11/22/03
Third Graders' Writing
(from State Standards)
 Return to 3rd Grade Statistics TOC
Compare to All Studies

A Statistical Perspective

     As of now, this study includes only the five samples from the 2001 Student Guide for Arizona's Instrument to Measure Standards. [See Grade 3, December 2nd in the KISS Grammar Workbooks.] Five is a very limited number of samples. However, even with such a small sample, we can probably recognize some things to think about, especially since we can compare these five samples to the ten samples from the 1986 study of the writing of fourth graders. (See the "Compare to All Studies" link above.) Perhaps the most important aspect of statistical perspectives may be to suggest what we should (and should not) be trying to teach students at different grade levels. This question is explored in the Introduction to December of Third Grade in the KISS Workbooks. The remainder of this document is primarily concerned with how the statistics were calculated and how they compare to those in other studies.
     One of the most interesting figures from this study is the 6.0 average number of words per main clause. This seems very low in comparison to the 7.60 and 7.67 reported in the Loban and O'Donnell studies. The difference may be the result of differences in topic, or in differences in the way in which these researchers counted data. For example, "garbles" such as those found in our sample #5 were, from what I can tell, simply omitted by Loban and O'Donnell. Unfortunately, transcripts of the writing that these researchers analyzed is not available for review. 
     Another major difference is that these researchers "corrected" the writing before analyzing it. Thus, for example, in sample # 1 in this study we find: 

 \F\[LAVFWhen it was my birthday] and [LAVFI was turning eight.] \-\I got a porcelain doll {from by Grandma.}
In the KISS analysis the opening fragment counts as a separate main clause, and thus this "sentence" counts as two main clauses with an average word count of nine. If the sentence is corrected and counted as one main clause, then it has a word count of eighteen. But treating it that way suggests that third graders can mentally process  eighteen-word main clauses. In other words, by first "correcting" the writing, the research obscures students' problems with subordinate clause fragments. One of the things that these researchers did not count is the average of the longest main clause that each student wrote. That average from the five samples is 10.6, and the highest, from sample # 2, is 15.
     The only other statistic for which a comparable study readily exists is the average number of subordinate clauses per main clause. O'Donnell reported that third graders write 18 subordinate clauses for every 100 main clauses. In his summary of the data, O'Hare gives no comparable figure for Loban's study. The five students in the Arizona sample averaged 13, but what is even more interesting is that only three of the five students used subordinate clauses. This type of data is simply not available from any other research study that I am aware of, but it would seem to be highly significant. If this sample is representative, it means that 40% of third graders rarely, if ever, use subordinate clauses in their writing. I might note here that some of my college Freshmen, in analyzing samples of their own writing, have found that they themselves did not use any subordinate clauses. The question is, of course, complex, and more data is obviously needed. My point here is that the individual results of students should be taken into account in any type of statistical study (and the writing samples themselves should be made available on the web).