# Would seeing Spider-Man: No Way Home decrease COVID-19 Cases?

In SNL’s cold open last night, “President Joe Biden” suggested that the COVID-19 surge we are seeing in the US is due to people seeing Spider-Man: No Way Home. If people would just stop seeing this film, he argues, cases will go back down! Interesting hypothesis, let’s take a looksy at the data, shall we?

I pulled the domestic box office data from the-numbers.com and put them in spiderman.csv (if you want to code along, you can do the same! My .csv has two columns: date and domestic_box_office) and US COVID-19 cases from the NY Times GitHub repository. Here’s a little code-along.

library(tidyverse)

col_types = cols(
date = col_date(format = "%m/%d/%y"),
domestic_box_office = col_double()
))

# get daily cases from cumulative cases
cases <- cases %>%
mutate(case = c(cases[1], diff(cases)))

d <- spiderman %>%
left_join(cases)

Now, let’s look at the relationship between daily domestic box office and new COVID-19 cases in the US.

Warning: The following is a very silly analysis. It is a joke. Please do not try at home!

lm(case ~ domestic_box_office, data = d)
##
## Call:
## lm(formula = case ~ domestic_box_office, data = d)
##
## Coefficients:
##         (Intercept)  domestic_box_office
##           3.559e+05           -3.691e-03

## Consistency

Have other studies shown a similar association? I’m going to say no…consistency is going to get rated as 👎.

## Specificity

Specificity means the exposure of interest (in this case Spider-Man domestic box office) only causes one thing (and that thing is a decrease in COVID-19 cases in this case). I’m going to go with no again. 👎.

## Temporality

Does the cause precede the effect? Kind of? It looks like sales are decreasing rapidly from Dec 7th through Dec 20 and cases appear to dramatically increase thereafter? But wait! Could the opposite be true? What if the surge in COVID-19 cases is causing a decrease in box office sales? (Acutally there may be some merit to this, but also we can just look at box office trends and see that they tend to decrease over time since lots of excited people go the first weekend and fewer as time goes on but I DIGRESS!) I’m going to give this a 🤷.

Do we see a dose effect? Indeed we do! As domestic box office decreases, COVID-19 cases increase, so can we infer the opposite, if domestic box office sales were to increase, would we see fewer COVID-19 cases? From our little window of data it looks like we can’t know (but reality check, if we had started the analysis from the day before, we actually have box office going from 0 to \$121,964,712 and we don’t see a precipitous drop in COVID-19 cases, so I’m going to give this one a 👎).

## Plausibility

Is this plausible? No, it’s silly. 👎

## Coherence

Is there a coherent argument to be made? Again, no. 👎

## Experiment

Was there a randomized trial // some attempt at estimating a causal effect to explore this relationship? Nope. 👎

## Analogy

Is there strong evidence for a similar exposure (something like Spider-Man box office sales) to cause a similar outcome (a decrease in COVID-19 cases in the US)? Nope. 👎

Ok, so using these criteria, let’s assess the overall likelihood of a causal effect:

Strength 🤷
Consistency 👎
Specificity 👎
Temporality 🤷