This entry is the first in the series exploring the effects of performance enhancing drugs on strength expression. In order to compare apples to apples, we must first quantify the differences between federations, starting with the effect of the deadlift bar.
One of the most obvious differences between powerlifting federations (besides the drug testing, bench press rules, weigh-in to competition lag…) is the barbell they use for the deadlift. There are two main choices:
- Power bar (aka “Stiff” bar)
- Deadlift bar.
There are a lot of differences between the two barbells. For example, a power bar has a diameter of 29mm and a deadlift bar has a diameter of 27mm, which makes it easier to grip. The deadlift bar also has more space between the sleeves (56” vs. 51.5”), is an overall longer bar (90.5” vs. 86.5”), and has a lower tensile strength (185psi vs 200+psi) than a power bar. The latter 3 qualities mean it’s easier to make the bar bend during the lift, thus making the lift easier.
Here’s a photo of deadlifter Cailer Woolam comparing the two bars taken from his video here (https://www.youtube.com/watch?v=9khgliEs3ws). You can see how the deadlift bar (pictured right) bends far more than the stiff bar (left) before the plates leave the ground, making the deadlift bar slightly easier.
The real question is how much easier?
The easy answer to compare the difference between the two bars is to test by having the same group of people do a 1RM test with both the power bar then the deadlift bar and see how they differ. However this gets difficult because you need to control for meet prep peaking, fatigue, and if you’re not doing them on the same day, you also need to check that the athletes are also not just getting stronger between the trials because they’re training anyway.
All those conditions make it practically an impossibility to accurately test using those methods. I guess you could do it with large sample sizes and just hope the differences average out, but then you’re left with trying to find a large group of powerlifters capable of hitting big deadlifts. However we can do it using the entire competition dataset with some assumptions, which I’ll outline below.
I’m going to use the data from http://www.openpowerlifting.org downloaded on 2/20/2020.
So to ensure apples-to-apples comparisons, let’s talk about some of the assumptions:
- I’m accepting openpowerlifting’s dataset as complete and correct. There might be some entries missing due to lifters asking to be removed, and there might be some competitions missing, but I believe that the imperfect information is insignificant enough to be ignored.
- “Drug tested” at face value. If a federation is listed as tested, I’m going to assume that there isn’t any significant number of people getting away with cheating. There might be more people getting away with cheating, but I will assume it’s not enough to significantly swing the analysis in any direction.
- The IPF attracts more drug-tested talent than any other federation. This one is probably the most controversial assumption I’m making. During the exploratory data analysis, the IPF competitions had a significantly bigger proportion of entries that were considered outliers, pushing the stiff bar values much higher. We’re going to take those out as that level of drug-tested competition isn’t present in other federations.
- The “average” (ie. not elite) lifters choosing to lift in an IPF (stiff bar) fed vs. IPL (deadlift bar) fed are the same. In other words, the “average” lifter just goes to whichever fed is more convenient and there’s no real reason one would choose to lift in a federation with a deadlift bar or a stiff bar.
Additionally, some filters need to be added. In order to get the most accurate comparison of populations we’re only looking at males who complete full power meets in the raw category. In other words, there are no “deadlift only” numbers in here, or equipped numbers. After all the filters we’re left with a dataset of:
|Stiff Bar Entries: 101504|
|Deadlift Bar Entries: 29370|
Note: The USAPL is HUGE and makes up for over half of the stiff bar entries.
First let’s look at the cumulative step histogram. As you can see, there are more people lifting more weight on a deadlift bar. This is great as it confirms the hypothesis that the deadlift bar makes the deadlift easier.
To interpret this graph, pick a weight on the x axis, then draw a line straight up to get the % of lifters whose deadlift is less than the x value. For example, if you draw a line straight up from 250kg, approximately 75% of lifters cannot deadlift that weight on a deadlift bar, and 80% cannot deadlift that weight on a stiff bar. In other words, a 250kg deadlift bar is in the top 25% of deadlifters on a deadlift bar, and in the top 20% of deadlifters on a stiff bar.
Clearly there’s a difference between the deadlift bar and the stiff bar. Now we just have to quantify both of those lines to get the difference between them.
Since we’re trying to find the “average” difference between a deadlift bar and a stiff bar, the next thing we’re going to do is remove the outliers which we’re defining as the quantiles between the 15th% and the 85th%. There just aren’t enough values on the top and bottom ends to get meaningful results.
So in order to model this data, we’re going to use the edges of the cumulative step histogram to generate points, then we’re going to log transform it to straighten it out so we can run a linear regression. After undoing those log transformations and plotting the regression back on the original cumulative step histogram we get:
This is excellent. Using the two regression lines, we can convert a 1 rep max deadlift from a stiff bar to an estimated deadlift bar (and vice versa).
Plugging my most recent stiff bar max of 250kg into the formula, it says my deadlift bar max is 255kg, which makes sense given my last competition on a deadlift bar was a real sandbagged 252.5kg.
The intersection of the two lines is around 165kg, which means that below that weight, there is no significant deadlift bar effect.
In English, this model is saying a 250kg stiff bar deadlifter is the top 77th percentile. The 77th percentile deadlift bar deadlifter pulls 255.5kg. Since the populations are roughly equivalent, we expect a 250kg stiff bar deadlifter to deadlift 255.5kg on a deadlift bar.
Based on the chart below, you can see the difference gets bigger and bigger as the lifter gets stronger. This is consistent when you look the two bars from a mechanical point of view. More weight means the bar bends more, which means the lifter doesn’t feel the entire weight of the bar until after the bar has bent a certain distance.
|Stiff Bar Weight (kg)||Projected Deadlift Bar Weight (kg)|
Github code can be found here: https://github.com/UBERCRUZER/powerliftingPEDs/blob/master/analysis.ipynb