Abstract

From high tides to crazy moods, the full moon is believed to cause a number of things. As the full moon approaches, full-term pregnant everywhere prepare hopefully to go into labor. Is this an old wives tales or something mysterious that we hear plenty about but don’t yet understand.

There is no evidence to such a phenomenon. This study is based to look into the birth data in US and attempt to link them with to full moons to see if there is any correlation. We will first try to identify some good data sources for the study. Based on the data, it will be filtered, formatted and grouped into appropriate categories for the study.

The prepared data will then be processed using statistical methods to reveal the facts around the myth and understand the rationality of the belief which has been around for a very, very long time. Many cultures around the world hold strong beliefs in this link and the folklore tends to stretch back forever.

Introduction

Myth or Fact: Do more Women Go into labor During a Full Moon ?

Compare the birth counts on regular days with the full moon days.
Do you see any difference? Can we draw any inference based on the data?

Null Hypothesis: There is no effect of Full Moon with Women going into Labor and child being born.

Alternate Hypothesis: Women go into Labor due to Full Moon, there are more births on full moon nights.

Below is the reference to an article on this topic:
_Ref: https://www.dukehealth.org/blog/myth-or-fact-more-women-go-labor-during-full-moon_

US births from 2000-2014, data source: fivethirtyeight.com

Time series of full moon occurrences, data source: Kaggle.

US Birth Data

Full Moon Days

Format Full Moon Data

Split Date description into Date, Month and Year

Highest Birth Count By Year (Top 6)

Group By Year, Sum and Sort

2007 birth count by month

## [1] "Total Births(mean) in 2007 is 144008.61"

## [1] "Mean Birth per day in 2007 is 12000.72"

Full Moon Births in 2007

## [1] "Mean Birth per full moon day in 2007 is 12934.00"

Merge and Compare the births counts (Regular vs Full Moon)

One of the most common statistical tasks is to compare an outcome between two groups.

Plot 2007 regular births and full moon births

T-Test Analysis

birth.t.test <- t.test(merge_by_month_data_2007$mean, merge_by_month_data_2007$births, data = merge_by_month_data_2007)
birth.t.test

## 
##  Welch Two Sample t-test
## 
## data:  merge_by_month_data_2007$mean and merge_by_month_data_2007$births
## t = 0.093272, df = 12.465, p-value = 0.9272
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -1570.097  1711.144
## sample estimates:
## mean of x mean of y 
##  12009.60  11939.08

# p-value
birth.t.test$p.value

## [1] 0.9271695

# confidence interval
birth.t.test$conf.int

## [1] -1570.097  1711.144
## attr(,"conf.level")
## [1] 0.95

High p-values indicate that the evidence is not strong enough to suggest an effect exists. It’s possible that the sample size is too small for the hypothesis test to detect it.

Lets try to increase the sample size and run the analysis again.

Birth data sample(2000 - 2013)

Birth count by month

## [1] "Sum of Births(mean) in 2000-2013 is 1911016.25"

## [1] "Mean Birth per day in 2000-2013 is 11375.10"

Full Moon Births (2000-2013)

## [1] "Mean Birth per full moon day in 2000-2013 is 11784.91"

Merge and Compare data

One of the most common statistical tasks is to compare an outcome between two groups.

T-Test Analysis

## 
##  Welch Two Sample t-test
## 
## data:  merge_by_monthyear_data_bigset$mean and merge_by_monthyear_data_bigset$births
## t = -0.34611, df = 187.91, p-value = 0.7297
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -423.1707  296.8438
## sample estimates:
## mean of x mean of y 
##  11381.14  11444.31

T-Test Results

# p-value
birth.2t.test$p.value

## [1] 0.7296505

# confidence interval
birth.2t.test$conf.int

## [1] -423.1707  296.8438
## attr(,"conf.level")
## [1] 0.95

Based on the T-Test the p-value is 0.729, which is a large value indicating weak evidence against the null hypothesis, so fail to reject it.

Data Visualization

Histogram of the average births

Data Visualization

Histogram of the full moon births

## Warning: Removed 153 rows containing non-finite values (stat_bin).

## Warning: Removed 2 rows containing missing values (geom_bar).

Scatterplot of average births by month

Scatterplot of Full Moon births by month

Scatterplot the births on any day of the week

Scatterplot the full moon births by day of the week

Conclusion

Based on the statistical analysis, there is no conclusive evidence to prove that there is an effect of Full Moon with more number of birth occurrences. Therefore, the Null Hypothesis holds true and the Alternate Hypothesis is rejected based on the above calculations.

The average birth count in both catergories is between 11000 and 12000. Although, looking at the Scatterplot of Full Moon births by month, we can see the high variance of birth counts from the mean. The birth counts range from 7000 to 15000 with concentration of counts on both end of the spectrum. This leads to believe that on Full Moon nights depending on some other ‘unknown’ variables, it can stimulate or defer women going into labor.

This can be looked into and studied further by using other statistical methods and/or adding other variables for analysis of this cultural myth, hoping to reveal the facts and understand its nature.

Questions & Answers