02640414.2011.624537

This article was downloaded by: [University of Edinburgh], [Jacki Thow]
On: 19 December 2011, At: 07:01
Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK
Journal of Sports Sciences
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/rjsp20
Comparison of modes of feedback on glide
performance in swimming
Jacqueline L. Thow
a
, Roozbeh Naemi
b
& Ross H. Sanders
c
a
Centre for Aquatic Research and Education, The University of Edinburgh, Edinburgh, UK
b
Faculty of Health, Staffordshire University, Stoke-on-Trent, UK
c
Department of Institute of Sport, Physical Education and Health Sciences, The University
of Edinburgh, Edinburgh, UK
Available online: 15 Dec 2011
To cite this article: Jacqueline L. Thow, Roozbeh Naemi & Ross H. Sanders (2012): Comparison of modes of feedback on glide
performance in swimming, Journal of Sports Sciences, 30:1, 43-52
To link to this article: http://dx.doi.org/10.1080/02640414.2011.624537
PLEASE SCROLL DOWN FOR ARTICLE
Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to
anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should
be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims,
proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in
connection with or arising out of the use of this material.

Comparison of modes of feedback on glide performance in swimming
JACQUELINE L. THOW1
, ROOZBEH NAEMI2
, & ROSS H. SANDERS3
1
Centre for Aquatic Research and Education, The University of Edinburgh, Edinburgh, UK, 2
Faculty of Health, Staffordshire
University, Stoke-on-Trent, UK and 3
Department of Institute of Sport, Physical Education and Health Sciences,
The University of Edinburgh, Edinburgh, UK
(Accepted 14 September 2011)
Abstract
The software product ‘‘GlideCoach’’ was recently developed to give quantitative and qualitative feedback on the glide
performance of a swimmer (Naemi & Sanders, 2008). This study compared the effect of feedback on glide performance from
GlideCoach with video and verbal feedback. Nineteen elite swimmers were randomly assigned to one of three groups: Group
1 and 2 included six swimmers and Group 3 included seven swimmers. All participants performed ten dives in each of five
sessions. Each group received one of three forms of feedback (video, video and verbal, and GlideCoach and verbal) for four
sessions. In the fifth, retest session, performed 4 weeks after the fourth session, all groups received GlideCoach and verbal
feedback only. This enabled the analysis of GlideCoach and verbal feedback on performance of the groups that had not yet
received this feedback and assessment of the retention ability for the group that had. Feedback resulted in all groups
recording an improvement, as indicated by effect sizes, for average velocity, glide factor (related to resistive drag), and initial
velocity (P 5 0.05). The improvement following the GlideCoach and verbal feedback was greater than that of the two other
feedback methods for all variables of interest (P 5 0.05), with effect sizes ranging from 1.0 to 2.5, compared with values less
than 0.6 for the other feedback methods. We conclude that GlideCoach feedback is effective in improving glide performance.
Keywords: Swimming starts, analysis, feedback, glide, GlideCoach
Introduction
During the 2000 Sydney Olympics, only 0.05 s
separated the swimmers finishing first and third in
the men’s 50 m freestyle event and just 0.5 s
separated all eight finalists in the same event in
2003 (Lyttle & Benjanuvatra, 2007). Thus, elite
swimmers are always striving to refine their techniques
to reduce the total time taken to complete an event
(Hay, 1986–1987). Starting time has been shown to
account for 0.5–11.0% of the total swim time (Vilas-
Boas & Fernandes, 2003), depending on the event and
distance (Hay, 1993; Mason & Cosser, 2000).
Many investigations of the start phase have been
conducted (Kirner, Bock, & Welch, 1989; Kru¨ger,
Wick, Hohmann, El Bahrawi, & Koth, 2003; Miller,
1984; Vilas-Boas, Cruz, Sousa, Conceiçaõ, Fer-
nandes, & Carvalho, 2003; Welcher, Hinrichs, &
George, 1999). The effect of dive start technique
remains unclear, as potential advantages gained
before entry to the water have not been found to
influence the duration of the start phase, commonly
recorded as the time from the gun to 7.5 m. The
longest part of the dive is the glide phase, the time
from entry until underwater kicking commences
(Guimara˜es & Hay, 1985). Guimara˜es and Hay
(1985) showed that the glide phase accounted for
95% of the total variance of start time between
individuals. During the glide, the swimmer adopts
streamlined postures to minimize resistive forces
without actively propelling the body (Naemi, Aritan,
Goodwill, Haake, & Sanders, 2008). Immediately
after entry, swimmers are moving faster than when
stroking in mid-pool. They are also moving faster
than can be sustained by underwater kicking. There-
fore, swimmers glide in a streamlined position
without kicking until kicking becomes beneficial.
In a biomechanical model (Figure 1), Sanders
(2002) highlighted that the prominent variable was
the average horizontal velocity over the glide
distance. Guimara˜es and Hay (1983) reported a
strong relationship (r ¼ 70.84) between average
glide velocity and start time. Average velocity is
dependent on the horizontal velocity at entry and the
Correspondence: R. Sanders, Department of Physical Education, Sport and Leisure Studies, The University of Edinburgh, Holyrood Road, Edinburgh EH8
8AQ, UK. E-mail: r.sanders@ed.ac.uk
Journal of Sports Sciences, January 2012; 30(1): 43–52
ISSN 0264-0414 print/ISSN 1466-447X online Ó 2012 Taylor & Francis
http://dx.doi.org/10.1080/02640414.2011.624537
Downloadedby[UniversityofEdinburgh],[JackiThow]at07:0119December2011

