Research Question: How Often Do I Pet My Cats?
My husband and I have three cats: Wilmer (3 years old, ginger), Calliope (2 years old, black), and Seamus (The Wee Kitten, 3 months old, tabby). They’ve all been rescued (at different times) from a feral colony in Jackson Heights, Queens, by a friend’s mother. And they’re all a bit shy around people. We’ve had cats previously who were much more outgoing and affectionate, so we’re always celebrating any time these cats let us interact with them. Which got me thinking: How often do I actually pet each of my cats? And how can I capture this in a dataset?
For one week (starting Monday October 18 through Sunday October 24, 2021), I kept track of every interaction I had with my cats that involved petting. I noted when each interaction began, approximately how long it lasted, and how many full-body strokes I pet the cat with. I also kept track of where in the apartment the interaction happened, if there was food involved, and if so what kind. I recorded who initiated each interaction and who ended it. I defined these fields as who touched who first and who walked away from the interaction, respectively. I also noted if the cat purred at all during the interaction as well as rated them on a scale of 1 to 3 (1 = not satisfying; 2 = somewhat satisfying; 3 = very satisfying). Lastly, I wrote down any notes about the interaction to better capture the moment.
Audience
The primary audience for this project is my household. Knowing how often we pet our cats is ultimately the most interesting to us; I hope that it also will prove beneficial to our cats, learning their patterns and more often doing things to improve our interactions and thus ensure many more (and more satisfying) interactions in the future.
The secondary audience is our friends and family who I’ve spoken with at length about this project. Even our vet is excited to hear the results. Any fellow cat-lovers out there who may wish to create their own dataset about their fur babies will also benefit from this post.
Description of the Visualizations
The first visualization I knew I wanted to create was a box and whisker plot (I just HAD to!). And given the very wide range strokes per interaction, I thought it would be best used to show this information for each cat. The least amount of strokes for each cat was 1, but they all had at least one interaction that exceeded 300 strokes, and Calliope came in on the last day of data collection with more than 800 strokes in a single interactions (it was glorious!). But the quartiles show that for all cats, most interactions occur between 1 and 100 strokes. To support this, I also created a pair of bar graphs displaying the total petting time and the total strokes per cat. I knew these would be different based on previous interactions. When Calliope wants to be pet, she will come to us and sit on our laps and she gets really into it; she will not let you sit there and pet her with only one hand, thus inflating her stroke count. Wilmer likes to walk and get pet, walking up and down the hallway with us in pursuit, or if he hops up on the couch for pets, he frequently hops down and back up again at various intervals, thus inflating the duration of his interactions while keeping the stroke count on the lower side. I’m very partial to my notes I took about each interaction as I think they reveal some of the struggles I had while creating the dataset and also show a little bit of personality of each of my cats, so I also created a chart that gives an overview of each interaction for each day. The bubble size corresponds to the length of the interaction, color corresponds to cat, and the tooltip also includes the number of strokes per cat and my notes.
On the second page of my storyboard, I look at petting across the days of the week. I duplicated the bar graphs from the first page, except I color-coded each bar according to the day of the week. Petting is wildly inconsistent across cats and across days. To further contextualize this, I kept track of how many hours I was awake each day, as well as how many hours I was actually in the apartment. I called this my “available petting time.” I also tried my hand at a calculated field so I could see how much of the available petting time I actually use to pet my cats. This last bar graph on the page is filterable by cat. I thought I pet my cats a fair bit over the week, but it looks like I only spent between 1% and 3% of my available time petting them. I know I have other obligations throughout the day, but still, these times could be higher!
On the third page of my storyboard, I examined agency and satisfaction. In the top two bar graphs, you can see a big difference in who starts an interaction and who ends it. I started approximately 2/3 of our interactions; the cats collectively ended more than 3/4 of our interactions. This is not surprising (sigh!). Next I compared my satisfaction ratings with their purring, assuming purring can be a sign of enjoyment for the cat. I only ranked 14 (out of 62) interactions as not satisfying (petting cats is generally a very satisfying endeavor!). The cats only purred in 12 of our interactions (less than 1/4). I had hoped this number would be higher, but it’s an imperfect metric (so I keep telling myself).
On the fourth and final page of my storyboard, I created a set of tree maps to see where in the apartment the petting is happening and how often and what kind of food is involved. By far the most petting happened in the living room. We spend most of our waking hours here–working, eating, relaxing–so this makes sense. And within the living room, most of the interactions took place on the couch, which again is where the most time is spent. I also feed them treats on the couch as I want to encourage them to be on the couch when we are, so this inflates my numbers here. In terms of feeding, more than half of our interactions occurred without food, which was a very heartening statistic (they like us, they really like us!). When food was involved, the trend mostly followed which food is eaten most: more pets for wet food because they eat this twice a day, the least pets for dry food, which they only eat once a day and usually my husband feeds them this. Treats are sporadic, but we use them to make the cats love us.
