Skip to main content

Generating hypotheses is a fundamental aspect of scientific discovery. I encourage all of you to form your own hypotheses based on analyzing the biomechanics of the exercises and exploring the available literature (click HERE to read the Hip Thrust Wiki Page).

The “Lee Hamilton” study is going to be a very big deal in S&C. Either my predictions will pan out and the hip thrust will be catapulted to the top of the list of best glute exercises and most functional exercises, or the critics’ predictions will hold true and hip thrusts won’t lead to any improvements whatsover and can be kicked to the curb.

It is quite clear to me that the critics have poor knowledge of sports science and biomechanical instincts, but that’s why research is so important. Rather than call each other names, cherry-pick anecdotes, generate conspiracy theories, find fake experts to make outlandish claims, shift goalposts and demand impossible evidence, or warp logic to mold to preexisting biases, we can conduct randomized controlled trials and see what the data reveals, then update our knowledge base and form new theories.

We are going to move forward in conducting a huge training study (longitudinal/training studies are rare in S&C because they require much more time and effort than cross-sectional/mechanistic studies) that is going to provide much clarity as to how well the three most popular lower body exercises (squats, deadlifts, and hip thrusts) compare with each other in regards to their transfer to performance and ability to grow the glutes. When I say “we,” I mean Lee Hamilton and his team of researchers, as they’ll be conducting the study. But my team (yours truly, Chris Beardsley, Andrew Vigotsky, Brad Schoenfeld, and John Cronin) will provide input and of course funding. This post will seem intimidating to many individuals who don’t have a good grasp of sports science terminology, but I’ll try to make good sense of it at the end in the conclusion section.

2

In order to measure all of the variables of interest pre and post intervention (before and after the six weeks of training), Hamilton and colleagues will be utilizing this technology and collecting these data:

  • Electromyography (EMG): Mean gluteus maximus EMG amplitude and Peak gluteus maximus EMG amplitude (between squat, hip thrust, and deadlift)
  • Tensiomyography (TMG): Gluteus maximus contraction time, Relaxation time, and Muscle displacement
  • Force plates: Max Isometric mid thigh pull force, Max horizontal pushing force, Force, RFD, Impulse, Power, and Work (between squat, hip thrust, and deadlift)
  • Linear position transducers: Barbell displacement and Velocity (between squat, hip thrust, and deadlift)
  • Isokinetic dynamometry: Isometric hip extension torque at 90, 45, and 0 degrees, Isokinetic hip extension torque at 60 deg/sec, 180 deg/sec, and 360 deg/sec
  • Ultrasound: Muscle thickness, Pennation angle, Fascicle length
  • Timing lights: 10m sprint, 40yd sprint, 5-10-5 agility
  • Radar guns: Maximum sprint velocity
  • Tape measures: Vertical jump height, Horizontal jump distance, Rotational medball toss distance
  • Goniometry: Hip and knee joint angular displacement (between squat, hip thrust, and deadlift)
  • Video capture: Hip and knee joint angular displacement, Segment angles
  • 1RM testing: Squat max, Hip thrust max, Deadlift max

We’re going to obtain a tremendous amount of valuable data which will require a serious amount of time to properly analyze all of it.

Essentially, we’re going to take inexperienced lifters and subject them to 6 weeks of 3X/wk training. One group will just full squat, one will just hip thrust, and the other will just conventional deadlift. Volume will be matched and periodized, and progressive overload will be utilized.

Below, I will post my predictions based on my current knowledge of prior research, pilot data, and strength training and sports science. I encourage you to make predictions so that when the study is finally published down the road, you can check to see how savvy your biomechanical instincts were.

1

Selected Predictions

Mechanistic Data

EMG (Gluteus Maximus Amplitude)

