Abstract
The study tested the hypothesis that separating stairway from escalator between pairs of origin and destination floors in a shopping center would increase the rate at which the stairway was used. The study site included 13 stairways and 12 pairs of escalators in seven connected shopping centers. A total of 33,793 pedestrians were counted ascending or descending over 35 days. Simultaneous flow rates up and down among centers correlated at r = .89. Distance between stairway and escalator accounted for 71% of variance in the ascending model and 21% in the descending model, with height between floors resulting in slightly diminished pedestrian volumes, and overall pedestrian volume accounting for less than proportional increases in stair climbing and descending. A 100% increase in distance between stair and nearest escalator yielded a 95% increase in stair climbing.
Keywords
Sedentary lifestyle is blamed as a leading cause of the continuous rise in death rates due to non-communicable diseases (World Health Organization, 2010) such as obesity, type 2 diabetes, hypertension, and heart disease. The promotion of a more active lifestyle has recently tended to focus on small measures taken in daily life, such as walking, bicycling, and climbing stairs. Urban environments are increasingly in multiple levels, potentially offering more opportunities for stair climbing in particular as part of daily routines. Efforts at promoting stair climbing have focused on motivational campaigns with some success as discussed below. Limited attention has been paid to the location of pedestrian facilities or their design as factors in increased stair climbing. Substantial variability in stair-climbing rates across facilities, environments, cities, and cultures suggests that the location of stair, escalator, and elevator facilities offers a largely untapped opportunity to increase the active transport option.
Architects and planners have recently tried to promote stair use by manipulating the materials, dimensions, location, and number of stair facilities, yet no known studies have undertaken to reveal the effectiveness of these strategies. This article is devoted to an examination of the effect of stair location with respect to the mechanical alternative on stair-climbing rates in several connected shopping centers. A review of the recent literature provides guidance for the design of the present study, as follows.
The Literature on Environmental Factors in Stair Climbing
Escalators are devices to move large numbers of pedestrians between building floors, with less exertion and without a break in the flow. Designers generally give a central position in the walking system to the escalator. A single escalator is implemented according to local building codes, but there is a wide variety of implementations of escalator and stair combinations (Strakosch, 1983). Although the addition of stairways to escalators is often necessary to help handle the heavy peak flows in transportation terminals, in most other circumstances, the facility design and location decisions are part of the larger planning effort, resulting in considerable variety in concept and provision level. The many approaches to the layout and local design of stairs and escalators lead to questions about these approaches and stair use. The following literature addresses the question of the effects of environmental interventions.
A review of the findings of the effectiveness of environmental modifications on stair-climbing rates found there were insufficient such studies to draw conclusions (Soler et al., 2010). In addition to the studies on environmental factors discussed above, two studies did report on stair use following non-structural design interventions (Boutelle, Jeffery, Murray, & Schmitz, 2001; Kerr, Yore, Ham, & Dietz, 2004). The following reports on the available literature in this area.
A considerable number of studies have been conducted on commuting behavior in the stair and escalator facilities of transportation terminals. Under conditions of heavy pedestrian traffic, individuals trade-off travel delay for personal exertion. Eves, Lewis, and Griffin (2008) reported two observation studies of stair choice at railway stations under variable passenger flow. As the number of passengers increases, so does the proportion of those opting for the stairs, with a plateau effect apparently reflecting stairway capacity. Lewis and Eves (2011) also observed a small but significant increase in stair climbing over elevator use as total traffic increased. In Hong Kong, 15% of commuters opted for the stairs when the delay due to congestion on the up escalator reached 17.4 s, but the same proportion of stair users tolerated 7.8 s of delay before descending the stairs (Cheung & Lam, 1998). It is all but certain that the greater effort required to walk up, about 220% of the effort to walk down (Teh & Aziz, 2002), accounts for the difference. Clearly, under conditions of moderate or heavy flow, it is important to account for overall pedestrian volume on the facilities.
Movement in a shopping center is more diffused and less time constrained but may respond to conditions of crowding. For example, Kerr, Eves, and Carroll (2001b) observed an increased rate of stair climbing as pedestrian volume increased in one shopping center stairway with adjacent escalators. Unlike transportation terminals where the stairway is invariably twinned with the escalator to enable distribution, shopping center designs include at least the following typical options: stairs paired with escalators, up and down escalators together or apart, stairs alone and escalators without stairs. Studies of stair climbing in shopping centers have focused on the impact of motivational campaigns (Webb, Eves, & Kerr, 2011). These studies have considered the impact of signs on stair climbing in various identifiable groups and under certain conditions of varying pedestrian density. Stair design and visibility are touched on, but otherwise, the environmental variables in this behavior remain largely unexplored.
