Home Articles Reader Opinion Editorial Book Reviews Discussion Writers Guide About TCRecord
transparent 13
Topics
Discussion
Announcements
 

Examining the Virtual Diffusion of Educational Resources Across Teachers’ Social Networks Over Time


by Yuqing Liu, Kaitlin T. Torphy, Sihua Hu, Jiliang Tang & Zixi Chen - 2020

Context: Individuals’ curation within social media provides a window into their sensemaking and conceptions of what is worth knowing. Within education, a majority of teachers use social media for professional purposes to access and share instructional resources.

Purpose: This work examines Pinterest.com and the intersection of influence across virtual and physical spheres as teachers choose and curate instructional resources. Setting: The study is conducted on 19 schools over five districts in three Midwestern states.

Participants: The sample consists of 108 elementary teachers in total: 34 early career teachers and 74 colleagues.

Research Design: This is a longitudinal observational study designed to repeatedly measure and track teachers’ online resource-seeking behavior over 52 weeks in the 2015–2016 school year.

Data Collection and Analysis: Resource curation data were collected for each teacher, as well as early career teachers’ egocentric school network and online network data. Using generalized linear growth modeling approach to examine relationships between teachers’ curation of resources, we identify differences in the impacts of teachers’ social networks across physical and virtual space.

Findings: Results indicate that teachers following one another within Pinterest have a higher rate of curating a resource, but Pinterest seems to act as a bridge between those less connected teachers within a school, with an even greater rate of curation for those teachers who do not closely work together. This seems to indicate that within the cloud of social media, Pinterest may be a conduit for information and resource distribution across schools.

Conclusion: As schools continue to seek improvement potential, leveraging social media connections and social capital within and outside the local context may prove useful for the flow of expertise and resources.  

Today’s connected world finds individuals interacting in person, through text, and in the online space. In social media, teachers may exchange ideas and advice with one another and connect to other relevant informational sources online. Through these actions, social media provides a living record and artifact of teachers’ decision making, sentiments regarding a particular resource or reform, and their access to and review of an instructional task.


In a social continuum of fluid virtual and physical interactions (Supovitz et al., 2015), notions of teacher social networks are no longer bounded by school building walls. Educational stakeholders have examined how teacher community and collaboration happen successfully within schools, districts, and communities (Jones et al., 2013). Since the report A Nation at Risk (Gardner, 1983), which revealed concerns about lagging school, teacher, and instructional quality, educational policy makers have attempted to incentivize teachers’ collaboration with one another. Yet, in virtual space, a majority of teachers are engaging with one another around resources created by teachers, for teachers (Torphy et al., 2017).


This study examines how the social networks and social capital that teachers develop and access relate to their curation of instructional resources within a prominent social media space, Pinterest.com. Though some research examines how teachers’ social networks shape their interactions and sensemaking of instructional practices, little work has explicitly parsed the impacts of social networks as they extend from the classroom to the cloud. This study will examine teachers’ curation of instructional resources as it relates to their collegial networks. Using social networks across physical and virtual space, this work examines the propensity to curate a particular resource after teachers’ networked peers have done so. Through this approach, we may better understand how network influence extends across physical and virtual space and may relate to teachers’ professional curation of instruction resources.


LITERATURE REVIEW


Teachers’ interactions with one another and within professional communities of scholars matter (Lave & Wenger, 1991). These interactions provide a basis for teachers’ practices and development over time (Coburn, 2001; Spillane, 1999). Some educational stakeholders have attempted to formalize teachers’ learning through enrollment in professional learning communities (PLCs; Achinstein, 2002). Within these communities, teachers regularly meet and discuss their instructional practices and orientations toward particular topics of professional interest (Mitchell & Sackney, 2000; Toole & Louis, 2002).


Research shows that these communities of scholars are an important factor for schools’ organizational health and quality (Desimone, 2002; Youngs & King, 2002). Furthermore, recent research indicates that teachers prefer to seek out advice and instructional resources from their local and global peers (Torphy et al., 2017). Advice seeking and exchanges provide opportunities to develop professionally and improve one’s practice (Baker-Doyle & Yoon, 2010; Frank et al., 2004, 2011; Lave & Wenger, 1991). In sum, teachers’ exchanges across local and global spaces may increase their social capital and impact their growth within their field (Spillane et al., 2012).  


TEACHERS IN SOCIAL MEDIA


Teachers learn through doing within their social context and through their professional preparation. They may form social network ties through formal professional communities and informal interactions, constituting the basis of their contextual experiences. These ties may extend across physical and virtual spaces and interactions (Baker-Doyle & Yoon, 2010; Frank et al., 2004). Virtually, teachers may connect with other like-minded individuals within social media to share ideas and resources (Wellman, 2001). These connections, whether across physical or virtual spaces, promulgate knowledge diffusion across social ties and facilitate teachers’ access to a large array of instructional resources. Knowledge diffusion across social networks, across virtual resource pools (VRPs) containing a great deal of instructional resources, or within the school in which teachers work blurs the boundaries between the physical and virtual worlds (Jimerson, 2014; Wellman, 2004).


Virtual engagement across online platforms, beyond the schoolhouse, allows teachers to connect in a way not previously possible. These interactions expand boundaries of connectivity and professional collaboration. For teachers, social media may relate to one’s personal and professional continuum. These platforms allow teachers to iteratively connect with one another and maintain a presence as a trusted source of professional knowledge (Hampton, 2016). Over time, teachers’ physical and virtual social ties may influence how they consider their profession and instructional orientation.


Seeking out resources online may allow teachers to avoid transaction costs within their physical space. For example, teachers who ask for help or advice within their school risk their colleagues’ awareness that they are not confident in a particular topic or field of instruction (Will, 2016). By accessing social media or VRPs, teachers are able to acquire help and instructional support while maintaining their reputation within school. Curation of resources—that is, resource seeking and acquisition—may be characterized as teacherpreneurial behaviors within VRPs, as teachers reach out to one another rather than larger organizational or state-sponsored sources of information (Torphy et al., 2017).


Resource curation and acquisition may be incidental or self-directed as teachers seek out particular instructional materials to meet a given need (Will, 2016). Research has found that social media is a frequent resource for teachers as they attempt to design their instruction and materials to standards-aligned resources (National Council of Supervisors of Mathematics, 2015). Across social media platforms, Pinterest and Teachers Pay Teachers are commonly consulted for mathematics and English language arts instructional materials (Opfer et al., 2016). Teachers spend a nonnegligible amount of time seeking out supplemental materials, with many reporting up to 12 hours per week acquiring online resources (Brown, 2019).


THE CASE OF PINTEREST


Pinterest, an image based social media platform, has become a major VRP in which teachers interact with instructional content and other educators (Opfer et al., 2016). Pinterest serves as a broker of information and facilitator of social networks between teachers and various VRPs, bridging teachers to broader virtual spaces and global networks (Burt, 2005). Teachers come together in socialized knowledge communities (SKCs), emergent communities of teachers who engage in professional problems of practice outside organizational purview (Torphy & Hu, in press). Within SKCs, teachers access and share classroom ideas and instructional tasks from other online sources, organizing them in their own Pinterest “scrapbook.” In addition, teachers may follow one another to get updates on pins and boards that they are interested in, customizing their newsfeed so they may be informed of others’ saved educational content. Situated within SKCs, teachers connect with their local and global networks to engage in different forms of informal learning (Schugurensky, 2000).


