Powerlifting at Junior Level
Selection Paradigm
Łukasz Płóciennik and Igor Ryguła
Department of Statistics, Academy of Physical Education and Sport, Górskiego Street 1, Gdańsk, Poland
Keywords: Powerlifting, Selection Model, Discriminant Analysis, Classification Functions.
Abstract: The variability of the sport result obtained in powerlifting (PL) causes a few profound problems within
coaching practice. One of them is the issue that concerns assigning individuals to particular training group
of fitness level. Simply, this process is called selection. Since PL does not have any scientific-based
selection algorithm we reckon, it is necessary to project it, with the idea to rationale the procedure. Thus the
aims of the study were to construct discriminant and classification functions. A group of thirty-two
powerlifters was selected for the investigation (22,397 yr ± 0,826). The average sport result was 331,449 ±
41,959 Wilks Points. Observation method and diagnostic survey were used to collect the data. During the
course of the multidimensional statistical analysis, Hellwig’s algorithm, multiple regression, and
discriminant analysis were utilised. The distances between stratified subdivisions of athletes were
maintained in 99%. The classification matrix of young powerlifting contestants indicates that all the athletes
were grouped adequately. Finally, for junior age category in PL, classification functions assign individuals
to specified subgroups statistically better than a priori rule.
1 INTRODUCTION
One of the most important aspects in professional
sport is the selection process. Due to methodological
progress in sports science, it is important to use
multidimensional techniques of data exploration
alongside the issue of talent identification and thus
selection. This kind of statistical analysis has been
presented by Ryguła (2003) and Maszczyk (2008).
In powerlifting (PL) the essential components of
sport mastery in all PL events (squat, bench press,
deadlift) were revealed broadly (Mayhew et al.,
1993); (Keogh et al., 2005); (Winwood, 2011). The
extension of cited research is the dilemma of
powerlifters selection.
The literature points out discriminant analysis
(DA) as a one of the most suitable analytical
methods in solving the problems concerning talent
identification and the selection process in sport.
The applicability of AD was exposed earlier on
the basis of many disciplines (Ryguła, 2003;
Magiera and Ryguła, 2007; Saavedra et al., 2010). It
is very true that it holds a privileged position in
identifying some key features of sports performance,
especially when its distribution is diversified among
athletes. Moreover, DA is suitable for prediction
group membership of a given individual (sports
selection) as well as to examine the structure of sport
result across a few homogenous divisions – classes
(Ryguła, 2003); (Magiera and Ryguła, 2007).
2 METHODOLOGY
2.1 Methods, Aims and Hypotheses
In the paper, observation method and diagnostic
survey were used. Several measurements and
assessments techniques of competitors’ personality
characteristics were implemented to gather the data.
The aims of the research were to construct
discriminant and classification functions for
homogenous groups. The goals implicate following
questions: (1) Which of the predictors will form the
optimum set of discriminate variables that
distinguish young powerlifters? (2) What will the
value of cumulative proportion of discriminate
functions be? (3) Will classification functions
identify powerlifters statistically better in
comparison with chance accuracy algorithm? These
questions concern two hypotheses:
116
Płóciennik Ł. and Ryguła I..
Powerlifting at Junior Level - Selection Paradigm.
DOI: 10.5220/0004605601160123
In Proceedings of the International Congress on Sports Science Research and Technology Support (icSPORTS-2013), pages 116-123
ISBN: 978-989-8565-79-2
Copyright
c
2013 SCITEPRESS (Science and Technology Publications, Lda.)
H1: Λ≠ 0;
H2: Press’s Q > χ
2
(α; df-1)
2.2 Participants
Thirty-two powerlifters participated in the research.
All subjects answered the powerlifting history
questionnaire and signed a consent form before
participation. The main precondition for
involvement in the study were: at least 4 years’
training experience in powerlifting drill, a positive
medical examination as well as a adequate level of
general and specific physical fitness.
The essential number of individuals was
established by the procedure proposed by Greń
(1976). The study was approved by the Bioethical
Committee for Scientific Research at the Regional
Medical Chamber (reference number KB - 102/11).
