Biomechanics of Long Cane Use

Abstract

Introduction:

The modern long cane has been used by people who are blind for traveling for decades. This article describes parameters surrounding the collection of over 10,000 trials of people walking with the long cane to detect drop-offs or obstacles.

Methods:

The data include 10,069 trials representing 101 different participants in 366 conditions over 11 studies spanning the 9 years from 2007 to 2016. Each of the studies investigated different participant or cane characteristics or both in terms of their effect on either drop-off or obstacle detection. Results of detection performance in these studies appear in other articles. This article describes biomechanical measures derived from 3-D motion analysis equipment used during the studies.

Results:

Initial treatment of the large data set indicated that participants tended to not center their cane arc laterally on their body, deviating up to about 20 centimeters from midline. Arc widths averaged almost a meter, and arcs were generally centered. Participants were generally poor at being in step or having consistent rhythm. Coverage rates averaged about 85%.

Discussion:

Although participants might have demonstrated artificially high skill performance due to being in a research study, data do offer insights into mechanical performance of skills. This survey of the data set indicates that not centering the hand holding the cane does not decrease body coverage less than about 85%. However, further analyses will be conducted to delve more deeply into all aspects of the data.

Implications for practitioners:

Basic cane skills can be taught with short sessions and massed practice. Novices can acquire basic cane skills on par with cane users who are blind, but individual differences exist and the interplay of biomechanical variables needs to more fully understood.

Beyond their own senses, the primary tool used by people who are blind to gather information about the environment around them and aid in independent navigation is the long cane. This being the case, it is critical that the operation of the user–cane system be understood as well as possible, so that the use of the cane can be optimized. Since being introduced in the 1940s, the design of the modern long cane has been changed in some notable ways. New materials such as graphite have been used for the shaft, segmented shafts that can be folded are common, and a range of interchangeable tips as well as grips of different material is available. However, basic techniques for the use of the cane have not changed appreciably, although some modified techniques (e.g., Uslan, 1978) have been suggested. The two-point touch and constant contact cane techniques remain the most common techniques (although which of the two is dominant is a matter of debate). Beyond basic cane design, the way that cane characteristics interact with user characteristics and with performance for critical performance outcomes such as drop-off and obstacle detection needs to be more fully studied.

There has been some work done in this area previously. The long cane can be modified to allow use by specific users who cannot wield the standard cane typically, such as by adding a secondary handle (Aquilante, Kern, & Courtney, 1996), but these types of targeted modifications have limited generalized impact. Of more general use is study of how different tips, grips, shaft materials, and cane techniques impact cane use. Kim and Wall Emerson (2018) found differences in obstacle detection for cane tip type (Bundu basher tip being better than marshmallow tip and constant contact cane technique being better than a modified constant contact technique). Constant contact was statistically better at detecting obstacles than two-point touch (Kim & Wall Emerson, 2014) and at detecting drop-offs (Kim, Wall Emerson, & Curtis, 2009). Another study found no difference for drop-off detection between a marshmallow roller tip and a marshmallow tip, but constant contact with a marshmallow roller tip was better than two-point touch with a marshmallow tip (Kim, Wall Emerson, & Curtis, 2010b). The benefit of constant contact over two-point touch for drop-off detection is more pronounced for less experienced cane users (Kim, Wall Emerson, & Curtis, 2010c).

In terms of cane characteristics, Rodgers and Wall Emerson (2005b) found that a cane that was 5% shorter or 10% longer than the length suggested by the sternum method for determining cane length led to significantly worse drop-off detection. However, Kim and Wall Emerson (2012) found that drop-off detection rate and 50% drop-off threshold were not different between standard length canes and canes that were 10 inches longer. A further study supported the earlier findings by showing that drop-offs were detected better with a cane of standard length than with a longer cane (Kim, Wall Emerson, & Naghshineh, 2017). This study also found that drop-offs were detected significantly more frequently when the cane arc was 3 centimeters wider than a person’s body as opposed to when the arc was substantially wider. Cane length and arc width were not found to have an effect on obstacle detection (Kim et al., 2017).

In looking at transmission of vibrational information along a cane shaft, Rodgers and Wall Emerson (2005a) found a less flexible cane shaft is better at transmitting vibrations. Further, lighter shafts transmit energy at higher natural frequencies. Although this finding suggests that a more rigid shaft might be better overall, Kim, Wall Emerson, Naghshineh, and Auer (2017) found that heavier canes facilitated drop-off detection when constant contact was used but that lighter canes were better when two-point touch was used. Similarly, flexible canes were better for drop-off detection but only when using two-point touch.