THE INTERSECTION OF PHYSICAL AND VIRTUAL SOCIAL NETWORKS WITHIN AND OUTSIDE THE SCHOOLHOUSE


Pinterest, a large VRP, is an unbounded social space where teachers may follow one another without physical barriers. In contrast, connections in teachers’ physical space are highly constrained by geographical distance and venues of common participation. In this article, we focus on the virtual connections that are also present in teachers’ physical network and consider an emergent phenomenon: bounded social networks in unbounded virtual spaces.


Next, we illustrate the intersection of teachers’ physical and virtual spaces. This image displays teachers’ interaction across physical and virtual social networks. Generally, research treats these spaces as mutually exclusive; however, research shows that teachers experience a social continuum in their professional intercommunication (Wellman, 2004).


In Figure 1, we see a teacher both influenced by and influencing the school in which she works. She is connected to her school-based colleagues and may influence their teaching beliefs and practices (as is evidenced by an arrow directed outward), or, vice versa, she may be influenced by her peers. In cases of network reciprocity, teachers may influence and be influenced by one another. Outside the school, knowledge acquired through social media and virtual connections may influence this teacher both directly and indirectly. Directly, she may virtually follow her close colleagues; indirectly, one of her colleagues may find resources on social media and share them with her at school. Alternatively, this teacher may connect virtually with some of her school-based colleagues, regardless of whether they interact within their school. Though teachers may not connect informally, membership within a district seems to impact online resources accessed (Torphy et al., 2018).  These connections may facilitate social network diffusion and changes in beliefs and behaviors over time.


Figure 1. Unbounded network influences in physical networks


[39_23306.htm_g/00002.jpg]



Because physical proximity within a school affects the formation of connection in virtual space, virtual tie strength can be defined in relation to the strength of teachers’ physical interactions, characterizing teachers’ online social capital and ability to access information and resources (Granovetter, 1973; Rost, 2011; Ruef, 2002). Strong ties, rich in support and trust, improve the efficiency of information diffusion between teachers and their close colleagues but reduce the possibilities of encountering new resources that have not been shared at school (Granovetter, 1983; Krackhardt et al., 2003). Weak ties, in contrast, facilitate information diffusion between teachers and less connected others, consequently increasing the chances of exchanging different types of information and perspectives (Granovetter, 1973).


Studies have found the critical role that weak ties play, relative to strong ties, in dissemination of information (Bakshy et al., 2012; Ruef, 2002). Pool (1980) commented that the development of a communication platform increases the number of weak ties one has in their network. He further stated that the utility of ties of differing strength depends on an individual’s intention in the resource acquisition process. Individuals in an insecure position in the work environment rely more on strong ties for information seeking, which are embedded in a larger cohesive subgroup (as cited in Granovetter, 1983). In contrast, weak ties have strength in promoting “boundary-spanning information flows” (Friedkin, 1982, p. 273). Optimal information diffusion and resource acquisition occur within a social network of strong ties and weak ties (Friedkin, 1982).  


Though a large proportion of teachers connect online to support professional learning and further develop their classroom—from instructional resources to organizational management—little research has been done on how their relationships influence the curation of online educational resources. In addition, empirical evidence suggests that early-career teachers’ (ECTs’) access to online resources leveraged by Pinterest is influenced by their informal physical and virtual networks (Chen et al., 2017). The emergent phenomenon of teachers’ social media engagement (N. Singer, 2017) indicates the importance of examining knowledge diffusion across teachers’ networks and how that shapes teachers’ professional mindsets. We conceptualize resource-pinning behavior as teachers’ curation of resources, archival, and, therefore, resource possession and tacit knowledge, which may be carried throughout a school year. Repeated pinning contributes to teachers’ potential to leverage their tacit knowledge to improve instruction and professional practices. Using a sample of ECTs and their colleagues across three Midwestern states, we examine collegial interactions of teachers both within and outside the school across VRPs to identify how teachers’ exposure to colleagues within social media impacts their own curated archive of identical educational resources.


Additionally, we explicitly consider the strength of ties as it relates to resource diffusion, and we define those relationships that intersect strong and weak ties as “streak” ties. Comparing weak, strong, and streak ties, we examine the influence of ECTs’ social networks on their own online educational resource archives. With time-ordered data on teachers’ resource curation and archival (possessing the resource), we employ negative binomial regressions to estimate the network influence on teachers’ archival of identical educational resources. Overall, we found that teachers’ access to instructional tasks and educational content is influenced by their colleagues’ pinning behavior within Pinterest. In addition, streak ties—those virtual connections between unassociated school colleagues—are more beneficial for online information diffusion, relative to close colleagues within one’s school.


RESEARCH QUESTIONS


In this article, we seek to answer two questions on how ECTs’ curated archives of educational resources are influenced by their networks in virtual spaces. Specifically, we ask:


1.

To what extent is colleagues’ pinning behavior influencing ECTs’ resource possession in social media?


2.

How does the virtual network influence vary by colleague type, defined by the tie strengths of ECTs’ in-school network?


SAMPLE AND DATA


SAMPLE


Our sample consisted of 34 ECTs and 74 colleagues from 19 schools in five districts across three Midwestern states (108 teachers in total). We used a convenience sampling method to identify school-based teacher colleagues with whom ECTs interacted on a regular basis around instructional discussion. ECTs were defined as teachers who were in their first four years of the teaching profession. All 108 teachers actively engaged in finding online educational resources on Pinterest in 2015–16 school year.


DATA COLLECTION


We used a combination of Pinterest longitudinal data and school survey data to assess the diffusion of teaching pins from colleagues to ECTs on one VRP, Pinterest.


Pinterest Data


Pinterest data were collected from September 8, 2015, to September 6, 2016. We chose this time interval to examine ECTs’ curation and archived resources across one academic school year. This coincided with teachers’ resource seeking within VRPs in preparation for the academic year. In addition, we constrained the instructional resources to the top 12 resources pinned by sampled ECTs, representing the diffusion pattern of the most prevalent teaching pins (see Appendix A for all 12 analyzed resources). These resources were pinned by at least six ECTs in our sample. Overall, these resources were pinned by 12 teachers on average, including ECTs and their colleagues (M = 11.83, SD = 2.89), with eight teachers pinning the least prevalent resource, and 17 teachers pinning the most prevalent. By restricting sampled pins with which we examined diffusion, our approach prevents creating an overdispersion of nonpins (or zeroes). This allows for a more precise estimate of resource diffusion across ECTs’ social network.


Physical and Virtual Network Data


We administered surveys to collect ECTs’ close colleague nominations in fall 2015 as representative of 2015–2016 egocentric physical social network data. This required reasonable assumptions, including that teachers’ close colleagues were relatively stable over a one-year period. We asked teachers to list up to 10 close colleagues with whom they had discussed mathematics instruction since the beginning of the school year. In a spring 2016 follow-up survey, we found that ECTs’ physical social networks were 59% similar. Through egocentric network data, we derived nodes of ego ECTs and nominated alters from the same school.


