In a daydream I came up with a somewhat weird challenge. To find a correlation between subjective and objective data in our lives. Basically, to find the objective inputs that are correlated with our subjective experience in a mathematical way.

Can you take a look at it and let me know if you have any feedback? Interesting? Or no.

What role does cognitive dissonance play in the relationship between your subjective ratings and the criteria of being "exceptional?" I personally like defining 1s & 3s as either below or above average because "exceptional" is tough to achieve subjectively.

I've been rating five areas, with a CBT psych doc friend, on a 1-3 scale for the past 6 weeks. I started "grading" weekly ratings by adding up scores each week and dividing by points possible to look for weekly trends.

flickr.com

Doing that to the data in your post is interesting.

It's tough to pin happiness, or even meaning, to a combination of variables. But the process of both reflection and analysis is rewarding in itself. :)

Thoughts?

-chris

I think you're right about using "above/below average" rather than exceptional for subjective ratings. In the end, it's all subjective, and the point is to use your own subjective judgement to rate your day, however you like, and then to simply try to come up with an objective function that correlates to the subjective rating. So I should make it clear that you have to use a subjective rating that resonates most with an actual measurement of which days you want to increase/decrease frequency of most.

As for your rating system, they're all subjective, so it is really a way to visualize your own judgements about your life rather than find causes or correlations.

If you're willing, I'd love for you to continue that rating system and then to also start recording objective data about your day, and seeing if you can find correlations between them. Is that of any interest?

Sorry for the late reply. To clarify, I rate the Self Care category objectively. There are three things that I track: 10k steps a day, going to bed by 11:30pm and following a specific diet. I give myself one point for each accomplished and no points if I do not accomplish any of these things... so that it fits on the same scale as the rest.

Steps, sleep and diet are the only things that I am interested in tracking objectively. Is that helpful at all?

-chris

I'll play the statistical devil's advocate here. You can't actually depend on the accuracy of simple Pearson's correlations for repeated measures (something measuring consistently over time within a single subject). Each measure is naturally correlated with those close to it - this is typically referred to as autocorrelation (en.wikipedia.org). You can do simple and complex adjustments for it, but I'll need to search a bit for how to easily do it in Excel.

Stats aside. I think this idea of subjective vs objective over time is super interesting and has high potential. I see experience sampling being a key to doing this at scale with specific questions. Possibly using something like Paco - code.google.com or Funf - funf.org

Definitely continue to push me on the statistical correlations stuff. What are the simple and complex adjustments that we need to make?

One obvious one is the ability to see if shifting the data of one series back a few days, or a week, strengthens the correlations.

Another one is to use various 3-day or 7-day or longer trends instead of individual points to find correlation.

All of these things are easy to do in Excel. If we can find a small set of statistical correlations that theoretically *could* prove correlation between subjective and objective measures, then we'd have something interesting.

And if we *can't* find that, then that's interesting too.

I'll check out your links too to learn more about this.

My wife is always hounding me to track my life in meaningful ways. I've spent much of my life tracking how others feel about me and the better they feel about me, or, the better I think they feel about me, makes my days go better.

This year has been all about starting over and getting to know myself again. I think that tracking in a simple way would be helpful.

Interesting discussion so far guys.

Buster, to Ernesto's point -- R has some pretty simple time series functions I've used to break down autocorrelation, seasonality, etc. If you have some files of data collected over the same time interval I can mess around a little with it.

Ernesto, I'm really interested in Experience Sampling. I wish there was an iPhone app that allowed me to use pre-built data gathering controls (or custom ones) + reminders + triggered reminders to build better data sets. Data + context is key as well. I fear data at the day level is just too course to measure cause and effect of fulfillment/meaning/happiness -- experience sampling could help provide better resolution and change.

Do you know of any iPhone equivalents of the Android apps you linked?

Objective: I've tracked my "firsts" for 13 years (e.g., first kiss, first time speaking at a conference, etc.) and number of workouts for the past 4 years.

Subjective: I tracked my productivity, enjoyment, and fulfillment on 1-10 Likert for 4 years.

Correlations:

Firsts-Fulfillment: 0.66 Firsts and fulfillment are strongly positively correlated. This finding supports intuition.

Exercise-Productivity: -0.70 Exercise and productivity are strongly negatively correlated. This finding contradicts intuition.

Just two of six correlations. The rest contradicted intuition, also. Confusing! But perhaps because I didn't account for autocorrelation yet?