Mean Activity

  1. Hip thrusts
  2. Deadlifts
  3. Squats

Peak Activity

  1. Hip thrusts
  2. Deadlifts
  3. Squats

Force Plate

Concentric Force 

  1. Hip thrusts
  2. Deadlifts
  3. Squats

Eccentric Force

  1. Squats
  2. Deadlifts
  3. Hip thrusts

Concentric Power

  1. Hip thrusts
  2. Deadlifts
  3. Squats

Linear Position Transducer

Barbell Displacement

  1. Squats
  2. Deadlifts
  3. Hip thrusts

Concentric Velocity

  1. Hip thrusts
  2. Deadlifts
  3. Squats

Architectural Adaptations

Ultrasound

Gluteus Maximus Muscle Thickness Changes

  1. Hip thrusts
  2. Squats
  3. Deadlifts

Pennation Angle Changes 

  1. Hip thrusts
  2. Squats
  3. Deadlifts

Fascicle Length Changes

  1. Squats
  2. Deadlifts
  3. Hip thrusts

Longitudinal Performance Adaptations

Force Plate

Isometric Midthigh Pull Force Improvements

  1. Deadlifts
  2. Hip thrusts
  3. Squats

Horizontal Pushing Force Improvements

  1. Hip thrusts
  2. Deadlifts
  3. Squats

Radar

Maximum Speed Improvements

  1. Hip thrusts
  2. Deadlifts
  3. Squats

Timing Lights

10 Meter Sprint Improvements

  1. Hip thrusts
  2. Deadlifts
  3. Squats

40 Yard Sprint Improvements

  1. Hip thrusts
  2. Deadlifts
  3. Squats

5-10-5 Agility Improvements

  1. Squats
  2. Hip thrusts
  3. Deadlifts

Tape Measure

Vertical Jump Improvements

  1. Deadlifts
  2. Squats
  3. Hip thrusts

Horizontal Jump Improvements

  1. Deadlifts
  2. Hip thrusts
  3. Squats

Medball Rotational Scoop Toss Improvements

  1. Deadlifts
  2. Hip thrusts
  3. Squats

Isokinetic Dynamometry

Isometric Hip Extension Torque at 90 deg Improvements

  1. Deadlifts
  2. Squats
  3. Hip thrusts

Isometric Hip Extension Torque at 0 deg Improvements

  1. Hip thrusts
  2. Deadlifts
  3. Squats

1RM Testing

Squat Strength Improvements

  1. Squats
  2. Hip thrusts
  3. Deadlifts

Hip Thrust Strength Improvements

  1. Hip thrusts
  2. Squats
  3. Deadlifts

Deadlift Strength Improvements

  1. Deadlifts
  2. Hip thrusts
  3. Squats

3

Conclusion

I’m quite certain that I’ll be wrong on numerous predictions, but that’s what makes research fun – learning via hypothesis testing.

Let me be the first to say that squats and deadlifts are indeed “harder” to perform than hip thrusts. You can check my Instagram and see that I do heavy squats and deadlifts and hip thrusts week in and week out, and squats and deads require more psyching up and lead to greater overall muscle activation than hip thrusts. However, getting better at sports requires the preferential strengthening of some muscles and ranges of motion (ROM) over others. Hip thrusts don’t strengthen the quads like squats or the hams/traps/grip/thoracic erectors like deadlifts, nor do they work you quite as well down deep in the stretch position, but nevertheless they will transfer very well to performance. Hip extension strength through a full ROM is paramount in sports, and the hip thrust is superior for strengthening end-range hip extension strength. This is incredibly important and under-appreciated in S&C

I was fairly confident with certain predictions, especially strength predictions due to specificity. But I completely guessed on other predictions, for example rotational power (I could make a good case for all three exercises), midthigh pull (this will come down to deadlifts versus hip thrusts IMO), and horizontal jump (this will come down to deadlifts versus hip thrusts IMO).

I’ve picked hip thrusts to lead to the greatest levels of gluteus maximus mean and peak EMG amplitude and concentric force, velocity, and power, as well as the greatest changes or improvements in gluteus maximus muscle thickness and pennation angle, horizontal pushing force, 10m and 40yd sprint, maximum velocity, isometric hip extension strength at a neutral (0 degree) hip angle, and max hip thrust strength.

I’ve picked squats to lead to the greatest levels of eccentric force and barbell displacement, as well as the greatest changes or improvements in gluteus maximus fascicle length, 5-10-5 agility, and max squat strength.

I’ve picked deadlifts to lead to the greatest improvements in mid thigh pulling force, vertical and horizontal jump, medball rotational scoop toss, isometric hip extension strength at a flexed (90 degree) hip angle, and max deadlift strength.

Summary of hip thrusts

I believe that hip thrusts will require the greatest amounts of electrical output to the gluteus maximus (due to the interaction between the nature of the unique EMG activation angle curve inherent to the gluteus maximus [see Worrell et al.] and the unique nature of the torque angle curve inherent to the hip thrust exercise).

Due in large part to the superior EMG activation in addition to the tension and metabolic stress they produce, I believe that hip thrusts will lead to the biggest gains in hypertrophy. Since the hip thrust is hardest at the top of the motion when the glutes are shortened, they likely will not lead to any changes in muscle (fascicle) length, but pennation angle might increase slightly.

I believe that due to the mechanical efficiency of the hip thrust exercise, during the rising (concentric) phase, it will yield the highest outputs of force (mass times acceleration), velocity, and power (force times velocity). However, since individuals tend to just let gravity do its thing during the lower (eccentric) phase, the hip thrust will yield the lowest eccentric outputs in force. I believe that the hip thrust will also involve the shortest range of motion out of the three exercises.

