Term Project Description

My project builds upon the study referenced below. I have proposed a potential follow-up experiment to extend the study’s findings. The project utilizes simulated data, structured and analyzed as if collected from a real experiment. While the outcomes do not provide scientific evidence, they showcase my skills in formulating and addressing relevant biological questions through statistics, experimental design, and programming.

Article citation:

Lak, M., Forbes, S. C., Ashtary-Larky, D., Dadkhahfar, S., Robati, R. M., Nezakati, F., … Tinsley, G. M. (2025). Does creatine cause hair loss? A 12-week randomized controlled trial. Journal of the International Society of Sports Nutrition, 22(sup1). https://doi.org/10.1080/15502783.2025.2495229

Brief statement on the findings from the original article that led to your followup experiemnt:

The article concluded that the supplement Creatine monohydrate led to no significant increase in DHT (dihydrotestosterone), which was an identifier for hair growth.

The Question

Does creatine monohydrate cause significant hair loss in young teens?

Disclaimer: This project analyzes simulated data. The questions and hypotheses are real, but the results and conclusions are not.

Rationale and Background:

Creatine monohydrate is one of the most popular and studied supplements, especially among athletes and people who work out. It’s commonly used to help build muscle, improve strength, and support athletic performance. As someone who goes to the gym often, I became interested in trying creatine. But before doing so, I wanted to understand not just its benefits but also any possible side effects. One side effect that comes up a lot — especially on social media and fitness forums — is hair loss. The idea that creatine causes hair loss started with a 2009 study. In that study, college athletes who took creatine showed increased levels of a hormone called dihydrotestosterone (DHT), which is linked to male-pattern baldness. Because of this, people assumed that creatine must lead to hair loss. However, more recent research has challenged this idea. A 2025 randomized controlled trial by Lak et al. found no significant increase in DHT after creatine use over 12 weeks. Even though the science doesn’t strongly support a link, many teens and young adults still believe that creatine causes hair loss and avoid it entirely out of fear. Since creatine is used more and more by teenagers — people like myself — I designed a simulated experiment to explore this concern. I tested whether daily creatine use could cause noticeable hair loss in teens. Although some hair growth or loss is normal during adolescence, I wanted to see if creatine had any measurable effect on accelerating hair loss or even increasing hair growth, as some users have claimed. My study uses paired data to compare individuals’ hair loss levels when using creatine versus a placebo. While the data are artificial, the setup reflects a real-world concern. The goal is to apply solid experimental design and statistical analysis to help answer a question that many young athletes wonder.

Hypotheses

A Statistical Null Hypothesis:

delta (creatine, placebo) = 0

A Statistical Alternative Hypothesis:

delta (creatine, placebo) != 0

Experimental Design

Variables:

First Variable

Supplement

Second Variable

Hair density (cm^2)

Sample size:

Sample size justification:

I chose a sample size of 60 participants for this experiment because based on the article that inspired me, 60 is a similar sample size in their experiment. Using a comparable number ensures that my simulated study somewhat aligns with real research–which makes it feel more real. Additionally, 60 participants reflects a realisitc number of people you would see on a daily basis inside a regular gym. Since creatine is most commonly used by people who work out, this sample models a real scenario.

Data Analysis Plan

Paired T-Test

I chose to do a paired t-test for my experiment because each participant were observed under both conditions, once while taking creatine and the other taking a placebo. This is paired data as each individual is their own control. By comparing the difference in hair loss on a single person with both supplements, variability is reduced. Since I control the supplement for the same person, a paired t-test is most appropiate.

Assumptions and Exploratory Data Analysis (EDA)

The assumptions for a paired t-test are that both variables are normally distributed.

#USE THIS BLOCK TO INPUT NECESSARY CODE.
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   3.5.2     ✔ tibble    3.3.0
## ✔ lubridate 1.9.4     ✔ tidyr     1.3.1
## ✔ purrr     1.0.4     
## ── 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

supplementData <- read.csv("rcpulido.csv")


head(supplementData)

deltaData = supplementData$Creatine - supplementData$Placebo

hist(deltaData)

Interpretation of EDA:

Luckily, the histogram of the differences between my creatine and placebo values showed that the data is approximately normally distributed. The shape of the distribution was symmetric and bell-shaped, with no clear outliers present. This visual check supports the normality assumption required for the paired t-test.

Primary Statistical Analysis

#USE THIS BLOCK TO INPUT NECESSARY CODE.
result1<- t.test(x = supplementData$Creatine, y = supplementData$Placebo, paired = TRUE)

result1

## 
##  Paired t-test
## 
## data:  supplementData$Creatine and supplementData$Placebo
## t = 24.688, df = 59, p-value < 2.2e-16
## alternative hypothesis: true mean difference is not equal to 0
## 95 percent confidence interval:
##  1.777319 2.090836
## sample estimates:
## mean difference 
##        1.934078

Data Visualization

supplementData |> ggplot(mapping = aes(x = "", y = deltaData)) + geom_boxplot(fill = "skyblue", color = "black") + theme_minimal() + geom_hline(yintercept = 0, linetype = "dashed", color = "red") + geom_point() + scale_y_continuous(breaks = seq(0.5, 4, by = 0.5, limits = c(0.5, 4)))

## Warning: In seq.default(0.5, 4, by = 0.5, limits = c(0.5, 4)) :
##  extra argument 'limits' will be disregarded

###Parts of this were developed with support from ChatGPT(OpenAI, 2025)

Conclusions

Their is a significant difference in hair loss levels between creatine and placebo. Because I used a paired t-test on my data, I was able to compare the hair loss score for each individual on my simulated data. The differences were greater than 0, and the resulting p-value was 2.2e-16 which is below my alpha level of 0.05. Thus, I reject the null hypothesis: creatine causes significnat hair loss in young teens who take creatine monohydrate regularly. In additional to the significant difference, the histogram shows no visual outliers that affect my data, which means my data is valid and can proceed with my paired t-test. This conclusion is a simulated dataset and doesn’t reflect real effects. In retrospect, hairloss is affected by many confounding variables including genetics, diet, and any underlying health conditions I couldn’t include in this expriment. In the future, if I were to redesign this expriment, I would control more variables such as diet, sleeptime, and athletic activity to get more accurate results and reduce bias in the participants I choose. This means, I would do a blind and random assignment.

###Explain that I incorrectly asked for data. as I set the mean as expected hair growth instead of differences

Citations

OpenAI. (2025). ChatGPT (July 31 version) [Large language model]. https://chat.openai.com/