**3. Analysis and Results**

*3.1. Research Question 1a: Do People Visually Engage with Street Edge Ground Floors more than Upper Floors along (i) Non-Pedestrianised and (ii) Pedestrianised Streets?*

To determine if participants visually engaged more with street edge ground or upper floors along both non-pedestrianised and pedestrianised streets, Welch two-sample *t*-tests were performed. This *t*-test was chosen because of unequal variance between the two samples. Visual engagemen<sup>t</sup> was focused overwhelmingly upon ground floors in comparison to upper floors along both non-pedestrianised streets and pedestrianised streets (Table 1).

**Table 1.** The amount of visual engagemen<sup>t</sup> with street edge ground and upper floors along non-pedestrianised and pedestrianised streets (mean ± standard error).


*3.2. Research Question 1b: Do Di*ff*erent Everyday Activities and Streets Walked Influence the Amount of Visual Engagement upon Ground Floors along (i) Non-Pedestrianised and (ii) Pedestrianised Streets?*

The effects of activity and street on participants' visual engagemen<sup>t</sup> with street edge ground floors were determined using linear mixed-effects models in the R statistical computing environment [49] ('lme4- package [50]). The fixed effects were 'activity' (optional or necessary) and 'street' (street id: 1–6 for non-pedestrianised and 7–11 for pedestrianised). To account for inter-participant variation in gaze behaviour, 'participant' (participant number 1–24) was entered as a random effect, which allowed different intercepts for each participant (i.e., a differing baseline level of engagemen<sup>t</sup> for each participant). *P*-values were simulated by comparing this model to a grand mean model using a parametric bootstrapping method ('pbkrtest' package; [51]) with 10,000 simulated generations. The goodness of fit for all mixed effect models was assessed using the 'R.squaredGLMM' function ('MuMin' package [52]) and marginal *R*<sup>2</sup> values (those associated with the fixed effects only) were high (non-pedestrianised street analyses: *R*<sup>2</sup> = 0.58, pedestrianised street analyses: *R*<sup>2</sup> = 0.55).

The activity being undertaken influenced the amount of visual engagemen<sup>t</sup> with street edge ground floors along both non-pedestrianised (see Figure 4A) and pedestrianised streets (see Figure 4B) (Table 2). The different streets walked influenced the amount of visual engagemen<sup>t</sup> with street edge ground floors along both non-pedestrianised (see Figure 4B) and pedestrianised streets (see Figure 4D) (Table 2).

**Table 2.** The influence of everyday activities and street walked upon visual engagemen<sup>t</sup> with street edge ground floors along non-pedestrianised and pedestrianised streets (mean ± standard error).


**Figure 4.** The influence of activity and street on the percentage of participants' visual engagemen<sup>t</sup> with street edge ground floors along non-pedestrianised and pedestrianised streets. Error bars represent 1 standard error.

### *3.3. Research Question 2a: Are there Di*ff*erences in the Amount of Visual Engagement upon Street Edges on Di*ff*erent Sides of the Street along (i) Non-Pedestrianised and (ii) Pedestrianised Streets?*

To determine if the amount of participant visual engagemen<sup>t</sup> with the different sided street edges varied along both non-pedestrianised and pedestrianised streets, Welch two-sample *t*-tests were performed. Visual engagemen<sup>t</sup> with street edges along non-pedestrianised streets was predominantly with the street edge on the walked compared with the opposite side (see Table 3a). The amount of visual engagemen<sup>t</sup> with street edges along pedestrianised streets did not vary across the street edges on the left and right sides (see Table 3b).

**Table 3.** (**a**) The amount of visual engagemen<sup>t</sup> with street edges on different sides of non-pedestrianised streets (mean ± standard error). (**b**) The amount of visual engagemen<sup>t</sup> with street edges on different sides of pedestrianised streets (mean ± standard error).




*3.4. Research Question 2b: Do Di*ff*erent Everyday Activities and Streets Walked Influence the Amount of Visual Engagement upon Street Edges on Di*ff*erent Sides of the Street along (i) Non-Pedestrianised and (ii) Pedestrianised Streets?*

To determine the effect of activity and street, linear mixed-effects models were again fitted to the data. The fixed effects were activity and street, with participant as a random effect. Marginal R<sup>2</sup> values (those associated with the fixed effects only) were high (analyses for walked non-pedestrianised edge *R*<sup>2</sup> = 0.54; opposite non-pedestrianised edge *R*<sup>2</sup> = 0.25; left side pedestrianised edge *R*<sup>2</sup> = 0.33; right side pedestrianised edge *R*<sup>2</sup> = 0.39).

The activity undertaken influenced the amount of visual engagemen<sup>t</sup> with both the street edges on the walked and opposite sides of non-pedestrianised streets (see Figure 5A and Table 4). Along pedestrianised streets, activity influenced the amount of visual engagemen<sup>t</sup> with both the street edge on the left and right sides (see Figure 5B and Table 4).

**Figure 5.** The influence of activity and street on the percentage of participants' visual engagemen<sup>t</sup> with street edges on different sides along non-pedestrianised and pedestrianised streets. Error bars represent 1 standard error.


**Table 4.** The influence of everyday activities and street walked upon visual engagemen<sup>t</sup> with street edges on different sides of the street along non-pedestrianised and pedestrianised streets (mean ± standard error).

The different non-pedestrianised streets walked influenced the amount of visual engagemen<sup>t</sup> with the street edge on the walked side, but not the street edge on the opposite side (see Figure 5C and Table 4). The different pedestrianised streets walked influenced the amount of visual engagemen<sup>t</sup> with both the street edges on the left and right sides (see Figure 5D and Table 4).

*3.5. Research Question 3: Are there Di*ff*erences in the Amount of Visual Engagement upon (i) Street Edge Ground Floors between Non-Pedestrianised and Pedestrianised Streets and (ii) Street Edge Sides between Non-Pedestrianised and Pedestrianised Streets?*

To determine if the amount of visual engagemen<sup>t</sup> with street edge ground floors varied between non-pedestrianised and pedestrianised streets, Welch two-sample *t*-tests were performed. The amount of visual engagemen<sup>t</sup> with the street edge ground floors did not vary between the non-pedestrianised and pedestrianised streets (Table 5).



The difference in the amount of visual engagemen<sup>t</sup> upon street edges of the same street was calculated and Welch two-sample *t*-tests were performed to determine if these differences varied between non-pedestrianised and pedestrianised streets. The difference in street edge visual engagemen<sup>t</sup> between the sides of non-pedestrianised streets was greater than the sides of pedestrianised streets (Table 5).
