The Elephant in the Room
 
This article will deal with a study that compared the effects of two start positions. Let’s begin by dealing with the elephant in the room: a weightlifter’s start position will be heavily influenced by the lifter’s proportion. However, as a starting point, lifters are taught that the bar should be roughly over the juncture of the metatarsal/phalangeal joints (the second joint of the big toe) and the shoulders should be positioned directly over or slightly forward of the bar. In addition, this start position will usually have the hips relatively low, at least lower than a typical powerlifting deadlift start position. Our elephant will make another appearance a bit later in the article.
 
The Hypothesis and Study
 
Mark Rippetoe has proposed we should set up for the clean and snatch in a similar position to that of a deadlift. If you read my previous article, you’ll remember that I criticized Mark Rippetoe for the idea that weightlifters should do low-bar back squats. You may be thinking that I have something against him. I don’t. As I said in my previous article, I am not here to criticize Mark Rippetoe or his Starting Strength program. From what I understand, many novice trainees gained a good deal of size and strength on this program. I have no issue with Starting Strength as a general strength program.
 
Back to the topic at hand. Rippetoe recommends starting the pull with the bar over the midfoot and the hips relatively high. He claims that this results in a more efficient first pull because it allows the bar to be lifted in a straight vertical line, which should result in more power being applied in the second pull because the lifter does not need to expend any energy in pulling the bar toward himself or herself. John Petrizzo, an assistant professor in Department of Exercise Science, Health Studies, Physical Education, and Sport Management at Adelphi University and a Starting Strength coach, conducted a study to determine if this “deadlift-like” start position would result in a larger power output. This is a review of that study.
 
How the Current Start Position Developed
 
Although Dr. Petrizzo’s study dealt with placement of the bar in relation to distance from the shin, for most lifters, starting with the bar over the midfoot (about one inch from the shins) will require a higher hip position because a lower hip position will require angled shins which will push the bar forward and away from the midfoot. The Starting Strength method of coaching the start position in the clean and snatch specifically tells the lifter to keep the hips fairly high (skip to 27:12 in the video). Therefore, my article will frequently refer to a “high hip” start position, but this is the same start position that Dr. Petrizzo tested.
 
Dr. Petrizzo believes that our current start position, which he calls the “traditional” or “TRAD” start position, is simply based on “the lasting influence exerted by the original rules that governed the sport subjective approach that has been used historically.” The “old rules” here refer to the rules in place before 1964 when the bar was not allowed to come into contact with the body during the pull phase. For you trivia buffs, this is how the term “clean” came into being – the bar had to be lifted “clean” of the body (i.e., no contact). He refers to the start position recommended by Rippetoe as the “experimental” or “EXP” start position.
 
In support of his contention that the TRAD position is based on, well, tradition, Dr. Petrizzo cited a study that dealt with similar start position issue involving the deadlift. The cited study made the same contention—the reason lifters set up for the deadlift with the bar over the second joint of the big toe was because they copied the start position of weightlifters before the 1964 rule change. This study, however, did not provide a citation for this proposition. I searched the Interwebz for some resources. The old York barbell courses from 1938 included the snatch as an exercise, complete with an image of the “correct start position,” but no detailed explanation on how to perform the lift. If anything, the image of the “correct start position” shows a position with relatively high hips which undercuts Dr. Petrizzo’s contention. I also searched the YouTubez for some pre-1964 weightlifting footage. This 1962 footage shows lifters using both low hips and high hips. I could spend all day watching old weightlifting footage and reading about weightlifting history. From my review of all of this, it does not seem that there was a set start position before 1964.
 
