Stepwise multiple regression is used to assess the extent to which a dependent variable can be predicted by the combination of variables. It computes and identifies which variables contribute to explaining and predicting the dependent variable. It generates the R, R Square values and so on to indicate how much variation the combination of variables or an outstanding variable account for. Can someone who holds Chaos or complexity theory to view how a foreign language is learnt use stepwise multiple regression? Does that make sense? For example, I am aware that there are too many factors, such as the learners' cultural backgrounds and reasons for learning, which influence second language learners’ attitudes towards their teachers’ instruction. These factors and perhaps with other unknown factors interact with one another in complex systems, and the interactions and their results are unpredictable. The purpose of using statistic methods is to understand part of the story between some factors. Besides, stepwise multiple regression does not intend to measure a factor by isolating it from others. Does this make sense?
Or, at least, he/she can use Chaos or complexity theory to discuss the findings?
I have a journal paper addressing language learning in terns of Complex Dynamic Systems, i.e. Chaos Theory. It does not specifically address multiple regression, but may have some elements of interest to you.
Article Environmental factors affecting Computer Assisted Language L...
I would be very sceptical about this, because complexity perspectives and multiple regression seem to be underpinned by very different epistemological beliefs. Multiple regression is underpinned by a linear understanding of causality. Complexity perspectives challenge this view (in fact, some people use Complexity theory interchangeably with 'non-linear science'. Multiple regression attempts to predict, but complexity holds that prediction is impossible. Stepwise multiple regression is all about removing variables, one at a time, in order to find the ones that are most likely to predict outcomes. Complexity theory argues that every variable, no matter how small, can have a disproportionate and unpredictable, and that it is therefore more productive to view the system as a whole.
Thank you very much, Achilleas Ioannis Kostoulas !
I understand the epistemology of two is different. However, I hope any explanation or bridge can be found because when you used questionnaire and collected attitude data, you still need to analyze them, right?