Vocabulary Data

A little experiment I ran. Two things I'm looking for...are word searches worthless? Are word banks helpful when doing vocabulary activities?

Days 1 and 2

Pre-test, Teach words, review

Days 3 & 4

Group 1--Word Search only

Group 2—Exercise (CRWD, Fill in) only

Group 3—Exercise & Word Search

Day 5

Post-test

On day 1, all students were given a vocabulary pre-test consisting of 10 words selected for low likelihood that many students would already know them. The mean for all pre-tests (80) was 1.31. The test was followed by a review (verbally, while words and meanings were projected on the board) of the correct definitions of the words.

Day 2 consisted of a very brief verbal and visual review of the words.

Day 3, Group 1 was given a Word Search with all 10 words; Group 2 a Crossword (without any word bank); Group 3 a Crossword with a Word Search as the word bank. Everyone had 5 minutes to complete as much work as they could.

Day 4, Group 1 was given a Word Search with all 10 words; Group 2 a Sentence completion task using the new words (without any word bank); Group 3 a Sentence completion task using the new words with a Word Search as the word bank. Everyone had 5 minutes to complete as much work as they could.

Day 5, everyone took a post-test of the same design as the pre-test.

I thought that word search only (Group 1) would yield the least improvement, as so little 'work with the word' would be done during the week. The exercise only (Group 2) would, I conjectured, would do better than Group 1, but not as well as exercise plus word search (Group 3) because the latter group would being doing more cognitive work with the definitions, and they would have the visual reminder of the words they were trying to learn.

In short, if Exercise + Word Search (as word bank) is a better learning device, then Group 3 post-test score improvements will be higher (or show more growth) than both Group 2 and Group 1. Also, if Word Searches generate no (or very little learning), then Group 1 post-test scores will be lowest and/or show the least growth. In other words, I expected scores improve with each successive group.

Comparative post-test scores and growth rates that trended in other directions, or no discernible difference among the groups' outcomes would serve as rejection of my hypotheses.

Results—

Descriptive Statistics

Mean of all pre-tests: 1.31

Group 1:

28 sets of tests taken

Pre-test mean-- .96 Post-test mean—3.00

4 test-takers achieved no change from pre- to post-test; 4 declined by 1 point each; 6 increased by only 1 (which I arbitrarily assume to reflect random 'error' as much as learning). 14 test-takers increased by 2 or more.

Group 2:

28 sets of tests taken

Pre-test mean—1.46 Post-test mean—3.8

3 test-takers achieved no change from pre- to post-test; 3 declined by 1 point each; 7 increased by only 1 (which I arbitrarily assume to reflect random 'error' as much as learning). 15 test-takers increased by 2 or more.

Group 3:

24 sets of tests taken

Pre-test mean—1.32 Post-test mean—3.54

3 test-takers achieved no change from pre- to post-test; 2 declined (1 by 1 point, 1 by 2 points); 7 increased by only 1 (which I arbitrarily assume to reflect random 'error' as much as learning). 15 test-takers increased by 2 or more.

Inferential statistics

I hope to run a chi-square test of independence to determine if the score change differentials among the group are statistically significant.

Both the test scores and the growth differentials are so similar across the groups that I suspect we will not find confirmation of my hypotheses.

Several factors may have worked against the processes I expected to find. First, while I wanted Group 3 to do the exercises with a word bank, the additional presence of a word search, appears to have distorted the test I was hoping to implement. Many Group 3 test-takers, indeed, nearly all of them on the fill in exercise, chose to do the word search, as I told them they could do whichever part they wanted for the five minutes. This effectively transformed Group 3 participants into something much more like Group 1 participants.

Second, the five minute work time did not allow enough time for students to 'work with' the words thoroughly enough—in any of the experimental circumstances—to generate the hypothesized differences in results. Five minutes proved too little for many in each of the groups to really accomplish much new work or learning with the words.

I hope, now, to sharpen the experiment by a) dropping the Word Search Group altogether and/or b) changing the Group 3 Word Search to a word bank, and conducting the whole process in the space of one day, so as to eliminate the effect of absenteeism (which required me to discard several test-takers' data) in the course of the experiment.