This system will be undergoing maintenance Monday, January 23 from 8:00 AM to 12:00 PM CST.

## Effects of Technology-Enhanced Language Learning on Second Language Composition of University-Level Intermediate Spanish Students

Cronbach's alpha. Cronbach's alpha (Cronbach, 1951) is a commonly utilized measure
of reliability. More specifically, alpha is a lower bound for the true reliability of a survey.
Mathematically, reliability is defined as the proportion of the variability in the responses to the
survey that is the result of differences in the respondents. That is, answers to a reliable survey
will differ because respondents have different opinions, not because the survey is confusing or
has multiple interpretations. The computation of Cronbach's alpha is based on the number of
items on the survey and the ratio of the average inter-item covariance to the average item
variance. Cronbach's alpha was run on a variety of measures utilized in this study such as the
Core Assessment Questionnaire that was used in part to elicit students' perceptions of the
usefulness of technology for composing in L2. Results are reported in the pertinent sections.
Missing Data. Statisticians address multiple approaches by which to handle missing data,
most commonly used is replacement of the missing data with the mean, median or mode.
Whether or not to replace missing values for analysis depends on the reason why they are
missing. The best reason for replacing data is when data is missing at random. During analysis of
Atajo logs it was found that in some cases students either neglected to turn on the logging feature
or there was a technical malfunction. This resulted in missing data for the time spent composing
and the inquiries made to the databases, two important issues in this study. To ascertain the best
choice for data replacement, an analysis of the distribution and measures of central tendency was
conducted. Commonly the mean is utilized, but the median is less sensitive to extreme scores
(the mean is pulled toward the skewness in distribution). It was determined that the mean (M=
39) was the best replacement for Composition 1 time spent composing for both groups. For
Composition 3 time spent composing, the control data was replaced with the median (Mdn = 39)
and the experimental data was also replaced with the median (Mdn = 37). As for the number of

90

Oxford, Raquel Malia Nitta. Effects of Technology-Enhanced Language Learning on Second Language Composition of University-Level Intermediate Spanish Students. Denton, Texas. UNT Digital Library. http://digital.library.unt.edu/ark:/67531/metadc4688/. Accessed January 19, 2017.