MATERIALES Y MÉTODOS
3.5 El procedimiento de la investigación
The following analyses m onitor the stability of the m easures taken for those children w ho visited the Institute on all three occasions and w ho continued to participate in intensive training.
Stability estimates for DSRS, self esteem and cohesion
Depression data set
The DSRS longitudinal data set comprised 246 young athletes. Before analysing the stability of these scores a series of independent t-tests were
conducted to establish w hether those children who completed all three visits had significantly different scores com pared w ith those w ho dropped out of the study and failed to retu rn for either visits 2 or 3. Results established there w ere no significant differences betw een any sport or gender group. N or w as there a significant gender difference betw een those w ho continued in the study, and those w ho dro p p ed out, for the sports population as a whole. M ean DSRS scores for the athletes w ho completed the survey stratified by visit, gender and sport are illustrated in table 54. Cross-sectional data from the com parison group have been included for com parative purposes.
Gender and group comparisons for the iongitudinai data set
A MANOVA was conducted to determ ine if gender or the type of sports participation had any effect on depression scores for visits 1, 2 or 3. The results established that there was no significant gender by activity interaction (Pillais (6,
4 7 6) = .49, p = n.s.), nor w as there a significant difference betw een male and
(Pillais (3^ 237) = 1.9/ p = n.s.). There was however a significant m ultivariate effect for type of sport (Pillais (9, 717) = 2.0, p = .03). A univariate F-test established
gym nasts ( x = 6.8) had significantly higher DSRS scores than tennis players ( % =
4.8) for visit 3 (F (3, 242) = 4.5, p = .004). However this was the only significant result, am ong four comparisons.
Table 54: Mean DSRS symptomatology score by group, gender and visit
F«ôtî>àïi (31) 5.1 ± 2.8 4.7 ± 3.6 4.7 ± 2.8 GyÀRastics MWe (16) 7.8 ± 4.2 6.4 ± 3.3 6.6 ± 3.8 6.4 ±3.7 6.1 ± 3.2 6.9 ± 3.9 SwIfuMfig . Male (32) 6.3 ± 3.6 5.8 ± 3.3 5.7 ± 4.1 4.9 ± 2.2 5.6 ± 3.6 6.3 ± 4.0 i* a e (% Q 5.8 ± 3.1 5.1 ± 3.5 4.8 ± 2.9 Female (41) 5.7 ± 3.0 5.1 ± 2.9 4.9 ± 3.1 Compmlso# gmiip Male (236) 7.9 ± 4.0 FémaM (241) 9.3 ± 4.5
Figures in brackets numbers of children. Values are means ± standard deviations.
Cohesion data set
The cohesion longitudinal data set comprised 155 young athletes. Before analysing the stability of these scores a series of independent t-tests were
conducted to establish w hether those children who completed all three visits had significantly different scores to those who dropped out of the study. Results established there were no significant differences between any sport or gender group. N or was their a significant gender difference betw een those who
continued in the study, and those who dropped out, for the sports population as a whole. Mean cohesion scores for the athletes who completed the survey
stratified by visit, gender and sport are illustrated in table 55. Cross-sectional data from the com parison group have been included.
G e n d e r a n d g ro u p c o m p a ris o n s fo r the lo n g itu d in a l d a ta s e t
A MANOVA was conducted to determine if gender or type of sports particiption had any effect on cohesion scores for visits 1 to 3. Results established that there was no significant gender by activity interaction (Pillais (6, 294) = -52, p =
n.s.), nor w as there a significant difference between male and female cohesion scores for the athlete population as a whole across the three visits (Pillais (3^ 146) =
.43, p = n.s.). Neither was type of sport significant (Pillais (9,444) = 1.1, p = n.s.).
Table 55: Mean family cohesion score by group, gender and visit
63.3 ± 8.8 64.2 ± 7.0 62.1 ± 8.4 63.0 ± 5.7 58.2 ± 7.9 55.6 ± 8.5 63.3 ±5.3 62.4 ± 5.1 60.3 ± 9.0 64.9 ± 8.3 61.3 ± 9.4 60.9 ± 8.2 62.5 ± 7.2 58.9 ± 7.8 58.7 ± 9.9 62.5 ± 7.2 61.2 ± 9.9 60.1 ± 11.1 61.4 ± 10.7 62.6 ± 10.3 61.5 ± 10.8 45.7 ± 9.8 46.3 ± 10.0
Figures in brackets numbers of children. Values are means ± standard deviations.