Studies on interventions at worksites have tended to show very modest effects, but stair location in these situations militates in favor of the mechanical alternatives (Eves & Webb, 2006). This pattern is less evident in shopping centers where there are prevailing reasons to position both escalators and stairways where they might be thought to get maximum use.
Given the apparent importance of exertion on the tendency to walk upstairs, there has been relatively little study of the effect of vertical displacement on choice (Eves & Webb, 2006). Dolan et al. (2006) reported on the effect of number of stairs in a limited set of studies although this variable was not included in the results of the original studies. As expected, the tendency to climb stairs was marginally reduced when the height increased. However, in a study of pedestrian bridge use, the height was not significant while time saving figured prominently in decisions (Räsänen, Lajunen, Alticafarbay, & Aydin, 2007). Because of these very limited findings as well as the important effect height is likely to have on stair climbing, the present study aimed at its inclusion as an independent variable.
Stair location and size may be important in making a larger part of the pedestrian’s immediate field of view, increasing the immediacy of the choice if not a preference for some stair metric. The significance of stair location was considered in a 10-site study of stairs and elevators on a university campus (Nicoll, 2007). Visibility of the stair and the directness of the pathway between stairway, and the topologically most central pathways in the buildings explained 53% of stair use. Although this result suggests the possible significance of stairway location, there are prevailing reasons why observed stair use would be different when the alternative was an escalator (Eves, 2008).
Fundamentally, does putting the stair and escalator options together equalize the chance of taking the stair or diminish it? Presenting the stair alternative next to the escalator may be thought to eliminate locational factors that lessen the utility of the stair. However, it seems equally likely that the stair option is somehow evaluated differently when immediately adjacent to the exertion-reducing device. Faced with an unequal choice, the individual must find other motivations to expend the energy to walk up, as all the literature on the topic reminds us. Is the stairway evaluated differently when it is presented separately from the escalator? If the stairway was presented en route to alternatives such as an escalator, it may be that it would appear more attractive, given a prior decision to travel to a higher floor, as suggested in the Nicoll (2007) study. Opting to take stairs when presented alone might be a spontaneous decision, particularly when the travel plan is complex or vaguely formed. For example, Eves, Olander, Nicoll, Puig-Ribera, and Griffin (2009) found a tendency toward stair use when stairs were substantially more visible to the traveler. Similarly, stairs visible to the pedestrian were more likely to be used than stairs that were hidden from view (Dolan et al., 2006). The visibility of prompts to take the stairs had similar positive effect on stair use (Kerr, Eves, & Carroll, 2001a). A number of possible reasons for such effects include response to salient environmental variables, path minimization, and global concepts of the path structure.
Aims and Hypotheses
Based on the findings with regard to stair climbing, and especially the missing studies with regard to stairway location, this study focused on stairway location as a factor in choice. Location was coded as the minimum walking distance between the stairway and the nearest corresponding escalator option to the same destination level. Overall directional pedestrian volume during the counting session in the center was treated as an independent variable to test the possible effect of leakage to the stairway option as a result of perceived or real congestion. Vertical distance between levels was also treated as an independent variable based on variability in required energy expenditure. Similarly, up and down choices were treated independently based on differences in energy requirements.
We expected greater distance from the escalator option to result in higher rates of stair climbing based on the idea that the stairway presented alone appears more attractive than the stairway presented adjacent the escalator. Higher overall pedestrian volume was expected to result in increased stair climbing because of perceived or real delay whereas increased height between levels was expected to result in diminished stair climbing.
Method
The seven linked shopping centers were selected in downtown Montreal, Canada, in the following way. The lower floor had to be connected to the upper floor by open stairways and escalators, such that the only way to travel between the two levels was using these devices. In Centers 3, 4, 5, and 7 (online appendix), the two connected floors are superimposed. In 2, the upper level leads to the street, whereas in 1 and 6, the connected floors are adjacent to each other. In all cases, there were also emergency exit stairwells behind fire doors in peripheral locations, most with alarm bars, and were accordingly eliminated as possible choices in normal circumstances. The centers were apparently laid out with a prototypical model in mind; for example, with a food fair at the lower level combined with shops, whereas the upper level housed goods shops and services. A sky-lit atrium connected the two levels as in Center 4 (Figure 1). In three centers, there were also several other floors above the two study floors, but these were ignored in the present study. The centers are connected to each other through a tunnel under the street and by walking outdoors and across the street, but because of the considerable distance and complexity of these trajectories, they were highly unlikely as options between levels within one center.
