The first set of data for My Quantified Mood was collected and analysed between May 1st-22nd. Being active in the entrepreneurial world I know to never use phrasings such as “failure” or “less than optimal”. Actually, after reading the book “Thinking fast and slow” I know that just entering these words into this text will prime your brain to conceive everything in a less positive way. But facts are that – primarily due to some data loss – no relevant analysis was possible. Right, so now that the lingustic damage is already done, let’s get to reporting the actual outcome.
The key takeaways from this first round of measuring were:
- Only use services with online backups. The app “Moodnotes“, used for collecting the mood data, was deleted from my phone after three weeks, and with it all the collected data.
- Food data doesn’t need to be on a super detailed level. Using “My Fitness Pal” for logging food gives extremely detailed output (such as dietary intake of potassium in milligrams). This provides a deceiving sense of detailed knowledge since it is almost impossible to log the food with the same precision.
- Measuring more “soft” parameters. I have been given input regarding the value of including more measures of e.g. the progress of long term goals and even more details on the productivity of the day.
The collection of data
The collecting of the data was mostly a breeze using the apps and services as presented in this post. Almost all of measuring was done automatically with apps and wearables, but especially the manual logging of food presented some hassles (and then, of course, the loss of the mood data).
The mood logging failed
Before this first trial I tried a variety of mood loggers, and found that MoodNotes was the app best fitting my needs. Well, so I thought. It has a nice interface and does what it promises, but it turns out that it doesn’t keep any backup copies. After three weeks of measurements, suddenly the app had disappeared from my iPhone. How did this happen?!? Was it my 2-year old nephew that managed to delete it when playing with my phone (if so, he has already got an impressive intuition regarding what others value the most), or was this a bug in the app? Contacting MoodNotes they were also puzzled and informed me that they unfortunately had no way of restoring my data. It was all stored locally on my phone, but then my already evil genius of a nephew deleted it. Given that it was the most important variable of the experiment, this messed everything up. I had a rough measurement of my mood logged each day in the app “Exist“, but with much fewer data points and just a three-level scale, which turned out to be too blunt.
The food logging was too detailed
The food logging was done with My Fitness Pal which is a neat service, but giving it a closer look there seems to be a lot of approximations done by the individuals when entering recipes into the system. This is combined with the difficulty of being sure of exactly what you are eating, and not to mention the exact amount (it’s hard to weigh your peas, they roll off the scale!). Having the exact amount of sodium intake each day may sound good, but if you’re not sure about the accuracy of that data, it lowers the quality of the whole data set. As with most thing: enter bad things and bad things come out. So for the next round of measurement there will instead be an update of the high level food logging options, such as “Lunch: high diary” and “Sugar free day”, which will be both more accurate and relevant.
…but most other things worked well
Besides those two issues, everything worked quite well. A few honorable / less honorable mentions:
- Automatic measurement of productive and non-productive time on my computer worked very well. Way to go RescueTime!
- Sleep and physiological measurements such as brain wave calmness when meditating (Go Muse!) and HRV in the morning, but also body fat and breathing patterns (Spire breathing sensor) did their job.
- The measurement of cognitive skills at Quantified-Mind might benefit from being more standardized, and this will be updated for the next round of measurements.
- Logging of water intake was ok, but I still look for an app that both measures the total intake of water, combined with caffeine, alcohol etc, and that provides all this in an export function.
- Time spent listening to music started out ok, but it seems like Zenobase no longer has a connection with last.fm to collect track data, so the analysis of e.g. genres listened to didn’t work out in the end.
Again, read this post to understand how all the logging and measurements were done.
Analyzing the data
As you know by now the analysis didn’t provide any relevant correlations, probably due to the loss of mood data, but at least I got an opportunity to try out running the data set through the tools.
Multivariate analysis with IBM SPSS
The data was primarily analysed using multivariate analysis in IBM SPSS. Given that the mood data was gone (the parameter I actually wanted to measure everything against) no proper analysis could be made. Using instead the more blunt data from Exist – where each day had gotten a score on a three point scale – nothing came out relevant. Reducing the data set to fewer parameters would have been needed even if the mood scores would have been intact, but even with just a couple of parameters left, no correlations could be found.
Using IBM Watson
I also entered the data into this tool that IBM recently made available for using the power of Watson to analyze data. Again, no relevant results came out.
Preparing for next round of measurements
Next attempt will be made in early fall 2016, and alterations to the method will be made and presented here in August. I hope you are enjoying your summer and I will see you soon!