Although cane and technique characteristics are important to study in order to fully understand the cane–user system, it is equally important to investigate user characteristics. An argument has been made that the two-point touch cane technique is biomechanically advantageous, if not optimal (Wall, 2002). Although novices to cane use can acquire the basic mechanics in a very short period of time (Wall & Ashmead, 2002b), when people with and without visual impairment but who are experienced in cane use are compared, both groups tended to have wider arcs and hand positions off midline with low measures of body coverage and rhythm (Wall & Ashmead, 2002a). This situation can occur when a wider arc, whether off center or not, leads to the cane spending more time to one side of a person’s body or the other instead of previewing the space through which a person will walk. Also, younger cane users are better at detecting drop-offs than are older users, and participants who lost vision early in life detected drop-offs statistically more reliably than those who lost vision later in life (Kim, Wall Emerson, & Curtis, 2010a).

Studies that tried to quantify safety and efficiency of cane use began to appear in the 1990s (e.g., Blasch & De l’Aune, 1992; Blasch, LaGrow, & De l’Aune, 1996; Bongers, Schellingerhout, van Grinsven, & Smitsman, 2002; LaGrow, Blasch, & De l’Aune, 1997). While previous work demonstrates how some characteristics of the cane or the user impact outcomes such as drop-off and obstacle detection, measuring how the cane and the user move together and how variations in movement affect outcomes is critical for understanding and perhaps optimizing long cane use. The fact that visually impaired people walk slower is well-documented (e.g., Beggs, 1991; Clark-Carter, Heyes, & Howarth, 1986b), and this speed is often slower than their preferred speed (Clark-Carter, Heyes, & Howarth, 1986a, 1987). This slowness is due to the need to pay more attention to the walking path for issues such as obstacle avoidance (Patla, Davies, & Niechwiej, 2004).

There has been some work on measuring the gait of blind people when using the long cane. Johnson, Johnson, Blasch, and De l’Aune (1998) studied seven sighted and five participants with visual impairments and found that, regardless of visual status, the cane tip tapped the ground outside where the foot landed. The only significant group difference was how fast the cane was being swung back and forth. Mason, Legge, and Kallie (2005) found that variability was low and similar for both blind and sighted walkers when walking slow, at a preferred pace, and quickly. Nakamura (1997) analyzed the gait of 15 late blind (meaning those who had lost vision later in life), 15 congenitally blind, and 15 sighted walkers and found that, compared to the sighted group, both blind groups walked slower, had a shorter stride, and spent longer in the stance phase of walking. Ramsey, Blasch, Akio, and Johnson (1999) found no differences in biomechanical measures between traditional and modified cane technique use but did find that gait velocity was slower and stride length was shorter when the cane user was anticipating a drop-off or was also engaged in an auditory task.

This article summarizes, in broad strokes, the approach and general results of a long-term project to collect biomechanical data on long cane use by blind people and sighted people who were blindfolded. The data include 10,069 trials representing 101 different participants (113 participants with people appearing in more than one study) in 366 conditions over 11 studies spanning the 9 years from 2007 to 2016. In each of the studies, a range of independent variables involving participant characteristics or cane characteristics or both was investigated in terms of their effect on either drop-off detection or obstacle detection. In each of these studies, data were also gathered on participants’ body and cane movements. These bio-mechanical data are the focus of this article and will be analyzed in more depth in a future publication. However, it was decided that an initial broad look at the data would be beneficial before a more intensive analysis was reported. It is unlikely that broad findings will change, even though a more nuanced understanding of the results will come with additional targeted analyses. Table 1 shows the primary independent and dependent measures in each of the studies, while Table 2 shows participant characteristics for each of the studies. In general, the participants who were blind were adults recruited from the community, and the sighted participants were orientation and mobility (O&M) graduate students who were familiar with the use of the long cane through their coursework. Some graduate students were also blind. For all studies, approval was obtained from the Western Michigan University Institutional Review Board prior to conducting the study, and informed consent was obtained from each participant before collecting data.

Table 1.

Characteristics of studies during which data were gathered.