Shopping Center 4 (Place Montréal Trust) layout at the lowest public level and the level immediately above, the same levels for all seven centers in this study.
In every case, the lower floor is a unified space, although with intervening kiosks, fountains, or seating areas, such that it was possible to walk between any pair of stairs and escalator at a single level. In all cases but one—the top of the rightmost stairway in Center 7—the nearest escalator option was visible from the bottom and top of the stairway. In all, there were 13 stairways and 12 pairs of escalators, with distances between the bottom and top of the stairway and the nearest escalator varying from 3.0 m to 72.8 m (Table 1). Height differences between the levels also varied from 12 steps to 37 (2.1-4.63 m). Eleven of the 12 escalators were of identical width, allowing one pedestrian to overtake a standing individual. The 12th escalator pair, also the shortest run, allowed only one individual to either stand or walk. Stairways varied considerably in design. All the stairways twinned with escalators were of a standard width of 2.0 m to 2.5 m, with intervening landings as required by the building codes. The standalone stairways were designed variously. One stairway in Center 7 was 5 m wide, while another 5 m wide stairway in that center was recently renovated in white ceramic tile and incorporated two landings. In Center 3, a twinned stairway is wrapped symmetrically around a structural column (online appendix).
Center Descriptions and Data Summary.
Note. CI = confidence interval.
Center names: 1 = La Baie d’Hudson; 2 = Promenades de la Cathédrale; 3 = Centre Eaton; 4 = Place Montréal Trust; 5 = Alliance Industrielle; 6 = Place des Arts; 7 = Complexe Desjardins.
Per 5-min block.
The centers themselves are part of the so-called Underground System of Montreal, which has 12 shopping centers connected by corridors passing beneath streets. Office buildings are located above or adjacent to the shopping centers. The clientele is a mix of office workers, people living elsewhere in the city or in the downtown on shopping trips or on leisure walks, as revealed in the following studies. Although the escalator alternative to the stairway is visible in almost all cases, it is nevertheless important to consider whether pedestrian movement constitutes an integrated system or discrete units.
Previous studies on two other shopping centers in the same downtown area provide some insights into the knowledge level of the visitors and the frequency of their visits. A questionnaire conducted in one of those shopping centers (Zacharias, 2000) revealed that 57% were on a walking excursion there more than once each week. In another study of 729 tracked visitors, goal-directed behavior predicted 37% of variance in spatial behavior (Zacharias, Bernhardt, & de Montigny, 2005), indicating quite a high level of purposeful behavior. A change in level was observed in 46% of the cases. Those studies are suggestive of the extent of knowledge of the centers by their denizens and presumably of the options for travel between floors available to them. More specifically, if they wished to reach a presumed destination on another floor, a high proportion of visitors are likely to have some knowledge of the alternatives. This study showed a high degree of familiarity with one of the shopping centers among the users, suggesting it is likely to be the case with the seven others, although we cannot be certain.
Procedures
Counts of moving pedestrians on stairs or escalator were conducted in 5-min blocks by a single observer, who rotated the order of counts over the recording time frame from 10:30 a.m. to 12:00 noon, and from 12:30 p.m. to 2:30 p.m., to ensure variability in the counts at any one site. A recording app and clock timer—NiftyCounterMax™ and ClockPro™—were installed on an iPad™ where up to four counters could be tapped simultaneously for the situations where up and down escalators were visible with a stairway. An initial test for the reliability of the recording protocol was carried out with 12 counts at locations with the highest flow levels and with 4 simultaneous counts. The first step at the bottom or top constituted the count cordon. The untrained observer’s counts correlated with the observer’s counts at .999, with 5 discrepancies in the counts out of a possible 24, amounting to one or two individuals, all cases of interpretation, not fact. We therefore accepted that the recording method enabled accurate counts at quite high overall volumes, requiring a visual fix on the scene and good eye–hand coordination. Because of the size of the centers and options that could not all be seen from a single vantage point, counts were conducted by the same observer simultaneously (Centers 1, 5, 6), in two successive time blocks (Centers 2, 3, 4) and in five (Center 7). Nearest escalators and stairways were counted together, but some of the counts were visually separated from each other and were counted up to 20 min later. The bigger question is how stable are the conditions experienced by the users over this longer time frame. To address this question, six successive 5-min counts were conducted for all flows at four locations. The 95% confidence band on the mean was .14 of its absolute value, dropping to .10 for a relatively high 5-min flow of 93 persons. The flows could be said to be relatively stable over 30 min, to the extent that 5-min time slots within that time frame would be interchangeable.
