Daily post

A word about chartjunk

And a few about correlation versus causation

On Thursday, pranksters ran rampant through Twitter, sporting fake (but purchased) verified check marks. One has been singled out as causing Eli Lilly stock to drop.

This stock quote is chartjunk.

Twitter display of Eli Lilly stock price
Brendan Keefe, Atlanta journalist, claimed correlation as causation:
“Real-world effects of this ‘verified’ tweet. Eli Lilly’s stock suddenly plummeted.”

The chart visually implies that Eli Lilly stock dropped almost 100% on 11 November 2022.

The design violates the visual “principle of proportional ink;” in this case, it exaggerates the difference in values.

A “filled” line chart [uses] shaded areas to represent values and should have an axis that goes to zero.

In addition, the Y-axis legend is on the right; most of us are accustomed to it being on the left, making it easy to skip. (I did the first time I saw the chart; I hurriedly showed it to my husband with exclamation marks in my voice. Later I shared a mea culpa.)

Moreover, the drop occurred on Thursday (10 November 2022) in the United States, not Friday the 11th. Of course it was Friday somewhere in the world. But not in the U.S., where it happened on Thursday afternoon, the 10th. Charts below detail date and time but not the time zone.

Big megaphones on Twitter erroneously claimed that the fake tweet caused Eli Lilly stock to drop: Brendan Keefe (51,000); Dr. Eric Feigl-Ding (721,000); Rafael Shimunov (61,700); Samuel Deats (68,600). These men decided that correlation was causation. None are economists.

Clearly, the buy-your-way-into-appearing-verified experiment failed. Bigly. But it didn’t cause stocks to plummet, much less drop.

On Friday, Eli Lilly stock did drop 4.45%, not almost 100%. Other pharma stocks also fell at the same time; for example, J&J fell 2.99%; Merck, 3.86%. The drop happened about two hours after the fake tweet.

stock price charts
Comparison of exaggerated (and not normalized) stock price movement, Eli Lilly, Johnson & Johnson and Merck.10 November 2022. @kegill

My chart of charts perpetuates chartjunk because each stock uses a different scale. Lilly is about $330-375; J&J, about $160-177; and Merck, about $95-105. Those charts should be lines, not shaded.

In the original chart, Lilly’s stock ranged about $350-370, exaggerating the drop even more than the Yahoo! charts.

Nevertheless, they all dropped at the same time. Something was in play as Friday’s U.S. holiday approached. I could find no financial analyst discussing the pharmaceutical market.

There is a slight correlation with the fake tweet: both happened on Thursday, within hours of one another. But a fake tweet about Eli Lilly giving away insulin would not affect J&J or Merck, except possibly to boost their comparative value.

Correlation does not mean causation.

Causation is hard to prove.

Remember: when something seems too good or too bad to be true, it probably isn’t.

Stop! Think before sharing. 🖖🏼

#scitech, #society, #communication
📷 123RF
Daily posts, 2022-2023 (295/365)

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.