Assessment 1: Critical Appraisal

Assessment 1: Critical Appraisal
Chosen study: Hsieh, Cheng-Hsuan

Part 1

• a This study had two purpose:
  

1. Investigate how triathletes use and perceive the effectiveness of their recovery strategies
   
2. Analyze the differences in recovery strategy usage between a normal training week and a week following a competition, to determine how these patterns might shift with participation in competitive events

• b
  In the introduction, the author give the assumption that recovery strategies are a key component of sport-related expenditures and a major concern for multisport athletes, because of their high demand of musculoskeletal consumption in their sports. Furthermore, he elaborates that triathletes at different levels and various events may make different decisions regarding recovery methods based on these factors. This aim is to collect and analyze the effectiveness of various recovery methods for endurance athletes and their respective advantages.
  However, I find this assumption to be somehow unclear. After reviewing the entire paper, in my opinion, the hypothesis should be that triathletes choose different recovery strategies and frequencies based on their levels, events, and timing.

• c
  First of all, the author first used the Shapiro-Wilk Test to make sure the normal distribution of the data. The method was chosen due to its accuracy for relatively small data. However, some samples were relatively large with the total number of 322, the method might be to sensitive to large data sets which may potentially affecting its accuracy.
  Second, Mutiple Chi-square Test was used to examine the associations between different variables, such as timing, events, and recovery strategies. The test also evaluated the response of different recovery strategies on preferred events and levels. Moreover, Pairwise, adjusting for inflation, were used to each distance to figure out the significant results. These methods are suitable to be applied on the data to find out the difference between different groups with different variables, and the weighting system is also well-suited for data sets with huge difference in magnitude or varying characteristics among the variables.
  Lastly, Wilcoxon Signed-rank Test was applied to compare the importance of recovery between different level triathletes. Because Traditional T-tests or ANOVA couldn’t be used on data which violates normality of residuals and equality of variances, this non-parametric test was applied instead.
  

• d
  Active recovery (51% during normal weeks, 65% post-competition), stretching (47% during normal weeks, 49% post-competition), and additional sleep (37% during normal weeks, 51% post-competition) are the most commonly used recovery strategies.
  In terms of recovery devices, foam rolling (35% during normal weeks, 39% post-competition), massage guns (20% during normal weeks, 20% post-competition), and compression garments (19% during normal weeks, 26% post-competition) are frequently utilized. Regarding the subjective effectiveness of recovery strategies, additional sleep (91% overall effectiveness), active recovery (88%), and professional massage (86%) show the most significant results.
  These findings address the first purpose of the study, identifying the most commonly used recovery strategies among triathletes and their perceived effectiveness, as well as the frequently used devices in combination. To sum up, triathletes prefer to select active recovery, stretching and additional sleep in their recovery process. After intense competition, their wilingness to use these recovery methods increases significantly. Among the subjective responses, additional sleep, active recovery, and professional massage receive the most positive feedback. This suggests that active recovery and additional sleep not only become more common choices for recovery activities in terms of timing but also tend to offer more effective subjective recovery outcomes.
  Long-distance athletes (Ironman and Half-Ironman) use IPC more frequently than short-distance athletes (Olympic distance or shorter, p=0.0296 and p=0.0296, respectively). Among Ironman event participants, hydrotherapy is preferred more than among Olympic distance participants. However, no significant differences were found between normal weeks and post-competition weeks.
  These results align with the second aim of the study, showing that triathletes’ willingness to use IPC and hydrotherapy depend on their prefer distance events. However, no significant differences were observed among the other variables. This may be related to insufficient data, a small sample sized, and uneven distribution of participants, which could account for the lack of significant differences in the analysis.