Other evidence of choice is in the simultaneous amplitudes of flows across all choices, more evidence of stability in the decision making of pedestrians, lending support to the assumption that each center operated as a system with all stair and escalator options involved. We compare the up and down flows below.
The stair counts were entered as the dependent variable in an ordinary least squares (OLS) regression model, performed separately for upward and downward movement, because of the expected difference in choice as discussed above. The minimum walking distance in meters between the foot of the stairway and the foot of the nearest escalator—obtained using the Footsteps™ pedometer installed on an iPod™—was entered as an independent variable for every count. The height between levels in individual centers was entered in meters. Individual 5-min counts, totaled for all channels within the round of counts conducted on a single day at a single center, represented the total directional pedestrian flow. There were 35 successive weekday counts for the 50 different stair and escalators flows, for a total of 1,750 individual counts. A total of 33,793 individual choices were recorded, with distributions and related center information summarized in Table 1.
Results
Climbing and descending rates varied as follows: Mean stair-climbing volume per 5-min block was 8.6 (SD = 12.2; max = 68), while the comparable figure for descending was 12.9 (SD = 13.3; max = 85). Mean up escalator volume was 35.1 (SD = 26.7; max = 141) with mean descending volume at 27.5 (SD = 20.4; max = 112). Although the detailed distributions are presented in Table 1, it can be seen that overall the counts are sufficiently distributed for further analysis. Overall, .23 of the total flow upward climbed stairs, while .27 of the downward flow was on stairs. The observed flows do not approach saturation. Cheung and Lam (1998) reported maximum escalator flow of 120 passengers/min at the study station in the Mass Transit Rail (MTR) of Hong Kong or 4.25 the maximum volume observed in our cases. Whereas maximum passenger flow obliges passengers to stand firm on the escalator, lower volumes with climbing passengers will not approach these volumes or the maximum design volume of 150/min. However, it remains possible that escalator passenger volume might enter into the decision to take the stairs in our case, either because it is not possible to walk up or down the escalator at these lower volumes or because the facility simply appears crowded to the approaching pedestrian. Therefore, we retain pedestrian volume as an independent variable in the subsequent analysis.
Up and down flows were fairly closely related within centers. Correlations between stair movement up and escalator movement up by center varied between .77 and .86 while movement down correlated between .39 and .82. As was pointed out in the pre-test, flows remained stable over 30 min within centers. For example, up and down movement across all centers was highly correlated (.89), lending support to the thesis that, because of the regularity of the observed choices across the environment, the flows constitute some form of rational decision making.
Regression Results
The dependent variable was the volume of pedestrians counted in each direction at every location in one 5-min time interval. The independent variables were distance between the foot or top of the stairway and the nearest foot or top of escalator, respectively; the total upward or downward flow over the recording period, entered in separate regression models; the height difference between floors, measured in meters.
Correlation coefficients were calculated for pairs of independent variables. Extremely weak values, varying from .000 to .042, make clear that covariance and any consequent inflation are not factors in this case. All three independent variables, for which we have theory as discussed at the end of the Literature section above, are retained in the subsequent analysis.
Hierarchical linear regression was chosen as the model to isolate the effect of total pedestrian volume on stair-climbing rates at all locations. A Poisson model, which might normally be used for count data, did not meet the requirement of equivalence between means and variances. In the up-movement Model I, total pedestrian flow alone was entered. The proportion of variance explained by pedestrian volume, that is real or perceived congestion, was a significant but relatively small .042. The regression coefficient is also positively signed, in agreement with the hypothesis for the effect of congestion. Distance and height are added next, for a total .71 of the variance in stair-climbing volume (Table 2). The distance coefficient is positively signed whereas height is negatively signed, in keeping with expectations. In this case, a 100% increase in height yields a decline in stair climbing of 50%. In Model II for downward movement, pedestrian volume also accounts for a significant but small proportion of the variance at .030. All variables result in .24 proportion of variance explained. In this model, both distance and height are signed as in Model I.
