Wednesday, September 7, 2011

Comparing Lexile to Accelerated Reader: A Guide for Parents

Reader Terri and I briefly discussed in the comment section the difference between AR (Accelerated Reader) reading levels and Lexile Framework reading levels. I've delved into this lately and thought many of you could benefit from learning about the two systems. Whether you homeschool or use traditional schools, you'll eventually encounter one or both of these leveling systems.

At one time school districts had little choice in adopting leveling systems designed to match readers to books. Most used Accelerated Reader or nothing at all, until a plethora of choices entered the educational market.

School districts that previously used Accelerated Reader are finding the Lexile Framework attractive because it's free--tight budgets are pushing AR use out. In addition, some standardized testing companies across the country are reporting student test scores using Lexile Framework numbers, making adoption of the Lexile Framework more attractive to school districts.

Another point increasing the attractiveness of the Lexile Framework is their involvement in developing Common Core Standards across the country. Here is an excerpt from the Lexile website explaining how and why they are involved in this endeavor. Full article here.

Any words in red are excerpts...not my words.

MetaMetrics is proud to be an "Endorsing Partner" of the Common Core State Standards Initiative. This historic endeavor was sought to establish a clear set of K-12 standards that would ensure all students graduate from high school "college and career ready." Initially, 48 states, 2 territories and the District of Columbia signed on.
The teams charged with drafting the Common Core Standards asked that we share our collective research on text complexity and the reading demands of college, careers and life in general. This research was conducted over the past 20 years using our widely adopted Lexile® Framework for Reading and is embodied in much of Common Core'sAppendix A. Today, Lexile measures are used at the school level in all 50 states, and 21 states report Lexile measures statewide on their year-end assessments. Each year, more than 30 million Lexile measures are reported from reading assessments and programs, representing over half of U.S. students.

To follow are some key points of our research which are fueling the need for common standards across the states.

  • The text complexity of K-12 textbooks has become increasingly "easier" over the last 50 years.The Common Core Standards quote research showing steep declines in average sentence length and vocabulary level in reading textbooks.
  • The text demands of college and careers have remained consistent or increased over the same time period. College students are expected to read complex text with greater independence than are high school students.
  • As a result, there is a significant gap between students' reading abilities and the text demands of their postsecondary pursuits. Research shows that this gap is equal to a Lexile difference between grade 4 and grade 8 texts on the National Assessment of Educational Progress (NAEP). (Read more orwatch a video about Lexile measures.)
Based on our research, the Common Core Standards provide text complexity grade bands and associated Lexile bands that are intended to put students on a college- and career-ready trajectory. These grade and Lexile bandsare the basis for determining at what text complexity level students should be reading—and at which grades—to make sure they are ultimately prepared for the reading demands of college and careers.

So, what is the difference between Lexile and AR?  And is it negligible or significant? To answer this question we have to understand how each system derives its levels. First we'll read about Lexile's system and then look at the AR system.

How is a text's Lexile measure determined? source here
Within the Lexile Framework, the readability of a text is determined by examining the whole text to measure such characteristics as sentence length and word frequency - characteristics that are highly related to overall reading comprehension. The word-frequency and sentence-length results are then entered into the Lexile equation to compute the Lexile measure of the book.
  • Word frequency is based on the frequency of the word in a corpus of over 300-million words taken from a variety of sources and genres. Knowing the frequency of words as they are used in written and oral communication provides the best means of inferring the likelihood that a reader would encounter a word, and thus become a part of that individual's receptive vocabulary.
  • Sentence length is determined by counting the number of words per sentence. Specific editing rules are used to ensure consistency of editing/analysis from text to text. Research has shown that sentence length is a good proxy for the demand that structural complexity places upon verbal short- term memory.

How is a text's Accelerated Reader Level determined?

The Accelerated Reading levels are based on a formula ( the ATOS) developed by the parent company, Renaissance Learning. The formula uses sentence length along with grade-level vocabulary lists to derive reading levels.

The paragraph below, describing ATOS vocabulary lists, is from a document distributed by Renaissance Learning.

Vocabulary list. (source here)
The ATOS  formulas use a unique measure of word difficulty compared to other readability formulas: grade-level difficulty of the words. This is computed by looking up the difficulty of the words in a book on a special vocabulary list containing more than 23,000 words developed specifically for ATOS. This new, improved graded vocabulary list reflects temporal change in the vernacular lexicon and incorporates the derivatives of words. Derivatives of words have been typically omitted from such lists in the past, or assumed to function at the same grade level as the root word, either of which might have skewed the outcome. The new list is a synthesis of several sources, including the revised Dale familiar word list (Chall & Dale, 1995), the Educator’s Word Frequency Guide (Zeno, Ivens, Millard, & Duvvuri, 1995), and the Renaissance word frequency corpus. Words from these lists and their derivatives were painstakingly reviewed by vocabulary experts to determine correct grade-level placements, which were then validated through comparisons to words used at various grade levels on major standardized tests.

Now that we've read the technical stuff, what can we take away from all this?

First off, the information about texts becoming easier over time concerns me. We do need a universal leveling system as a tool to ensure that students are adequately prepared for college. Lexile isn't perfect, but if it's widely distributed and well understood by educators and parents, it will certainly help prepare students for college.

It's important to note that the Lexile levels use a 75% comprehension rate in their figures (link is to a comprehensive article on Lexiles). This 75% rate assumes a student is reading the text under the direction of a parent or teacher. A 75% comprehension rate means the text is challenging but not frustrating. New vocabulary is being presented which should be explained by a parent or teacher for maximized learning.