2.3 Investigation Procedure
The study protocol consisted of seven test and seven
retest days divided into two areas: general and
specific. During the first day (meeting),
anthropometric measurements were made. In the
course of the two consecutive days a general fitness
test (EUROFIT) and tests measuring the maximum
power of the upper limbs and the whole body were
examined. On the fourth day the efficiency of the
cardiovascular system, the reaction time
measurements and a psychological test (NEO-FFI)
were executed. All of the aforementioned procedures
were included in the general part of the diagnosis
and between each meeting an interval of 24 hours
was set. A retest was carried out immediately after a
two day break after the last test in the general
examination. Subsequently, with an interval of 48
hours, the second session of tests (powerlifting
specific) was carried out. The sport result was
assessed firstly. Next, after three days, specific
speed was tested, and after a further two days,
specific endurance was assessed. As in the case of
the general part of the examination, after collecting
the data from the second block of tests (powerlifting
specific), with a 48 hour break, a retest was
performed.
Measurements were taken during the transition
phase of the annual training schedule, in the
afternoon (3 PM), except for anthropometric
measurements, which were performed in the
morning, before breakfast. Each test was
accompanied by a standard warm-up, along with a
movement explanation and its demonstration.
2.4 Measurements and Variables
Independent variables were obtained by measuring
different athletes’ characteristics in the areas
outlined below. Their detailed descriptions have
been documented in doctoral thesis of Płóciennik
(2012). All subjects undertook a comprehensive set
of test, which include assessment in the following
domains:
Anthropometric Dimensions. In order to obtain the
structural status data of the powerlifters, research
was performed by the same person using the tools
recommended by the International Society for the
Advancement of Kinanthropometry (ISAK) and by
applying the assumptions of sport anthropometry
(Drozdowski, 1998). Particularly, the height was
measured with a portable stadiometer (Model 214,
Seca Corp., Hanover, MD, USA) and weight was
measured with Tanita scales (model BC-418, Tanita
Corp, Tokyo, Japan). During skinfold thickness
examination, a Harpenden caliper (Gima, Milan,
Italy) was used. In measuring muscle circumferences
we utilised a fibreglass tape. Other features of the
body structure, such as skeletal dimensions – bone
breadths, width or lengths were determined with a
small anthropometer.
The obtained results, according to formulas
proposed in the literature (Drozdowski, 1998);
(Mahyew et al., 1993); (Shephard, 1991); (Watson et
al., 1980), were used to determine the components of
body mass (adipose and muscle tissue) in total as
well as in percentage values. Also basic
anthropometric indices and silhouette proportions
were computed. Namely, trunk length to stature
ratio, upper to lower limb length ratio, Quetelet II
index, chest depth to chest width ratio, acromio-iliac
index.
Maturity Offset. The formula described by Mirwald
et al., (2002) was adapted;
Anaerobic and Aerobic Capacity. The maximum
oxygen uptake (aerobic capacity) was defined by
McArdle`s equation (McArdle et al., 1972). The
maximum anaerobic work (MAW), as an expression
of anaerobic–non–lactate capacity, was diagnosed
according to guidelines published by Drabik (2007);
The Measurement of Overall Physical Fitness. The
EUROFIT test battery was applied with a standard
concept (Council of Europe, 1988);
The Muscle Power Indices. The testing procedures
ware described by Council of Europe (1988);
Salonia et al., (2004); Mayhew et al., (2005);
The Measurement of Specific Physical Fitness. The
PowerliftingatJuniorLevel-SelectionParadigm
117
number of correct performed movements made
within 15 seconds in each of the three PL events was
the basis for the assessment of the specific speed.
Rules for performing the trials were based on the
regulations of the International Powerlifting
Federation (IPF) and assumptions of anaerobic
capacity test (ACT 5/15) (Bolach and Jacewicz,
2008). Fundamentally, ACT 5/15 test meets the
main conditions for assessing the speed skills in PL.
According to the IPF, athletes had three rounds in
each event at their disposal. Rest between attempts
was as much as three minutes long. With respect to
the results of the powerlifting events, the load was
adjusted to 50% repetition maximum (RM). It
equalled the initial intensity of the ACT 5/15 test.
The time was measured with an accuracy of 1/100
second with a standard electronic timer.
Specific endurance was determined by counting
subsequent repetitions in each of the PL events until
exhaustion (Forbes, et al., 2007). Athletes carried
out tests with a load of 70% RM (Forbes et al.,
2007). After warming up, the subjects performed
one attempt for each trial. Whole procedure was
accomplished according to the principles of the IPF;
The Measurement of Movement Technique. The
frequency of movements represented the indicator of
movement technique I (IMT
I
). Data from the fifth
and fifteenth (last) second of specific speed tests
were subjected to evaluation:
IndicatoroftechniqueI
averagefrequencyofmovementsin5seconds
fromallevents
averagefrequencyin15secondsfromallevents
2
The technique of movement is connected with an
athlete’s somatic and energetic potential. Thus,
keeping in mind PL requirements, a suitable
construction of the indicator of movement technique
II (IMT
II
) was designed:
IndicatoroftechniqueII
musclemasskg
upperbodypower lowerbodypower
The Measurement of Personality. NEO-FFI
Personality Inventory was used in the Polish version
(Zawadzki et al., 1998), based on the original
inventory by Costa and McCrae (1992).