change in velocity during the glide. In turn, the
change in velocity depends on the forces resisting the
forward motion of the swimmer.
The forces that resist a swimmer’s motion are
known collectively as ‘‘resistive drag’’. When there
are no actions designed to produce propulsion, as in
the glide phase, the resistance is termed ‘‘passive
drag’’. Passive drag consists of three components:
frictional drag, form drag, and wave drag (Lyttle,
Blanksby, Elliott, & Lloyd, 2000). The ability to
minimize passive drag and maintain body velocity
over time is known as ‘‘glide efficiency’’ and can be
quantified as a ‘‘glide factor’’ (Naemi & Sanders,
2008). This is influenced by hydrodynamic char-
acteristics of the body, such as height, mass and
shape, as well as depth, glide trajectory, speed, body
position, and pre-glide phases (Lyttle et al., 2000).
The optimal time for the glide phase relies on the
swimmer maximizing average horizontal velocity.
Thus, swimmers strive for a high initial horizontal
velocity at entry and to remain as streamlined as
possible to minimize the loss in horizontal velocity
while gliding underwater. The initial horizontal
velocity is affected by preceding actions performed
during the block and flight phases, such as altering
body positions or the generation of force and velocity
before take-off (Lyttle & Benjanuvatra, 2007).
Variables such as the size of the ‘‘hole’’ at the surface
through which the body passes during entry (Lyttle &
Benjanuvatra, 2007) can influence changes in
velocity following entry. The main aim is for the
swimmers to achieve and maintain high initial entry
velocities (Arellano, Pardillo, De la Fuente, &
Garcia, 2000) by minimizing deceleration through-
out the glide (Hay, 1986–1987).
To perform optimally, swimmers must learn to
maximize their initial velocity at entry and to
minimize their resistance during the glide. Magill
(2004) defines learning as a change in the capability
to perform a skill resulting from a relatively
permanent improvement in performance. A key
element of learning is ‘‘feedback’’, a broad term
describing information about the performance of a
skill that can be received intrinsically or extrinsically
(Hodges & Franks, 2002; Perez, Llona, Brizuela, &
Encornacioń, 2009). Everyone experiences intrinsic
feedback received from our senses while performing
skills. However, it is believed that athletes learn and
perform skills better when they receive extrinsic
(augmented) feedback (Hodges, Chua, & Franks,
2003). There is always some sort of feedback, so it is
difficult to assess the relative contributions of the
practice itself and the particular type of feedback
received. However, the latter can be assessed by
performing controlled studies in which the amount
of practice and volume of feedback are kept constant
but the type of feedback is different across matched
groups.
In swimming, extrinsic feedback depends on
coaches providing information because intrinsic
Figure 1. Biomechanical model of the glide adapted from Sanders (2002).
44 J. L. Thow et al.