Study	Dependent measure	Primary independent measures	Year
Phase 1	Drop-off detection	Cane technique	2007
Phase 2	Drop-off detection	Cane technique, cane tip	2008
Phase 3	Drop-off detection	Cane length, cane technique	2009
Phase 4	Drop-off detection	Cane length, cane technique	2010
Cane weight	Obstacle detection	Weight distribution	2015
Cane material	Drop-off detection	Cane weight, flexibility, technique	2014
Arc width	Obstacle detection	Arc width	2015
Obstacle 2013	Obstacle detection	Cane technique	2013
Obstacle 2014	Obstacle detection	Cane technique	2014
Bundu 2016	Obstacle detection	Cane tip	2016
Obstacle 2016	Obstacle detection	Cane technique	2016

Table 2.

Participant demographics.

Study	Participants with useful data	Gender (male/ female)	Age, M (SD)	Cane use in years, M (SD)	Cane training (years), M (SD)	Primary technique(TT, CC)
Phase 1	15 Blind	10 M, 5 F	36 (14.28)	10.67 (10.91)	4.97 (6.02)	2 TT, 14 CC
Phase 2	17 Blind	8 M, 9 F	42.06 (14.57)	12.23 (12.72)	3.47 (5.83)	9 TT, 8 CC
Phase 3	10 Sighted, 1 blind	2 M, 9 F	35.36 (10.6)	0.16 (0.1)	0.17 (0.14)	7 TT, 4 CC
Phase 4	5 Sighted	3 M, 2 F	35.8 (8.17)	0.16 (0.23)	0.16 (0.23)	0 TT, 5 CC
Cane weight	8 Sighted	I M.7 F	25.5 (3.82)	0.33 (0)	0.33 (0)	0 TT, 8 CC
Cane material	13 Blind	8 M, 5 F	43.92 (11.74)	16.85 (14.94)	3.67 (5.38)	4 TT, 9 CC
Arc width	3 Blind	2 M, 1 F	52.0 (17.78)	28.67 (17.01)	4.7 (7.2)	1 TT, 2 CC
Obstacle 2013	1 1 Sighted, 2 blind	7 M, 6 F	32.64 (12.96)	1.08 (2.63)	0.3 (0.1 1)	7 TT, 6 CC
Obstacle 2014	10 Sighted	2 M, 8 F	29.0 (8.21)	0.22 (0.12)	0.22 (0.12)	4 TT, 6 CC
Bundu 2016	10 Sighted, 2 blind	5 M, 7 F	25.5 (4.12)	1.42 (3.11)	1.59 (3.11)	3 TT, 9 CC
Obstacle 2016	6 Sighted	1 M, 5 F	33.17 (12.42)	0.14 (0.1)	0.14 (0.1)	0 TT, 6 CC

Note. M = mean; SD = standard deviation; TT = touch technique; CC = constant contact.

Method

All studies were designed for drop-off or obstacle detection, with biomechanical data being collected for future analysis. This section will outline the drop-off and obstacle procedures and then describe the procedures relevant to collecting biomechanical data.

Drop-off detection

Six platforms made of wood and covered with carpet (each 2.4 meters long, 1.2 meters wide, and 20 centimeters high [94.5 inches long, 47.2 inches wide, and 7.9 inches high]) formed a walkway of 9.8 meters (32 feet) in length (see Figure 1). The first half of the walkway (4.9 meters or 16 feet) was 1.2 meters (47.2 inches) wide and the second half was 2.4 meters (94.5 inches) wide. Two carpetted plywood boards (0.8 meters long and 1.2 meters wide [31.5 inches long and 47.2 inches wide]) on wooden framing set against the end of the walkway were used to vary the drop-off depth (2.5 centimeters to 17.5 centimeters with 2.5 centimeters increments or 1 inch to 7 inches with 1-inch increment) between trials.

Figure 1.

Drop-off detection walkway.

Each participant had their vision obscured with sleep shades and wore headphones (Radio Shack Full-Size Stereo Headphone 33–1225) connected to an MP3 player (iPod Touch 5th Generation). Participants listened to a steady beat of 90–110 beats per minute played over white noise. The speed of the beats was set to match each participant’s relaxed walking pace. The intent was to help participants walk at a consistent pace in all trials. Since participants were asked to walk as they would typically, the immediate surroundings were not likely to affect cane use. However, it is possible that the recording devices and headphones might have had some unmeasured effect on how participants walked.

Participants began each trial at the center of the walkway, at a randomly chosen starting point between 4.3 meters (14 feet) and 9.1 meters (29 feet, 10 inches) from the drop-off. When the experimenter tapped a participant’s shoulder, the participant walked toward the drop-off using the cane technique prescribed for that trial, following a randomized schedule. Once a participant detected the drop-off, the participant stopped and verbally indicated their detection. An experimenter trailed each participant to provide assistance in case of a participant stumbling. After each trial, the participant was guided to the next starting point through an irregular path to prevent knowledge of the distance between the drop-off and the next starting point. Participants were able to practice before beginning data collection in each condition. Participants completed from 16 to 48 trials in a given experimental condition, depending on the study.