I believe that due to the horizontal vector and torque angle curve inherent to the hip thrust, in addition to the superior glute activity and moderate hamstring activity, they will transfer best to sprinting and horizontal pushing force, and they’ll improve torque production the most at a neutral hip extension angle (end-range hip extension strength). Obviously they will maximize hip thrust strength due to the law of specificity, but they will also transfer better to squats than deadlifts and better to deadlifts than squats.

Summary of squats

Due to the interaction between the nature of the unique EMG activation angle curve inherent to the gluteus maximus [see Worrell et al.] and the unique nature of the torque angle curve inherent to the squat exercise, I believe that the squat will yield the lowest gluteus maximus EMG amplitudes. However, due to the muscle damage they produce in addition to moderate activation and tension, they will still produce good gains in gluteus maximus hypertrophy (just not as much as hip thrusts). And since squats are hardest down low in a stretched position (flexed-range hip extension), they will slightly lengthen the muscle fibers (increased fascicle length).

Since squats require a reversal from the eccentric to concentric phase in the air (hip thrusts and deadlifts can almost be dropped to the ground), they’ll lead to the highest levels of eccentric force (mass times acceleration during the lowering phase). Squats will utilize the greatest range of motion in terms of barbell displacement out of all three exercises.

Due to the knee extension demands and heavy eccentric/lowering component inherent to the squat, I believe that they will transfer best to agility. Due to specificity, squats will transfer best to squat strength.

Summary of deadlifts

Due to the movement specificity between deadlifts and vertical jumps, I believe that deadlifts will maximize vertical jumping ability. However, even though deadlifts are axial in nature, I believe that they’ll slightly outperform hip thrusts in horizontal jumping, due to the superior hamstring activation inherent to the deadlift. Even though deadlifts activate the hamstrings better than hip thrusts, I predict that hip thrusts will still be better for speed improvements due to the increased end-range hip extension demands.

I picked deadlifts to transfer best to isometric midthigh pull force due to the similarity in movement pattern, but hip thrusts will strengthen this test very well too due to the similarity in hip extension joint angle and torques. I also picked deadlifts to transfer best to rotational power (rotational medball scoop toss) since they utilize the glutes and the erectors very well and seem to involve more overall core activity, but I can see hip thrusts and squats leading to good improvements as well.

Deadlifts will transfer best to deadlifts in accordance with the law of specificity, and they’ll also improve isometric hip extension torque at 90 degrees (flexed range hip extension strength) to the greatest degree.

Bigger picture

Most important, I predict that all three exercises are needed to help maximize performance in athletes. I can’t wait to see how it all pans out!

18 Comments

  • Ron says:

    I think the only one I disagree with (with a lot of undecided’s) is the deadlift over the squat in the standing vertical jump measurement. Supposedly, a standing vertical jump is 40% hip, 40% knee and 20% ankle. I think this favors the squat a lot more in the knee or quad and most likely more in the hip or butt. It took me a while to be able to feel my butt working when I lift and it really started when I switched to full squats. In my deadlifts before this I really felt it more in my hamstrings and back and a lot less butt. Now, with my deadlifts, I try to think of grabbing a bar and doing a leg press into the floor. I feel it mainly in the butt with some quad. Still nowhere near 50/50. We’ll see……..eventually.

    • Ron, I always thought the same thing too, but in a pilot study we’re trying to publish, the deadlift outperformed the squat (non-statistical due to the study being underpowered). So this influenced my decision. I do think the quads are very important in the VJ, but I think the deadlift mimics the movement pattern a bit more, so I’m up in the air too. Nice hearing your thoughts.

  • Lollo says:

    Hi Bret,

    This sound very interesting and I wish you all the best with the study!
    I do have a few thoughts on this..

    Have you thought about that some people are more quad VS glute dominant,and how this can affect the study? Let say most of the people are quad dominant and that group will perform squat or deadlift, I am sure the data will show that gluteus maximus wont fire as much as for those who perform hip thrust.

    Also, technique and body structure, angles will make a big different how you perform the exercise and what muscle will fire more.

    What are your thoughts on this?

    Regards,

    Lollo

    • Lollo – this all washes out when you have ample sample sizes in your studies. I think we intend on having 20 randomly allocated subjects in each group so each group should have the same characteristics, and when the data is averaged out, outlier effects will be minimized.