feedback does not always relate to external evaluation
(Perez et al., 2009). Extrinsic feedback has mainly
been verbally supplied by a coach on poolside due to
the constraints of environmental conditions in
swimming (Perez et al., 2009). With technological
advances, additional methods can be used, particu-
larly video feedback (Rucci & Tomparowski, 2010).
Research has shown that both video and verbal
feedback alone are more effective than intrinsic
feedback. More recently, studies have analysed the
effects of combining verbal attention cues as well as
video information on error correction strategies to
improve performance, but results are unclear (Rucci
& Tomparowski, 2010).
The content involved in providing extrinsic feed-
back can also influence its effectiveness in learning
skills. This can be separated into two categories:
knowledge of results (information about the outcome
of a performance) and knowledge of performance
(information about the movement characteristics of a
skill) (Magill, 2004). Both forms are thought to
improve skill learning more than using intrinsic
feedback (Rucci & Tomparowski, 2010). It is
believed that different feedback types are suited for
specific skills or stages of learning depending on the
desired outcome (Tzetzis & Votsis, 2006). However,
equivocal results suggest that the most effective
method of supplying feedback remains unknown
(Hodges & Franks, 2002; Tzetsis & Votsis, 2006).
Naemi et al. (2008) developed user-friendly soft-
ware called GlideCoach to provide immediate feed-
back on glide performance for use by coaches and
athletes during aspects of swimming performance.
The feedback via GlideCoach includes information
about postures, initial velocity, glide factor, and
average velocity, with comparisons of those values for
previous attempts, as well as displays of velocity–time
graphs and video replay of performances (Figure 2).
The glide factor is a measure of the ability to
maintain velocity over the glide (Naemi et al.,
2008). It is obtained by fitting a logarithmic mathe-
matical function, derived from an equation of motion
of the body during a horizontal rectilinear glide, to
the digitized displacement of the hip during the glide
(Naemi et al., 2008). The use of automatic tracking
during digitization allows the data to be processed
quickly while maintaining accurate information. This
allows quicker provision of feedback on performance
compared with other manual digitization processes
(Naemi et al., 2008). This enables GlideCoach to
provide both forms of augmented feedback: knowl-
edge of results and knowledge of performance
(Magill, 2004).
The effectiveness of GlideCoach in providing
feedback on glide performance to swimmers has
not been compared with other forms of feedback.
Thus, in this study we compared the effectiveness of
feedback of the data output by GlideCoach with
video and verbal feedback. It was hypothesized that
feedback via GlideCoach would yield greater im-
provements in the glide phase of a swimming start
than either viewing video replays of the glide phase
alone or a combination of video and verbal feedback.
Methods
Participants
Nineteen elite swimmers (age 19.5 + 2.6 years, mass
71.7 + 12.4 kg, height 179 + 10.6 cm) volunteered
to participate in this study. All participants were in
full training from the same club, under the same
general conditions, and competing at national and/or
international standard. Each individual completed an
information data form and provided informed con-
sent. The study received approval from the Uni-
versity of Edinburgh Ethics Committee.
Experimental design
In this study, the amount of practice and the volume
of feedback were controlled by keeping these
constant across three groups. The total duration of
feedback was equivalent across the groups. Partici-
pants were aware that the glide phase was being
analysed but were not aware of the different feedback
protocols. Participants in Group 1 individually
viewed a video replay of their glide performances
using a media player, during which no other feed-
back was provided. Group 2 received similar video
feedback combined with verbal feedback about the
glide performance from a coach. Group 3 received
both video and verbal feedback from a coach
together with quantification of glide performance
variables using the GlideCoach software. The video
feedback in Group 3 used the replay software
incorporated within GlideCoach. For Group 2 and
Group 3, verbal feedback was provided in relation to
the video recordings of glide performances. This was
a summary of each session, provided to each
participant individually and immediately before the
next session, using the relevant feedback method, to
reinforce the development of glide skills. It consisted
of a short list of positive and negative points
identifying correct or incorrect technical actions,
determined by applying biomechanical principles
and coaching methods. Participants were then given
three technique points used to guide and improve
aspects of their glide performance and overall dive.
This prevented overloading swimmers with too
much information. The same coach provided all
verbal feedback throughout the study and was
unfamiliar with the participants to remove bias and
subjectivity, ensure consistency in feedback, and
Feedback on glide performance 45

prevent experimenter bias. Immediately before ses-
sion 5, all participants received feedback of session 4
using GlideCoach as well as a review of the feedback
given during sessions 1–4.
Data were collected over an 8 week period. In the
first week (Week 1), swimmers were separated into
three groups (Group 1, n ¼ 6; and Group 2, n ¼ 6;
Group 3, n ¼ 7), based upon starting performance
times. Starting performance time was defined as the
time from the starting signal to the tips of the fingers
passing a marker placed 7 m from the starting wall.
Each dive start followed standard FINA (2009)
competition regulations and swimmers maintained
the glide position to the 7 m marker. Swimmers were
ranked on the basis of the average performance times
for six dive trials, timed using a stopwatch.
Each individual performed a total of 50 dives,
separated into five sessions (session 1 to session 5) of
10 dives each. To prevent sources of participant
error, the first four sessions (sessions 1–4) were
performed in a single week by each of the groups.
For example, Group 1 completed sessions 1–4 in
week 2, Group 2 performed them in week 3, and
Group 3 in week 4. After completing sessions 1–4,
each participant had a 4 week break during which no
dive training was performed, before performing the
final retest session, session 5. That is, Group 1
completed session 5 in week 6, Group 2 in week 7,
and Group 3 in week 8. Magill (2004) indicated that
4 weeks was a sufficient time to determine whether
the initial skills learned were retained or forgotten
using GlideCoach by Group 3. Retest session 5 was
completed using the same protocols as sessions 1–4.
A pilot study analysing various combinations of dive
repetitions and rest durations was conducted prior to
the investigation. This ensured the protocol was
time-efficient and did not induce a fatigued state, as
indicated by low heart rates and perceived exertion
measures, allowing participants to focus upon their
glide performances.
Data collection protocol and recording
Before performing, body markers were placed on five
anatomical joint centres on the left side of the body,
using black marker pen: wrist (styloid process of
ulna), shoulder (head of humerus), hip (greater
trochanter of femur), knee (lateral epicondyle of
femur), and ankle (lateral malleolus of fibula).
Participants completed a warm-up swim of 200–
400 m before performing ten dives at maximum
effort. A rest of 45–60 s between dives was permitted
to minimize the effect of fatigue, as established
during the pilot study.
Figure 2. Image of the data GlideCoach software can provide for feedback (Permission from the Centre of Aquatics Research Centre).