Neuroticism, extraversion, openness to experience,
agreeableness and conscientiousness were measured.
The raw data was used in the analysis;
The Measurement of Reaction Time. Reaction time
was obtained with means of computer tests (Klocek
et al., 2002);
The Measurement of Hemodynamic Parameters.
Stroke volume and cardiac output (SV, Q) were
calculated according to Starr`s concept (as cited in
Woźniak et al., 1986, p. 126).
Ultimately, the measurements of 45
characteristics were made so that in the further part
of our study they served as 44 independent variables
and one dependent variable Y (table 1).
2.5 Statistical Analysis
All data were primarily studied through descriptive
statistics. Pearson’s product moment was computed
to screen the linearity across the matrix of
independent variables (X) and to assess the
relationship between each predictor variable (x
i
) and
Y – the sport result. Strictly, for defining errors in
performed test, we used the LoA technique (Altman
and Bland, 1983). In order to select the optimum
combination of model parameters, Hellwig’s
algorithm was adapted. Its description is given by
two formulas:
h
j
= │r
j
2
│ / 1+∑
i
≠j
│r
i
j
│ (1)
H
∑
h
,
(2)
where h
i
– is the individual capacity of information
for the i-th explanatory variable (x
i
), r
0i
is the
correlation coefficient of the i-th explanatory
variable with the dependent variable (Y); r
ij
is the
linear correlation coefficient between i-th and j-th
explanatory variable; H is the overall capacity of
information of carriers (independent variables) for a
given combination.
Since this analytical method
does not take into consideration statistical
significance of variables, we ran multiple regression
analysis and therefore checked P-values. Finally, to
construct selection model in PL at junior age
category, we applied multiple discriminant analysis
– DA (Ryguła, 2003). Briefly, one of the main goals
of DA is to derive mathematical functions for strata
membership of new cases. There are as many
equations as subgroups under investigation.
Therefore, we computed three linear classification
formulas for group of weak (W), medium (M) and
elite (E) sport results (Equations: 3, 4, 5).
Statistical analyses were made on a standard PC
using the STATISTICA software (Release 10.0).
3 RESULTS
Data exploration was initiated from descriptive
analysis. We postulate to present all variables, which
were taken in the investigation (table 1).
icSPORTS2013-InternationalCongressonSportsScienceResearchandTechnologySupport
118
Table 1: Descriptive characteristics of parameters tested in 21-23-year old powerlifters.
No. x
i
/ Y Units M SD CV As Cu-3
Y- Sport result
Wilks
Points
331.449 12.659 41.959 0.127 -0.588
1. Age years 22.397 3.688 0.826 0.000 -0.968
2. Body mass kg 83.488 12,720 10.620 0.579 0.701
3. Axillary chest circumference at maximum
inhalation
cm 110.331 5.506 6.075 0.296 -0.696
4. Arm circumference cm 36.428 7.752 2.824 -0.183 -0.799
5. Tight circumference cm 61.941 6.433 3.985 -0.058 0.158
6. Trunk length to stature ratio Point 30.985 4.852 1.503 0.097 -0.116
7. Upper to lower limb length ratio Points 86.259 4.139 3.570 0.916 0.780
8. Quetelet II index Points 26.341 8.556 2.254 0.742 0.955
9. Stroke volume ml 65.254 9.886 6.451 0.457 0.024
10. Total body balance n 3.156 84.445 2.665 1.097 0.627
11. Lower body power m 2.520 4.641 0.117 0.006 -0.278
12. Upper body power m 7.996 7.420 0.593 -0.142 -0.723
13. Total body power m 15.434 6.218 0.960 0.723 0.779
14. Upper limb speed s 7.803 6.572 0.513 0.249 -0.815
15. Hand grip force kg 60.094 14.466 8.693 0.533 -0.361
16. Upper arms isometric endurance s 38.856 35.038 13.614 0.668 0.229
17. Simple reaction time s 0.270 7.985 0.022 -0.150 -0.274
18. Choice reaction time s 0.444 8.777 0.039 0.244 -1.185
19. Specific speed n 47.188 7.901 3.728 0.034 0.649
20. Indicator of technique I Hz 1.142 8.454 0.097 -0.393 -0.379
21. Indicator of technique II a.u. 4.946 10.995 0.544 0.002 0.120
