https://github.com/agahkarakuzu/template-a5-chaudhuri-practical

https://github.com/agahkarakuzu/template-a5-chaudhuri-practical

Repository

https://github.com/agahkarakuzu/template-a5-chaudhuri-practical/blob/master/

# Demonstrating *p*-hacking with real data
*Paramita Saha Chaudhuri, PhD (paramita.sahachaudhuri@mcgill.ca)*

In this week's reading from [Simmons and colleagues (2011)](http://journals.sagepub.com/doi/10.1177/0956797611417632), we learned about "researcher degrees of freedom."
That is,

> exploratory behavior [that] is not the by-product of malicious intent, but rather the result of two factors:
> - ambiguity in how best to make these decisions and
> - the researchers desire to find a statistically significant result.

Here, we will see the difficulty of these degrees of freedom when analyzing data.

## Background
[Hypoxic Ischemic Encephalopathy (HIE)](https://www.birthinjuryguide.org/birth-injury/types/hypoxic-ischemic-encephalopathy-hie/) is a type of brain damage that occurs when an infant's brain doesn't receive enough oxygen and blood.
It is a dangerous condition that requires immediate medical intervention.

HIE affects 1.5 out of 1000 newborns in developing countries.
While there are some treatment options—such as [therapeutic hypothermia](https://www.birthinjuryguide.org/birth-injury/treatment/neonatal-therapeutic-hypothermia/)—that improve neurological outcomes, at least half of these infants still experience adverse effects in their early childhood.
There is therefore a pressing demand for reliable biomarkers to guide early clinical decisions and to help counsel parents.

The [included data](https://github.com/reprocourse/template-a5-chaudhuri-practical/blob/master/1-longitudinal-minimal%20data%20set-V2.csv) is a slightly altered version of data [from a published article](https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0184593).
You will find definitions for all listed variables in the included [variable descriptions](https://github.com/reprocourse/template-a5-chaudhuri-practical/blob/master/variable_descriptions.md) file.


## The present study
You and your collaborators are interested in investigating [copeptin](https://www.ncbi.nlm.nih.gov/pubmed/18291667) and [neuron-specific enolase (NSE)](https://www.mayomedicallaboratories.com/test-catalog/Clinical+and+Interpretive/80913) as two novel biomarkers to predict long-term neurodevelopmental outcome after birth asphyxia.
For the present study, there is a primary aim:

**Aim 1:** Are copeptin and/or NSE associated with favorable neurological outcomes at two years post-HIE?

To investigate this aim, we have sketched out [an analysis plan](https://github.com/reprocourse/template-a5-chaudhuri-practical/blob/master/analysis_plan.md).
After you have examined the listed relationships, we encourage you to conduct additional exploratory analyses to find associations between long-term neurodevelopmental outcome and any biomarker or other characteristic available in the dataset.

Please update [the analysis plan](https://github.com/reprocourse/template-a5-chaudhuri-practical/blob/master/analysis_plan.md) as you complete the assignment.
Please also record all code that you generate in the associated [`hie_analysis.py`](https://github.com/reprocourse/template-a5-chaudhuri-practical/blob/master/hie_analysis.py) script.

## Useful resources

Previously introduced  Python statistics packages are likely to be very helpful for this assignment.
In particular, the [SciPy lecture notes on Statistics in Python](http://www.scipy-lectures.org/packages/statistics/index.html) and the [`statsmodels` getting starting guide](http://www.statsmodels.org/stable/gettingstarted.html) are good references for statistical analysis in Python.

Owner

GitHub Events