To characterize the intersection of ECTs’ physical and virtual networks, we collected data on ECTs’ egocentric virtual networks, identifying whom they were following on Pinterest in mid-August 2016 to be the 2015–2016 virtual network data. For those people whom ECTs were following on Pinterest, we constrained alters to be teachers who were nodes in our physical network sample, either nominated by them directly as close colleagues, or indirectly nominated by other ECTs who were also in the same school or the same district. This represented the bounded network, defined by connections across physical and virtual space, in an unbounded social media space.


An ECT, on average, followed 150 people on Pinterest. Fewer than four of 150 were teachers with whom an ECT connected across physical and virtual spaces. We further decomposed these cross-space ties based on the relationship strength reported in physical teacher surveys. We found that 30.84% of ECTs’ virtual ties were to those whom they nominated as close colleagues, 50.14% of virtual ties were to those with whom they worked in school but did not nominate as a close colleague (i.e., unassociated school-based colleagues), and 19.02% of virtual ties were to those with whom ECTs worked outside school within the same district (see Table 1).


Table 1. An Average Early-Career Teacher’s Virtual Tie Decomposition Based on Relationship Strength in Physical Connection

 

Tie to close colleagues

Tie to unassociated school-based colleagues

Tie to teachers outside school in the same district

Total

Frequency

1.07

1.74

0.66

3.47

%

30.84

50.14

19.02

100


Figure 2 represents the triadic relationship between teacher network connections across space (network tie between an ECT and a colleague) and the teacher–resource interaction in VRPs (teachers pinned word-building resource).


Figure 2. Representation of virtual tie influence of close colleagues on an ECT’s resource possession over time



[39_23306.htm_g/00004.jpg]


In Figure 2, we depict an ego ECT that nominates an alter colleague as a close colleague with whom he frequently interacts regarding instruction at his school. The ego ECT also follows his alter colleague on Pinterest, establishing a virtual connection. The arrow of the tie, between alter and ego, represents the online resource flow direction. Above, his colleague pins a word-building resource at Week 15. Subsequently, we observe the ECT pinning the same resource at Week 23. In our following analysis, we estimated the impact of one’s exposure to physical colleagues within social media and the resource diffusion that occurred across ties.


METHODS


Using generalized linear growth models, we examined teachers’ curated resource archives, with the network exposure term as a time-varying predictor. We translated a one-time resource-pinning behavior to an increase in teachers’ instructional resources archived within social media and hence an increase in their tacit knowledge, which can be carried throughout a school year. We observed that 26.47% of ECTs repinned at least one top resource within a school year. For teachers, online resource acquisition may not be considered purposeful learning. Teachers may not intentionally review what is in their curated resource portfolio each time before the next search of resources for lesson planning. Thus, teachers may pin more than once the same resource until they either (a) mastered the instructional task, or (b) were certain that the resource in need was stored in their curated archive. Because teachers perceive their career and professional development as an ongoing process, instances of repeated pinning may contribute to teachers’ potential to leverage their tacit knowledge to improve instruction and professional practices. Through repeated, cumulative, and iterative resource-seeking processes, teachers may be able to enhance their familiarity with the materials, internalize the instructional resources, and enact them in the classroom with more flexibility and across different contexts.


OUTCOME


We chose to model teachers’ curated resource archive as the outcome variable to study ECTs’ resource-seeking behavior. Though teachers may not pin the same resource in a weekly manner, their curated archive (i.e., the resources teachers pinned since the beginning of the school year) was a constant source for them to draw supplemental teaching materials from and can be used to reflect the current status of teachers’ resource possession for a given week. This measure was either monotonically increasing or constant from the beginning to the end of the school year. In addition, we assumed that teachers had different thresholds of maximum pinning of a particular resource, which was contingent on both an individual teacher’s familiarity with the content and resource, and the cognitive demand of the resource. Once the threshold was reached, teachers stopped pinning the same resource. Thus, we censored a given teacher’s curated resource archives variable during the week in which a teacher last pinned a particular resource, and viewed it as the moment teachers reached their individual threshold. Data after the threshold were set to missing (see Figure 4 and Table 2). We considered several alternative conceptualizations of teachers’ resource pinning behavior and its subsequent model specifications, including a conventional hazard model, a multi-spell hazard model, and a negative binomial regression model. See Appendix B for rationale of alternative outcomes and the corresponding models.


PREDICTORS


Time


We chose to examine the linear growth of resource possession across weeks. We used the discrete time framework to record teachers’ resource possession over 52 calendar weeks. Week zero was set to be the first week of the 2015–2016 school calendar year.


Network Exposure


ECTs were exposed to the resources their connected-colleagues pinned through at least two mechanisms. First, teachers browsing pins within their newsfeed personalized by Pinterest— in which Pinterest automatically displays recommended resources—may encounter pins that they seek, pins that Pinterest recommends, and those pins that colleagues have pinned and archived in the past. Alternatively, ECTs can directly access colleagues’ educational boards to search their pinned resources.


We defined teachers' network exposure term as the sum of alter colleagues’ pinning behavior of a particular resource at week t-1 or earlier, [39_23306.htm_g/00006.jpg], where [39_23306.htm_g/00008.jpg]    was the

network tie from ECT i to colleague i’ across physical and virtual space;  [39_23306.htm_g/00010.jpg] was

colleague i’s pinning behavior of resource j at week t-1 or earlier. We counted all previous pinning behavior of colleague i’ and summed it across all colleagues whom an ECT was connected with. This term comprised his or her network exposure. For example, if an ECT followed Colleague B and Colleague C on Pinterest, and Colleague B had pinned resource R once, while Colleague C had pinned the same resource three times up to Week 12, then the potential network exposure from Colleague B and C to the ECT on pinning resource R at Week 13 was 1 + 3 = 4 (see Figure 3). Network exposure varied over time, as an ECT dynamically experienced different levels of exposure to colleagues’ pinning behaviors.


Figure 3. Network exposure to colleagues’ pinning behavior

[39_23306.htm_g/00012.jpg]


To better understand the online diffusion mechanism in conjunction with teachers’ varying degrees of connection within school, we further divided virtual network ties and calculated the corresponding exposure term into three mutually exclusive groups: ECTs’ close colleagues, unassociated school-based colleagues, and teachers outside one’s school, in the same district.


Diffusion Dynamics


Table 2 and Figure 4 displayed the diffusion dynamics of alter colleagues’ previous resource pinning to an average ego ECT’s curated resource archive through network exposure. To show how we coded ECTs’ resource curation and archive as they relate to the resource pinning of their colleagues, we present the following example.


Consider an ECT who started with no exposure to an instructional resource at the beginning of her school year. If the ECT had colleague(s) who pinned resource j twice in the following week(s), she was exposed twice to the same resource j and hence was coded a 2 for her network exposure. Afterward, if she pinned resource j once, she was given a 1 in the outcome for her resource possession—that is, curating and archiving the resource. Subsequently, if her colleagues continued to pin the same resource twice, she received two more opportunities for exposure, increasing her total exposure level to 4; if she pinned it again thereafter, her resource possession would be 2 at week t. We illustrate the dynamics between virtual network exposure and ECTs’ response to the exposure as reflected in the resource possession in Table 2.


