The bar chart in that article, where the y axis apparently starts at 14% and change - it's really hard to think of journalists as sources of truth when they use such blatantly deceptive graphics.
Sorry, but you (and parent) seem to be misinformed on all counts. This chart is both entirely normal and makes complete sense for anyone with a financial background, as can rightly be assumed for a Bloomberg or WSJ reader.
- Financial charts commonly don't start at 0%, e.g. look at any stock chart ever. No one ever assumes it does. This is to maximize clarity of differences, otherwise a lot of financial charts would have very small differences on top of very long bars, which would be much harder to read
- Percentage change is the correct metric when the title is "growth slows". If direct revenue were shown then readers would need to compare slopes rather than than heights, which is much harder, confusing, and less clear. (Direct revenue would be appropriate for a chart titled "revenue drops", but this story is about growth, not revenue, so that would be the wrong chart.)
- YoY percentage change is a normal metric to use on a quarterly or monthly basis precisely because it removes seasonal effects, see [1]. There is nothing nonsensical about it whatsoever.
This article and chart assumes a reader who is financially literate. It is not deceptive in any way, but rather follows best practices.
Despite the dripping holier-than-thou tone of your post, I believe it is you who are misinformed:
> Financial charts commonly don't start at 0%, e.g. look at any stock chart ever.
Which is why those types of charts usually are line charts, or candlestick charts. The only reason to do a chart like this as a bar chart is to deliberately overemphasize the size of the difference between the quarters. A bar chart with a non-zero baseline is a horrible way to display this kind of data.
> This is to maximize clarity of differences, otherwise a lot of financial charts would have very small differences on top of very long bars, which would be much harder to read
Well, using a chart with a zeroed axis seems to work just fine for the very next chart in the article.
It’s not holier than though, it’s a simple factual rebuttal.
The purpose of the chart is to highlight change. showing a 0-100% range for data that shifts within a couple of percentage points makes the change difficult to see
The second chart starts at zero because the lower bounds of the data are much closer to zero. It still tops out at 70% to highlight the change.
It doesn’t matter if it’s a line or bars, the data would look nearly flat with a 0-100% range — and for what reason?
It is you who has assigned a special value in your mind to 0 and 100. There is nothing constraining real data to those limits. Growth can be negative or above 100. Those values are no more magical than 12 or 27 or 1444.829.
> the data would look nearly flat with a 0-100% range — and for what reason?
To show honestly that the change is barely noticeable.
An even more useful chart would have a y-axis from -50% to 50%. Then readers could see clearly that revenues were still growing, just not quite as fast as last period and well above any risk of contraction.
>To show honestly that the change is barely noticeable.
Any claim that the chart is misleading would have to show that it is disproportionate relative to the impact of these "barely noticable" changes on discounted future cash flows, which is the rational basis for stock valuation. These changes, if sustained, are potentially very significant.
>Then readers could see clearly that revenues were still growing
But the target audience already knows that because it's a simple headline number. The chart zooms in on a particular aspect that matters a great deal to that audience - for good reason.
>using a chart with a zeroed axis seems to work just fine for the very next chart in the article.
It actually has an axis at around -6, hence why the bars are floating above the axis. Looks like they do this to make it clear which graphs are full range and which are clipped, as the clipped ones touch the axis.
There would be no value to a chart that shows quarter-over-quarter growth, especially in a seasonal industry such as advertising. There is always a drop from Q4 to Q1 in ad revenue, simply because companies spend heavily on advertising around holidays.
I find it frustrating that so many of the suggestions in this thread for cleaning up these charts would actually lead to much more deceptive (even clickbaity) visualizations.
I agree. My point was by showing sequential quarters in the x-axis, that comparison is very hard to make. Instead the reader is probably more focused on quarter to quarter change.
A better graph would be q1 revenue of the past 10 years in the x-axis.
> Also, charts like that are very common in financial publications. The target audience knows how to read them.
Yea, I've never really understood this complaint. Except in publications targeted only at the most ignorant or innumerate audiences, it's ludicrous to claim that the only magnitude of change that can matter is one that's visible on a 0-100 scale.
Ignorant and innumerate audiences do exist, but this isn't a general-purpose argument against ever having enough respect for your audience that you can use the scale that lets you most accurately convey the data.
So you are saying reporters don't have any say in what images appear as part of their reports?
I'm not saying "any" say, but in my experience working for two major newspaper companies, it's not a given. There are other people who decide that. Again, it's not a blog.
How is the WSJ's credibility any better than a blog?
>The Wall Street Journal is the most trusted news source in the country, according to the index, with 57.7% of Americans trusting it.
All polls are biased based on how you ask questions.
https://www.forbes.com/sites/tonysilber/2018/10/03/the-wall-...
I'd say little more than half the poll is about as irrelevant as a blog. Either you trust something because of confirmation bias or you don't.
Disregarding blogs as somehow being less credible, holds traditional media to a higher standard which they have chosen to abandon for partisan reporting over the past 10 years.
Other random sources for comparison/counterpoints:
https://mediabiasfactcheck.com/wall-street-journal/https://poorvucenter.yale.edu/writing/using-sources/principl...
How is the WSJ's credibility any better than a blog
If you seriously don't know the difference between a 129-year-old newspaper that has won countless journalism awards, has broken world-changing news, is respected both among its peers and by millions of readers versus some rando blog, then I really don't know what to tell you.
WSJ takes huge legal risks in order to provide honest, ground-breaking reporting that makes society better. They were responsible, for example, for the Theranos exposé.
It doesn't matter who does it. What matters is as the end user reading that article this is pretty misleading. So from that standpoint, the journalist writing the article is implicated by default since he/she literally writes their name on the article. And frankly they should feel responsible for it.
Maybe I'm an outlier, but I don't find properly labeled charts deceptive at all. There is no hard rule that all charts must have an axis at zero, and charts that fit this template are not necessarily more valuable or informative than those that don't.
> There is no hard rule that all charts must have an axis at zero
Starting at something other than zero is like, the #1 thing that cable news channels do to make charts flashed on the screen for a few seconds look exaggerated.
And for a chart like the one shown in the article, there's no reason to not start at 0 unless you're trying to exaggerate the differences in values.
>Starting at something other than zero is like, the #1 thing that cable news channels do to make charts flashed on the screen for a few seconds look exaggerated.
In this case however it is not being flashed on the screen for a few seconds, so we have plenty of time to read it and notice the axis labels, then argue in the comments that the label that we read is deceptive because people might not read it.
Also, if labelling the y axis on this graph as something other than zero is deceptive, should the x axis start at the beginning of time?
... Or show the difference more clearly. It's about visual dynamic range. Nothing wrong with it. Maybe the problem is people assuming that every x and y axes always start at 0.
Why should it start at specifically zero though? Growth can be, and often is, negative. Shouldn't it start at -50 to be sure? The point of the chart is exactly to show ("exaggerate") the difference in values!
0 is merely a convenient numerical value that also has business meaning - no growth. You can then contrast any numerical value with this, 10% meaning it grew by a fraction of .1, and so on. For most percentage quantities, it usually makes sense to start at 0 if that has some business meaning.
Cable news channels have a problem of not labeling their axes at all, or having proportions that don't match the labels that are there. Like pie charts that show <40% as being roughly half. And then not providing enough time for viewers to fully understand the data.
> There is no hard rule that all charts must have an axis at zero
Of course, but the charts are a vehicle to show casing data, and this specific one intentionally misleads viewers. I'm not sure if it was due to ill-intention or ignorance, but charts like that go against everything charts are for in the first place.