22. Maturity offset years 3.612 27.732 1.002 0.220 0.108
23. Quantity of fat tissue kg 14.869 15.619 2.322 -0.195 -0.697
24. Height cm 178.175 6.848 46.890 0.549 1.113
25. Percentage of fat tissue % 12.255 1.490 2.219 0.226 -0.770
26. Cardiac output l 4.594 0.341 0.116 1.377 3.649
27. VO
2
max ml/kg/min 46.243 2.089 4.362 0.789 1.100
28. Chest depth to chest width ratio Points 71.913 5.913 34.968 0.044 -1.148
29. Acromio-iliac index Points 67.003 4.599 21.152 1.225 2.619
30. Body surface m
2
2.035 0.162 0.026 0.401 0.577
31. Total body water l 49.385 4.444 19.751 0.488 0.656
32. Flexibility cm 13.563 7.691 59.157 0.213 -0.692
33. Total endurance n 71.219 13.937 194.241 1.150 1.947
34. Abdominal endurance n 29.438 2.602 6.770 -0.082 0.673
35. Agility s 19.567 2.067 4.271 0.919 1.008
36. Maksimal anaerobic power kJ 2.109 0.327 0.107 0.684 0.523
37. Specific endurance n 50.750 4.363 19.032 -0.118 -0.207
38. Neuroticism Points 15.313 6.508 42.351 0.374 -0.527
39. Exstraversion Points 29.781 6.057 36.693 0.232 -0.425
40. Openness to experience Points 24.969 5.642 31.838 0.499 0.658
41. Agreeableness Points 27.563 6.101 37.222 -0.772 0.098
42. Conscientiousness Points 33.156 5.419 29.362 -0.109 -0.171
43. Quantity of muscle tissue kg 52.107 7.329 53.719 0.145 -0.202
44. Percentage of muscle mass % 62.344 2.728 7.440 0.433 0.220
*Presented data are expressed as mean (M), standard deviation (SD), coefficient of variation (CV), asymmetry index (A
s
), kurtosis (Cu-3).
**Note: results for variables x
11
-x
13
were obtained in general fitness tests (see information in paragraph 2.4). Thus, original units were
shown.
The reason for this is that main statistics, like
mean and standard deviation are very informative
about the data distribution. In turn, appropriate range
of values for kurtosis and skewness enable to
perform multidimensional data exploration (table 1).
At this moment it is also necessary to report the
errors in accomplished measurements. They were
ranged from 93.75% to 100% limit of agreement
(LoA).
The further (advanced) statistical analysis we
began from choosing the optimum combination of
variables (H
max
) – equations: 1, 2. It included
following dimensions: age (x
1
), axillary chest
circumference at maximum inhalation (x
3
), trunk
length to stature ratio (x
6
), upper to lower limb
length ratio (x
7
), Quetelet II index (x
8
), total body
balance (x
10
), lower body power (x
11
), indicator of
technique I (x
20
).
PowerliftingatJuniorLevel-SelectionParadigm
119
At this stage of study relevant issue is to test H0:
b
i
= 0 and H0:
∑
b
b
= 0. The t and F statistics
ware essential in falsification procedure (table 2).
From the data, it appeared all of the predictors are
statistically significant as well as the whole model.
Calculations indicate that R
2
is very high and S
e
rather low. Straight, it means that constructed
function adequately describes sport result in PL for
junior age and is good enough to incorporate it into
coaching practice.
Nevertheless, the range of Y variable was high
and equalled 166.079. This situation shows that the
researched group of sportsmen did not represent a
homogenous structure. Such an occurrence
facilitates the performance of a discriminant
analysis. In the very beginning of the computational
process in DA, powerlifters were stratified into
independent groups (subdivisions). This was done
through establishing sport result categories.
Consequently, we have grouped athletes into three
classes: n
w
- weak = 12 individuals, with sport result
range: 250-299 Wilks Points; n
m
- medium = 10
individuals, with sport result range 300-349 Wilks
Points; n
e
- elite = 10 individuals, with sport result
lower limit >350 Wilks Points. Since three athletes
outperformed 400 Wilks Points, the last interval is
open.
Bearing in mind that DA has many restrictions
(Bates, 2005), discriminant functions were computed
from stepwise algorithm – the backward variant.