Table 2. Data Structure on the Censored Teacher’s Resource Possession and the Network Exposure

Week

Outcome: ECT’s censored resource possession of resource j

Network exposure to colleague’s pinning behavior of resource j

0

0

0

15

0

1

23

1

1

24

1

1

38

1

2

39

1

2

40

2

2

 

46

 

3

 

51

 

3


We may visualize the resource diffusion dynamic as represented by a leading increase on colleagues’ pinning behavior and a lagged increase on an ECT’s curated resource archives over time (see Figure 4).  


Figure 4. The dynamic of resource diffusion from colleagues to an ECT

[39_23306.htm_g/00014.jpg]


In this example, an average ECT has been exposed to a resource that colleagues pinned at Week 15, which led to the ECT’s curated resource possession and archive, totaling 1, at Week 23. From Week 24 to Week 37, no action was detected on either the ECT’s or colleagues’ resource pinning activity. The diffusion dynamics resumed at Week 38 as colleagues continued to pin the identical resource a second time, followed by an increase in the ECT’s possession of the same resource at Week 40.


MODELS


We addressed Question 1, fitting a negative binomial regression model in both a fixed effect model with teacher and resource dummy variables (see Model 1-1), and cross-classified random effect model with teacher and resource random components (see Model 1-2) to accommodate the nesting structure on the observation of ECT i’s possession of resource j at week t. Thus, we estimated the time and network effect accounting for the variations on ECTs’ tendency to possess a resource and the variations on common resources’ tendency to be possessed. The number of observations before censoring the outcome by a teacher’s threshold on a resource was 21,216 (t = 52, I = 34, j = 12, N = 52 X 34 X 12 = 21,216) and 14,746 after the censoring.


Model 1-1. Fixed-effect model for overall network exposure


[39_23306.htm_g/00016.jpg]


If it is Week 8, and the last time the teacher pinned the resource was Week 7, the value of resource possession is set to missing. If it is Week 8 and the last time the teacher pinned or repinned the resource was Week 10, the value is set to teachers’ resource possession at Week 8. We used the same censoring technique for all subsequent models.


Model 1-2. Cross-classified random effect model for overall network exposure




[39_23306.htm_g/00018.jpg]



In Model 2, we investigated Question 2 in an alternative model and only fitted it using the cross-classified random effect model.


Model 2. Cross-classified random effect model for virtual network exposures to physical colleagues of strong ties, streak ties, and weak ties


[39_23306.htm_g/00020.jpg]


RESULTS


We found a significant network effect on ECTs’ online resource curation and archive. Using a fixed effect model, we estimated an increase in the log of the mean count of ECTs’ resource possession by 0.51 for each unit increase in ECTs’ network exposure. Therefore, for a 1-unit increase in the virtual exposure to colleagues’ pinning behaviors on an identical resource within Pinterest, the mean count of ECTs’ resource possession was multiplied by the value of 1.67 (see Table 3 and Figure 5). The estimated network effect remained significant and stable in the random effect model, with an increase of 0.49 in the log of the mean count of ECTs’ resource possession given a 1-additional-unit increase in network exposure (see Table 3). To fully partial out the time effect from both ECTs’ resource possession and their exposure to the same resource in their network, we fitted a model, using a set of week dummies, with no assumption on the relationship between time and ECTs’ pinning behavior. Results were consistent across models


  ( [39_23306.htm_g/00022.jpg]=0.55, SE = 0.14). We quantified the robustness of the network effect, using the Frank et al.


(2013) approach to sensitivity check. Robustness indices indicated that 50% of the data must be due to bias to invalidate the inference (Rosenberg et al., 2018). This is more robust than 75% of studies reported in the Frank et al. (2013) article. See Appendix C for alternative model estimate results.


Incidentally, there was an overall linear growth effect of time on ECTs’ curated resource archives over 52 weeks. The mean count of ECTs’ resource possession within these archives was multiplied by 1.03 each week. For the model using a set of week dummies for nonlinear time effect, we found March, April and June, July to be significantly positive, compared with the first week of September. This result is consistent with our observations that teachers find more online resources in March and April for test preparation, and more resources in June and July for lesson planning for the next school year.


Table 3. Summary of the Negative Binomial Regression Analysis of ECTs’ Resource Possession as a Function of Overall Network Exposure


Independent variable

Model 1-1

Model 1-2

 

B

SE B

[39_23306.htm_g/00024.jpg]

B

SE B

[39_23306.htm_g/00026.jpg]

Week

0.03***

0.004

1.03

0.03***

0.004

1.03

Network exposure

0.51***

0.13

1.67

0.49***

0.13

1.63

Note. B represents a regression coefficient in log scale. Model 1-1 is the fixed-effect model, including teacher and resource dummies for estimating correct standard error on the network effect. Model 1-2 is the cross-classified random-effect model with teachers and resources as the random components.

***p < .001.


Figure 5. The network effect on ECTs’ online resource possession



[39_23306.htm_g/00027.jpg]



Note. This was a concrete illustration of the effect of the ECT’s network exposure to this colleague’s pinning behavior of the phonics resource on her resource possession [39_23306.htm_g/00029.jpg] The ego ECT was 1.67 times higher in her possession of the phonic instructional resource for each additional exposure she received via network connection with her alter colleagues, through which the colleague’s pinning behavior of the same phonic instructional resource (word building resource) was observed and diffused.


Regarding the heterogeneity of network influence, we found that ECTs responded differently to the pinning behavior of different colleague types, based on their degree of connections in the physical space. Those unassociated school-based colleagues (i.e., colleagues whom ECTs did not nominate as those with whom they most frequently discussed instruction) had the strongest influence on ECTs’ resource possession. In fact, the virtual network influence of those unassociated school colleagues was greater than the influence of ECTs’ close colleagues. Results showed that the log of mean count of ECTs’ resource possession increased by 1.12 for each additional exposure to the resource pinned by those unassociated school-based colleagues (see Table 4). That is, for a 1-unit increase in the virtual exposure to unassociated school colleagues’ pinning behaviors on the same resource, the mean count of ECTs’ resource possession and archival was multiplied by a value of 3.06. This was triple the influence that ECTs’ closest colleagues had on the log of mean count of their resource possession and archival, with an estimate of 0.34, for each unit increase in the network exposure. Correspondingly, for a 1-unit increase in the virtual exposure to close colleagues’ pinning behaviors on the same resource, the mean count of ECTs’ resource possession and archival was multiplied by a value of 1.40. Finally, we examined the impact of the pinning behavior of teachers outside ECTs’ school, within district, and found no significant influence on ECTs’ resource possession.


Table 4. Summary of the Negative Binomial Regression Analysis of ECTs’ Resource Possession as a Function of Separate Network Exposures


Independent variable

Model 2

 

B

SE B

[39_23306.htm_g/00031.jpg]

Week

0.03***

0.004

1.03

Network exposure to close colleagues

0.34*

0.15

1.40

Network exposure to unassociated school-based colleagues

1.12***

0.26

3.06

Network exposure to teachers outside school in same district

0.11

0.88

 

Note. B represents a regression coefficient in log scale.

*p < .05. ***p < .001.


