ABSTRACT
The last few chapters have presented inferential statistics, such as t-tests and the ANOVA, that are used to compare group means. However, if researchers aim to understand how different variables might affect language learning or test performance, then another kind of inferential statistic is required. For example, Jia, Gottardo, Koh, Chen, and Pasquarella (2014) investigated the relative effect of several reading and personal background variables on reading comprehension of ESL learners in Canada, including word-level reading ability, vocabulary knowledge, length of residence in Canada, enculturation in the mainstream culture, and enculturation in the heritage culture. Not only were the researchers interested in the effect of each variable, but they were also interested in the relative importance of those variables. For example, they were interested in finding out which variable had the strongest impact on reading comprehension, and which combination of variables best explained reading comprehension. The statistical procedure used to answer these questions is known as multiple regression. In order to illustrate how multiple regression is performed, this chapter begins with a description of simple regression.
