Skewed Science

This hits the nail on the head when it comes to people with personal biases.

http://www.youtube.com/embed/GI0sSgLUbXk

The Secret to Good Health PT.2: The Data - BOOZE is GOOD?

...Continued from part 1.

I then graphed the relationship between the variables and the output and found these conclusions.

Figure 4 Correlation Study: Hours of Sleep Effect on Change in BF%

Figure 4 illustrates that as you get more sleep bodyfat percentage goes up.  Coefficient of T-Statistics is -.0046, R2 is 1.9%, R value is 0.137.  Therefore the data supports that sleep is bad for you. This is not true and illustrates how data can be corrupt.

Figure 5 Correlation Study: Exercise Effect on Change in BF%

Figure 5 illustrates that as you get more exercise bodyfat percentage goes down.  Coefficient of T-Statistics is -3E-.06, R2 is 0.15%, and R value is 0.039.  The data supports that exercise is good for you.  However there factor in this relationship seems to be a flat line correlation (Y=-3E=06x+.0029) therefore there is no evidence to suggest any correlation.

Figure 6 Correlation Study: Booze Effect on Change in BF%

Figure 6 illustrates that as you drink more alcohol your bodyfat percentage goes down.  Coefficient of T-Statistics is -0.0003, R2 is 0.25%, R value is 0.5.  Therefore the alcohol is good for you.  As much as this study spits in the face of common logic I personally believe that the driver between this correlation is how alcohol is accompanied with peer bonding and other moral builders which could explain the positive benefits.

The Secret to Good Health (Pt. 1)

I attempt to answer the age old question what is best for my health using myself as (N=1) experiment.  In other words what is better: Sleep or exercise or diet?  The output that I wish to measure is “bodyfat percentage”.  The main reason I select this aspect to measure is how it strongly correlates to longevity a due to a good profile in lean body mass and linkage to good insulin sensitivity.  I the test subject being of moderate bodyfat (male 10-15%) is therefore a perfect candidate to test (since having a 4% bodyfat and losing bodyfat would be considered unhealthy).  I am not a medical professional… I just attempt to read the scientific numbers and this report is purely to illustrate the demonstration of using statistics.

Variables:  As stated before bodyfat change will be the measured output.  An increase will be seen as a negative mark against my health and a decrease will be viewed as positive event.  The three measures that have been recorded is 1) hours of sleep averaged in a month, 2) amount of exercise I average each day (measured in calories) and, 3) Alcoholic Beverages consumed in a total month (measured as a unit of 12oz beer, 8oz wine, 1.5oz hard liquor).

The change in bodyfat percentage was put on a chart to analyze the validity or the data and the distribution.

Block number	Cell boundaries	Frequency
1	-1.057%	3
2	-0.563%	2
3	-0.070%	7
4	0.424%	3
5	0.918%	3
6	1.412%
block size	0.494%

Figure 1 Chart of Changes in Bodyfat Change

Figure 2 Chart of Changes in Bodyfat Change

Figure 2 illustrates a normal distribution evenly centered about the median of the date.  Most of the measurements were in the 3^rd block which represented data of -0.070% to 0.424% change in bodyfat.

Month Sampled	30-day Sleeping average	Estimated amount of Exercise (Calories)	Alcoholic Beverages	Change in Body Fat
Sum	143.2	6454.3	455.3	0.0314
Standard deviation	0.2	84.2	11.4	0.0071
Sample mean	8.0	358.6	25.3	0.0017
Median	7.9	358.3	25.0	0.0021
Coefficient of Variation	2.7	23.5	45.1	407.1271

Figure 3 Analysis of Variables and Output

I then graphed the relationship between the variables and the output and found these conclusions.

Stay tuned for the data...

Control Charts in the Workplace

“A control chart is merely a graphical record of data taken from a repetitive process…” Physically the control chart is a data distribution turned sideways with the horizontal axis being successive tests, days, distance, or some other indication of order.” An example is shown below with the math to derive the points.

Control charts are a very practical idea because they depict what to accept and reject in a very simplistic way that someone on a assembly line can understand.  Formulas and scripts can be written to update the control chart.

Now why don’t we use this useful too?  I think it is because leaders are unfamiliar and to distant from their academic years to do the raw math to computing these limits.  The math is very simple but those who have the capability and education stay away from using such tools.  Also there is a stigma against using math and statistics for the same reason people distaste economist and their predictions.  As a manage/supervisor the greater leverage to increasing productivity is not to come at you employees with numbers and stats but more of the softer relationship building skills.  Although keeping score with statistics can create a sports team mentality… statistics as your primary form of communication creates disconnection and resentment amongst the workforce.

Bad Science: Correlation and Regression

Analyzing the quality product of a road surfaced based on some quality measure in either the material itself or compaction process makes sense and is proven by statistics to have a positive correlation.  Typically these correlations are very precise where the line of best fit can be used to estimate the quality of the final product.  I think these are very useful in conveying statistical findings to the field personal in the form of heuristics or “rules of thumbs”.  One of these useful ones is that for every 1% deviation in compaction spec equates to 10% off the life the road.  This rule of thumb can be used to send home a point to workers that their attention to quality is magnified in the service life of the final product and this is why quality in their work means so much. To use statistics in this manner is a powerful management tool to foster anotomony in employees.

Observational studies do not establish cause and effect it is a logical fallacy. It can be used to generate hypothesis that then can be tested to establish cause and effect.  This explains that some statistics can fit these lines of best fit but it is not the major contributing factor to the final outcome.  In other words for example we all know that “people who eat breakfast are less likely to be obese” and this can be proven statistically but is that the mechanism that is factoring to obesity? Perhaps people who eat breakfast go to sleep on time and aren’t snacking and playing video games to the wee hours of the morning?

The Craigslist Penis Effect

This idea is not novel by any means but it is a piggy back on someones elses idea that people out there suck.

"Basically The Craigslist Penis Effect describes situations where everyone else is so horrible that, by being even half-decent, you can dominate everyone else and win. These moron men on Craigslist would be better served writing 5 half-decent responses, testing to see which got the best response, and then sending it out instead of a picture of their generally mediocre manhood. I did exactly that for my friend on JDate and ended up getting very good at introducing girls to him. (Jewish women, beware of my Ashkenazi skills.)"

Call it doing your own research and a little personal accountability but again things aren't easy, they are simple. In 99.9% if you put a little effort into it such as 1 minutes or 5 minutes you beat the mean by leaps and bounds.

80/20 rule

I know blogging about work is a big no no but today was just a true case of pareto's law. (80 percent of your efforts come from 20 percent of your problems) So couple days ago I passed out these forms that I asked everyone to write their emergency information so that incase something bad happens I know who to contact. So two days pass and five verbal reminders spaced evenly throughout the meeting and out of my sample group of 43 participants, guess how many emergency contact sheet I don't get back??? 9!!!!

Fyi 43 people x 20% = 8.6 people