Data were recorded using one camera (JVC KY-
F32 colour camera), recording at a sampling rate of
50 Hz and positioned 10 m perpendicular to the
participant’s glide path underwater. The camera
encompassed a field view of 9 m horizontally, there-
by ensuring at least 1.5 s of glide performance was
available for analysis (Naemi et al., 2008). A second
camera was located 5 m out of the water directly
above the pool lane and swimmers’ glide trajectories.
Its sole purpose was to ensure participants’ glide path
did not deviate from the line used to calibrate their
speed. The T-line at the bottom of the pool and the
wall from which participants started allowed nine
reference points with known coordinates to be
identified and digitized for calibrating the Glide-
Coach system. The live video for both cameras was
displayed on a poolside monitor to allow clear
analysis of the dive performances. The video data
were stored as AVI files.
Data analysis
The video files for all individuals were analysed to
obtain quantitative and qualitative variables by using
GlideCoach. Each AVI file was first trimmed using
the software to commence when a participant’s body
was completely underwater and reached a horizontal
position, decreasing the effect of bubbles on the
accuracy of digitization. The software allows auto-
matic tracking with the capacity to digitize lost points
manually with accuracy assured by using a zooming
function. This was enhanced by the clear view of
joints and joint markers. The digitized coordinates
were converted to displacement data using a
standard two-dimensional direct linear transforma-
tion algorithm incorporated within the GlideCoach
software, in conjunction with the digitized reference
points on the calibration line. The software output
values of glide performance parameters included the
dependent variables to be analysed: initial and
average velocity and glide factor.
Statistical analysis
The three dependent variables and performance
times were obtained for each individual over all 50
dives. The raw scores from each participant were
summarized for each session (ten dives) into means
and standard deviations and presented as descriptive
statistics. In addition, effect sizes of individuals
between the test sessions were used as measures of
meaningful changes and deemed suitable as
they normalized for differences in magnitude and
within-participant variability of the scores (Coe,
2000).
The effect sizes were calculated for each partici-
pant using means and pooled standard deviations of
each individual’s scores from each session, according
to equations (1) and (2) (Coe, 2000):
effect size ¼ M1 À M2ð Þ=SDpooled; ð1Þ
where M1 and M2 represent each individual’s mean
raw score for a session, and
SDpooled ¼
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
n1 À 1ð Þs2
1 þ n2 À 1ð Þs2
2
n1 þ n2À2
;
s
ð2Þ
where n1 and n2 are the number of dives in the
sessions being compared (i.e. 10) and s1 and s2 are
the standard deviations of a swimmer’s score in the
sessions being compared. These values were then
averaged for each group, per session and used to
determine group effect sizes for session 2, session 3,
session 4, and session 5 relative to session 1. This
enabled a measure of cumulative learning across the
feedback sessions to be obtained. Effect sizes were
also calculated for retest session 5 relative to session
4. This provided an indication of the effect of
GlideCoach feedback compared with the effect of
feedback given for Group 1 and Group 2 in sessions
1–4, and as a measure of retention for Group 3. The
effect sizes were interpreted according to Hopkins
(2002).
Although the amount of change due to practice
and the different types of feedback was quantified, it
was also of interest to determine whether differences
between groups at equivalent stages of practice and
feedback were significant. To analyse statistically
significant differences within groups and between
sessions, paired t-tests were conducted. To compare
differences between each pair of groups for each
session, independent t-tests were conducted. Ses-
sions were compared from session 2 to session 5
(changes relative to session 1) and for retest session 5
relative to session 4. To test for significant differ-
ences within groups and between session 4 and
session 5, paired t-tests were used. This enabled the
statistical assessment of the effect of GlideCoach
combined with verbal feedback for those who had
not yet received it (i.e. Group 1 and Group 2), and to
assess the retention skills for Group 3. Each t-test
was conducted using individual mean raw scores
and also individual mean effect sizes as sources of
input.
Given that the study focused on specific mean-
ingful comparisons to develop an overall ‘‘picture’’,
rather than relying on single significant results to
support the hypothesis, a P-value of 0.05 was applied
for all comparisons without correction for the
number of comparisons. This minimized the risk of
making both Type I and Type II errors and ensured