To present a concrete example for better interpretation of the findings, imagine Ms. Hernandez, an ECT who uses Pinterest to curate educational resources online. Assume that Ms. Hernandez is connected with one close school-based colleague on Pinterest, who pinned a resource in Week 1. Given this, our results suggested that the mean count of Ms. Hernandez’s curation and archival of the same resource in Week 2 is 1.45, [39_23306.htm_g/00033.jpg]assuming a baseline of zero. In contrast, instead Ms. Hernandez is connected with one unassociated school-based colleague, who pinned a resource in Week 1. Everything else being equal, the mean count of her possession of the

same resource in Week 2 would increase from 1.45 to 3.16 [39_23306.htm_g/00035.jpg]. Therefore, if we conceptualize Ms. Hernandez’s online resource-seeking process as incidental learning and a desire to maximize her learning experience within her virtual social network, it would be more effective for her to virtually connect with streak ties, those unassociated school-based colleagues.


CONCLUSIONS


Resource flows within Pinterest may be a modern-day actualization of information diffusion through weak ties (Granovetter, 1973). Moreover, the strength of one’s ties within physical spaces adds a layer of complexity to understanding online resource flows. Strong ties provide teachers with support and redundant information, which reinforce particular teaching practices (Reagans & McEvily, 2003). In contrast, weak ties provide teachers with novel information and potentially the resources missing from teachers’ close collegial networks. Streak ties, the intersection of strong ties and weak ties, provide teachers with novel but context-relevant information, offering teachers opportunities to exploit the existing resources (i.e., the expertise of their school-based colleagues within VRPs, among an infinite catalog of instructional resources).  


Positive network effects, along with existing network paths, indicate that resources are not distributed evenly (Frank et al., 2008). Existing network paths vary by teacher, shaping the individual experience of online resource seeking and the meso-level resource flow, with both opportunities and constraints inherent in the network structure (Frank & Zhao, 2005). The difference in existing network path presents inequitable access to instructional resources across teachers and schools (Rogers, 2010). Consequently, ECTs are exposed to varying resources of potentially unequal instructional quality. This uncoordinated information diffusion may further exacerbate inequitable school resources.


LIMITATIONS


In this article, we do not examine connections that exist only in virtual spaces. Given that physical network ties account for approximately 3% of sampled teachers’ virtual networks, it is possible that exposure and diffusion mechanisms may differ or be more or less salient, depending on the origins of one’s network ties. Future work will examine how resources diffuse across primarily virtual networks over time.


Additionally, we found that the exposure to unassociated colleagues within one’s school has a stronger effect on ECTs’ resource possession. One factor that may contribute to this result is how we collect our network data. We use a snowball sampling approach, surveying our ECT sample and asking teachers to nominate with whom they discuss mathematics most frequently. We examine same-school ECTs’ nomination of close colleagues as the basis of our school-based unassociated social network. An ECT may not nominate colleagues within their school, yet may connect with them via VRPs. Because we examined all educational resources accessed by ECTs rather than those only related to mathematics, it may not be surprising that those unassociated same-school colleagues have the strongest effect on ECTs’ resource possession. Though this is possible, our sample comprises elementary school teachers who are generalists, teaching all subjects within their classroom. Therefore, it is reasonable to assume that those colleagues from whom ECTs seek advice related to mathematics are largely the same trusted individuals with whom they discuss other instructional topics.


DISCUSSION


Insufficient resources are one of many challenges ECTs face within their classrooms. This may be a particular problem for those ECTs who do not have a close collegial network within their school. Pinterest, a prominent educational VRP, provides an opportunity for teachers to seek out relevant materials while preserving their reputation as knowledgeable in their profession. However, the vast volume of resources within Pinterest may not be fully aligned with the curriculum adopted in ECTs’ schools. This potential misalignment between district expectations and available online resources may create uncertainty in content selection. Thus, ECTs may benefit from guidance on how to select quality online resources. Virtual connections to same-school colleagues within VRPs represent a trusted community to procure instructional content.


PRACTICAL IMPLICATIONS


Individual teachers’ resource-seeking behavior and network connections determine the resource flow structure, which consequently impacts the organizational performance in complex school systems. We next propose three points regarding both individual teachers’ behaviors and collective actions, with the aim of achieving better alignment of instructional content and cohesive instructional resources.


BOOSTING RESOURCE FLOW THROUGH ACTIVE ENGAGEMENT IN VIRTUAL RESOURCE POOLS


Teachers who seek educational resources online with the intention of applying the material for classroom teaching should be strongly encouraged to share their modification and experiences of classroom enactment in the virtual platform. The extra steps of commenting and sharing accelerate the resource flow, nurture new tie creation between teachers with similar interests (McPherson et al., 2001), and increase teachers’ capacity to apply the resources from multiple perspectives in their classrooms. In fact, a variety of functions on the Pinterest virtual platform allow teachers to actively interact with resources, such as the “Add photo” button, which encourages teachers to upload photos of their modification of the same resource from their classroom. Teachers are strongly encouraged to take initiative and provide more feedback on their application of resources in their classroom, including commenting on the quality or usefulness of the resources. This action helps direct other educators to select high-quality resources and be inspired to plan their teaching with better use of the resource.


LEVERAGING EXISTING EXPERTISE WITHIN A SCHOOL ON A VIRTUAL PLATFORM


Curated resource archives reveal, to some degree, the amount of tacit knowledge or expertise a teacher possesses. Though Pinterest is a platform that affords teachers the potential to seek out expertise that transcends school boundaries, teachers may also find it useful to reach out to other colleagues within their school for expertise. Healthy dynamics of resource flows rely on a combination of exploring resources outside, and exploiting existing resources within, schools. These actions extend individual teachers’ expertise, as evidenced by resources and tacit knowledge they curated online, to their same school colleagues in the virtual space (Daly, 2010). Resources and collegial ties beyond school boundaries provide teachers with opportunities to further exploit the within-school expertise. Thus, teachers are strongly encouraged to take advantage of the existing expertise with the most relevance to their school context before exploring other available resources in the virtual space.


ESTABLISHING AN ONLINE PROFESSIONAL LEARNING COMMUNITY WITHIN SOCIAL MEDIA


A growing body of educators on Pinterest have created collaborative boards where teachers may jointly contribute to curate relevant resources that are available online (Hooks, 2015). We suggest that teachers from the same school can extend their collaboration from class to cloud and increase the flow of resource diffusion with broader scope and easier access. Though teachers often curate educational resources individually, examining resources within social media and through their virtual collegial networks, they may also engage with collaborative boards. Collaborative boards within Pinterest provide teachers the opportunity to engage with one another iteratively and concurrently, creating an inclusive VRP community. The co-constructed knowledge may in turn improve individual teachers’ performance and facilitate organizational learning in a larger social system (Daly, 2010).


Current distribution of educational resources leaves a stratified system of access to high-quality, up-to-date educational materials. Social media and teacher communities extending across physical and virtual space may be able to leverage affordances to connect with a conglomerate of instructional resources and professional wisdom. We suggest that ECTs and experienced teachers collaborate outside their school-based social groups. ECTs may diversify their professional connections to school colleagues within virtual spaces to leverage the diffusion of same-school colleagues’ professional knowledge, while colleagues may take the initiative to introduce new resources through curating and archiving instructional content from those outside their primary social group. These approaches each increase ECTs’ exposure to diversified teaching resources, with the aim of more consistency and equity in online educational resources acquired within a school.