Self esteem data set
The self esteem longitudinal data set comprised 195 young athletes. Before analysing the stability of these scores a series of independent t-tests was
conducted to establish whether those children who completed all three visits had significantly different scores to those who dropped out of the study. Results established there were no significant differences between any sport or gender group. Nor was there a significant gender difference between those who
a whole. Mean scores for the athletes who completed the survey stratified by visit, gender and sport are illustrated in table 56. Cross-sectional data from the com parison group have been included for comparative purposes.
Gender and group comparisons for the longitudinal data set
A MANOVA was conducted to determine if gender or group had any effect on global self esteem scores for visits 1 to 3. Results established that there was a significant gender by activity interaction (Pillais (6,374) = 2.3, p = .03). Thus
data analysis indicated that there was no significant difference between male and female self esteem scores for the athlete population as a whole across the three visits (Pillais (3,186) = 1 9 , ^ = n.s.). Neither was type of sport significant (Pillais (9, 56 4) = .80, p = n.s.). A series of t-tests were carried out to determine the location of
the interaction effect. The results established that male swimmers (.? = 25.0) had significantly higher self esteem than females {x = 22.8) for visit 1 (t = 2.38, DP 57, p = .0 2), and that male tennis players ( % = 26.0) had higher scores than female
players {x = 23.9) for visit 2 (t = 2.45, DP 73, p = .01). Across all visits males had higher global self esteem scores than females (visit 1: t = 2.51, DP 193, p = .01: visit 2: t = 2.56, DP 193, p = .01: visit 3: t = 2.76, DP 193, p = .006).
Table 56: Mean global self esteem scores by group, gender and visit
Football (24) 25.5 ± 2.7 25.4 ± 2.8 25.3 ± 2.8 Gymmtsbm M ale (9) 24.4 ± 2.5 23.7 ± 3.7 25.1 ± 3.5 Female (28) 23.5 ±3.6 24.9 ± 3.1 23.2 ± 4.6 M ale (27) 25.0 ± 3.7 24.6 ± 4.2 24.1 ± 4.6 Fem ^e (32) 22.8 ± 3.2 23.2 ± 3.7 23.2 ± 3.8 Temtia M ale (41) 24.3 ± 4.1 25.9 ± 2.9 25.9 ± 3.5 Female (34) 24.2 ± 3.5 23.9 ± 4.4 24.4 ± 4.6 Compaxisoii group M ale 23.0 ± 4.2 Fem ale 21.1 ±4.7
S u m m a ry o f fin d in g s
The results of the analysis of the longitudinal data set for DSRS, family cohesion and global self esteem are almost identical to the cross-sectional
findings reported the first results section (chapter 5). With the exception of DSRS scores the results remain relatively constant across visits and do not appear to be affected by type of sport or the gender of the young athlete. The reason for the betw een sports difference in depression scores w ould appear to be the result of a reduction in mean DSRS for tennis players, from 5.7 for visit 1 to 4.8 for visit 3, com pared to a relatively stable pattern of results for young gymnasts. Overall these data indicate little in the way of group or gender effects over time, and therefore subsequent analyses investigating the stability of m easures taken will treat the athletes as belonging to a homogeneous group.
S ta b ility o f th e a th le te d ata a c ro s s visits
A Pearson correlation matrix was calculated to gauge the stability of individual athletes' self esteem, family cohesion and DSRS scores across visits. The results are show n in Table 57.
Table 57: Stability estimates across visits
Global self esteem (195)
Male (101) .41 .32 .54
Female <95) .54 .52 .69
DSKS (246) .46 .38 .53
Cohesion (179) .62 .78 .62
Note: all estimates significant at p< .001; numbers of subjects given in parentheses
As can be seen in Table 57, the stability estimates were all significant, ranging in value from .34 to .78. The greatest stability of scores was for family cohesion.
To explore the stability of these variables in greater detail repeated
tested the hypothesis that there w as no significant difference in scores across visits. For the purpose of the following analyses tw o contrasts w ere available. The tw o contrasts, t2 and t3, represent the effect of time: t l represents the m ain effect of visit 3, t3 the m ain effect of visit 2. M ultivariate tests of significance established that adolescents' perceptions of family cohesion reduced over time (Pillais (2,153) = 9.6, p < .0005). Inspection of specific contrasts indicated a
significant difference betw een visit 3 and baseline (F (i, 154) = 19.0; p < .0005). For global self esteem and DSRS scores MANOVA established there w as no
significant tim e effect (Pillais (2,193) = 1.4, p = n.s. ; and Pillais (2,224) = 1.3, p = n.s.
respectively).
STABILITY OF GROUP EFFECTS ACROSS VISITS - ATHLETES VERSUS THE