeria, Singapore, Norway, and Denmark. This distribution means norms are currently most representative of Indian, American, and German populations, and should be applied cautiously elsewhere.

Score Distribution

Across the full scale, the average score was 51.36% (SD = 15.22), with a median of 49 — close enough to the mean to suggest a reasonably symmetric distribution. That's backed up by the shape statistics: skewness of 0.56 and kurtosis of 0.43, both mild enough that the data doesn't show strong departures from normality.

Scores ranged widely, from a low of 8% to a high of 98%, giving an observed range of 90 points. The 95% confidence interval for the population mean sits between 49.78 and 52.93.

Percentile benchmarks:

PercentileScoreInterpretation10th35%Very Low25th40%Below Average50th49%Average75th59%Above Average90th73%Elevated

Reliability

Internal consistency, measured with Cronbach's alpha, came in at 0.706 — generally considered "acceptable" by conventional psychometric standards, though below the 0.80+ threshold typically preferred for high-stakes individual assessment.

Item-level analysis tells a more nuanced story. A handful of items pull real weight — Q11 posted the strongest item-total correlation (r = 0.52, rated "Excellent") — while a notable cluster underperforms. Ten of the twenty items were flagged for review, including several rated "Problematic" (r below ~0.20). The weakest item, Q20 ("Very large environments can sometimes feel psychologically unsettling"), actually had a negative correlation with the total score (r = -0.01), meaning it isn't tracking the same underlying construct as the rest of the scale and is a strong candidate for revision or removal.

The Four Dimensions

The MSSS-20 breaks down into four subscales, and they don't move together:

DimensionMeanSDSize Fear59.94%24.05Environmental Anxiety51.73%22.80Object Avoidance49.81%20.46Visual Discomfort44.97%18.28

Size Fear scores highest by a clear margin, while Visual Discomfort scores lowest — a 15-point gap between them. This spread suggests megalophobia, as measured here, isn't a single uniform experience: people report more discomfort around the sheer scale of large objects than around visually processing them, with avoidance behavior and environmental anxiety sitting in between.

Bottom Line

This is a young but actively growing dataset. The sample size and country spread give it real value as a preliminary normative reference, and the overall distribution and reliability numbers are respectable for an early-stage instrument. That said, the geographic concentration (nearly 45% of respondents from just India and the US), the age skew toward younger participants, and the ten flagged items — especially Q20's negative correlation — mean the scale would benefit from revision and a larger, more balanced sample before its norms are treated as definitive.

All data was collected anonymously and voluntarily; no personally identifying information is retained.