Maze tasks measure sentence-level comprehension, not passage-level comprehension, study says

iStock_000025474831XSmallThe maze task is one of the most popular methods teachers use to check on their students’ reading proficiency and comprehension in elementary school.

But, a new study in School Psychology Review warns that while the maze may be popular, it is not a reliable measure of reading comprehension. Students can correctly complete maze tasks with only sentence-level (not passage-level) comprehension and can potentially give correct answers on 66% of maze items simply by guessing or eliminating the more obvious “distracter” word choices, write researchers from the University of Georgia .

“Despite nearly 30 years of research, teachers and administrators apparently are not convinced that CBM-R is an adequate predictor of students’ comprehension,” the authors write.

There is a clear need, given the results of this study, for another measure of comprehension to be developed to serve as a screener alongside CBM-R (Curriculum Based Measurements-Reading) measures, they write.

Scrambled maze tests vs. intact maze tests

In this study of 225 3rd, 4th and 5th-graders, students performed almost as well on scrambled maze tests (90.01%) as on intact maze tests (92.66%), suggesting that students need no more than sentence-level comprehension to complete the task.

For the study, researchers used 4 maze probes from AIMSweb, which provides schools with 30 maze probes for progress monitoring at each grade level. To develop the scrambled maze tests, researchers randomly drew sentences from 3 of the maze tests to create a scrambled probe. Students took the tests in their homerooms during one 20-40-minute session.

“Although context may benefit students when completing the maze, it seems that passage-level context was not necessary for them to select 90% of the target words accurately,” the researchers write.

Researchers also examined how students performed when the missing word in the maze task was a “function” word such as prepositions and conjunctions as opposed to a “content” word. Previous research indicates students only need phrase-level comprehension to replace a function word in a maze task and often can answer correctly by selecting the function word from among the 3 word choices.

Function words vs. content words

Other studies have found that students are more likely to answer correctly when the deleted word in a text is a function word. In this study, there was little to no difference in students’ performance between function words and content words.

Researchers determined that because there was no control over the words that students had to select from (every 7th word is given a multiple choice option on the maze task), there was no ability to select more function words over context words. Control over the selection of the target words may have resulted in a more accurate reading assessment.

The maze is a timed, silent reading and fill-in-the-blank reading comprehension practice. In a maze, the first sentence of the passage is given and every seventh word after that is replaced with three word choices including:

  • target word: the word that completes the sentence
  • distracter word: the same part of speech as the target word but not meaningful to the sentence and
  • far distracter word: a word that is not meaningful to the sentence and not the same part of speech.

Most of the support for using maze tasks in school today is based on research done on the cloze procedure. The cloze procedure requires students to self-generate target words instead of selecting from three word choices. Since the maze only requires that students select from one of three choices, research suggests that the maze may require less comprehension than the cloze.

Based on this research, students’ maze scores may simply be an indication of how far they read silently in three minutes. The maze did not measure reading comprehension beyond what is measured by oral reading fluency, the researchers write, nor did it measure comprehension beyond the sentence level. Both the cloze and maze could be improved if there were greater control over target words, instead of the current practice of every 7th word being the target word.

Maze tasks can also be improved with better distracters that are both meaningful to the sentence and are the same part of speech as the target word.

“The Impact of Context and Word Type on Student’s Maze Task Accuracy,” by Stacy-Ann A. January and Scott P. Ardoin, University of Georgia, School Psychology Review, 2012, Vol. 41, No. 3, pp 262 – 271.

One Response to “Maze tasks measure sentence-level comprehension, not passage-level comprehension, study says”

  1. Yvonne

    I am so glad I found this article! I would love to have more scholarly data that supports that the Maze is not appropriate for comprehension at the contextual level. I have been discouraged with this as an RtI assessment intervention progress. It is not fittingly designed to support what a child should know at ages 3rd grade and up in order to be successful in comprehension at the learning level in the classroom.


Leave a Reply

  • (will not be published)