In the 1970s, the Soviet sports juggernaut enlisted the help of biomechanists to determine optimal pulling technique. An excellent summary of their findings can be found on Sean Waxman’s blog. One such biomechanist was R.A. Roman who wrote “The Training of Weightlifter.” Dr. Roman specifically recommended that the bar start over the metatarsal-phalangeal joint (the second joint of the big toe). Roman explained the logic of this as follows: "The movement of the barbell in this manner is more advantageous than a strictly vertical elevation of the weight. This action minimizes the horizontal distance between the bar and the hips, decreasing the moment force, allowing for proper utilization of the leg extensors. If a straight barbell path is used, the common center of gravity will be shifted forward towards the toes. This causes a large increase in the moment forces of all the working joints and a decrease in the body’s ability to create the acceleration, speed and power necessary for an efficient lift.” Roman stated that “the movement of the barbell in this manner is more advantageous than a strictly vertical elevation of the weight.” Contrary to Dr. Petrizzo’s contention, it seems that the TRAD start position is not simply a holdover from pre-1964 based on “this is the way we’ve always done it” but is based on a biomechanical analysis of optimal pull mechanics. Moreover, Roman specifically addressed why a vertical first pull is not ideal, and we should instead start the bar forward of the midfoot and then pull back: "The movement of the barbell in this manner is more advantageous than a strictly vertical elevation of the weight. This action minimizes the horizontal distance between the bar and the hips, decreasing the moment force, allowing for proper utilization of the leg extensors. If a straight barbell path is used, the common center of gravity will be shifted forward towards the toes. This causes a large increase in the moment forces of all the working joints and a decrease in the body’s ability to create the acceleration, speed and power necessary for an efficient lift.” In “The Training of Weightlifter,” Roman again wrote that “[t]his direction of the barbell’s movement has a big advantage over a strictly vertical trajectory. The shifting of the bar towards the torso (during the knee extension) reduces the ‘toppling over’ of the moment force of gravity over the barbell.” Not including this analysis by Roman is a big omission from the study.
 
Methodology
 
The study had 10 participants–seven male and three female. The participants were required to have at least two years of weightlifting experience. Three participants had competed at the national level. The weight used for the test was 90 percent of 1RM. The 1RM was self-reported. The lowest 1RMs were about 44 kgs and 61 kg for the snatch and clean respectively. These seem low. To be fair, Dr. Petrizzo noted that the study had a few very petite female lifters. Still, these low numbers indicate that some of the lifters were likely inexperienced.
 
In order to be consistent, the TRAD position had all of the lifters start with the bar over their metatarsophalangeal (MTP) joints. Here’s where our elephant returns. Given that a lifter’s proportions influence start position, it was possible that a few of the lifters’ normal start position was somewhere other than with the bar directly over their MTP. While I understand the study needed uniformity, the test should also have allowed the lifters to start in their normal starting positions as the “control group.” Without data from this control group, it is possible that some of the lifters’ TRAD start position was not their normal start position. The study also had a set warm up that every lifter had to use. Again, this could have been different from some of the lifters’ normal warm up.
 
Results
 
Less Horizontal Displacement
 
The study found that both the TRAD and the EXP start positions had horizontal displacement, but the EXP position reduced horizontal displacement by 40.3 mm and 31.9 mm for the snatch and clean respectively. The study used Dartfish software to trace bar paths, but the horizontal displacement measurements were taken only for the first pull. We know that in the TRAD position, the bar moves back toward the lifter. I wear a size nine weightlifting shoe. If I started with the bar over my midfoot instead of my MTP (where I normally start) the bar would move back about 90 mm. The study essentially confirmed what could have been measured with a ruler. However, the study’s hypothesis was that reducing the horizontal displacement would make the lift more efficient. In order to test this, the study measured power outputs for each start position.
 
Power Output
 
The study found that the EXP start position resulted in higher average power outputs of 2.2 percent and 2.6 percent for the snatch and clean respectively. This margin equaled about 15 watts and about 20 watts higher for the snatch and clean respectively. However, this result was not statistically significant. In this podcast Rippetoe and Dr. Petrizzo discussed the study. Rippetoe stated that while a (roughly) 2 percent increase in power may not be statistically significant in the academic world, it would be significant on the competition platform since the difference between first and second place is usually less than 2 percent. Dr. Petrizzo agreed. Unfortunately, this is not what is meant by the term “statistically significant.”
 