Due to the analysis, from the verified set of variables
(H
max
), discriminant model comprised of five
predicators (age (x
1
), axillary chest circumference at
maximum inhalation (x
3
), upper to lower limb length
ratio (x
7
), lower body power (x
11
), indicator of
technique I (x
20
)). The total discriminant power of
these variables (Wilks Lambda Λ) reached the value
of 0.068. Based on this result, we can say that
parameters in the model should be considered as
highly adequate for developing a discriminant
functions. Now, respecting theoretical assumptions,
the verification of H0 is of particular interest –
variables do not discriminate powerlifters. To test
this, discriminant functions – u
1
and u
2
, had to be
constructed. The value of empirical Chi-square
statistics was large enough to accept H1 only in the
case of u
1
. Thus, ultimately u
2
was not analysed.
In the model, the variables with the highest
discriminatory power, in order of importance, were
as follows: axillary chest circumference at maximum
inhalation (x
3
), upper to lower limb length ratio (x
7
).
According to computation, the lowest weight in
the function u
1
had x
1
(0.34822), and following the
guidelines (Bates, 2005) it has been removed from
further analysis. This move resulted in obtaining
adequate high significance for all predictors in
discrimination model; Λ= 0.089 (F
(8, 52)
= 15.192;
p<0.0000).
Due to the findings that are placed in table 3,
cumulative proportion totalled 0.99. Therefore, after
reducing number of dimensions to u
1
hyperspace,
distances between subclasses were maintained in
99%. Besides determining the optimum hyperspace
(discriminant functions) that separates athletes
divisions, DA is also helpful in classification
function computation. By means of DA it is possible
to construct classification functions for each of the
established subgroups. They should be recognized as
the fundamental instruments of diagnosis process in
the selection procedure (model) of young
powerlifters.
Table 2: The coefficient weights of sport result predictors for junior age category powerlifters.
R= 0.977; R
2
= 0.954; F(8,23)= 59.888 p<0.00000; S
e
= 10.426
n=32 b* Stand. error b* B Stand. error b t(23) P
Intercept -583.816 82.17859 -7.10424 0.000000
x
1
0.119476 0.051814 6.069 2.63179 2.30587 0.030476
x
3
0.235655 0.090051 1.628 0.62197 2.61689 0.015415
x
6
0.176696 0.081615 4.932 2.27785 2.16500 0.041008
x
7
0.154085 0.063421 1.811 0.74543 2.42956 0.023335
x
8
0.149064 0.068286 2.775 1.27132 2.18294 0.039504
x
10
-0.128212 0.061340 -2.018 0.96566 -2.09019 0.047849
x
11
0.129987 0.058501 46.631 20.98667 2.22196 0.036405
x
20
0.214541 0.068699 93.248 29.85937 3.12290 0.004781
* Names and order of variables are the same as in table 1.
**Note: the parameter R reflects the multidimensional zero-order correlation coefficient. Consequently, R
2
indicates the amount of
explained variation by the regression equation. Abbreviation S
e
stands for standard error of estimation; F is a common test, which in the
analysis of multiple regression is utilised for measuring the significance of all parameters in the model. Finally, statistics b* and B are
standardized and unstandardized coefficient weights respectively.
icSPORTS2013-InternationalCongressonSportsScienceResearchandTechnologySupport
120
Table 3: The weights of the first discriminant function
after x
1
exclusion.
x
3
x
7
x
11
x
20
Cumul. prop.
-0.717340 -0.802297 -0.548476 -0.500043 0.990285
* Names and order of variables are the same as in table 1.
In our research, a priori probability was set in
accordance to group sizes. In terms of raw data,
classification functions have the following structure:
W= -2716.19+13.19x
3
+28.26x
7
+542.63x
11
+361.58x
20
(3)
M= -2935.00+13.68x
3
+29.23x
7
+561.34x
11
+395.93x
20
(4)
E= -3262.18+14.55x
3
+30.72x
7
+586.22x
11
+424.38x
20
(5)
By performing classification matrix investigation
(table 4), misclassified observations have been
identified. Equations 3, 4, 5 predicted correctly
100% cases; Press’s Q = 64> χ
2
(α; df-1)
= 44.99.
Table 4: Classification matrix.
n=
32
Assignment
correctness
percentage
A priori
prob.
p=.31250
A priori
prob.
p=.37500
A priori
prob.
p=.31250
W 100,000 10 0 0
M 100,000 0 12 0
E 100,000 0 0 10
T 100,000 10 12 10
*Abbreviations: W- weak sport results group; M- medium sport
results group; E- elite sport results group; T- total classification
accuracy derived by the equations.