Each trial was labeled a “detection” if the participant stopped and verbally indicated a detection before a portion of either foot went over the edge of the drop-off. Each trial was labeled a “miss” if the participant stepped off the drop-off or would have stepped off the drop-off without intervention from the experimenter. Intervention was required to avoid injury, especially when large drop-offs were used. Interrater reliability across trials was generally about 98%.

Drop-off detection studies used absolute drop-off detection threshold (50%) and drop-off detection rate as measures of participants’ drop-off detection performance. The 50% absolute drop-off detection threshold for each experimental condition was calculated by fitting a cumulative normal distribution curve to the data points to estimate the drop-off depth that was detected in 50% of the trials (Gescheider, 1997). This threshold measure is a standard measure in psychophysical studies to indicate the point between stimulus intensities that are detected by a participant and stimulus intensities that are not detected by a participant. Drop-off detection rate was calculated by dividing total detections by total trials in each experimental condition and drop-off depth. Depending on the study, possible drop-off depths included 2.5 centimeters (1 inch), 7.6 centimeters (3 inches), 12.7 centimeters (5 inches), and 17.8 centimeters (7 inches).

Obstacle detection

Obstacles used in the obstacle detection studies were cylindrical objects made of Styrofoam and linoleum (see Figure 2). Obstacles were constructed in four diameters (5.1 centimeters or 2 inches, 15.2 centimeters or 6 inches, 25.4 centimeters or 10 inches, and 35.6 centimeters or 14 inches) and four heights (2.54 centimeters or 1 inch, 7.62 centimeters or 3 inches, 12.7 centimeters or 5 inches, and 17.8 centimeters or 7 inches) for 16 differently sized obstacles. The smallest obstacle diameter and height reflected the smallest obstacle that might reasonably present a tripping hazard. The largest diameter obstacle was approximately the largest while still maintaining the possibility of not being detected by a cane sweep. The tallest obstacle height was chosen to represent a reasonably tall obstacle (similar to the largest drop-off in the drop-off studies) but not so tall that the cane would always contact the obstacle (such as a pole). Due to time considerations in study design, obstacles of only these four diameters and heights were constructed and not all obstacles were used in all studies. These objects were presented on a participant’s walking path either at the mid-line of the path or slightly off to the side (approximately 15 centimeters [6 inches] off either to the left or right from the midline) following a randomized schedule. A rail of 6.1 meters (20 feet) in length and 91 centimeters (35.8 inches) in height, built with polyvinyl chloride pipes, was placed beside the walking path for participants to trail with the free hand, allowing participants to walk consistently along the midline of the intended walking path (see Figure 3). This measure was designed to ensure that the participant encountered the obstacle at the intended relative position from the sagittal plane of the body. Forcing participants to travel a path with the obstacles somewhat in front of them was used, so that the relative effect of cane techniques and movements could be assessed without the confounding variable of participants veering and missing obstacles because of their veer.

Figure 2.

Range of obstacles used in obstacle detection studies.

Figure 3.

Approach set up for obstacle detection studies.

Participants wore sleep shades and headphones connected to an MP3 player for all trials. Each participant was positioned at the midline of the obstacle detection walking path and properly aligned with the path before the beginning of each trial. Since the participants would not be able to anticipate an obstacle based on the distance walked, the starting distance between the participant and the obstacle was randomly varied from 3 meters (9.8 feet) to 6 meters (19.6 feet). When the experimenter tapped a participant’s shoulder, the participant approached the obstacle on the indicated walking path. Upon detecting the obstacle, the participant stopped immediately and said “obstacle.” After each trial, the participant was guided to the next starting place. A research assistant changed the size and location of the obstacle for the next trial following a randomized schedule. Each participant completed from 12 to 48 trials in a given experimental condition, depending on the study. A trial was labeled as a miss if the participant failed to detect the obstacle with the cane before contacting it with the foot or other part of the body. Interrater reliability across trials was generally about 98%. The obstacle detection rate was calculated by dividing total detections by total trials.

