Choosing a Metric

readscore supports nine readability metrics. They are not interchangeable. Each formula was built for a specific context, makes different assumptions, and targets a different kind of reader. Picking the wrong metric gives you a number that means nothing.

This guide helps you choose.

Quick Selector

Find your situation in the left column. The right column tells you which metric to start with.

My text is…	Use this metric
For young readers (grades 1–3)	Spache
For children (grades 4+)	Dale-Chall
For a general adult audience	Flesch Reading Ease
For health communications or patient materials	SMOG
For technical documentation or manuals	ARI
For military or government documents	ARI or Linsear Write
I need a US grade level number directly	Flesch-Kincaid
I need the most widely recognized single metric	Flesch Reading Ease
My text has fewer than 30 sentences	Anything except SMOG
I want to avoid syllable counting	ARI or Coleman-Liau

SMOG and Flesch-Kincaid Are Not Comparable

Do not mix SMOG and Flesch-Kincaid scores in the same analysis

SMOG and Flesch-Kincaid target different comprehension thresholds:

• SMOG was designed for 100% comprehension — the grade level at which virtually all readers can understand the text.
• Flesch-Kincaid was designed for 75% comprehension — the grade level at which most readers can follow it.

On the same text, SMOG consistently scores 2–4 grade levels higher than Flesch-Kincaid. If your text scores grade 8 on FK and grade 11 on SMOG, both scores are correct. They just answer different questions.

If you report one to stakeholders, use only that metric. If you report both, explain what each measures.

What Each Metric Counts

The inputs a metric uses tell you a lot about what it can and cannot detect.

Metric	Sentence length	Syllables per word	Character count	Word list
Flesch Reading Ease	Yes	Average	—	—
Flesch-Kincaid	Yes	Average	—	—
Gunning Fog	Yes	Count (3+)	—	—
SMOG	—	Count (3+)	—	—
ARI	Yes	—	Letters + digits	—
Coleman-Liau	Yes	—	Letters only	—
Dale-Chall	Yes	—	—	3,000 familiar words
Spache	Yes	—	—	Primary-grade word list
Linsear Write	Yes	Weighted	—	—

Character-based metrics (ARI, Coleman-Liau) count letters instead of syllables. This makes them fully deterministic — two systems always agree — and better at handling technical jargon where syllable counting is unreliable.

Word-list metrics (Dale-Chall, Spache) check each word against a list of familiar words. This captures vocabulary difficulty more directly but requires the word list to match your audience and era. Both lists were built in the mid-20th century, which affects modern vocabulary coverage.

Syllable-based metrics (Flesch, Flesch-Kincaid, Gunning Fog, SMOG) treat longer words as harder. This holds up well across general prose but breaks down for text with many short technical terms or long common words.

Score Types

All metrics except Flesch return a US grade level or education level. Flesch returns an ease score that runs from 0 to 100.

Metric	Score type	Extra fields
Flesch Reading Ease	0–100 ease score (higher = easier)	ease (text label)
Flesch-Kincaid	US grade level	—
SMOG	US grade level	—
ARI	US grade level	ages (age range)
Coleman-Liau	US grade level	—
Gunning Fog	Years of formal education needed	—
Dale-Chall	Raw score mapped to grade bands	—
Spache	US grade level	—
Linsear Write	US grade level	—

Grade levels are US-specific

All grade levels in this library are calibrated for the US school system. Grade 8 means the reading level of a US 8th grader (age 13–14). International users can use ARI’s ages field as a more portable alternative — age ranges translate across educational systems better than grade names do.

Approximate mapping: grades 1–5 = primary school, grades 6–8 = middle school, grades 9–12 = secondary/high school, college = undergraduate.

Minimum Text Requirements

Metric	Minimum	What happens if too short
Most metrics	100 words (default)	ValueError on construction
SMOG	30 sentences	ValueError by default; pass ignore_length=True for a warning

You can lower the 100-word default with Readability(text, min_words=N), but scores from short texts are less reliable.

Same Text, Different Results

Running all nine metrics on the same paragraph shows how much the scores can vary. The paragraph below is from this library’s README — a description of the Gunning Fog index, written at roughly a 10th–12th grade level.

from readscore import Readability

text = """
In linguistics, the Gunning fog index is a readability test for English writing.
The index estimates the years of formal education a person needs to understand
the text on the first reading. For instance, a fog index of 12 requires the
reading level of a United States high school senior (around 18 years old).
The test was developed in 1952 by Robert Gunning, an American businessman
who had been involved in newspaper and textbook publishing.
"""

r = Readability(text)

print(r.flesch().score)          # ~52.4  → "fairly_difficult" → grades 10–12
print(r.flesch_kincaid().score)  # ~11.8  → grade 12
print(r.ari().score)             # ~11.9  → grade 12, ages [17, 18]
print(r.coleman_liau().score)    # ~12.1  → grade 12
print(r.gunning_fog().score)     # ~13.2  → college level
print(r.dale_chall().score)      # ~7.4   → grades 9–10
print(r.linsear_write().score)   # ~11.0  → grade 11

# SMOG raises ValueError — this paragraph has fewer than 30 sentences
# r.smog()  →  ValueError: SMOG requires at least 30 sentences. 6 found.

# Spache runs but is outside its valid range — designed for grades 1–3
print(r.spache().score)          # ~4.1   → grade 4 (not meaningful for adult text)

The Spache result illustrates why metric selection matters. Spache was designed for primary school text. Applying it to adult prose produces a score that is technically computed but practically meaningless.

Combining Multiple Metrics

Running several metrics and comparing the results is a valid approach when:

• You want to reduce the effect of any single formula’s quirks on your assessment
• You want to flag text where metrics disagree significantly, which can indicate unusual sentence structure, heavy jargon, or formatting that confuses the parsers

r = Readability(text)

grade_estimates = [
    r.flesch_kincaid().grade_level,
    r.ari().grade_level,
    r.coleman_liau().grade_level,
]
average = sum(grade_estimates) / len(grade_estimates)
spread = max(grade_estimates) - min(grade_estimates)

print(f"Average grade estimate: {average:.1f}")
print(f"Spread across metrics: {spread} grades")

A large spread (4+ grade levels) between metrics on the same text is a signal to look more closely. It often means the text has features that one formula handles differently — heavy technical terms, very short sentences with long words, or non-prose formatting.

Do not mix Flesch ease scores with grade-level metrics

Flesch Reading Ease produces a score from 0 to 100 on an inverted scale (higher = easier). It cannot be averaged with grade-level metrics without conversion. Keep Flesch scores separate or use Flesch-Kincaid for any analysis that combines metrics.

Metric Strengths at a Glance

Metric	Particularly strong when…
Flesch Reading Ease	Communicating ease to non-technical audiences; legal compliance
Flesch-Kincaid	Reporting US grade level; government and education contexts
SMOG	Health communications; when 100% comprehension matters
ARI	Technical manuals; avoiding syllable-counting errors on jargon
Coleman-Liau	Batch processing; deterministic results at scale
Gunning Fog	Business and journalism contexts; identifying “foggy” prose
Dale-Chall	Educational materials; validating vocabulary for grade 4+ audiences
Spache	Primary school materials; grades 1–3 only
Linsear Write	Military and government technical documents