In order to test robustness of the group
membership prediction, the formulas were also
verified along the validation sample. Four
contestants composed of the validation dataset. Their
data are reported in brackets: n
33
[105.8 81.949 2.43
1.022], n
34
[107.2 85.095 2.38 1.00], n
35
[106.3
81.989 2.55 1.144], n
36
[114.5 84.989 2.51 1.133].
Multiplying the individual’s score by the
classification coefficient for each variable in the
equations (3, 4, 5), we obtained the same accuracy
of prediction as in the case of training group.
4 DISCUSSION
The study was established by performing a
multidimensional analysis. The findings showed that
an optimum combination of independent variables in
powerlifting in the junior age category includes only
eight predictors out of forty-four. These are: age,
axillary chest circumference at maximum inhalation,
trunk length to stature ratio, upper to lower limb
length ratio, Quetelet II index, total body balance,
lower body power, indicator of technique I. To
obtain their diagnostic value, multiple regression
coefficients were computed. In the light of
factography, on the basis of weight factors, each of
the dimensions in the H
max
set strongly influence
sport result. Subsequent analysis (table 2) proved
that the stochastic parameters of biometric model for
sport result in powerlifting satisfy the requirements
of coaching practice. It fulfils coincidence criterion
(Hellwig, 1969): sign r
(xiY)
= sign a
i
(sign of
regression coefficient). The determination index
equalled 0.954 points, S
e
was low and amounted to
about 10.5 Wilks Points.
From coaching practice viewpoint above means
that the biometric model can be used as a basis for
effective prediction of dependent variable – Y, e.g. if
axillary chest circumference at maximum inhalation
is increased by 1-cm then the value of Y variable
(sport result in PL) will increase by 1.628 Wilks
Points, assuming that the other variables from the
regression model remain unchanged (table 2).
As it was presented in many research, stepping
forward from multiple regression analysis to
discriminant analysis, the structure of sport result
can be studied profoundly (Magiera and Ryguła,
2007); (Ryguła, 2003).
Our study demonstrated that the best set of
variables, which discriminate powerlifters consists
of four predictors: axillary chest circumference at
maximum inhalation, upper to lower limb length
ratio, lower body power, indicator of technique
I. All
of them are important in distinguishing young
powerlifters. According to the evidence, 99% of the
phenomenon we investigated has been explained;
Wilks Lambda was only 0.09 points and satisfied the
significance criterion at P≤ 0.05. Thus in the
spotlight of the statistical theory, H1 holds true.
In the area of strength sports disciplines, there is
lack of applicative research demonstrating
discriminant analysis. It should be pointed out that in
this domain, only Fry et al., (2006) have presented
comprehensive model of selection that was based on
DA. In their study the global Wilks Lambda
equalled 0.664, and percentage of correct
classifications was fairly high – 88.55%.
If specific physiological demands are taken into
consideration, other papers regarding scientific
approach to selection problem in sport were run for
disciples much different than powerlifting. Namely,
handball (Ignacik, 2008); (Ryguła, 2003), sport
climbing (Magiera and Ryguła, 2007), javelin
(Maszczyk, 2008), swimming (Saavedra et al.,
2010). Aforementioned experiments, when
PowerliftingatJuniorLevel-SelectionParadigm
121
comparing results, have one main thing in common
– appropriately high value of classification
correctness. It was always greater than the
calculation based on chance accuracy algorithm.
In the presented research, the total number of
correctly identified athletes has a value of 100%.
Basic statistics in the assessment procedure of
powerlifters classification effectiveness was Press’s
Q test. Its empirical result was much higher than the
table value of Chi-square. Therefore at the 95%
confidence, the inequality described with H2 has
been proven positively. Hence, according to the
analysis and statistical theory the model of selection
we projected reached significantly better results in
athletes’ assignment than chance accuracy
procedure. Consequently, it may be said that on the
basis of the study, the classification paradigm can be
usefully applied to support the process of recruiting
athletes in PL.
Normally, discriminate analysis in the science of
sport is run on training and testing set. In this
manuscript research was extended to validation
sample. After assigning output to variables in
equations: 3, 4, 5, the total accuracy of prediction
equalled also 100%. Subsequently, it suggests that
constructed model is plausible and satisfies the
requirements of effective selection in powerlifting
for junior age category.