Biomechanical data collection and cleaning

For all trials in all studies, participants had an infrared marker light attached to each shoulder; the elbow, wrist, and index finger of the arm and hand holding the long cane; the lateral midpoint of their torso; the end of each foot; and the top and tip of the long cane. These marker lights were monitored in three dimensions 100 times per second by an Optotrak Certus motion analysis system (Northern Digital, Inc.). The camera unit of the Optotrak system was always positioned, so that participants were walking toward it but beyond the drop-off or obstacle that participants were trying to detect. Figure 4 shows the Optotrak Certus unit.

Figure 4.

Optotrak Certus unit.

Each trial resulted in a matrix of X, Y, and Z coordinates for each marker light, taken 100 times per second for the duration of the trial. Trials varied from 4 to 10 seconds long, depending on the walking speed of the participants. During postprocessing, unreliable data from the start and end of the file were deleted. Only clear data from the middle of a participant’s walk were preserved. Data files were clipped to begin and end when the cane arc was at the left or right extreme. Due to how some participants lifted their feet, moved their arms, or swung the cane in front of their bodies, most trials demonstrated points where certain marker lights were not registered by the camera. This lack of detection was most common for the foot markers, which tended to point toward the floor, as a foot was lifted before taking a step. A computer program was written to approximate the missing values based on the trends and patterns of the surrounding data. The resulting cleaned and complete data files were processed by a second computer program to derive outcome measures (see Table 3).

Table 3.

Biomechanical outcome measures.

Measure	Definition
Hand offset	Average distance from cane hand to middle of body
Arc width	Width of cane arc between left and right extremes
Walking speed	How fast a person walked, on average
Arc center	How centered a person’s arc was compared to the middle of their body
	−∞ = arc entirely to left of person, −1 = centered, 0 = arc totally to right of person
Rhythm	Correspondence of cane arcing and footsteps. Measure potentially ranged from −.5 to ≤ .5 with 0 = in rhythm and .5 perfectly out of rhythm. Positive means cane tap lags closest footfall. Negative means cane tap leads closest footfall
In step	Correspondence of cane arc endpoint (right or left) with trailing foot. Measure potentially ranged from − 1 to ≤ 1
	+ means cane tap lags closest opposite footfall, − means cane tap leads closest opposite footfall
	0 = perfectly in step (i.e., right tap, left foot coincide) +1 = perfectly out of step (i.e., right tap, right foot coincide)
Coverage	Percentage of space participant’s body moved through that was previewed by the cane. Measure potentially ranged from 0 to 100

Results

Due to the very large number of trials, participants, conditions, and measures contained in this data set, this article will only deal with a very brief overview of the data. A future document is planned, which will contain a fuller description and statistical treatment of the data. Means and standard deviations from all 10,069 trials for main outcome measures are shown in Table 4. These data show that when a range of people was sampled using the long cane in a variety of tasks and conditions, they tended to have their cane-holding hand about 14 centimeters to the right of midline, used a cane arc of just under a meter, centered their arcs fairly well, and were generally very poor at being in step or having consistent rhythm. However, given these apparent deficiencies in ideal cane technique, the average percentage of participants’ bodies previewed by the cane was relatively high. Outcome measures, separated out by vision category, gender, and age, are shown in Tables 5 and 6.

Table 4.

Overall biomechanical outcome measures.

Measure	Mean	Standard deviation
Hand offset	14.2 cm (5.6 in.)	5.6 cm (2.2 in.)
Arc width	83.2 cm (32.8 in.)	23.0 cm (9.1 in.)
Walking speed	89.4 cm/s (35.2 in./s)	19.7 cm/s (7.8 in./s)
Arc center	−1.10	1.35
Rhythm	−0.04	0.14
In step	−0.06	0.28
Coverage	85.59%	13.23%

Table 5.

Means and standard deviations for main outcome measures by vision.

	Blind, low vision (n = 4,324)		Sighted (n= 5,745)
Measure	Mean	Standard deviation	Mean	Standard deviation	t (p)
Hand offset	16.6 cm (6.5 in.)	5.3 cm (2.1 in.)	12.4 cm (4.9 in.)	5.1 cm (2.0 in.)	40.09 (<.00l)
Arc width	83.5 cm (32.9 in.)	25.5 cm (10.0 in.)	82.9 cm (32.6 in.)	20.9 cm (8.2 in.)	1.32 (.19)
Walking speed	82.0 cm/s (32.3 in./s)	24.1 cm/s (9.5 in./s)	95.0 cm/s (37.4 in./s)	12.8 cm/s (5.0 in./s)	−32.27 (<.00l)
Arc center	−1.26	1.56	−0.97	1.15	−10.43 (<.00l)
Rhythm	−0.05	0.17	−0.04	0.13	−4.02 (<.00l)
In step	−0.07	0.35	−0.05	0.21	−2.39 (.02)
Coverage	86.20%	13.53%	85.13%	12.98%	3.98 (<.00l)

Table 6.