Acknowledgments


Research reported in this article was supported by the Center for Business and Social Analytics at Michigan State University, the National Science Foundation, and the William T. Grant Foundation under award numbers NSF REAL-1420532, WT Grant-182764. We thank and acknowledge the Teachers in Social Media team for their thoughts and work on various topics surrounding social media in education. Specifically, thanks to Dr. Kenneth Frank for his contribution to the intellectual theory and ongoing support. Additionally, thank you to Dr. Jiliang Tang for his collaboration in building an excellent data archive. Finally, thanks to the network, instructional content, and computer science team, including (listed alphabetically) Ibrahim Ahmed, Brianna Canedo, Nicole Donzella, Kim Jansen, Andy Jurasek, Hamid Karimi, Maggie Keech, John Lane, Amanda Opperman, Branda Peck, and Zhiwei Wang.


References

 

Achinstein, B. (2002). Conflict amid community: The micropolitics of teacher collaboration. Teachers College Record, 104(3), 421–455.

 

Baker-Doyle, K. J., & Yoon, S. A. (2010). Making expertise transparent: Using technology to strengthen social networks in teacher professional development. In A. J. Daly (Ed.), Social network theory and educational change (pp. 115–126). Harvard Education Press.

 

Bakshy, E., Rosenn, I., Marlow, C., & Adamic, L. (2012, April). The role of social networks in information diffusion. In Proceedings of the 21st international conference on World Wide Web (pp. 519–528). ACM.

 

Brown, D. (2019, July 1). How former teachers are making $100,000 or more, and you can too. USA Today. https://www.usatoday.com/story/tech/2019/07/01/teachers-quit-jobs-make-100-000-pinterest-selling-lesson-plans/1547697001/

 

Burt, R. S. (2005). Closure, trust, and reputation. In Brokerage and closure (pp. 93–166). Oxford University Press.

 

Chen, Z., Torphy, K., Hu, S., & Liu, Y (2017, May). Informal instructional resource acquisition and diffusion within Pinterest: A study of early career teachers' physical and virtual social networks. Paper presented at the International Network for Social Network Analysis Conference, Beijing, China.

 

Coburn, C. (2001). Collective sense-making about reading: How teachers mediate reading policy in their professional communities. Educational Evaluation and Policy Analysis, 23(2), 145–170.

 

Daly, A. J. (Ed.). (2010). Surveying the terrain ahead. In A. J. Daly (Ed.), Social network theory and educational change (pp. 259–274). Harvard Education Press.

 

De Nooy, W. (2011). Networks of action and events over time. A multilevel discrete-time event history model for longitudinal network data. Social Networks, 33(1), 31-40.

 

Desimone, L. (2002). How can comprehensive school reform models be successfully implemented? Review of Educational Research, 73(3), 433–479.

 

Frank, K. A., Maroulis, S. J., Duong, M. Q., & Kelcey, B. M. (2013). What would it take to change an inference? Using Rubin’s causal model to interpret the robustness of causal inferences. Educational Evaluation and Policy Analysis, 35(4), 437–460.

 

Frank, K. A., Muller, C., Schiller, K. S., Riegle-Crumb, C., Mueller, A. S., Crosnoe, R., & Pearson, J. (2008). The social dynamics of mathematics coursetaking in high school. American Journal of Sociology, 113(6), 1645–1696.

 

Frank, K. A., & Zhao, Y. (2005). Subgroups as a meso-level entity in the social organization of schools. In L. Hedges & B. Schneider (Eds.), Social organization of schools (pp. 279–318). Sage.

 

Frank, K. A., Zhao, Y., & Borman, K. (2004). Social capital and the diffusion of innovations within organizations: The case of computer technology in schools. Sociology of Education, 77(2), 148–171.

 

Frank, K. A., Zhao, Y., Penuel, W. R., Ellefson, N. C., & Porter, S. (2011). Focus, fiddle and friends: Sources of knowledge to perform the complex task of teaching.  Sociology of Education, 84(2), 137–156.

 

Friedkin, N. E. (1982). Information flow through strong and weak ties in intraorganizational social networks. Social Networks, 3(4), 273–285.

 

Gardner, D. P. (1983). A nation at risk: The imperative for educational reform: A report to the nation and the Secretary of Education, United States Department of Education. The National Commission on Excellence in Education.

 

Granovetter, M. S. (1973). The strength of weak ties. American Journal of Sociology, 78(6), 1360–1380.

 

Granovetter, M. (1983). The strength of weak ties: A network theory revisited. Sociological Theory, 1, 201–233.

 

Hampton, K. N. (2016). Persistent and pervasive community: New communication technologies and the future of community. American Behavioral Scientist, 60(1), 101–124.

 

Hooks, M. (2015). Pinterest: A tool for lesson planning. Mathematics Teacher, 108(6), 466–468.

 

Jimerson, J. B. (2014). Thinking about data: Exploring the development of mental models for “data use” among teachers and school leaders. Studies in Educational Evaluation, 42, 5–14.

 

Jones, N. D., Youngs, P., & Frank, K. A. (2013). The role of school-based colleagues in shaping the commitment of novice special and general education teachers. Exceptional Children, 79(3), 365–383.

 

Krackhardt, D., Nohria, N., & Eccles, B. (2003). The strength of strong ties. In R. Cross, A. Parker, & L. Sasson (Eds.), Networks in the knowledge economy (pp. 216–239). Oxford University Press.

 

Lave, J., & Wenger, E. (1991). Situated learning: Legitimate peripheral participation. Cambridge University Press.

 

McPherson, M., Smith-Lovin, L., & Cook, J. M. (2001). Birds of a feather: Homophily in social networks. Annual Review of Sociology, 27(1), 415–444.

 

Mitchell, C., & Sackney, L. (2000). Profound improvement. Swets & Zeitlinger.

 

National Council of Supervisors of Mathematics. (2015, March). Mathematics education in the digital age. https://www.mathedleadership.org/docs/resources/positionpapers/NCSMWhitePaper1.pdf

 

Opfer, V. D., Kaufman, J. H., & Thompson, L. E. (2016). Implementation of K–12 state standards for mathematics and English language arts and literacy. RAND Corporation.

 

Pool, I. (1980, May). Comment on Mark Granovetter’s “The strength of weak ties: A network theory revisited.” Read at the 1980 meeting of the International Communications Association, Acapulco, Mexico.

 

Reagans, R., & McEvily, B. (2003). Network structure and knowledge transfer: The effects of cohesion and range. Administrative Science Quarterly, 48(2), 240–267.

 

Rogers, E. M. (2010). Diffusion of innovations. Simon and Schuster.

 

Rosenberg, J. M., Xu, R., & Frank, K. A. (2018). Konfound-It! Quantify the robustness

of causal inferences. http://konfound-it.com

 

Rost, K. (2011). The strength of strong ties in the creation of innovation. Research Policy, 40(4), 588–604.

 

Ruef, M. (2002). Strong ties, weak ties and islands: Structural and cultural predictors of organizational innovation. Industrial and Corporate Change, 11(3), 427–449.