Summing up, paradigm we developed is
adequate tool for young coaches for optimization of
the selection procedure. We claim it is worth of
further investigation.
5 CONCLUSIONS
(1) The most important determinants for the
powerlifters discrimination model are axillary chest
circumference at maximum inhalation, upper to
lower limb length ratio, lower body power, indicator
of movement technique
I; (2) According to the value
of cumulative proportion, the first discriminant
function maintain the distances between
subdivisions of powerlifters in 99%; (3) As per Q
Press’s test, classification functions are identifying
powerlifters statistically better from a priori
procedure.
ACKNOWLEDGEMENTS
This research work was supported by the system
project “InnoDoktorant – Scholarships for PhD
students, V
th
edition”. Project is co-financed by the
European Union in the frame of the European Social
Fund. Hence, authors would like to thank
Pomeranian Voivodeship governor’s office.
REFERENCES
Altman, D. G., Bland, J. M., 1983. Measurement in
medicine: the analysis of method comparison studies.
Statistician, 33, 307-17.
Bates, R. A., 2005. Multivariate research methods. In R.
A. Swanson & E. F. Holton III (Eds.). Research in
organization. Foundations and methods of inquiry (pp.
115-142). San Francisco, USA: Barrett-Koehler, Inc.
Bolach, E., Jacewicz, K., 2008. Modyfikacja testu
Wingate w ocenie wydolności wysiłkowej kadry
Polski niepełnosprawnych ciężarowców [Modified
Wingate test for estimation of exercise capacity of
Polish disabled weight-lifters]. Fizjoterapia, 16(3), 57-
68.
Costa, P. T., Jr., & McCrae, R. R. Personality Inventory
(NEO-PI-R) and NEO Five-Factor Inventory (NEO-
FFI) manual. Odessa, FL: Psychological Assessment
Resources.
Council of Europe. 1988. EUROFIT: Handbook for the
EUROFIT Tests of Physical Fitness. Rome:
Secretariat of the Committee for the Development of
Sport within the Council of Europe.
Drabik, J., 2007. Ocena poziomu sprawności specjalnej
[An assessment of the level of specific fitness]. In J.
Drabik, M. Resiak, E. Prętkiewicz-Abacjew (Eds.),
Sześcioletnie dziecko gdańskie w latach 1995 – 2004:
Szkic antropometryczny: Trendy i standardy [A 6-
year-old child of Gdańsk 1995 – 2004:
Anthropometric draft: Trends and standards] (pp. 14-
16). Gdańsk, Poland: Wydawnictwo Akademii
Wychowania Fizycznego.
Drozdowski, Z., 1998. Antropometria w wychowaniu
fizycznym [Anthropometry in physical education].
Poznań, Poland: Wydawnictwo Akademii
Wychowania Fizycznego.
Forbes, S. C., Candow, D. G., Little, J. P., Magnus, Ch.,
Chilibeck, P. D., 2007. Effects of Red Bull energy
drink on repeated Wingate cycle performance and
bench – press muscle endurance. Journal of Sport
Nutrition and Exercise Metabolism, 17, 433-444.
Fry, A. C., Ciroslan, D., Fry, M. D., LeRoux, C. D.,
Schilling, B. K., 2006. Anthropometric and
performance variables discriminating elite American
junior men weightlifters. Journal of Strength and
Conditioning Research, 20, 861-866.
Greń J., 1976. Statystyka matematyczna. Modele i zadania
[Statistical mathematics. Models and exercises].
Warszawa, Poland: PWN.
Hellwig, Z., 1969. Problem optymalnego wyboru
predykant [The problem of variables’ optimum
choice]. Przegląd Statystyczny, 3-4, 221-238.
icSPORTS2013-InternationalCongressonSportsScienceResearchandTechnologySupport
122
Ignacik, J., 2008. Informacyjność wybranych kryteriów
doboru kandydatów do szkolenia w piłce ręcznej
młodych szczypiornistów. [Informativeness for the
selection criteria of young handball players]. Zeszyty
Metodyczno – Naukowe. Studia Doktoranckie AWF
Katowice, 25, 319-333.
Keogh, J. W. L., Hume, P. A., Mellow, P., & Pearson, S.
(2005). The use of anthropometric variables to predict
bench press and squat in well–trained strength athletes.
In Q. Wang (Chair), 23
rd
International Symposium on
Biomechanics in Sports. Symposium conducted at the
meeting of the International Society of Biomechanics
in Sports, Beijing, China. Retrieved from
http://w4.ub.unikonstanz.de/cpa/article/view/ 841.