that the power of the study was maintained. All
statistics were performed using the Microsoft Excel
package.
Results
Results are reported for the three dependent
variables of average velocity, glide factor, and initial
velocity in terms of: the change between sessions,
across feedback groups, and the comparison of the
magnitude of changes between session 4 and
session 5.
Between sessions and groups
Effect sizes as well as means and standard deviations
of raw data for all groups, in each session, are shown
in Table I.
Overall, the results highlighted an improvement in
average velocity, glide factor, and initial velocity.
Figure 3 shows the average velocity (a) and the
magnitude of change relative to session 1 as
indicated by effect sizes (b) across sessions 1–5.
All groups improved average velocity in session 5
relative to session 1, with large effect sizes. However,
only Group 3 had improved average velocity relative
to session 1 and prior to session 5, with a large effect
size (1.15) in session 4. In contrast, the differences of
Group 1 and Group 2 from session 1 to session 4
were not significant (P 4 0.05). Average velocity was
significantly different between Group 1 and Group 2
during session 5 (P 50.05).
Figure 4 shows the glide factor (a) and the
magnitude of change relative to session 1 (b) across
sessions 1–5. All groups showed large improvements
by session 5. However, for Group 1 and Group 2,
most of the improvements in glide factor occurred in
session 5 following the feedback using GlideCoach.
In contrast, Group 3 improved steadily and showed
modest changes in glide factor by session 2 and large,
significant changes (P 50.05) by session 4. The large
improvements in Group 3 by session 4 were signi-
ficantly different from those of Group 2 (P 50.05).
However, while Group 3 had larger improvements in
glide factor than Group 1 by session 4, the difference
was not significant (P 4 0.05). In terms of raw glide
factor data, all groups showed significant differences
between session 5 and the pre-test, session 1
(P 50.05).
Figure 5 shows the initial velocity (a) and the
magnitude of change relative to session 1 (b) across
sessions 1–5. All groups recorded large increases in
initial velocity by session 5 and most of that
improvement occurred during session 5 for all
groups.
Comparison of the fourth and final session
Effect sizes, including means and standard deviations
of group raw data for session 4 and session 5, the
Table I. Mean (+ standard deviation) of the raw data and calculated effect sizes for each variable during each session (relative to the pre-test:
session 1).
Group Session 1 Session 2 Session 3 Session 4 Session 5
Average velocity
1 mean + s 1.73 + 0.17 1.67 + 0.23 1.72 + 0.18 1.71 + 0.15 1.85 + 0.16*
mean effect size 0 70.8 0.79 70.03 2.28*
2 mean + s 1.73 + 0.23 1.75 + 0.2 1.69 + 0.2 1.70 + 0.17 1.81 + 0.18
mean effect size 0 0.16 70.53 0.05 1.6*
3 mean + s 1.74 + 0.16 1.76 + 0.13 1.72 + 0.15 1.77 + 0.15 1.84 + 0.09*
mean effect size 0 0.1 70.73 1.15 1.01
Glide factor
1 mean + s 4.29 + 1.9 4.17 + 1.25 4.58 + 1.05 4.57 + 0.67 5.18 + 0.79*
mean effect size 0 70.73 0.25 0.63 2.2*
2 mean + s 4.51 + 0.76 4.45 + 0.67 4.48 + 0.76 4.41 + 0.65 4.89 + 0.71*
mean effect size 0 70.02 70.12 70.35 1.01*
3 mean + s 4.63 + 0.53 4.90 + 0.65 4.88 + 0.58 5.21 + 0.68* 5.24 + 0.85*
mean effect size 0 0.63 0.41 1.01* 1.52*
Initial velocity
1 mean + s 2.42 + 0.21 2.33 + 0.28 2.34 + 0.26 2.34 + 0.25 2.48 + 0.25
mean effect size 0 70.46 0.21 70.01 1.09*
2 mean + s 2.35 + 0.31 2.38 + 0.26 2.29 + 0.23 2.35 + 0.21 2.50 + 0.28
mean effect size 0 0.07 70.53 0.24 1.43*
3 mean + s 2.34 + 0.22 2.36 + 0.15 2.30 + 0.22 2.34 + 0.25 2.45 + 0.15
mean effect size 0 0.16 70.73 0.64 1.13*
*P 50.05.

