Estimating the effects of writing beliefs, writing processes and drafting strategies on the development of subjective understanding

The problem-solving accounts of writing (Flower & Hayes, 1981; Bereiter & Scardamalia, 1987) are the prominent underpinning theories within writing research. These models assume that writing is a problem-solving activity - the writer combines planning, translation and revision processes to achieve their rhetorical goals, and expert writing is assumed to be a knowledge transforming process in which rhetorical problem-solving leads to both higher quality text and the development of understanding. The dual-process model of writing (Galbraith, 2009; Galbraith & Baaijen, 2018), accepts the idea that writing is a problem-solving process, but claims in addition that spontaneous text production also leads to the development of understanding. These two different processes have conflicting effects on text quality, which the dual process model claims can be resolved by using a revision drafting strategy. The essential contrast, therefore, between the problem-solving models and the dual-process model is over whether spontaneous text production is simply a knowledge telling process (as in the problem-solving models) or is an active knowledge constituting process (as in the dual-process model). Only two published studies thus far have examined whether the dual-process view is an appropriate depiction of the components involved in writing in relation to the development of subjective understanding. Therefore, this project attempted to further investigate the model’s role in writing to learn by carrying out three studies: Study 1 examined individual differences in writing beliefs; study 2 described the development of reproducible measures of writing processes identified from keystrokes; study 3 used the methods developed in these two studies to carry out an experimental investigation of the effect of drafting strategies, writing beliefs and writing processes on the development of subjective understanding. Study 1 used exploratory structural equation modelling