  • Chris says:

    That sounds like a nice little project. 🙂
    I love the heap of measurments – the more, the better. Id add a 5, 8 or 10RM to the 1RM test – its easily done and provides more insight. And contrary to others (Stoppani *cough*) you are aware of error inflation with that many variables.
    Well, you sure wont convince those critics regardless what youre doing. They dont understand science, they dont understand the topic. I wouldnt care about them. I dont care.
    However, every researcher has open or implicit goals as to what a study should show. And can design the study accordingly to facilitate desired outcomes. Youre obviously very invested and commited to hip thrusts being very valuable. I do think evidence favours your position, but I still know this researcher team influences the study a bit pro-hip thrust. To make it clear: I dont accuse you of any wrongdoing, thats just how research works. So Id love to see some researcher who are a bit more neutral to or even critical of hip thrusts to come up with similar studies.

    Hey Bret, I even discovered a first clever trick to ensure favourable outcomes for the HT: You use bgeinners as participants who of course profit more from the easier learning curve of hip thrusts. As I said, no accusation, because beginners are a valuable population because of several reasons (availability, less inference of former training, homogenity, expected effect size –> power).

    Still, Id also love to see similar studies done with experienced lifters who are proficient and trained with all three exercises cancelling out the different technical difficulty of the lifts.

    Predictions: I mainly go with yours, after all youre the nerd on this. Maybe in the 10 yards HT not first, and in the VJ squat first – previous research in yo face! 🙂

    And please grab as many beginners as you can lay your hands on in a ten km perimeter of the lab – it is so useful with these many variables to have a big N. After all, preparation and setup is the majority of the work – you dont have any blood work so the total costs are pretty similar whether you take 10 or 30.

    All the best for the study – you guys are doing a great job!

    Greetings from Germany, I hope our sports scientists finally get on their asses and publish more (internationally)!

    • Hi Chris, thanks for your response.

      1. I don’t see why a 5, 8, or 10RM would be necessary; we’re interested in maximum strength gains so the 1RM tests and force plate tests will accurately predict gains in force production in multiple vectors and patterns.

      2. The error inflation topic is controversial. We usually do Bonferroni corrections to address this, but in this case, since we’re measuring potentially 50 things (I listed 24 in my predictions), you’d have to divide alpha (.05) by 50, yielding a new p value of .001 in order to achieve significance. This heavily prevents the risk of committing a type I error at the expense of massively increasing the risk of committing a type II error. There are other corrections we could go with, or we could publish multiple papers with the various data grouped together (EMG paper, force plate paper, ultrasound paper, performance paper, etc.). But it’s definitely a controversial topic in statistics and I personally don’t like post hoc corrections as a) they discourage scientists from gathering comprehensive data, b) the topic can be easily addressed in the limitations section of the paper, and c) the tests will be determined a priori so we can’t be accused of p-hacking of going on a fishing expedition.

      3. Funny you should mention the learning curve; our recently published performance paper on front squats vs. hip thrusts had subjects who had previously performed the front squat for a year (but not the hip thrust). Going by your assessment, this would make it more fair for the front squat group, but others would claim that this yielded an advantage for the hip thrust. I do agree that there are different learning curves associated with the different lifts, but I wonder how rates of strength gains compare over the course of 6 months or 1 year or 2 years between the 3 exercises; they may not be very different.

      4. We’re definitely all biased whether we realize it or not, but I think my studies are indeed very fair as I’m aware of my biases and I want to produce legit studies. So I don’t think I influence the outcomes with my designs whatsoever in terms of providing an advantage for squats or hip thrusts, etc., especially since I have a team who looks over everything that’s done (they aren’t invested or biased and they’re pure scientists seeking the truth). In the end, I want to publish long-lasting research that will be reproduced and built upon, and I only seek the truth as deep down I love squats and deadlifts just as much as hip thrusts and I want to know the real science behind how everything works – for the betterment of my own training, my training of clients, and my online programs.

      However, your points are taken, and I’m glad you posted your thoughts. We will definitely strive to obtain ample subjects to enable us to achieve statistical power and not commit any errors in our recommendations.

      • Chris says:

        Thanks for your answer, Bret.
        2. I think your point c) is the most important one that makes it legit to procede this way.
        3. I probably described it misunderstandingly: If you put strength training novices on free weight exercises, the gains within a short time frame are in part determined by the learning curve of the exercise. Hip thrusts are easy to learn (which is a good thing), thus enabling good strength gains within a short time frame. With SQ and DL, you cant use your actual capacities because youre occupied with technique. Its exactly what happened in this study referenced by Lyle: http://www.bodyrecomposition.com/research-review/comparison-strength-muscle-mass-increase-young-women.html/ So exactly what you said: When comparing pre-post measures for a training time of one year, this wouldnt be a bias. Neither with lifters who are experienced in the lift theyre grouped into. But within 3 months AND strength training beginners, this logically favours the easier exercise – the hip thrust. So yes, in the Front Squat vs HT study, this also favoured the HT participant, because youre looking at change (delta) measurements.
        4. I give you every benefit of the doubt and even how sincere and classy as a person I have known you – albeit not in person, but only virtually – in the last four years following your great work, youre still human. There are hundreads of cases of well-meaning, sincere researchers who have nevertheless – without any conscious manipulation whatsoever – produced results that couldnt be replicated by people who were neutral to the topic. So I stand by my wish to see more HT research coming from other researchers.