difference between each session, and the P-values
obtained from the repeated measures t-tests are
shown in Table II.
Figure 3 shows the increased change in average
velocity, from session 4 to session 5, by Group 1
and Group 2 (P 50.05). The greatest differences
in changes were for Group 1 between session 4
and session 5. Significant increases in glide factor
for Group 1 and Group 2 occurred between
session 4 and session 5 (P 50.05). Initial velocity
for Group 1 and Group 2 changed significantly
between session 4 and session 5 (P 50.05).
Although Group 3 improved initial velocity be-
tween session 4 and session 5, it was not
significant (P 4 0.05).
The improvements in gliding ability by Group 1
and Group 2 are emphasized by the raw data scores
for session 4 and session 5 (Table II). Although there
was no significant increase in average velocity, glide
factor or initial velocity between session 4 and
session 5 for Group 3, the raw data scores tended
to indicate a small improvement in performance for
each variable building on the high levels of perfor-
mance achieved up to session 4.
Figure 3. (a) Mean average velocity results of each group for each session. (b) The magnitude of changes of average velocity, indicated by
effect sizes relative to session 1.
Figure 4. (a) Mean glide factor results of each group for each session. (b) The magnitude of changes of glide factor, indicated by effect sizes
relative to session 1.

Discussion
The main aim of this investigation was to compare
the effectiveness of three different types of feedback
on variables related to glide performance in swim-
ming starts. The three types of feedback were: self-
observation by swimmers of videos of their glide
performances (Group 1); observation of video
combined with verbal feedback by a coach (Group
2); and both video and verbal feedback by a coach
together with a quantification of average velocity,
glide factor, and initial velocity using newly devel-
oped GlideCoach software (Group 3).
The overall results of analysis between sessions
and groups highlighted improved average velocities
and glide factors in all groups. The largest improve-
ments for each group and variable occurred after the
GlideCoach and verbal feedback intervention. In
fact, a significant change by Group 1 and Group 2
was evident only in session 5, the first and only
Figure 5. (a) Mean initial velocity results of each group for each session. (b) The magnitude of changes of initial velocity, indicated by effect
sizes relative to session 1.
Table II. Results of paired t-tests assessing changes due to each feedback method compared with GlideCoach for each group between session
4 and session 5.
Group Session 4 Session 5 Difference Paired t-test
Average velocity
1 mean + s 1.71 + 0.15 1.85 + 0.16 0.14 0.01*
mean effect size 70.03 2.28 2.31 0.01*
2 mean + s 1.70 + 0.17 1.81 + 0.18 0.11 0.08
mean effect size 0.05 1.6 1.55 0.04*
3 mean + s 1.77 + 0.15 1.84 + 0.09 0.07 0.42
mean effect size 1.15 1.01 70.14 0.03*
Glide factor
1 mean + s 4.57 + 0.67 5.18 + 0.79 0.61 0.03*
mean effect size 0.63 2.2 1.57 0.02*
2 mean + s 4.41 + 0.65 4.89 + 0.71 0.48 0.02*
mean effect size 70.35 1.01 0.36 0.02*
3 mean + s 5.21 + 0.68 5.24 + 0.85 0.03 0.13
mean effect size 1.01 1.52 0.21 0.43
Initial velocity
1 mean + s 2.34 + 0.25 2.48 + 0.25 0.14 0.04*
mean effect size 70.01 1.09 1.1 0.02*
2 mean + s 2.35 + 0.21 2.50 + 0.28 0.15 0.06
mean effect size 0.24 1.43 1.19 0.01*
3 mean + s 2.34 + 0.25 2.45 + 0.15 0.11 0.29
mean effect size 0.64 1.13 0.53 0.16
*P 50.05.