Hierarchical Linear Regression for Distance Between Nearest Options, Total Pedestrian Volume, and Height Between Floors as Predictors.
p < .01.
In the down-movement models, it was expected that distance would have less impact on the decision to take the stairs rather than the escalator. Faced with the choice of stair and escalator together, it was reasoned that descending the stairs presented a lesser challenge than climbing the stairs from the bottom in those locations. Overall, the impacts of the predictor variables is much less in the down-movement model, at about the rate of reduction expected by relative energy expenditure (Teh & Aziz, 2002). For example, the ratio of ascending pedestrians to total flow on stairways without paired escalators is .49 whereas ascending pedestrians represent .27 of the stairway flow on stair–escalator pairs.
Discussion
Deciding to walk up stairs or ride an escalator has psychological, environmental, and situational components. Although the decision to take stairs is popularly thought of as spontaneous, or not yet understood, it can also be manipulated. The walk up stairs can be successfully reframed as a contribution to personal health and the environment (Dolan et al., 2006; Eves et al., 2009; Lewis & Eves, 2012). The visibility of printed messages was significant in overall decision making (Olander, Eves, & Puig-Ribera, 2008) and more effective than verbal intervention (Olander & Eves, 2011). Whether increases in stair climbing are maintained after motivational campaigns end remains a matter for debate, although most studies show a sustained increase over baseline.
Also with significant psychological dimensions is the path choice itself. There has been some investigation of how path choices are conceptualized but little on how perceived physical exertion affects path choice. Perfect knowledge of the local environment would enable an optimal choice in terms of physical exertion, if physical exertion were actually salient in experience. Other considerations have entered the plan and execution of the path that include dimensions of the self, perceptions of the environment, and conceptualization of the trip.
Environmental factors have been explicitly identified as having an impact on stair climbing, including the visibility of the stairway and its width. The effectiveness of stairway width merits further investigation based on preliminary results, along with the impact of landings, materials, and stair configuration. These local effects also need to be disentangled from location. In the present study, for example, it was impossible to know the proportion of observed choices that constituted distance or time-optimized paths between some origins and presumed destinations. Similarly, we do not know whether the trip would have been optimized had the traveler taken an escalator at some other location instead of the stairway. Based on the present study, facility location would be a good candidate for further investigation. Escalators are often planned for the most central and accessible locations especially in shopping centers where it is generally believed that offering effort relief will promote vertical movement. It will be argued that escalators promote ascent more than do or could stairways. So it would also be worthwhile to examine the effect of providing only broad and elaborately designed stairways for most of the locations for vertical movement as was done in the new Porta Nuova business and residential area of Milan. Accordingly, a central and highly visible location to the stairway while eliminating the escalator entirely is another approach worth investigating, as was done at the Swiss Air terminal at Zurich International Airport. These and other similar experiments in promoting stair-climbing merit study.
Situational variables may play into the decision to walk up or stand on an escalator. We have already seen that in the time-constrained commuting terminal, travelers are more likely to take the stairs when their travel is impeded. Shopping environments are also not all alike, with the present shopping centers patronized in large part by office workers, whose leisure time travel behavior might well be different from that of the dedicated shopper in a suburban setting. Small groups in verbal exchange will conceive the stair and escalator choice somewhat differently based on the spatial needs of their interaction, for example.
Conclusion
This study set out to test the hypothesis that separating the stairway option from the escalator by some distance would result in an increase in the number of stair users, all else being equal. The relatively large sample of stairways and escalators, the same time frame, and identical counting methods rendered the possible effect of design highly unlikely. Flows on stairs and escalators in the same time frame within centers and between choices were highly related, strongly suggesting that movement operated as a single system within centers and among the available choices.
Distance, height between floors, and total pedestrian flow collectively account for 71% of the variance in stair use in ascending pedestrians, while the same variables account for 24% in descending pedestrians. In these models, distance alone accounts for most of the variance. All else being equal, increasing distance between a stairway and the nearest escalator by 100% increases the likelihood of climbing the stairs by 95%. Increasing pedestrian flow had a significant albeit marginal effect on stair-climbing rates in our study. This may be due to the fact that overall pedestrian flow was far from saturation levels, without impedance on travel speed in particular.
Footnotes
Acknowledgements
The authors wish to thank three anonymous reviewers and the editor for their helpful suggestions.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The Peking University supported this study.