When you are analyzing yourself using personal data, there are several ways you analyze yourself. The depth of your analysis is dependent on your purpose, and desire. The deeper you go with the analysis, the more you’ll get to know about yourself. However, this requires a lot of data collection and time-investment (if this can’t be automated). Because the more complex analyses, are just, complex.

The list below will state all the analysis types that I am aware of. The list is ordered based on complexity, from basic to very complex

  • The current state of you
  • You over time
  • The connection between two variables
  • The relation between various variables

The current state of you
All the measures that are “now” define the current you. A wearable that shows you your current heart rate on the display is an examen of this.  There is no real analysis involved with the measure itself. However, this quick view makes it very easy to you to change your behavior directly.

You over time
In this case, you look at one variable and see how it changes over time. Within an app or in excel you can look at your graphs and investigate how your sleep or step rates changed over the past week. However, there are also statistical methods that can analyse whether you changed over time. Time series analyses or repeated measures might be helpful methods to investigate this. This type of data analyses can give you a lot of insight into which direction you are heading, or where you came from.

Connection between two variables
For this type, you need to measure at least two variables over time, for example your steps and your sleep quality. Suppose, that every day you walk a lot during the day,  you sleep very well at night. In this case you can look at the connection between your steps and sleep quality. Here you can apply correlation, this analyses compares the variation of both variables and sees if they are relating to each other. If there is a positive correlation this means that your assumption is true, the more you walked, the better you slept. But there can also be a negative correlation; this would mean that the more steps you take, the worse you sleep. However, a famous saying is: correlation does not mean causation. This means that a correlation is not necessarily directly the truth. It could be that you when you do walk a lot, you always award yourself with a nice beer at the end of the day. This beer actually might be the cause of your better sleep quality, and not your walking activity.

But there are also some more advanced methods that investigate the relationship deeper. How about the effects of two consecutive nights of bad sleep on your fitness level? As far as I know, there are no methods to analyse this yet, but auto-correlation is a start.

Example of my correlations between various variables

Relation between various variables With correlation, you take two variables. However, there are analysis methods that combine multiple variables to say something about you. If you take more variables, you can do more things and make stronger predictions. You can interpret which variables have the strongest relation, or how each variable relates to one another. You can look at control variables, or you can predict things. Regression analyses is an example where several variables are taken to predict a certain variable. Mixed models, principal component analyses, and multi-level analyses are methods that you can use as well. However, many of these methods are not made for n=1 studies, and are quite advanced. Thus, you should learn about the methods, and use the correct data before interpreting your outcomes. The difference between methods for analyzing are quite immense. The way you look at, or analyse your data does really make a difference. Which method you use depends on your data. The outcomes can be interpreted in many ways, where many ways of these interpretations are wrongly.

Share on FacebookShare on Google+Tweet about this on TwitterShare on LinkedIn

Leave a Reply

Your email address will not be published. Required fields are marked *