Klocek, T., Spieszny, M., Szczepanik, M., 2002.
Komputerowe testy zdolności koordynacyjnych
[Computer tests of coordination abilities]. Warsaw,
Poland: Centralny Ośrodek Sportu.
Magiera, A., Ryguła I., 2007. Biometric model and
classification functions in sport climbing. Journal of
Human Kinetics, 18, 87-98.
Maszczyk, A., 2008. Określenie wartości diagnostycznej i
predyktywnej wybranych cech budowy ciała,
sprawności fizycznej ogólnej i specjalnej 14 – 15-
letnich oszczepników [The determination of diagnostic
and predictive value of selected morfological
parameters, general and specific finess among 14-15-
year old javelin throwers]. Zeszyty Metodyczno –
Naukowe. Studia Doktoranckie AWF Katowice, 25,
23-37.
Mayhew, J. L., Bird, M., Cole, M. L., Koch, A. J.,
Jacques, J. A., Ware, J. S., Fletcher, K. M., 2005.
Comparisons of the backward overhead medicine ball
throw to power production in college football players.
Journal of Strength and Conditioning Research, 19,
514-518.
Mayhew, J. L., McCormick, T. P., Piper, F. C., Kurth, A.
M., Arnold, M. D., 1993. Relationships of body
dimensions to strength performance in novice
adolescent male powerlifters. Pediatric Exercise
Science, 5, 347-356.
McArdle, W. D., Katch, F. I., Pechar, G. S., Jacobson L.,
Ruck, S., 1972. Reliability and interrelationships
between maximal oxygen intake, physical work
capacity and step-test scores in college women.
Medicine and Science in Sports, 4, 182-186.
Mirwald, R. L., Baxter-Jones, A. D. G., Bailey, D. A.,
Beunen, G. P., 2002. An assessment of maturity from
anthropometric measurements. Medicine and Science
in Sports Exercise, 34, 689-694.
Płóciennik, Ł., 2012. The diagnostic value of criteria for
selection among 19-23-year-old powerlifters
(Unpublished doctoral dissertation). Academy of
Physical Education and Sport, Gdansk, Poland.
Ryguła, I., 2003. Proces badawczy w naukach o sporcie
[Research process in science of sport]. Katowice,
Polska: Uniwersytet Śląski
, Katowice.
Saavedra, J. S., Escalante, Y., Rodriguez, F. A., 2010. A
Multivariate Analysis of Performance in Young
Swimmers. Pediatric Exercise Science, 22, 135-151.
Salonia, M. A., Chu, D. A., Cheifetz, P. M., Freidhoff, G.
C., 2004. Upper-body power as measured by medicine
ball throw distance and its relationship to class level
among 10- and 11-year-old female participants in club
gymnastics. Journal of Strength and Conditioning
Research, 18, 695-702.
Sampaio, J., Janeira, M., Ibáñez, S., Lorenzo, A., 2006.
Discriminant analysis of game – related statistics
between basketball guards, forwards and centres in
three professional leagues. European Journal of Sport
Science, 6, 173-178.
Shephard, R. J., 1991. Body composition in biological
anthropology. London: Cambridge University Press.
Watson, P. E., Watson, I. D., Batt, R. D., 1980. Total body
water volumes for adult males and females estimated
from simple anthropometric measurements. American
Journal of Clinical Nutrition, 33, 27-39.
Winwood, P. W., 2011. Strongman: Strength and
conditioning practices, and the inter – relationships
between strength, anthropometrics and performance
(Master’s thesis). Auckland University of Technology,
Auckland, USA.
Woźniak-Kozak, E., Laurentowska, M., & Tokłowicz B.
(1986). Wybrane próby czynnościowe określające
wydolność układu krążenia [Selected functional tests
determining cardiovascular capacity]. In W. Rożynek-
Łukanowska (Ed.), Wybrane ćwiczenia z fizjologii
wysiłku fizycznego i sportu [Selected exercises from
physiology of exercise and sport] (pp. 81-131).
Poznań, Poland: Wydawnictwo Akademii Wychowania
Fizycznego.
Zawadzki, B., Strelau, J., Szczepaniak, P., & Śliwińska M.
(1998). Inwentarz Osobowości NEO – FFI Costy i
McCrae [Costa`s and McCrae`s Personality Inventory
NEO - FFI]. Warsaw, Poland: Polskie Towarzystwo
Psychologiczne.
PowerliftingatJuniorLevel-SelectionParadigm
123
