Crying: Can our tears be parsed?
By Robin Weis
I’ve always considered myself to be a strong person in the presence of others, able to work through emotionally challenging situations with steadiness and composure. At the same time, I’ve always cried a lot in private, sometimes at the tiniest or most unreasonable provocation. I wasn’t sure how or why I was able to be so stoic in some situations and so uncomposed in others, so I tracked my crying for a year and a half to gain a better understanding of my emotional reactions.
Gathering the data
A “cry” began when the first physical tear fell and ended when the tears stopped. Each cry was assigned an intensity rating on a numeric scale, with 1 being the least intense and 5 the most intense. The scale was based loosely on physiological cues: 1-3 solely involved tears, and 4-5 also involved changes to my voice or breathing.
Every time I cried, I would make a note in my phone that looked something like this:
3:45pm 2 3 phone with mom
This meant that I cried for 2 minutes starting at 3:45PM at an intensity of 3 while on the phone with my mother. The idea was to track just enough context to recall the event when I transferred my phone logs to a Google spreadsheet, which I did at the end of every day that I cried. This way, logging was minimally disruptive in the moment, and I could flesh out more information once distanced from the situation.
I wound up keeping track of the date, time, duration, location, and context of each cry (including who I was around and what I was feeling). I used freeform inputs instead of pre-set categories because I didn’t want to limit the type of data I was collecting or force myself to make an analytical judgment while I was upset. This of course meant that my data was almost entirely unstructured, but I was willing to parse through it later.
Analyzing the data
The first time I tried to categorize the cries was a year after I started logging them. The categories I came up with were messy and inconsistent: some of them were emotions and some of them were situations. I threw it all into an ugly but insightful pie chart and continued logging.
Image: First draft of crying categories.
Six months later, I tried to categorize the logs again. I wanted to iterate through my first set of categories to see if I could label each cry with both a situational context and an emotion. This time, I found that analyzing the data while still trying to log it became disruptive: when I started crying, I immediately began thinking about what “type” of cry it was, and how it would fit in the grand scheme of the taxonomy. This was clearly an unhealthy interference with the way I was experiencing my emotions, so I stopped logging crying altogether.
Now that I was done gathering the data, there was nothing left to do but finish analyzing it. I relied on my memory to group cries based on the likeness of the feelings I remembered experiencing. For example, I would start with a large group of cries that were generally related to one emotion (like sadness) and then tease out the “type” of sadness: one group of memories had to do with loss, another with feeling hurt, another with self-assessment. Using memory for this process meant that I had to relive a lot of these experiences a number of times, which was painful and difficult, to say the least.
Image: Filtering through the categories with a physical card sorting exercise.
Once I parsed the memories in a way that felt satisfying, I looked for the right word to describe the emotion each group had in common. Sometimes it was easy to think of a word that fit, and sometimes I had to use a thesaurus or recruit help from friends. There was only one feeling that I couldn’t capture with the English language, and it was a specific (and vile) combination of rage and despair. I wound up using the word “apoplectic,” which means that you’re so angry you become incapacitated. It doesn’t capture anguish in the way that I was looking for, but it at least alludes to the helplessness I remembered feeling in those moments.
Displaying the data
I used Excel for the vast majority of the statistical and graphical analysis. I made charts of as many different relationships as I could think to measure, and often the interpretation of one graph would lead to questions that I’d try to answer by making another graph. I kept only the graphs that showed something of substance: if the punchline of a graph could be described in a sentence or statistic, I summarized it with words or left it out instead of displaying it visually. In the end, I cut about half of the graphs I created.
When it came to visualizing the relationships between the things that made me cry, I considered using a taxonomy, but I also wanted a way to show magnitude. Sankey diagrams are typically used to show flows of materials, but I thought it would be a good tool to show the complexity and size of the relationships between the contexts and emotions of my crying. I wound up using the Google Charts API to implement it.
Image: Vox.com adaptation of the interactive Sankey diagram.
Lessons and other considerations
After everything was said and done, there were a few things I learned both about my behavior and the process of tracking personal, qualitative data:
- I realized that I wasn’t being “tough” when I was around others, I was actually afraid of being vulnerable. I’ve since become more comfortable crying in front of people, and now I consider myself to be strong because I’m able to express my feelings freely (and work through them independently, if need be). Interestingly, crying in front of other people more often has led me to become more cognizant of my needs, since it is conscientious to let someone else know how, or if, they can help when you’re upset, which forces me to think about and articulate the type of support that would be helpful, if any.
- It’s a bad idea to track and analyze your self-data at the same time -- you’ll probably wind up altering your behavior, which would “contaminate” the very thing you’re trying to gain insight to.
- It’s okay to be a little imprecise. Operationalizing and categorizing behavior is meant to help someone gain a better and more objective understanding of their behavior, not a perfect or complete one. Just make sure you are careful with your interpretations.
Explore the full project here.
About the author
Robin Weis is a writer and UX designer who has gathered and analyzed life-logging data for more than 5 years. She shares her projects on www.robinwe.is.