
Supporting mathematics instructional change through professional-development message networks
Charles N. Hayward, Sandra L. Laursen
International Journal of STEM Education
Network design
Nodes
the 281 listserv messages exchanged by workshop participants and facilitators during the follow-up year
Ties
directed coded response links from each message to earlier messages it answered, thanked, extended, or revisited
Methods
Messages and response functions were manually coded, arranged as conversation networks, and analyzed for parents, children, ancestors, descendants, reach, timing, and sender role using R network packages and group comparisons.
Source: International Journal of STEM Education
Reviewed summary

A year-long listserv following an inquiry-based mathematics workshop linked qualitative coding with SNA. Thirty-five participants and 11 facilitators exchanged 281 messages; question and follow-up patterns sustained discussion, and 47 percent of messages included a community-building function. Treating messages rather than people as nodes reveals how support unfolds through conversational sequences and shows which facilitation moves help a professional-learning community remain responsive.
Thirty-two of 35 participants posted, 81 percent of responses arrived within two days, follow-up messages revived older threads, and nearly half of all messages contained primary or ancillary community building. The study examined one workshop cohort and one closed listserv, excluded private conversations, relied on interpretive coding of response functions, and did not compare instructional or student outcomes with a control group. The network patterns describe support processes, not causal effects of professional development.
SNA
How SNA was used
The study defined the 281 listserv messages exchanged by workshop participants and facilitators during the follow-up year as nodes and directed coded response links from each message to earlier messages it answered, thanked, extended, or revisited as ties. Messages and response functions were manually coded, arranged as conversation networks, and analyzed for parents, children, ancestors, descendants, reach, timing, and sender role using R network packages and group comparisons.
Source
International Journal of STEM Education, 5, Article 28
DOI: 10.1186/s40594-018-0120-9