A student's independent reading level would be about 90% comprehension, so subtract 250 Lexiles from your child's Lexile reading level to find a book your child can comprehend without help, for independent reading time (subtracting 250 gives you a 90% comprehension rate).  Adding 250 Lexiles gives you a 50% comprehension rate.

School districts give students reading tests to assign Lexile reading levels. If you homeschool, find a book your child can read to you with 90% accuracy (10 errors per 100 words--don't count dialogue, just prose). Next, look up the Lexile level for that book, and then add 250 Lexiles to that number. The resulting number will be your child's individual reading Lexile. As you're recording/counting your child's reading errors, make sure the errors are truly words your child doesn't know the meaning of, and not words she just can't pronounce correctly. Pronunciation of words doesn't correlate with comprehension, most of the time. For example, when you're reading the Bible, does it bother you when you can't pronounce a Hebrew or Greek name?  It doesn't lessen your comprehension of the passage, right?

Once you have your child's reading Lexile, help him pick books within -100 Lexiles and +50 Lexiles of the target number. For example, if your child's reading Lexile is 800L, she should read books between 700L and 850L.

I would expect the AR system to break down across the genres of literature because its 23,000-strong, grade-leveled word bank simply can't encompass enough words to adequately cover all genres and time periods. The Lexile system, with its 300-million-strong word bank, would seem to give a more accurate picture of vocabulary difficulty, however, this system would break down when a text contained infrequent words that were nonetheless easy to decode.

Another crucial point is that leveling systems tell you nothing about the quality of a text. Quality of writing is most important to me; I now use leveling numbers as reference only. To adequately prepare for college a student must learn to construct grammatically correct, smooth-flowing, complex sentences. Poor quality literature does not prepare students for the demands of college-level reading and writing, no matter how high the Lexile or AR level is. Parents and teachers must always open a book and read it to decide if it's worthy. Leveling systems are not shortcuts....just tools.

Lastly, if your child is highly interested in birds, she'll be able to comprehend non-fiction bird text well above her reading level, due to increased motivation. Topics students find uninteresting are harder for them to comprehend, due to lower motivation.

Don't say no very often to material your child is highly interested in, just because the reading level seems too low. Reading should remain a pleasure. Perhaps you could assign something harder for a prescribed amount of time...say twenty or thirty minutes, and then allow your child to read his favorite material?

Here are some comparisons of the two systems using many Newbery and classic titles. At the bottom of this list is another look at how Lexile levels compare to grade levels.

The Sign of the Beaver (an AR level of 4.9 refers to 4th grade, 9th month)
AR = 4.9
Lexile = 770

Caddie Woodlawn
AR = 6.0
Lexile = 890

The Adventures of Huckleberry Finn (unabridged)
AR = 6.6
Lexile = 980

Farmer Boy
AR = 5.2
Lexile = 820

The Wheel On The School
AR = 4.7
Lexile = 710

The Cricket in Times Square
AR = 4.9
Lexile = 780

Old Yeller
AR = 5.0
Lexile = 910

The Courage of Sarah Noble
AR = 3.9
Lexile = 610

Little Women
AR = 7.9
Lexile = 1300

The Hobbit
AR = 6.6
Lexile = 1000

Ginger Pye
AR = 6.0
Lexile = 990

The Moffats
AR = 5.2
Lexile = 800

Thimble Summer
AR = 5.7
Lexile = 810

Pippi Longstocking
AR = 5.2
Lexile = 870

Island Of The Blue Dolphins
AR = 5.4
Lexile = 1000

Turn Homeward, Hannalee
AR = 4.9
Lexile = 830

Sing Down The Moon
AR = 4.9
Lexile = 820

To look up an Accelerated Reader level, click here.
To look up a Lexile level, click here.

Source for the following tables can be found here.


GradeReader Measures, Mid-Year
25th percentile to 75th percentile (IQR)
1Up to 300L
2140L to 500L
3330L to 700L
4445L to 810L
5565L to 910L
6665L to 1000L
7735L to 1065L
8805L to 1100L
9855L to 1165L
10905L to 1195L
11 and 12940L to 1210L

Data for the first column of text measures came from a research study designed to examine collections of textbooks designated for specific grades (MetaMetrics, 2009). The "stretch" text measures (defined in 2010 through studies related to the development of the Common Core State Standards for English Language Arts) in the second column represent the demand of text that students should be reading to be college and career ready by the end of Grade 12.


GradeText Demand Study 2009
25th percentile to 75th percentile (IQR)
"Stretch" Text Measures
25th percentile to 75th percentile (IQR)
1230L to 420L220L to 500L
2450L to 570L450L to 620L
3600L to 730L550L to 790L
4640L to780L770L to 910L
5730L to 850L860L to 980L
6860L to 920L950L to 1040L
7880L to 960L1000L to 1090L
8900L to 1010L1040L to 1160L
9960L to 1110L1080L to 1230L
10920L to 1120L1110L to 1310L
11 and 121070L to 1220L1210L to 1360L
Notice that there is considerable overlap between the grades. This is typical of student reading levels and texts published for each grade. In addition, the level of support provided during reading and reader motivation have an impact on the reading experience. Students who are interested in reading about a specific topic (and are therefore motivated) often are able to read text at a higher level than would be forecasted by the reader's Lexile measure.

1 comment:

Terri said...

Thanks so much for all your research! This post will be a great reference for me in the future.