A Brief Look at Statistical Significance
 
The Placebo Effect
 
Discussing statistical significance can get complicated (and boring) in a hurry, so let’s start with something most of us have heard about – the placebo effect. For those who may not know, the placebo effect occurs when you give a person with an illness something that looks like medication but doesn’t actually do anything, but you tell the person that the fake medication will actually make them feel better. Some people will actually feel better on this fake medication, while others will not. This is why research on drugs always compares the effect of the real drug with a placebo group. Researchers want to make sure that it is the actual drug that makes people feel better and not just the placebo effect. As mentioned, some people do not respond to the placebo effect, and even those who respond may do so differently. This means that even among the placebo group, a certain percentage will get better. In order for a study to state that a new drug actually works, the researchers hope that the group that received the actual medication had a higher percentage who got better. But how big a difference should this be? Also, it is not as if researchers can predict which study participants respond to the placebo effect, so the number of placebo responders in either group is based purely on chance. So how can the researchers know that the group receiving the actual medication didn’t, based purely on chance, have a higher percentage of participants who responded to the placebo effect? This is what a test of statistical significance attempts to determine.
 
For purposes of the study, there could have been slight a placebo effect in that when the lifters were placed in the EXP start position, some may have thought that they were using a cool new lifting technique that would change the weightlifting world, and this caused them to pull with a bit more enthusiasm. But there are other chance factors that likely came into play. Suppose I ask a lifter to perform five singles with 80 percent of her 1RM. She can warm up however she wants, use her normal start position, and rest as long as she wants between reps. I also make sure that her surroundings remain as consistent as possible – no changes in temperature or music (if any is playing). Given these conditions, what is the likelihood that our lifter will perform each lift with exactly the same power output? This is not very likely, since humans do not come with a keypad where you can punch in the exact power output you want. Given that we tried to make the conditions for each lift as consistent as possible, there is no reason for our lifter to use different levels of power other than random human error. These same random variations in power output could have resulted in the average power output of the EXP to be higher.
 
Another source of error in studies is researcher error. For example, a researcher could misread a number on one of the instruments or “fat finger” a number when entering it into the database. I am not accusing the researchers in the study of performing poor workout. I am simply saying the reality that mistakes can be made. There can also be errors in how the equipment measured the data. The study used TENDO Power and Speed Analyzer to measure power outputs. Again, I am not accusing the manufacturer of the TENDO unit of making a poor product. However, I am sure even the manufacturer will admit that the device has a margin of error plus or minus some number.
 
Statistical Significance
 
As with all studies, various “random noise” could have crept its way into the start position data to make it appear as if the EXP start position allowed a lifter to produce more power, when in reality the difference was due solely to the random noise and not because of the EXP position.  This is where the test of statistical significance comes into play. Calculating statistical significance causes brain damage and is best done by statistical software. For our purposes, we just need to understand the concept: a result is statistically significant if the calculation shows that the result of an experiment was not due to the random noise but is instead likely to be attributable to a specific cause. The study ultimately found that the 2 percent increase in average power output when lifters used by the EXP was not statistically significant. This means that the EXP start position provided the lifters with no advantage for increasing their power output; the difference was simply due to the “random noise” errors.
 
The definition I used above used the phrase “likely to be attributable.” This is because the calculation of statistical significance is itself based on probability. Recall in our discussion about a drug study that researchers have no way of predicting how many placebo responders accidentally found their way into the study. Statistical significance is therefore based on a “confidence interval.” The study used a 95 percent confidence interval. This means that there is a 95 percent chance that the difference in power output between TRAD and EXP was based on random noise and only a five percent chance that the EXP start position actually allowed the lifters to express a higher power output.
 
It would have been helpful if the study had included a table listing the power output data for all of the lifters along with their bodyweight. Given how light some of the 1RMs were, this study may have included lifters that still had very novice and perhaps inconsistent technique. If the increased power output readings were fairly even among all of the lifters, this could have given some support to the idea that the EXP start position did have some effect on power. If the higher power outputs were concentrated among the more novice lifters, this would have further supported the idea that the EXP start position made no difference at all.
 
Do You Want to Gamble with Your Technique?
 
For the “too long, didn’t read” crowd, here is the simple conclusion: a study with a few design flaws found that there is a five percent chance that changing your start position for the snatch and clean to one that is more like a deadlift – bar over midfoot and high hips – can increase your power output by around two percent. Over 40 years ago, Soviet biomechanists advised against this type of start position. Are you willing to learn how to lift from a different start position based on these facts?