I started running for the first time in my life one year ago this month. I have been kind to myself and not set any speed goals. But now that a year has gone by I am curious if I can increase speed by focusing on a metric called “strikes per minute” (SPM) which is defined by the number of times your foot hits the ground during one minute. From what I have read, a “good” SPM is 160, with elite runners at 170 SPM.
I have tracked my runs and speed for the >6 months using Apple Watch, which captures average speed, SPM, and distance. I have >6 months of data or me running normally and my average SPM is 153. I will compare that with what I do during the project period where my goal is to increase SPM.
How do I plan to do it? I am going to create a customized music playlist I will run to featuring music timed to ≥160 BPM (beats per minute). The goal is to sync my foot falls to the beat of the music. As a former punk rock drummer I am pretty decent at locking in to timing. I hope I can sync my feet and grow my SPM, and see if my average mile time improves. I plan to use Apple Watch and Spotify as tools and log data into spreadsheets.
Considering my average of 153 SPM is slow, and 160 SPM is “good” I am curious what a statistically significant improvement should look like. Is hitting 160 the goal? Is it to get to 160 SPM faster (within the first few weeks)? Is it to go over 160 SPM?
Do you think there’s a question you want answered here? I guess that might be… “Does music affect my running speed?”
It’s nice to open a project log with a question. (Not required, but feel free to edit your post to add it!)
One thing to think about is doing “AB testing” – that is, testing between two different conditions. Maybe it’s already implicit here! It sounds like the “A” is what you’re currently doing, and “B” is trying the playlist.
Main issue might just be: keep track of when you’re doing A or B.
I don’t think you’ll need to analyze the music itself (so this is totally overkill) – but @gedankenstuecke has done some work getting & using data from Spotify. He built a Spotify import-to-Open-Humans tool and this data analysis notebook, which mentions some interesting metadata Spotify provides like “energy” and “valence”.
I am happy to edit my post title to start with the question. I really like your proposed question–it is nice and simple! My mind keeps thinking of all the things I can track and compare and not the big picture, which is the question!
Additionally, I like the idea of A/B testing. What I might do is make my run more structured: stick to the SAME ROUTE (distance) for the entire research period to make sure I have the same obstacles and hills to face. The A/B test can be between the playlist which is programmed to improve SPM, and then perhaps a “auto” playlist. Or, I can design an alternate playlist that has a variety of tempos of songs.
Oh yeah – same route, that sounds valuable to make consistent! For AB testing I wonder if you should aim for a contrast where you’re pretty sure to see an effect? Then you’ll know how obvious the effect is / what it looks like, and can start testing less dramatic things.
Ok, getting nerdier here. I have heard of “adaptive clinical trials” in cancer where if you see a person is NOT doing well on a drug you switch them to a new arm. If after a month of running I don’t see improvement with the playlist with the 160 BPM I might try another angle. I still need to define what improvement is… is it SPM or faster average mile time?
Another reason same route is important is because if I fluctuate and run 2 miles one day and 3.5 miles another day, the warm up period of a 1/2 mile at the beginning where I am slower will drag down the average mile time for a 2 mile run vs. 3.5 mile run.
My method to create the playlist so far is to Google “songs at 160 bpm” (beats per minute) and add those songs to the playlist. I am planning to verify the songs bpm by testing them against a metronome tool. Perhaps as a final touch I will adjust them with the extension you suggest to make sure ALL SONGS are at the right tempo.
Yeah, if the music played is recorded through Spotify it might be possible to try to see whether any other of the Spotify-metadata correlate (e.g. the “mood” of the music being played!) But I agree, starting with some A/B testing would probably the way to go.
I think the problem with this would probably be that you’re likely to improve speed regardless of the intervention, just by getting regular training? So any signal of the adaptive approach might be hidden behind expected improvements. Thus, somewhat regularly switching between two conditions might be better in terms of signal?
I have started prototyping my self-research project.
The plan as it is in my mind right now:
I will A/B test running with a playlist featuring music at 160 BPM.
I will run the same 5k route during this time period to make sure route and elevation are the same (controlled environment). I chose a 5k because it is a challenging (but not unattainable) next step for me to be running this distance on a regular basis. (My baseline data showed I was running an average of 2.25-2.75 miles before starting the experiment). Additionally, a 5k is a distance used for most entry level fun runs and races. I think this distance will make the project more interesting to others who run 5ks.
I will alternate the playlist every other run and track the data. So far I have done 3 runs.
I started a Google Spreadsheet to track my data. I am learning to use some of the graphing features. I want to learn how to show three sets of data in one graph and ultimately embed a live version of the graph into this log. (I just need to spend time learning how to bring in the “before” data, and the two lines of the A/B test, in one graph.)
Songs are validated as being at the right tempo (at or above 160 BPM) by tapping a metronome app to the beat of the song (MetroTimer).
I needed ONE SONG as a warm up song (i.e., at a slower pace). I picked a song that is around 5 minutes in length, and I start every playlist run with this song. The rest of the playlist runs at random after the warm up song. The warm up song is “Tom Sawyer” by the band Rush, off of the album Moving Pictures. I imagine others who replicate this project in the future can pick their own favorite warm up song.
Observations: I am finding certain songs on the 160 BPM playlist are easier to run to than others.
I’m really interested to see what results you get! I watch videos while using a stationary bike and noticed that when at fast-beat song came on the background, I would unconsciously speed up. I’m very curious whether you can take advantage of that kind of effect to drive long-term improvement.
Have you thought about how to disentangle the direct effect of the faster-paced music from effects stemming from you knowing which playlist is which? Is your memory or sense of rhythm good enough that you can distinguish between the 160 bpm playlist and the “random” playlist?
Storing a listening history on Open Humans takes a while after setting it up, but if you keep using Spotify it’ll grow (Spotify only gives access to the last 50 songs, but once you’ve set it up we update it to keep a full archive going forward).
Another caveat: Looking at my own list in the notebook I see a lot of songs where I highly doubt that they are actually 160 bpm
Checking in. I am still tracking my runs and A/B testing the special playlist vs. running with random songs. While I trying to hold off on running a real data analysis (keeping it a surprise from myself!) I have observed two things that are making me question everything:
If I run after a particularly challenging strength training workout (the day before) the 160bpm playlist really doesn’t do anything to help me. My body is sore and tired and likely should rest.
If I run with someone else I run faster (with an increased SPM) regardless of which music I listen to. WHY?!! Likely because I subconsciously want to impress this person or not appear too slow in comparison. Seriously, my best time so far was using a random playlist but being followed by my husband.
I am definitely logging observations along with Apple Watch data. I am making a note when I run with others. I don’t do it too often. Should I throw in a few other runs with people just to see what happens?
He built a 