I present this as a scientific article to make it easier for others to replicate or try on themselves. If you want the good stuff right now, skip to the results section. (No hard feelings, promise 😉)
- Writing out my thoughts improves my mood
- I get better insights when writing than talking to a microphone
- Either journaling method takes about the same time
Table of Contents
- What am I studying? (Interventions)
- What am I tracking? (Outcome-measures)
Every day I track my time to an accuracy of 5 minutes. This allows me to review what I did on any day and to improve how I spend my time. Part of this review process is daily journaling. I’m a medical student, and we dictate everything into a microphone. Could dictation be faster for my journaling or even improve it?
I was on the fence, so I decided to study it.
I pre-published my protocol here.
What am I studying? (Interventions)
Every day I journal following this template:
I’m comparing writing out my thoughts in Day One (written, W) to verbalising it to my recorder (verbal, V).
What am I tracking? (Outcome-measures)
Daily journaling serves two purposes for me:
- Improve well-being
- Generate ideas
It improves my well-being by allowing me to re-iterate my accomplishments and get into a self-reflective state.
And, being reflective, I consider which actions matter and to notice processes that I can improve.
Quality of Insights
The major reason I do daily reviews is to improve my life over time. The quality of insights I generate is thus paramount in deciding how I’m going to journal. I manually collected all additions to my to-do list during my journalling sessions and added them to a spreadsheet. For analysis, I rated them at a -5 to +5 scale like the below:
I squared each rating for outcomes, then summed them to get the insight-rating for each day. I rated insights at the end of data collection (ie. at day 10 of the experiment), without knowledge of which insights were either written or verbal. This should remove my personal bias from the outcome data.
I believe journalling has an impact on my mood, so this is something I’d have to compare. I used Reporter to sample my mood when I woke up, 3 times randomly during the day, and when I went to sleep. The outcome is all mood-entries on the day of review, based on the scale below.
To do cost/benefit analysis, I also need an idea of the costs — so I measured time spent.
How will I do it? (Study design)
A randomised-controlled-trial comparing a written diary vs. a verbal diary. Days were cluster-randomised with the following clusters:
Cluster A: 0 1 0 0 1
Cluster B: 1 0 1 1 0
In essence, I ran a random-number generator to pick which of the above sequences to run first. This resulted in me doing the following sequence:
W V W W V V W V V W
Looks like I’ll have to stick to a written diary! Both insight quality and mood appears to be better for a written diary.
I collected data from the 11th of October to the 21st of October (10 days). I adhered 100% to the randomisation.
All analysis was done in Excel. Data was assessed graphically for normality and P-values calculated with the student’s t-test (as all categories had n ≤ 30).
Deviations from protocol
- I didn’t send my insights to friends or family because I had a hard time explaining each insight
- I changed my journalling to be in the morning. To adjust my to this, the mood outcome become mood on the day of journal-entry rather than the day after.
What would I change?
I will use power-calculations to figure out how many days I need to sample to be sufficiently certain of the outcomes. However, I did define P < 0,25 as my goal and found it for 2 outcomes, so my intuition for the required saple size was accurate.
What are my next steps?
When I journal I switch between two modes: pure contemplation, where all my self-dialogue is in my head, and stream-of-consciousness writing. I have no idea which of those is the best for me, so I want to compare them like the above.
Where to go from here?
Not every human is the same, and definitely not in their self-reflection practice.
I urge you to run a similar experiment on yourself rather than generalise from this post.
My time-consumption was 1.5 hours for protocol, data collection, and analysis. You can do it even faster: You have the study-design and my excel-spreadsheet for analysis right here.