Data and Design Decisions
From the moment I conceived of this project, my husband did not agree with some of my parameters. Namely, who initiates the petting. Wilmer very rarely will touch us first, but he has this look and sometimes a small head nod he gives us when he wants to be pet. So his initiation numbers may be artificially low, but I didn’t know how to account for this, and neither of the other cats share this behavior, and I couldn’t define the parameter differently for each cat. My initiation numbers are likely even higher than what I recorded: when I give them treats, I shake the jar to entice them over, and each cat generally touched me first when they came to me to get them. On an even more basic level, the first thing many people asked about when I told them about this project is whether tracking petting would make me more likely to pet the cats. If I’m thinking about it (and I was as I knew I wanted to get to a minimum of 50 lines of data), does that make me more likely to do it? On the hand, it was really challenging recording all of these data, especially when there were multiple cats being pet more or less simultaneously.
And there were times where I noted that I ended an interaction because it was after midnight and I wasn’t sure how to record data that was technically the next day even though I wanted it to count as the same day, or when Wilmer was being pet in the hallway and then walked into another room and clearly wanted to keep being pet. Would that be two interactions, or should I have allowed for an interaction to continue across multiple locations? In the first case, Dr. McSweeney said I could add a new column and call it day so that I could track my days as I experienced them. For the second issue, I just ended these interactions because I didn’t know how else to handle them in the moment. Do the overthinking/over-petting and early termination/avoidance of petting balance each other out? I’m not sure. I was also over-eager in thinking I could accurately time my interactions with the cats. Many interactions were brief, and clearly less than a minute. So I decided than any interaction that was clearly less than 1 minute would be noted as 0.5 minutes to make things easier. Some interactions involved brief pauses, for instance Wilmer walking while being pet, or a cat hopping up and hopping down from the couch. When these pauses were assuredly less than a minute, I decided to track it all as one interaction rather than multiples. If the pause was longer, or one of us left the room, then I considered that interaction to be over. Whenever multiple cats were being pet at more or less the same time, I wrote this in my notes column, as I think any of the data at these time could be more prone to human error. It’s really hard to pet three cats consecutively while they eat treats out of your lap and keep track of time and strokes.
The dataset also only reflects my petting the cats, not my husband. To get a fuller picture of how we interact with our cats, I’d also need to collect his data, but I was afraid that would turn it into a competition, so I kept it to just me, and I expressly picked a week where he had to work in his office two of the days. (I also picked this week as the previous week I was recovering from hip surgery and was therefore less mobile.) On Friday, we also had the vet visit because Seamus is still getting all of his kitten shots. The other cats hid for several hours from us after they were here as they’re terrified of the vet. I’m sure all of these affected every aspect of my data collection.
In terms of design, I knew it would be very easy to color code the data as each of my cats has very distinct coloring, and I wanted to match them. I had hoped to create custom pattern fills so I could give Seamus and Wilmer’s data stripes, but all of the tutorials I found explaining this had older versions of Tableau and seemed to require creating faux calculated fields. I honestly struggled enough with creating real calculated fields, so I couldn’t quite make it work and so put pictures of my cats in the storyboard pages instead. For the data that more focused on days of the week, location, and food. My overall color choices don’t quite go together, but I thought it was important to make this distinction. I chose to use a lot of tree maps and area maps as a lot of my data were strings and ordinal numbers, and because these are new visualizations for me. I also utilized Tableau’s alias feature for many columns of my data as I wanted most of my tooltips to read as sentences, so I was able to easily create aliases that conveyed the information more clearly.
Next Steps
I’ve noted all of the ways in which the week I tracked data was perhaps not reflective of every week. To learn more about any trends in cat petting, I would ideally like to track it for longer and I’d also have my husband keep track of his interactions. Is petting going up? It feels like it since the temperatures have dropped. If I could go back in time, I’d love to have been able to compare this data with petting data of previous cats who were much more affectionate. Perhaps I’m not giving these babies enough credit because I miss the others?
Within the data I’ve collected, I would like to further analyze the locations of the apartment alongside the satisfaction and purring data–are there any trends? Similarly with food, are there more purrs with or without food? Were my interactions generally longer or shorter when food was involved, and how did my satisfaction levels change when food was involved?
