Discussion of Paper 2

Paper 2 found that PD interventions had a medium effect on process quality, a large effect on structural quality, and no effect on teacher knowledge. Furthermore, we found significant effects for phonological awareness and alphabet knowledge, but not for vocabulary. These findings are only partially in line with a previous meta-analysis of the effects of training on child-care providers by Fukkink and Lont (2007). They found that training had positive effects on teacher processes as well as knowledge. Furthermore, they did not find significant effects for child outcomes. However, Paper 2 is limited in its comparability with the meta-analysis by Fukkink and Lont. Our investigation dealt exclusively with PD interventions that focused on language and/or literacy, and our outcome for teacher knowledge was also limited to teachers’ knowledge of language and literacy development. Research has previously found that preschool teachers may have low levels of language and literacy knowledge (Cunningham, Zibulsky, & Callahan, 2009).

69 Although we found that PD had some positive effects on child outcomes, we did not find that process quality effect sizes mediated children’s gains. This finding calls into question the role of process quality in children’s development of language and emergent literacy skills. Previous research has found that process quality is associated with child outcomes (e.g., Mashburn et al., 2008; Pianta et al., 2005), but our meta-analysis indicates that improvements in process quality might not result in equal effects in children’s outcomes.

Burchinal et al. (2010) found that associations between process quality and child outcomes were stronger when quality was higher. It is therefore possible that the process quality in our included studies was still too low (despite statistically significant improvements) to demonstrate strong relations to children’s outcomes. It should be noted, however, that the sample of studies that included child outcomes was small (5-6 studies per outcome). It is entirely possible that a larger sample would reveal significant associations.

Our analysis of PD formats revealed that courses were beneficial when they were combined with at least one other format of PD, but coaching was effective in isolation or combined with more formats. This finding could be due to coaching being a more effective form of PD, but it should be noted that the intensity of coaching interventions was higher than the intensity of courses, which introduces a possible confound. However, Neuman and Wright (2010) compared a course-based PD intervention against a coaching-based intervention with the same intensity. They found no significant effects on process quality, but coaching did benefit structural quality more than courses.

We also found evidence that the number of PD formats was related to quality outcomes. This more-is-more finding suggests that teachers benefit from multiple learning platforms when they receive PD, and we find support for this notion from studies such as Landry et al. (2009), who also found that comprehensive PD models were more effective than models consisting of fewer formats. In addition to the cumulative effect of multiple formats

70 of PD, we also found that PD duration was a significant predictor of quality. Grace et al.

(2008) was the included study with the longest duration (three years), and the effects of their study were some of the highest. This indicates that PD interventions are most effective when they have a long-term framework.

Paper 2 was limited in a number of ways. Several experimental studies that otherwise met inclusion criteria had to be excluded because they failed to present the data needed for estimating effect sizes. The study was also limited by the small subsets of studies that included outcome data for receptive vocabulary, phonological awareness, and alphabet knowledge. Another cautionary point is that the review consisted mostly of studies conducted in North America. This is not to say that we should assume that the learning processes of preschool teachers in other countries are fundamentally different, but the need for experimental research from countries other than Canada and the United States is needed to better understand how PD interventions function in other linguistic and cultural contexts.

Future research should work towards identifying the underlying factors and processes that make PD interventions successful. Sheridan, Edwards, Marvin, and Knoche (2009) recommended a paradigm shift in which researchers moved beyond evaluating the overall effect of PD interventions, and investigated the conditions under which PD works, and for whom. We find support for this recommendation in our own meta-analysis. Although we were able to identify evidence of factors that potentially explain variation in the effects of PD, future research should more systematically investigate the effects of factors such as intervention intensity and duration. Furthermore, researchers might also consider studying more the effects of processes that underlie PD experiences such as quality of PD delivery (i.e.

how well courses are taught), or even take a more theoretical approach, and investigate the psychological processes teachers experience while receiving PD interventions. Social cognitive theory (Bandura, 1977b, 1986) cites a number of factors that are important when

71 learning new skills, such as attention to the learning target, and ability to retain information.

A greater understanding of how or if these factors support or facilitate learning could lead to interventions that are more effective.