Correlation between CLOZE and CARI Tests

Image retrieved from http://www.russell.com/us/glossary/analytics/correlation.htm

Defining what Correlation means

According to the dictionary correlation refers to "the degree to which two or more attributes or measurements on the same group of elements show a tendency to vary together". When a correlation between two variables is established, the researcher can assert that there is a relationship between then: if one variable changes, then the other will also change.

The correlation coefficient corresponds to a statistical value, ranging from negative one (-1) to positive one (1), which describes such relationship. The closer this coefficient gets to 1, the stronger the relationship between the two variables. It is important to warn the reader that correlation does not mean causation. In other words, if the correlation coefficient shows a strong relationship between two variables, the researcher can not argue that one variable causes the other or vice-versa. In addition to find out the correlation coefficient, the researcher needs to verify through a test the significance of that correlation. For the purposes of this study, the program SPSS was used to calculated both the correlation coefficient and its significance.

Testing the Relationship between the Students' scores for CARI Test and the Cloze Test

■ Pre-tests



According to the results of the SPSS program there is a strong significant relationship between these types of tests. Pearson Coefficient equals to 0.733; two-tailed critical value equals to 0 which is less than the alpha significance level (0.05)  

■ Post-tests

According to the results of the SPSS program there is a positive significant relationship between these types of tests. Pearson Coefficient equals to 0.539; two-tailed critical value 0.008 which is less than the alpha significance level (0.05)

Interpreting the Results

• There is a positive relationship between the CLOZE test results and the CARI test results, which means that if a person does well on one test, he/she will in average do well on the other test.
• Learners whose reading level according to the CLOZE test corresponds to frustration (below 43% of correct responses) during the pre-stage also achieve low results in the CARI test.
• The correlation between these two type of tests weakens from the pre-stage to the post-stage; indicating therefore, that there is much more variability among the results at the end of the project. In other words, learners whose reading level according to the CLOZE test corresponds to frustration (below 43% of correct responses) during the post-stage might achieve competent results on the CARI test. Considering that the CLOZE tests determines, in my opinion, a linguistic ability, this result might indicate that under the CLIL framework learners with a low linguistic ability might achieve competent reading comprehension levels, if that comprehension is measured through open questions and the reader has a set of reading strategies.
• If the Pearson coefficient (r) is squared and written as a percentage, pre-stage = 54%, post-stage = 29%, it can be concluded that the 54% of the variability observed in the CARI test can be explained by the results on the CLOZE test during the pre-stage. On the other hand, 29% of the variability observed in the CARI test can be explained by the results on the CLOZE test during the pre-stage. This information might indicate that during the post-stage there were other variables besides linguistic ability contributing to the results of students reading comprehension in a content area such as motivation, awareness of different reading strategies, and learners' training.
  
  
  Useful Links
  ■ http://www.biddle.com/documents/bcg_comp_chapter2.pdf
  ■ http://www.socialresearchmethods.net/kb/statcorr.php