Comparing students' performance on sentence completion and passage comprehension

Introduction

NGRT provides a standard age score derived from the test as a whole which gives the most accurate and reliable indication of a student’s reading ability. However, it is possible to look at sub-scale scores for each section of the test in order to determine if students have particular areas of strength or areas for further development in their reading skills.

The scores for the sub-scales are presented using stanines rather than standard age scores. Stanines minimise the over interpretation of small, insignificant differences among test scores. As a rough guide, if the sub-scale stanines are the same or differ by one, this indicates similar performance on the two sub-scales. If the stanine score differs by two or more, it may be an indication of significantly different performance on the two sections of the test.

Therefore it is important to recognise that any significant differences in the two subscale scores may only be indicative of a difference in performance and would benefit from further investigation. For example the Suffolk Reading Scale (SRS) could be used to confirm students reading ability at sentence level and the York Assessment of Reading Comprehension (YARC) could be used to confirm their ability in reading comprehension.

Sentence completion

The sentence completion section of the NGRT assesses students’ reading skills at sentence level. These skills are slightly different from those used when reading a whole text and answering retrieval and inferential questions about that text. In the sentence completion questions students have much less contextual information to use in order to determine the missing word. Therefore students who rely heavily on contextual cues to draw even simple inferences may score less well on this section of the NGRT. The lack of contextual cues will mean that students have to rely more heavily on their word knowledge (vocabulary) in order to successfully answer the more difficult sentence completion questions where the vocabulary may be considered challenging for students in that age range. Therefore students with a more extensive vocabulary may perform better on the sentence completion questions.

In sentence completion questions the students also need to rely more heavily on syntactic cues (the structure of a sentence) and their own knowledge of grammatical features when deciding which of the five possible options best completes the sentence. Some of the distracters that are used in the sentence completion questions have been chosen as they are semantically (in meaning) or graphically (in written form) similar to the target word; however using any of these distracters would make the sentence grammatically incorrect. Therefore the student is required to read both the sentence and all five of the answer options very closely in order to determine the correct answer.

Although students need to be able to make single local inferences in order to answer the sentence completion questions they do not need to employ any of the more complex global inferencing skills required in the passage completion section of the test. Therefore a student who has difficulty understanding longer texts or making complex inferences may perform significantly better on the sentence completion section of the NGRT.

Passage comprehension

In the passage comprehension section students need to use a variety of reading skills to read up to three passages and answer a number of questions on each passage. In reading and answering the texts students are expected to make global as well as local inferences. Each text is accompanied by a range of question types including:

  • Context comprehension
  • Retrieval
  • Inference and deduction
  • Organisation of texts
  • Writer’s use of language
  • Writer’s purpose and viewpoints
  • Social, cultural and historic tradition.

The scores students obtain on this section of the test are influenced by their ability across a wide range of reading skills. Students who are very good at decoding but poor at inference and deduction may obtain a significantly lower stanine score on this section of the test.

The majority of students will achieve scores that are within one or two stanines: for example 2:3 or 7:6 and for exceptionally good readers 9:8 or 7:9. It is likely that just a small number of students will need additional investigation as indicated above but this aspect of NGRT now enables screening and progress tracking for large groups of students and more meaningful individual profiling from a single test.