Grade-8 Analogies: Topic Matching Masquerading as Reasoning
Mr. Alvarez is reviewing a 20-item verbal analogies subtest completed by Priya, a grade-8 student. Each item gives a word pair and asks her to choose the option that completes a second pair with the same relationship. Priya answered 9 of 20 correctly in 14 of the 20 allotted minutes. Sorting her responses by relation type gives a sharp picture: she was correct on 6 of 6 synonym, antonym, and part-whole items; 2 of 7 function and cause-effect items; and 1 of 7 items where the correct option had to preserve the direction of the stem relationship. Of her 11 errors, 8 were options that shared a topic with the stem pair but expressed a different relationship. On finger : hand :: toe : ___, for example, she chose shoe rather than foot. She left nothing blank, and her reading comprehension scores are average for her grade. Mr. Alvarez wants to know whether this is a vocabulary problem, a reasoning problem, or a habit, and what to teach first.
1. Take the item finger : hand :: toe : ___ with options foot, shoe, nail, leg. Explain what each wrong choice would tell you, and why the distractor set matters more than the total score.
Read the worked answer
The stem relation is part-to-whole: a finger is a part of a hand.
- foot - correct. Same relation, same direction, same level of hierarchy.
- shoe - topically contiguous with toe and foot but not a whole that contains a toe. Choosing it means she matched the topic and ignored the relation. Given that 8 of her 11 errors share this profile, this is the dominant error.
- nail - a part of a toe, so the relation is right but the arrow points the wrong way. Choosing it means she knows part-whole and lost direction. Completely different remediation.
- leg - a whole, but not the immediate whole; too large a level. Choosing it means she is reasoning at the wrong grain size.
That is the argument for item-level review. All three wrong answers are plausible-sounding and all three imply different instruction. A total score of 9/20 tells you she is below the midpoint; it tells you nothing about what to teach. The distractor design is doing diagnostic work that the score cannot, which is why you read the chosen options, not just the marked ones.
2. Priya says, 'I just pick the word that sounds like it goes with the first two.' Give the procedure you would teach and explain why it beats telling her to 'think about the relationship.'
Read the worked answer
Bridge-sentence procedure
- Turn the stem into a full sentence with no blank: A finger is part of a hand.
- Say the same sentence with the third term and an empty slot: A toe is part of a ___.
- Generate your own answer before reading the options - here, foot.
- Find your word among the options. If it is there, verify the rest anyway.
- Test the remaining options in the same sentence. A toe is part of a shoe is false. A toe is part of a nail is inverted. Reject both.
- Check direction: if the stem were hand : finger, the answer would have to be a part of a toe.
Why it works. 'Think about the relationship' asks her to introspect on an intuition - the exact process that is already failing. The sentence converts an association task into a verification task with a fixed output format, and it constrains direction and hierarchy automatically, because an ungrammatical or false sentence is immediately audible.
Add a short relation checklist she can search deliberately: part-whole, category-member, function, cause-effect, degree, synonym/antonym, object-user. Naming the relation type out loud before scanning options gives her something to look for instead of something to feel.
3. If the procedure is working, which errors should fall first, and what would you monitor over four weeks?
Read the worked answer
Predicted order of improvement. Direction-reversal errors should drop almost immediately, because the bridge sentence fixes the arrow - 'a toe is part of a ___' cannot produce nail without the sentence going false. Topic-matching errors should drop next but more slowly, because they are a habit, and habits resurface under time pressure, especially on function and cause-effect items where the relation is abstract and the topical lure is strong.
What should not move quickly: items where a stem word is genuinely unknown. No procedure fixes an unknown word, and those items should stay wrong. If they do, that is a signal to separate vocabulary support from strategy instruction.
Monitoring plan. Weekly 10-item probe, scored by error category - topic-match, direction, unknown word, no attempt - rather than by total. Add one transfer check in a different format, such as deciding whether two pairs share the same relation, to see whether the procedure generalizes beyond four-option items. Target by week 4: topic-match errors at 2 or fewer of 10. A rising total score with a flat topic-match count is a warning sign - she may simply be meeting easier items, not a better procedure.