session these groups received GlideCoach and verbal
feedback (P 50.05). Group 3 continued to show
improved performance and retention of skills in
session 5. In the present study, session 5 was
performed after 4 weeks without further dive practice
or intervention following session 4. This proved to be
a sufficient time to determine whether the use of
Glidecoach could elicit a further improvement in
glide performance compared with standard modes of
feedback or retain previously developed skills
(Magill, 2004). These results provide strong evi-
dence that GlideCoach is a valuable feedback tool in
the initial learning, retention, and application of
gliding skills compared with other feedback methods.
The three dependent variables studied (average
velocity, glide factor, and initial velocity) have been
identified as the main variables affecting glide
performance (Sanders, 2002). Initial velocity de-
pends upon pre-glide actions such as body position
and angles, and momentum (Lyttle & Benjanuvatra,
2007). Thus by considering these three variables, the
combined effect of practice and feedback on the glide
phase could be assessed independently of the effect
on overall start performance.
To improve overall glide performance, swimmers
strive to reduce resistive drag so that velocity is
maintained. The glide factor output from Glide-
Coach is a sensitive measure of that ability.
GlideCoach provides visual information as well as
data on the positioning of body segments to improve
overall streamlining (Lyttle & Benjanuvatra, 2007;
Sanders, 2002). Improvement in posture through the
use of GlideCoach and verbal feedback was reflected
in a higher average velocity and by higher values of
glide factor. This was evident in session 5 where the
average velocity and glide factor of Group 1 and
Group 2 increased significantly following Glide-
Coach and verbal feedback. Further research is
required to ascertain the effect of specific postural
changes on glide performance.
In the present study, a retention test was used to
reduce the possibility of misinterpreting improved
performance (Magill, 2004). A month without
further practice and intervention following session
4 did not cause a reduction in performance of Group
3 and did not prevent Group 1 and Group 2 from
making very rapid gains in performance following
feedback from GlideCoach. These findings empha-
size that the changes were not due to continuous
practice of the dive. Rather, the evidence strongly
suggests that GlideCoach offered the swimmers
feedback that was effective in improving their
performance.
GlideCoach and verbal feedback may have im-
proved glide and dive performance for several
reasons. First, this method combined verbal and
video information, provided by a coach and the
software’s user-friendly graphic and visual images,
allowing participants to evaluate and correct perfor-
mance errors using this information (Guadagnoli,
Holcomb, & Davis, 2002; Rucci & Tomparowski,
2010). Second, the video software enabled partici-
pants to visually compare incorrect and correct glide
trials and identify the differences, a process shown to
be better than normal practice methods (Boyer,
Miltenberger, Batsche, & Fogel, 2009). Third,
GlideCoach provided information on knowledge of
results and performance-related kinematic variables,
particularly average velocity and glide factor, focus-
ing the participant’s attention on internal limb
movements and the external effects of actions
(Naemi et al., 2008). Research has shown that
focusing individuals’ attention on the external effects
of actions greatly improves skill development
(Hodges & Franks, 2002; Wulf & Prinz, 2001).
These factors and the dependence of feedback
effectiveness on a number of factors, including the
individual’s level of knowledge and skill, may explain
why Group 1 and Group 2 did not significantly
improve their glide performance from session 1 to
session 4 (Hodges & Franks, 2002). Also, the
variation in results may be due to differing adapta-
tion rates to the feedback and practice as participants
established a more effective technical glide position,
despite being elite athletes (Guadagnoli et al., 2002).
However, more work is necessary to gain insights
into the most effective combinations of feedback
(Hodges & Franks, 2002).
Although the present study was designed to enable
assessment of the effect of three types of feedback on
learning to optimize the glide phase, there was no
‘‘no-feedback’’ condition. Thus, the relative con-
tribution of practice to learning could not be assessed
completely. Future work could include a control
group to quantify the relative effects of practice and
learning.
In conclusion, this study provides strong evidence
that GlideCoach and verbal feedback can improve
glide performance. Also there is evidence that the
skills attained with feedback using GlideCoach
and verbal advice are retained after an absence of
diving trials for 4 weeks. GlideCoach proved more
effective in improving glide performance than other
standard modes of feedback commonly used by
coaches.
References
Arellano, R., Pardillo, S., De La Fuente, B., & Garcia, F. (2000).
A system to improve the swimming start technique using force
recording, timing and kinematic analysis. In Y. Hong & D. P.
Johns (Eds.), Proceedings of the XVIII Symposium of the
International Society of Biomechanics in Sports (pp. 609–613).
Hong Kong: The Chinese University of Hong Kong.

