Feature Article - December 2018 |

by Do-While Jones |

Misleading math can lead to incorrect conclusions about evolution.

Nearly every article about evolution in the professional literature is full of calculations. **Many times evolutionists (usually unknowingly) draw incorrect conclusions based on those calculations.** To illustrate this, we will use actual data and perfectly valid mathematics to show there is a connection between the temperature in Moscow, Russia, and the number of runs scored by the Los Angeles Dodgers baseball team.

To be perfectly clear, **we are not saying that evolutionists intentionally misuse math** to prove something that isn’t true (although in very rare cases an evolutionist might). We suggest that **evolutionists who want to find mathematical proof of evolution will not question a calculation which seems to prove their point.** And, to be fair, we will point out a classic example where **creationists have done the same thing.**

Furthermore, **although we will prove the Russian connection using verifiably correct data and accurate, impartial calculations done by an Excel spreadsheet, we do not really believe that Russia affected the number of runs scored by the Los Angeles Dodgers. We are simply using a silly example to make our point.**

Finally, **we are going to try to make this explanation as enjoyable and understandable as we can for people who don’t like math.**

The phrase “statistics and probability” is sometimes used as if statistics and probability are the same thing. There is an important difference. **Statistics help us understand what happened in the past. Probability tells us what may happen in the future.**

Las Vegas is built upon statistics and probability. Statistics tell the casino owners how much money was bet on roulette tables in previous years. Based on those statistics, casino owners can estimate the amount of money that will be bet on their roulette tables next year, and probability will tell them how much money they can expect to win next year if that much money was actually bet on the tables.

**The classic mistake some creationists make is that they try to use probability to predict the likelihood of something that happened in the past.** You no doubt have heard **the “Tornado in a Junkyard” probability argument.** You can’t make probability calculations about what happened in the past. (Well, technically, you can make the calculations—but they are meaningless.) **Probability only works for future events. Probability can’t tell what happened in the past.**

Consider this example: What is the probability that, sometime in the future, a man will be born on February 26? That’s easy to calculate. The probability someone’s birthday is February 26 is 1/365, which is 0.0027. ^{1} The probability that someone will die on September 12 of any given year is also 1/365. The probability that someone born on February 26 will die on September 12 is 0.0027 times 0.0027, which is 0.0000075. Those are pretty slim odds.

Similarly, the probability that a woman will be born on June 23 and will die on May 15 is 0.0000075. The probability that those two people will marry is 0.0000075 times 0.0000075, which is 0.000000000056. Since the probability that anyone’s anniversary is March 1 is also 1/365, the probability those two people will get married on March 1 is 0.0027 times 0.000000000056, which is 0.00000000000015. That’s not very likely.

Furthermore, what are the odds that they will get married in Franklin, Kentucky? There are so many places in the world to get married, the probability is just too small for me to calculate. **Therefore, the probability that (in the future) somebody will be born on February 26 and marry someone born on June 23 on March 1 in Franklin, Kentucky, and will die on September 12 after his wife dies on May 15 is too small to imagine.**

**That incredibly low probability does not prove that Johnny Cash did not marry June Carter Cash on March 1, 1968.** As improbable as it was, it did happen. Probability tells the likelihood of something happening in the future, but cannot be used to compute the likelihood that something happened in the past.

That’s why **we have never used the improbability argument against abiogenesis** (the natural, spontaneous generation of the first living cell). **Low probability for a past event is never a valid proof.**

Yes, the probability that chemicals will spontaneously combine to form a living cell is so small that it is reasonable to believe that it will not happen in the future; and **that is good reason to believe it didn’t happen in the past—but it isn’t proof that it didn’t happen** in the past. **The real proof** that it didn’t happen in the past **is that scientists have been unable to cause it to happen in the laboratory.** The reason scientists haven’t been able to cause it to happen in the laboratory is **because one has to violate natural laws to do it.** The scientific proof against abiogenesis is impossibility—not improbability. Scientific research has shown abiogenesis cannot happen through natural processes.

Just as creationists sometimes misuse probability, **evolutionists sometimes misuse correlation to prove causality.**

Mathematicians might cringe at the way we will oversimplify this; but here is **a simple explanation of correlation: When one thing increases, another thing increases, too.** Let’s illustrate that with some simple numbers.

X | Y1 | Y2 | Y3 |
---|---|---|---|

1 | 9 | 20 | 51 |

2 | 14 | 22 | 46 |

3 | 19 | 24 | 41 |

4 | 24 | 26 | 36 |

5 | 29 | 28 | 31 |

6 | 34 | 30 | 26 |

7 | 39 | 32 | 21 |

8 | 44 | 34 | 16 |

9 | 49 | 36 | 11 |

10 | 54 | 38 | 6 |

Correlation | 1 | 1 | -1 |

X is the independent variable plotted on the X axis. It goes from 1 to 10. The values in the Y1 column are dependent values which go from 9 to 54. They are plotted as black diamonds. The values in the Y2 column are plotted as red squares, and the values in the Y3 column are plotted as yellow triangles.

