by Lawrence M Lesser
At my first checkup
with this doctor, he scans the labs,
declares me diabetic for life
like 1 in 10 Americans.
Just like that: normal to diabetes,
skipping prediabetes. He
pulled the trigger not from my A1C, but from
a tolerance test, possibly false
with too few carbs,
a pitfall I learned not from him or the lab,
but frantic Googling
this diagnosis.
This internist is also a diabetologist,
so diabetes may be his null,
as the Anderson-Darling null is normality,
with default to launch protocol.
So is my Type 2
a Type II error?
He may not care
I’m a researcher:
I’d love a CGM to show in real time
what spikes my sugar and its CV or my time out of range.
He says that’s too complex –
it’s glucometer and finger pokes for me. So back home I
pull the trigger on my lancing device,
and the pain’s less than bracing
for the prick.
Do I have tolerance for this? The monitor’s MOE
exceeds half the window of
what’s normal. Whatever
label may stick, I want to change.
Given how unmindful I’ve been with
my potbelly and what goes in it,
there’s power in turning carbs to servings
and each plate into a pie chart, reducing portions
of things like pie.
Note: this poem was written before the diagnosis was deemed incorrect and overturned
lyrics by Jamie Tan Xin Yee, Joelyn Chong, Deston Tang, Christine Sia, Nellie Lee, Josiah Tan, and Lee Yi Yuan
may be sung to the tune of "LOVE" by Bert Kaempfert and Milt Gabler
Let us teach you a thing or two
Stats is more than just concepts, for you
Stats, is very useful, can be found around us
Here’s a simple song for you to get to know it better
F, tests for different variances
Chi, looks for a relationship, you see
Z, checks for both means and true proportions
And T, is used when sigma is unknown to us, so...
One, decide on your hypothesis
Two, ascertain your significance level
Three, identify which test you should be using, and
Four, complete the steps and analyse your results, and
These, are four basic tests, for stats
All, can be done in a few simple steps
Stats, is more than numbers so make sure to analyse it
Come, let me show you what to do
Come, let me show you what to do
Come, let me show you what to do!
by Lawrence Mark Lesser
When doing hypothesis tests,
There's no threshold value that's best.
It's very contrived
To use point-oh-five
To say if you should be impressed!
Lyrics copyright by Dennis K Pearl
May sing to the tune of "When I'm Sixty-Four" (Beatles)
When collecting data, try to be fair
Many years from now
Will you still have a null value defined
Test to construct, with a p to find?
If I get one ‘n three-quarters for zee
Would you lock the door?
Where should p lead you, will you still need me
When I'm point oh-four?
If the science changes
Shouldn’t the test?
I could work in stages
Take mu as random, don’t draw me a line
State your point of view
Indicate precisely what your mean will say
Information wasting away
Give me your posterior, and its form
changing evermore
Where should p lead you, will you still need me
When I'm point oh-four?
A sign test has its pluses and minuses but the Wilcoxon does first things first.
Larry Lesser
I was in denial about using a non-parametric test - but the signs were all there.
Dennis Pearl