Means and standard deviations for main outcome measures by gender.

	Male (n= 4,099)		Female (n= 5,970)
Measure	Mean	Standard deviation	Mean	Standard deviation	t (p)
Hand offset	15.0 cm (5.9 in.)	5.5 cm (2.2 in.)	13.6 cm (5.4 in.)	5.6 cm (2.2 in.)	−12.75 (<.00l)
Arc width	83.6 cm (32.9 in.)	23.6 cm (9.3 in.)	82.9 cm (32.6 in.)	22.6 cm (8.9 in.)	−1.61 (.11)
Walking speed	92.2 cm/s (36.3 in./s)	20.9 cm/s (8.2 in./s)	87.5 cm/s (34.4 in./s)	18.5 cm/s (7.3 in./s)	−11.78(<.00l)
Arc center	−1.14	1.32	−1.06	1.38	2.90 (.004)
Rhythm	−0.04	0.16	−0.04	0.13	−0.46 (.65)
In step	−0.08	0.29	−0.05	0.27	4.68(<.00l)
Coverage	86.39%	13.19%	85.04%	13.23%	−5.04(<.00l)

The data and statistical comparisons shown in Table 5 indicate that visually impaired participants were not significantly different from blindfolded, sighted participants on arc width or how well they were in step. However, visually impaired participants did tend to have their cane-holding hand more to the side, walk more slowly, have an arc slightly more off-center, and be slightly better in rhythm with slightly better body coverage.

The data and statistical comparisons shown in Table 6 indicate that men and women were not significantly different from each other on arc width or being in rhythm. However, men did tend to have their cane-holding hand more to the side, have a slightly wider arc, walk faster, have an arc more off-center, and be slightly more off-rhythm but have slightly better body coverage.

Discussion

This data set constitutes the most comprehensive and exacting investigation to date into how people use the long cane when traveling without vision. A full analysis of the data set will elucidate long-standing questions in the field of O&M such as how much having a person’s hand centered affects body coverage or whether a wider arc improves body coverage, perhaps at the expense of rhythm or being in step. These questions will have to remain questions for a little while longer. However, this initial look at the data has demonstrated some interesting findings. Not surprisingly, most people tend to walk with the hand holding the cane not centered laterally on their body. This ranges from about 10–20 centimeters (3.9–7.8 inches). Arc widths were generally slightly wider than a cane user’s shoulder width, and arcs were generally centered (although all participants were expecting to find either a drop-off or an obstacle, which might have artificially improved technique). Participants were also generally poor at being in step or having consistent rhythm. Given that coverage rates averaged about 85%, it remains to be seen whether this level of coverage is sufficient to provide adequate safety for a traveler. Further analyses will investigate what biomechanical components contribute most strongly to body coverage and how each of these biomechanical measures impacts drop-off and obstacle detection.

Limitations

In this initial treatment of a large data set, group comparisons were made that were somewhat artificial. For example, when comparing cane use between blind participants and sighted participants who were blindfolded, both groups contained people using different cane techniques, different types of canes, and performing different tasks. Further analyses will partition groups into more precise subgroups to investigate whether differences identified at this survey level remain when probed more deeply.

As mentioned earlier, all data in this data set reflect people using the long cane to attempt to find an obstacle or identify a drop-off. This task might have artificially improved cane use. However, since most participants completed many trials, often repeating a task for up to an hour, it is unlikely that an artificially high level of skill performance was maintained for the duration. Further analyses will investigate the change in biomechanical measures over the course of a set of trials for each participant to check for practice effects, fatigue, and skill degradation.

Practical implications

Differences found between groups were not surprising (e.g., older participants tended to walk more slowly), but any statistically significant differences reported need to be matched with a clinical perspective. A very small difference in a given measure might be statistically significant in this data set, given the large sample size. There is evidence, even from this initial treatment of the data, that instruction and practice in O&M skills lead to basic skills relatively quickly. Although experienced cane users do perform at a higher level than novices, such as in their ability to detect drop-offs (Kim, Wall Emerson, & Curtis, 2010), this initial analysis indicates the speed with which basic cane skills can be acquired, even if most cane users do not adhere to classic description of proper cane use.