 

Schugurensky, D. (2000, October). The forms of informal learning: towards a conceptualization of the field. Paper presented at the New Approaches for Lifelong Learning annual conference, Toronto, Canada. http://hdl.handle.net/ 1807/2733

 

Singer, J. D., & Willett, J. B. (2003). Applied longitudinal data analysis: Modeling change and event occurrence. Oxford University Press.

 

Singer, N. (2017, September 2). Silicon Valley courts brand name teachers, raising ethics issues. The New York Times. https://www.nytimes.com/2017/09/02/technology/silicon-valley-teachers-tech.html?mcubz=1

 

Spillane, J. (1999). External reform initiatives and teachers’ efforts to reconstruct their practice: The mediating role of teachers’ zones of enactment. Journal of Curriculum Studies, 31(2), 143–175.

 

Spillane, J. P., Kim, C. M., & Frank, K. A. (2012). Instructional advice and information providing and receiving behavior in elementary schools: Exploring tie formation as a building block in social capital development. American Educational Research Journal, 49(6), 1112–1145.

 

Supovitz, J. A., Daly, A. J., & del Fresno, M. (2015). #COMMONCORE project: How social media is changing the politics of education. University of Pennsylvania ScholarlyCommons. http://repository.upenn.edu/hashtagcommoncore/1

 

Toole, J. C., & Louis, K. S. (2002). The role of professional learning communities in international education. In K. Leithwood & P. Hallinger (Eds.), Second international handbook of educational leadership and administration (pp. 245–279). Spring Science & Business Media.

 

Torphy, K., & Hu, S. (in press). Social media in education: Curation within and outside the schoolhouse. In Handbook on social media analytics: Advances and applications. CRC Press.

 

Torphy, K., Hu, S., Liu, Y., & Chen, Z. (2017, May). Teachers turning to teachers: Teacherpreneurial behaviors in social media. Paper presented at the American Education Research Association Annual Conference, San Antonio, TX.


Torphy, K., Liu, Y., Hu, S., & Chen, Z. (2018, April). Teachers turning to teachers: Social networks and curation of instructional resources in social media. Paper presented at the American Educational Research Association Conference, New York, NY.

 

Wellman, B. (2001). Computer networks as social networks. Science, 293(5537), 2031–2034.

 

Wellman, B. (2004). Connecting communities: On and offline. Contexts, 3(4), 22–28.

 

Will, M. (2016, October, 5). Looking for more support, new teachers turn to online communities. Education Week. http://www.edweek.org/tm/articles/2016/10/05/looking-for-more-support-new-teachers-turn.html?r=657076764

 

Willett, J. B., & Singer, J. D. (1995). It’s déjà vu all over again: Using multiple-spell discrete-time survival analysis. Journal of Educational and Behavioral Statistics, 20(1), 41–67.

 

Youngs, P., & King, M. (2002). Principal leadership for professional development to build school capacity. Educational Administration Quarterly, 38(5), 643–670.

 


APPENDIXES


Appendix A

Twelve Common Resources Pinned by at Least Six Early-Career Teachers for the 2015-2016 School Year

1. Pocket Chart–Word Building

2. Flexible Seating–1

                                   [39_23306.htm_g/00037.jpg]

[39_23306.htm_g/00038.jpg]                               

3. Silent Games

4. Implement a Reading Program

                                   [39_23306.htm_g/00039.jpg]

                               [39_23306.htm_g/00040.jpg]

5. Flexible Seating–2

6. Kindergarten Writing Checklist

                                   [39_23306.htm_g/00041.jpg]

                               [39_23306.htm_g/00042.jpg]

7. Twenty Frames

8. Classroom Management Strategies

                                  [39_23306.htm_g/00043.jpg] 

                     [39_23306.htm_g/00044.jpg]          

9. Alternative Seating in Classroom

10. Student Contract on Flexible Seating

                                [39_23306.htm_g/00045.jpg]   

                               [39_23306.htm_g/00046.jpg]

11. Become a Positive Teacher

12. Growth Mindset

                                   [39_23306.htm_g/00047.jpg]

                               [39_23306.htm_g/00048.jpg]

Note. Resources refer to images posted on Pinterest by a particular user that are either pinned from an outside website or an original upload. Resources can contain single or multiple images taken from different angles on the same activity from the same setting. Pinning behavior of any of the pictures is regarded as pinning the same resource.


Appendix B

Descriptions of Alternative Approaches, Rationales, and Models

The main analytical approach in the article conceptualizes teachers’ pinning behavior as knowledge possession with a threshold of maximum repeats. This approach best represents our theoretical understanding of how teachers use Pinterest to acquire educational resources. Alternatively, we propose three other conceptualizations of teachers’ pinning behavior: Approach 1 investigates the first occurrence of teachers’ resource-pinning activity; Approach 2 studies the repeated occurrence of teachers’ resource-pinning activities; and Approach 3 studies teachers’ resource possession without adjusting for individual maximum pinning threshold. Table B1 displays the data structures of three outcome variations in alternative approaches. Models B1-1 and B1-2 display the specification of fixed-effect and random-effect model for testing the network exposure effect.


Table B1. Data Structures of Three Alternative Conceptualizations of the Outcome and the Network Exposure

Week

First occurrence of teachers’ resource-pinning activity

Repeated occurrence of teachers’ resource-pinning activities

Teachers’ resource possession with no adjustment on individual maximum pinning threshold

Network exposure

0

0

0

0

0

1

0

0

0

0

15

0

0

0

1

23

1

1

1

1

24

 

0

1

1

 

38

 

0

1

2

39

 

0

1

2

40

 

1

2

2

41

 

0

2

2

 

46

 

0

2

3

 

52

 

0

2

3


Approach 1. Discrete time hazard model: We conceptualize the first occurrence of teachers’ resource-pinning activity as teachers’ transition from the state of null to another state of full accomplishment (De Nooy, 2011; J. D. Singer & Willet, 2003). In this model, teachers can only be in one of two possible states: absence or occurrence of pinning. This outcome variation contains no data on teachers’ repeated pinning, given that the transition of pinning state indicates the completion of a resource pinning cycle. All data after first pinning are set to missing. If the resource has not been pinned by the ECT in the one-year time interval, data are censored at Week 52. We favor the conceptualization of teachers’ resource possession over the conceptualization of an occurrence of teachers’ resource-pinning activity because one-time pinning may not be enough for teachers to grasp certain resources in a school academic year. Truncation of subsequent pinning activity ignores the natural continuity in teachers’ resource-seeking behavior.


Approach 2. Multiple-spell discrete time hazard model: This model conceptualizes teachers’ repeated pinning of a particular resource as a sequential occurrence of a disparate event (Willet & Singer, 1995). In this conceptualization, teachers do not inherit their prior pinning experience and the associated memory to the next sequence. After each sequence ends, the outcome is set back to zero until the next pinning behavior occurs. The last sequence is censored at Week 52. This approach has its limitations; pinning activity is more than a binary switch on teachers’ resource state. Though the impact of previous pinning may not be substantial, it should be accounted for and reflected in teachers’ level of resource possession in the following weeks. Thus, turning teachers’ resource state to zero ignores the impact of previous pinning experience on teachers’ current level of resource possession.