• e
  During data collection, 60% of responses on the effectiveness of different recovery methods were marked as N/A, which limits the statistical results to being only indicative. Additionally, the survey didn’t include clear definitions to distinguish between different frequencies or recovery methods, making it difficult to eliminate subjective misjudgments and misunderstandings by the participants. Furthermore, there is no way to confirm if the recovery methods selected by participants align with the methods they actually use. This also leads to potential significant errors in the reported subjective effects.
  As for the subjective effectiveness, participants may using multiple methods simultaneously might occur cumulative effects that could alter their personal feeling at that time, and they can’t distinguish the effects are based on which methods. In addition, the survey also lacked a precise definition of recovery effects, whether related to reduce muscle soreness, decrease psychological stress, or improve performance on their further training. Future research should aim to clarify the definitions of recovery methods, frequency, and subjective responses to enhance more accuracy and objective statistical results.
  There were huge disparities in the number of participants at different levels, with only 12 elite athletes, and an imbalance in gender distribution, with males making up 66% of the total participants. Additionally, there were differences in the number of long-distance and short-distance participants. These factors may generate a variety of bias to influence the accuracy of the results.
  A large portion of the data came from subjective questionnaires, which are susceptible to recall bias, social desirability bias, emotional state bias, and representative issues, affecting the overall reliability on the statistics.
  

• f
  I will include more objective and precise definitions for recovery methods, frequency, and response in the distribution. Moreover, I will try my best to refer to other more credible questionnaires to ensure the data that we collected can be less error. Except the questionnaires will be filled by testers with standardized criteria to ensure consistency when interviewing and recording responses from participants. In addition, the injury history , other health related factors, and other exercise habits will be include to gather more ojective data which may affect the effectiveness of the recovery methods significantly. Efforts will also be made to balance the number of participants across different regions, levels, distances, genders, and ages. This approach will allow for a more detailed examination of how different groups choose their recovery methods and it’s effectiveness.

Part 2

R coding

2.1

#import data
library(tidyverse)

## ── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
## ✔ dplyr     1.1.4     ✔ readr     2.1.5
## ✔ forcats   1.0.0     ✔ stringr   1.5.1
## ✔ ggplot2   3.5.1     ✔ tibble    3.2.1
## ✔ lubridate 1.9.3     ✔ tidyr     1.3.1
## ✔ purrr     1.0.2     
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag()    masks stats::lag()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors

library(ggplot2)
library(patchwork)
data<-read.csv("Frequency_long2.csv")
data<-data|>
  rename(
    'Active recovery'=Active_recovery,
    'Additional sleep'=Additional_sleep,
    'Compression garments'=Compression_garments,
    'Cryotherapy chambers'=Cryotherapy,
    'Float tank'=Float_tank,
    'Foam rolling'=Foam_rolling,
    'Hydrotherapy'=Water_immersion,
    'Massage gun'=Massage_gun,
    'Paracetamol/other painkiller'=Paracetamol,
    'Professional massage'=Professional_massage,
    'Topical magnesium'=Topical_magnesium
    
  )
#transform the data
df<-data |> 
  mutate(across('Active recovery':'Topical magnesium',~ case_when(
    .=="Almost never"~0,
    .=="Rarely"~1,
    .=="Sometimes"~2,
    .=="Often"~3,
    .=="Almost always"~4)
  )) |>
  mutate(Level=ifelse(Level==1|Level==2,'1','2')) |> 
  group_by(Level,Time) |> 
  summarise(across('Active recovery': 'Topical magnesium',~sum(.)/(n()*4),.names = " {.col}"))

## `summarise()` has grouped output by 'Level'. You can override using the
## `.groups` argument.

df_long<- df |> 
  pivot_longer(cols = starts_with(" "),
               names_to = "methods",
               values_to= "percentage")

# seting the data for transverse transform
df_new <- df_long %>%
  mutate(percentage = ifelse(Time == "competition", percentage, -percentage))
correct_levels <- c(
  ' Topical magnesium',
  ' Stretching',
  ' Sauna',
  ' Professional massage',
  ' Paracetamol/other painkiller',
  ' Massage gun',
  ' IPC',
  ' Hydrotherapy',
  ' Foam rolling',
  ' Float tank',
  ' Cryotherapy chambers',
  ' Compression garments',
  ' Additional sleep',
  ' Active recovery'
)
df_new<- df_new |> mutate(
  methods=factor(methods,levels=correct_levels),
  Level=factor(Level,levels=c('2','1')))

