Network analysis approach to Likert-style surveys

Likert-style surveys are a widely used research instrument to assess respondents’ preferences, beliefs, or experiences. In this paper, we propose and demonstrate how network analysis (NA) can be employed to model and evaluate the interconnectedness of items in Likert-style surveys. We explore the advantages of this approach by applying the methodology to the aspects of student experience scale datasets and compare the results to the principal component analysis. We successfully create a meaningful network based on survey item response similarity and use modular analysis of the network to identify larger themes built from the connections of particular aspects. The modular NA of the network of survey items identifies important themes that highlight differences in students’ overall experiences. Our network analysis for Likert-style surveys methodology is widely applicable and provides a new way to investigate phenomena assessed by Likert-style surveys.

Figure 1

Visualization of generating a network based on Likert-style survey responses. (a) Bipartite network of students (light gray) and response selections (dark gray). (b) Projection of bipartite network onto the response selections, weighted by number of students connecting. (c) Network of response selections for just Q1 and Q2. (d) Adjacency matrix representation of Q1 and Q2 response network with the negative temperature relationships boxed in blue, the positive temperature relationships boxed in red, and the inverse relationships boxed in gray. The same color scheme is used in the remaining panels. (e) Adjacency matrix of the survey item network after collapsing the response network with significant edge weights in bold. (f) A full survey item network with all edges included. (g) A backbone survey item network with only significant edges.

Figure 2

Representations of the full network of ASES items. (a) Adjacency matrix of ASES items represented by a heat map. (b) Network of survey items arranged in a circle with edge weights corresponding to the absolute value of the similarity. Both the adjacency matrix representation and the network representation use two scales: one for items that share similar responses that ranges from red (mutual agreement) to blue (mutual disagreement), the second for items that share dissimilar responses ranging from black (most dissimilar) to white (no correlation in responses).

Figure 3

The ASES similarity network. The node colors indicate the principal component that the survey item was loaded into; mentoring and research experience, $E$ (orange), professional development, $D$ (teal), social and academic integration, $I$ (green), financial support, $S$ (purple), or not loaded (white). The shaded regions represent the clusters identified by the fast-greedy clustering algorithm; $C_1$ (green), $C_2$ (orange), $C_3$ (teal), or $C_4$ (purple). The edge thickness represents the amount of similarity. The edge color indicates the edge temperature, i.e., whether the edge is created via mutual agree (red) or disagree (blue) answers. The dashed edges indicate connections that span across clusters.

Figure 4

The hierarchical structure of the partition provided by the fast-greedy cluster detection algorithm, represented as a dendrogram. Each node name is listed along the bottom. Each of the four clusters is identified by colored boxes around the respective branches; $C_1$ (orange), $C_2$ (green), $C_3$ (teal), or $C_4$ (purple).

Figure 5

A visual representation of the four clusters in which each term represents a particular survey item. The colors correspond to those used to shade clusters in Fig. 3: $C_1$ (green), $C_2$ (orange), $C_3$ (teal), or $C_4$ (purple). The font size of each term reflects the relative importance of that survey item to the respective cluster, as determined by the hierarchical clustering shown in Fig. 4. Complete definitions of the items are presented in Table 2.