        • Chris, regarding point #3, this is exactly how I understood it. I can understand the logic and I realize that many experts share this sentiment, and I do agree with it if you start off with an advanced variation. For example, if I take a beginner and throw them right into barbell squats and deadlifts, they’re form may be horrendous and they’d have to hold back in order to improve upon their technique. But if you started off with goblet squats and box squats for higher reps, they can work the muscles just fine (if you EMG’d these folks you’d see very high muscle activation). But since the adolescent males in the training study already had a year of practicing the front squat but no practice with the hip thrust, some could see this as a benefit for the front squat group (they were no longer in the “learning phase” but they had laid off from the exercise), while others would see this as a benefit for the hip thrust group (they could get “beginner gains” from the hip thrust but not the front squat). I hope that makes sense. And if this were true, then it would infer that we should start all beginners off with machines and gradually transition to free weights so they can work their muscles hard all the while, but I don’t do this with my clients; they all start off with variations of squats, deads, hip thrusts, lunges, presses, pulls, etc.

          Regarding point #4, you’d be a fool not to think this way. Hell, I mention this in my Glute Lab Seminars and in my Hip Thrust wiki post (you always want to be patient and wait for research to be duplicated especially if the only lab pumping out research has a vested or biased interest in the outcome). I have heard rumors of a European lab submitting an EMG study for publishing involving the hip thrust, and there’s a case study conducted by another American researcher that showed similar performance findings (but slightly different in the case of the vertical jump, and it measured more variables) as my study. But there will be much more in time. So you’re just being a good scientist and I agree with you.

          Cheers, BC

  • Nick Carey says:

    Hey bret, Awesome work to all involved, when can we expect results roughly?

  • Anoop says:

    Hey Bret,

    This is very interesting! And looking forward to it.

    The only thing I would suggest is to do a sample size calculation using a main outcome ( like vj or HJ) than just picking a random number of 20. The rest of the outcomes could be secondary. This will ensure there is no multiplicity problems or p-hacking.

  • Graham says:

    Hi Bret,
    A couple of observations/queries
    1) squat vs deadlift for VJ; as well as squat depth, another variable to consider is how “low” the participant crouches in their VJ and the relationship of this when comparing specific joint angles. A study released in the past couple of months suggested 1/4 squats have more immediate carryover to both VJ and sprint speed than 1/2 or full squats
    2) a while (3 years lol) ago you released figures regarding glute activation on kneeling squats; what are your thoughts on the likely increase or decrease in this number if you were to turn the movement into more of a good morning/squat hybrid (if you can imagine the bottom position looking like a Z where the hamstrings touch the calves and abdomen touch the quadriceps)
    Regards,
    Graham

  • Logan Edwards says:

    My main concern with this study is about the participants. How are they going to account for the drastic variation in genetics. The study might include some individuals who are genetically predisposed to jumping or sprinting power but have lived sedentary lifestyles. These individuals could skew the results drastically, so hopefully the number of participants is large so we can eliminate the extremes.

    I personally don’t know much about all the angles, forces, elongation etc. you bring up. I do know that for myself, the squat translates extremely well to increased deadlift numbers while the opposite is not true. To me, the squat feels like it translates to increasing power for a larger variety of activities, so I expect it to perform better than you predict on a number of tests.

    It will all be interesting to see!

    • Logan, great question. If you conduct a power analysis beforehand, you can get an idea of the number of subjects you’ll need to achieve good statistical power, and this should wash out the effects of genetic variation. We intend on having ample subjects in our study.

  • Mark says:

    I can’t get my head around pennation angle
    Looking at this picture : https://en.wikipedia.org/wiki/Gluteus_maximus_muscle#/media/File:Sobo_1909_294.png , will fibers be more horizontal or vertical if pennation angle increases?

  • mg says:

    Was this study ever done and if so can you share the results?

Leave a Reply

SIGN UP FOR THE FREE NEWSLETTER

and receive my FREE Lower Body Progressions eBook!

You have Successfully Subscribed!