2.2

P1<-ggplot(df_new, aes(x = methods, y = percentage, fill = Level)) +
  geom_bar(stat = "identity", position = 'dodge', width = 0.5) +
  coord_flip() +
  scale_y_continuous(breaks = seq(-.75, .75, by = .25),labels = abs, sec.axis = dup_axis(name = NULL)) + 
  scale_fill_manual(values = c("1"="black","2"="gray"))+
  labs(x = NULL,  
       y = NULL) + 
  theme(strip.placement = "outside",
        strip.background = element_blank(),
        panel.grid.major.y = element_blank(),
        panel.grid.minor.y = element_blank(), 
        legend.position = "none") + 
  geom_hline(yintercept = 0, color = 'white', size = 2)

## Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
## ℹ Please use `linewidth` instead.
## This warning is displayed once every 8 hours.
## Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
## generated.

P1

2.3

df_new2<-df_new |> 
  mutate(percentage=ifelse(percentage<0,-percentage,percentage))
P2<-ggplot(df_new2, aes(x = methods, y = percentage, fill =interaction(Time, Level), color = interaction(Time, Level), group = interaction(Time, Level))) +
  geom_bar(stat = "identity", position = "dodge", width = 0.7, alpha = 0.5, color = 'white') +
  geom_line(aes(linetype = Time), size = 1) + 
  geom_point(size = 3) +
  scale_y_continuous(breaks = seq(-.75, .75, by = .25), labels = abs) +
  scale_fill_manual(
    values = c(
      "competition.1" = "blue",
      "normal.1" = "lightblue",
      "competition.2" = "black",
      "normal.2" = "red"
    ),
    labels = c(
      "competition.1"="Elite & competitive - After competition week",
      "normal.1"="Elite & competitive - Normal week",
      "competition.2"="Recreational & novice - After competition week",
      "normal.2"="Recreational & novice - Normal week"
    )
  ) +
  scale_color_manual(
    values = c(
      "competition.1" = "blue",
      "normal.1" = "lightblue",
      "competition.2" = "black",
      "normal.2" = "red"
    ),
    labels = c(
      "competition.1"="Elite & competitive - After competition week",
      "normal.1"="Elite & competitive - Normal week",
      "competition.2"="Recreational & novice - After competition week",
      "normal.2"="Recreational & novice - Normal week"
    )
  ) +
  labs(x = NULL, y = NULL, fill = "Type of Week and Level", linetype = "Type of Week") +
  theme(
    panel.grid.major.y = element_blank(),
    panel.grid.minor.y = element_blank(),
    legend.position = "bottom"
  )
P2

PlotA and PlotB

combined_plot <- P1 + P2 + plot_annotation(title = "Plot", 
                                            tag_levels = 'A')
combined_plot

In the figures above, Plot A shows the data I have created from the paper. Note that I’m not sure how the author calculated the frequency, so I used my own way to calculate it. I set “almost never” to 0, “rarely” to 1, “sometimes” to 2, “often” to 3, and “almost always” to 4, and summed the scores for each recovery method to calculate it’s percentage. The data shows overall similar results, but there are still some differences. Plot B is my attempts to improve the presentation. These were done by:

• Changing the horizontal bar chart back to a standard bar chart, which provide better comparison of data across different weeks.

• Using contrasting colors to make the lines, points, and bars more clearly represent different variables. The different colors also allow for quick differentiation of data levels, even in overlapping areas.

• Organized all bars into a standard charts and scale for better comparison of frequency differences between various recovery methods

• Adding the original charts with lines and points. Although the points on the graph can be difficult to interpret due to overlap when comparing a single method with identical data, there are no instances of data being too close. Therefore, the visual representation can intuitively show frequency changes within the same method due to different variables.

• The most notable aspect of the line graph is that it allows us to estimate the area occupied below the lines under different variables, providing a general understanding of whether athletes choose different recovery methods or if there are frequency differences under varying conditions.