**The Y1 and Y2 values increase as X increases, so they have a positive correlation. The Y3 values decrease as X increases, so Y3 is negatively correlated to X.**

Because all the values fall on a perfectly straight line, the correlations for Y1 and Y2 are +1, and the correlation computed for Y3 is -1. **The fact that the Y1 line is steeper than the Y2 line is irrelevant. Correlation has to do with how well the values fit on a straight line—not how steep that line is.**

Let’s take the Y1 values and corrupt them to create the Y4 values below.

X | Y4 |
---|---|

1 | 9 |

2 | 8 |

3 | 22 |

4 | 24 |

5 | 16 |

6 | 40 |

7 | 44 |

8 | 32 |

9 | 79 |

10 | 54 |

Correlation | 0.85785 |

Because they don’t fit perfectly on a straight line, the correlation drops to 0.85785. If none of the values fit a straight line at all, the correlation would be 0.000.

The Excel spreadsheet calculates correlation according to the standard mathematical definition of correlation, which we don’t want to try to explain. All you need to know is

- Correlation values range from +1 to -1
- Positive correlations (regardless of magnitude) mean that Y increases as X increases
- Negative correlations (regardless of magnitude) mean that Y decreases as X increase
- Correlations very near 0 mean that the values of X and Y are not correlated
- Correlations near +1 or -1 mean that the values of X and Y are definitely correlated

Now that the tutorial is over, we can examine how the weather in Moscow, Russia, influenced the number of runs scored by the Los Angeles Dodgers last season.

The Internet can be used to find the scores of every regular season game the Dodgers played,^{2} and the daily high and low temperatures at the Sheremetyevo Station in Moscow.^{3} We copied the high temperatures for 162 Dodger dates, and the runs scored on those days (with two entries for the double-headers) into an Excel spreadsheet and analyzed the data. We placed the spreadsheet on our webpage ^{4} so you can examine all the raw data and calculations if you like.

We showed you the graph of the data at the beginning of this essay. Here it is again:

Looking at the graph, you can see that the temperature in Moscow at the start of the season was around 50 degrees (Fahrenheit), it got hotter in the summer, and cooled off in the fall. (Duh!) There is no obvious trend in the baseball scores.

If you plot the runs scored versus the temperature in Moscow, the entire data set looks like this:

This was frankly surprising to me. I expected the data points to be bounded by a rectangle rather than a triangle because the number of runs scored has nothing to do with the temperature in Moscow. Because the data isn’t a rectangle, **the correlation coefficient** isn’t zero. It **is +0.027**, which is a slightly positive correlation.

**The day of the week should not matter—but it seems to.** If we just look at all the games played **on Thursdays, the correlation coefficient is +0.504** and the scatter plot of the Thursday data looks like this:

**The significant thing is not the day of the week—it’s the fewer number of data points.** Two data points will always fall on a perfectly straight line, and will always have a correlation of exactly +1, 0, or -1 regardless of whether or not they are correlated. The more points you add, the better the chance that not all the data points will fall on a straight line. **Fewer data points generally result in higher correlations.**

If you take all the Dodger data and divide it into seven data sets according to days of the week, and compute the correlation coefficient for each of the individual days, they are all as far, or farther, from zero than the correlation coefficient for all days combined because each day has fewer data points. Four are negative, and three are positive.

Weekday | Correlation |
---|---|

All Days | 0.027 |

Sundays | -0.027 |

Mondays | -0.026 |

Tuesdays | 0.194 |

Wednesdays | -0.159 |

Thursdays | 0.504 |

Fridays | 0.060 |

Saturdays | -0.164 |

**When evolutionists analyze their data statistically, they are likely to find correlation because there aren’t very many samples to analyze.** There just aren’t very many relevant fossils. In the beginning, there weren’t very many genomes that had been decoded, either. There still aren’t very many genomes that have been decoded—but the more that are decoded (as we expected) the worse the DNA analysis fits the evolutionary theory.

Sometimes evolutionists erroneously try to correlate things to prove causality.

**The Thursday scores and temperatures really are correlated.** That does not prove that the temperature in Moscow on Thursdays affects the number of runs scored by the Dodgers that day. In fact, **it is the number of runs that the Dodgers score on Thursdays that causes the temperature to rise in Moscow!** The solution to the Moscow warming problem is to get the umpires to give the Dodgers just one swing each time at bat so the Dodgers will score fewer runs, which will make it cooler in Moscow! (Hopefully, no politician will read that last statement and pass a stupid law based on it.)

Seriously, if A and B really are correlated, it might be because A causes B; but it could be because B causes A; or an unknown process C causes both A and B; or there might be no causal relationship at all—it’s just a coincidence. **Correlation does not prove causality.** But math is so seductive when it confirms a pet theory, it is hard not to latch onto it.

Quick links to | |
---|---|

Science Against Evolution Home Page |
Back issues ofDisclosure(our newsletter) |

Web Site of the Month |
Topical Index |

Footnotes:

^{1}
Yes, I know that birthdates, death dates, and anniversaries aren’t equally likely; but let’s pretend they are so as not complicate the example too much.

^{2}
https://www.cbssports.com/mlb/teams/LAD/los-angeles-dodgers/schedule/regular/

^{3}
https://www.wunderground.com/history/monthly/ru/moscow-sheremetyevo/UUEE/date/2018-3

^{4}
http://scienceagainstevolution.info/v23i3.xls, the printable version is at http://scienceagainstevolution.info/v23i3xls.pdf