Boyer, E., Miltenberger, R., Batsche, C., & Fogel, V. (2009).
Video modelling by experts with video feedback to enhance
gymnastic skills. Journal of Applied Behaviour Analysis, 42, 855–
860.
Coe, R. (2000). What is an ‘‘effect size’’? A guide for users. Available
at: http://www.cemcentre.org/ebeuk/research/effectsize/ESguide.
htm (accessed 20 February 2009).
FINA (2009). Directory of rules and regulations. Available at: www.
fina.org (accessed 1 April 2009).
Guadagnoli, M., Holcomb, W., & Davis, M. (2002). The efficacy
of video feedback for learning the golf swing. Journal of Sports
Sciences, 20, 615–622.
Guimara˜es, A., & Hay, J. (1983). A quantitative look at swimming
biomechanics. Swimming Technique, August/October, pp. 11–
17.
Guimara˜es, A., & Hay, J. (1985). A mechanical analysis of the grab
starting technique in swimming. International Journal of Sport
Biomechanics, 1, 25–35.
Hay, J. (1986–1987). Swimming biomechanics: A brief review.
Swimming Technique, November/January, pp. 15–21.
Hay, J. (1993). The biomechanics of sports techniques (4th edn. pp.
347–356). Englewood Cliffs, NJ: Prentice-Hall.
Hodges, N., Chua, R., & Franks, I. (2003). The role of video in
facilitating perception and action of a novel coordinated
movement. Journal of Motor Behaviour, 35, 247–260.
Hodges, N., & Franks, I. (2002). Modelling coaching practice:
The role of instruction and demonstration. Journal of Sports
Sciences, 20, 793–811.
Hopkins, W. (2002). A new view of statistics. Available at: www.
sportsci.org/resource/stats/index.html-2k (accessed 10 March
2009).
Kirner, K., Bock, M., & Welch, J. (1989). A comparison of four
different start combinations. Journal of Swimming Research, 5
(2), 5–11.
Kru¨ger, T., Wick, D., Hohmann, A., El-Bahrawi, M., & Koth, A.
(2003). Biomechanics of the grab and track start technique. In
J. C. Chatards (Ed.), Biomechanics and medicine in swimming IX:
Proceedings of the IX International Symposium on Biomechanics
and Medicine in Swimming (pp. 219–223). Saint-Etienne,
France: University of Saint-Etienne.
Lyttle, A., & Benjanuvatra, N. (2007). Start right: A biomechanical
review of dive start performance. Available at: http://coachesinfo.
com (accessed 12 November, 2008).
Lyttle, A., Blanskby, B., Elliott, B., & Lloyd, D. (2000). Net forces
during tethered simulation of underwater streamlined gliding
and kicking techniques of the freestyle turn. Journal of Sports
Sciences, 18, 801–807.
Magill, R. (2004). Motor learning and control: Concepts and
applications (7th edn.). New York: McGraw-Hill.
Mason, B., & Cosser, J. (2000). What can we learn from competi-
tion analysis at the 1999 Pan Pacific Swimming Championships?
In R. H. Sanders & Y. Hong (Eds.), Proceedings of XVIII
International Symposium on Biomechanics in Sports, Applied
Program: Application of Biomechanical Study in Swimming (pp.
75–82). Hong Kong: The Chinese University of Hong Kong.
Miller, J. (1984). Starting techniques of elite swimmers. Journal of
Sports Sciences, 2, 213–223.
Naemi, R., Aritan, S., Goodwill, S., Haake, S., & Sanders, R.
(2008). Development of immediate feedback software for
optimising glide performance and time of initiating post-glide
actions. In M. Estivalet & P. Brisson (Eds.), The engineering of
sport 7 (pp. 291–300). Berlin: Springer.
Naemi, R., & Sanders, R. (2008). A ‘‘hydro-kinematic’’ method of
measuring glide efficiency of a human swimmer. Journal of
Biomechanical Engineering, 130 (6), 9–16.
Perez, P., Llona, S., Brizuela, G., & Encornacioń, A. (2009).
Effects of three feedback conditions on aerobic swim speeds.
Journal of Sports Science and Medicine, 8, 30–36.
Rucci, J., & Tomparowski, P. (2010). Three types of kinematic
feedback and the execution of the hang power clean. Journal of
Strength and Conditioning Research, 24, 771–778.
Sanders, R. H. (2002). New analysis procedures for giving feed-
back to swimming coaches and swimmers. In K. Gianikelis, B.
R. Mason, H. M. Toussaint, R. Arellano, & R. H. Sanders
(Eds.), Scientific Proceedings of XX International Symposium on
Biomechanics in Sports, Applied Program: Swimming (pp. 1–14).
Caceres, Spain: University of Extremadura.
Tzetzis, G., & Votsis, E. (2006). Three feedback methods in
acquisition and retention of badminton skills. Perceptual and
Motor Skills, 102, 275–284.
Vilas-Boas, J. P., Cruz, J., Sousa, F., Conceiçaõ, F., Fernandes,
R., & Carvalho, J. (2003). Biomechanical analysis of ventral
swimming starts: Comparison of the grab start with two track-
start techniques. In J. C. Chatards (Ed.), Biomechanics and
medicine in swimming IX: Proceedings of the IX International
Symposium on Biomechanics and Medicine in Swimming (pp. 249–
253). Saint-Etienne, France: University of Saint-Etienne.
Vilas-Boas, J., & Fernandes, R. (2003). Swimming starts and
turns: Determining factors of swimming performance. In P.
Pelayo (Ed.), Actes des 3e`mes journeés spećialiseés de natation.
Lille: Editions Publibook.
Welcher, R., Hinrichs, R., & George, T. (1999). An analysis of
velocity and time characteristics of three starts in competitive
swimming. In Proceedings of XVII Congress of the International
Society of Biomechanics, Calgary, Canada.
Wulf, G., & Prinz, W. (2001). Directing attention to movement
effects enhances learning: A review. Psychonomic Bulletic and
Review, 8, 648–660.

02640414.2011.624537

Recommended

Recommended

More Related Content

What's hot

What's hot (19)

Viewers also liked

Viewers also liked (20)

Similar to 02640414.2011.624537

Similar to 02640414.2011.624537 (20)

02640414.2011.624537