Approach 3. Negative binomial regression model: This approach uses the complete teacher resource possession data before censoring. The pinning outcome is teachers’ possession of a particular resource in a given week. All data are kept from Week 1 to Week 52. The limitation of this model is the assumption of a uniform pinning threshold for each teacher on any resource. This approach does not consider that information in each pin asks for a different level of cognitive demand, and teachers with different levels of content knowledge and resource familiarity may have different thresholds for the maximum pinning of a particular resource. Thus, we favor the censored teachers’ resource possession data adjusted for the threshold of each teacher’s maximum pinning of a particular resource over the uncensored teachers’ resource possession data.


Model B1-1. Fixed-effect model for alternative pinning outcome


[39_23306.htm_g/00050.jpg]


Note. Link function [39_23306.htm_g/00052.jpg] for hazard model in Approaches 1 and 2 is logit, and log for the negative binomial regression model in Approach 3. The multiple-spell discrete time hazard model also includes the event occurrence sequence as a covariate to model the repeated occurrence of teachers’ resource-pinning activity.


Model B1-2. Cross-classified random effect model for alternative pinning outcome


[39_23306.htm_g/00054.jpg]


Note. Link function [39_23306.htm_g/00056.jpg]for hazard model in Approaches 1 and 2 is logit, and log for the negative


binomial regression model in Approach 3. The multiple-spell discrete time hazard model also includes the event occurrence sequence as a covariate to model the repeated occurrence of teachers’ resource pinning activity.

                                                                  


Appendix C


Alternative Models Estimate Results


Table C1. Summary of the Logistic Regression Analysis on Discrete Time Hazard Model of Early Career Teacher’s First Occurrence of Resource Pinning Activity as a Function of Overall Network Exposure (N = 13,787)


Independent variable

Model B1-1

Model B1-2

 

B

SE B

[39_23306.htm_g/00058.jpg]

B

SE B

[39_23306.htm_g/00060.jpg]

Week

0.03**

0.01

1.03

0.023**

0.008

1.023

Network exposure

-0.15

0.38

 

-0.10

0.35

 

Note. B represents a regression coefficient in logit scale. Model B1-1 is the fixed-effect model that includes teacher and resource dummies for estimating the correct standard error of network effect. Model B1-2 is the cross-classified random-effect model with teachers and resources as the random components.

**p < .01.


Table C2. Summary of the Logistic Regression Analysis on Multiple-Spell Discrete Time Hazard Model of Early-Career Teacher’s Repeated Occurrence of Resource-Pinning Activities as a Function of Overall Network Exposure (N = 21,216)


Independent variable

Model B1-1

Model B1-2

 

B

SE B

[39_23306.htm_g/00062.jpg]

B

SE B

[39_23306.htm_g/00064.jpg]

Week

0.026**

0.008

1.026

0.023**

0.009

1.023

Network exposure

-0.07

0.27

 

-0.03

0.25

 

Sequence

-0.25

0.18

 

-0.001

0.17

 

Note. B represents a regression coefficient in logit scale. Model B1-1 is the fixed-effect model that includes teacher and resource dummies for estimating the correct standard error of network effect. Model B1-2 is the cross-classified random-effect model with teachers and resources as the random components.

**p < .01.


Table C3. Summary of the Negative Binomial Regression Analysis on Teacher’s Resource Possession as a Function of Overall Network Exposure (N = 21,216)


Independent variable

Model B1-1

Model B1-2

 

B

SE B

[39_23306.htm_g/00066.jpg]

B

SE B

[39_23306.htm_g/00068.jpg]

Week

0.06***

0.002

1.06

0.06***

0.002

1.06

Network exposure

0.21***

0.04

1.23

0.21***

0.04

1.23

Note. B represents a regression coefficient in log scale. Model B1-1 is the fixed-effect model that includes teacher and resource dummies for the estimating correct standard error of network effect. Model B1-2 is the cross-classified random-effect model with teachers and resources as the random components.

***p < .001.


The network effect is positive and significant when we conceptualize the pinning behavior as a knowledge possession that can be carried over time (see Tables 3 and C3). In contrast, the network effect is not significant when ECT’s pinning behavior is conceptualized as a transition from one state to another (see Tables C1 and C2). In addition, the magnitude of the network effect in the uncensored negative binomial regression is less strong than the network effect in the censored regression, where the latter is presented in the main analytical approach (see Tables 3 and C3). The mean count of ECT’s resource possession is multiplied by 1.23 for each unit increase in the network exposure, assuming that teachers have no upper bound on maximum pinning of a particular resource, in comparison with the estimate of 1.67 in the censored regression model.





Cite This Article as: Teachers College Record Volume 122 Number 6, 2020, p. 1-34
https://www.tcrecord.org ID Number: 23306, Date Accessed: 5/25/2022 11:50:01 AM

Purchase Reprint Rights for this article or review
 
Article Tools
Related Articles

Related Discussion
 
Post a Comment | Read All

About the Author
  • Yuqing Liu
    Michigan State University
    E-mail Author
    YUQING LIU is a doctoral candidate in the College of Education at Michigan State University. Her research focuses on the diffusion of innovation, gossip networks in organization, and the impact of students, schools, and policy contexts on teachers' online resource curation. Her recent paper, “Innovation Diffusion Within Large Environmental NGOs Through Informal Network Agents,” was published in Nature Sustainability.
  • Kaitlin Torphy
    Michigan State University
    E-mail Author
    KAITLIN TORPHY is the lead researcher and founder of the Teachers in Social Media project at Michigan State University. She has expertise in teachers’ engagement across virtual platforms, teachers’ physical and virtual social networks, and education policy reform. She has published work on charter school impacts, curricular reform, and teachers’ social networks, and she has presented work regarding teachers’ engagement within social media at the national and international levels. Dr. Torphy earned a PhD in education policy and a specialization in the economics of education from Michigan State University in 2014 and is a Teach for America alumna and former Chicago Public Schools teacher.
  • Sihua Hu
    Michigan State University
    E-mail Author
    SIHUA HU is a postdoctoral fellow on the COHERE project at Northwestern University. Her research examines various dimensions of teaching quality and how teaching quality is related to mathematics teachers’ social networks within physical and virtual spaces. Dr. Hu earned a PhD in mathematics education from Michigan State University and was a co-PI for an American Education Research Association conference convened in October 2018 on social media and education.
  • Jiliang Tang
    Michigan State University
    E-mail Author
    JILIANG TANG is an assistant professor in the computer science and engineering department at Michigan State University. Before that, he was a research scientist in Yahoo Research and earned his PhD from Arizona State University in 2015. His research interests including social computing, data mining, and machine learning and their applications in education. He was the recipient of the 2019 NSF Career Award and runner-up for the 2015 KDD Best Dissertation, and he has received numerous awards for his papers from top data science conferences, including WSDM2018 and KDD2016.
  • Zixi Chen
    Michigan State University
    E-mail Author
    ZIXI CHEN is a PhD candidate at the College of Education, Michigan State University, majoring in measurement and quantitative methods. She is interested in learning individuals' social-emotional behaviors in online space using computational social science and traditional quantitative methods.
 
Member Center
In Print
This Month's Issue

Submit
EMAIL

Twitter

RSS