And frankly, I’ve always found the whole “margin of error” thing more than a bit confusing, so again, I appreciate you breaking that down as well. I could be wrong, but my suspicion is that the average reader of a mainstream publication like the NYT doesn’t fully understand the meaning of “margin of error,” either.

Finally, I feel the need to point out to you — and for that matter, to everyone else who happens to read this comment — that the last two paragraphs of your “Part 2″ comment pretty much perfectly encapsulate *everything* I was trying to say myself. Especially your last paragraph, because I think you’re absolutely right that until 20 or 30 years pass (or maybe 10 or 15; who really knows?), it simply won’t be possible to parse this sort of data in studies or reports or papers without coming across as at least fairly confusing or, yes, even contradictory-seeming.

In the meantime, we have what we have, I guess — the Pew Research Center reports and so on — and I won’t say that there isn’t any inherent value to those sorts of reports. However! When the mainstream media chooses over and over again to treat them as indisputable fact, or manna from heaven, say, it’s kind of hard not to chuckle a bit.

Anyway, thanks for reading, Gary.

]]>You say:

Even The New York Times‘ blog post about the report, for instance, included a line of fairly confusing info: “Sales of dedicated e-book readers may have peaked last year,” writes tech reporter David Streitfeld, “but the percentage of Americans owning one rose to 19 percent from 10 percent.”

Wait … what?! The first part of that quote seems to imply that a significantly smaller number of people bought dedicated e-readers in 2012 than in years past. But the second part seems to imply that the percentage of people who own a dedicated e-reader nearly doubled in 2012.

—————————

This is not impossible. Imagine the following possible scenario. [Note - All numbers are completely made up. I really wish I knew how many readers were sold, but the manufacturers won't release that information.]

Year 2008 – readers sold 1 million – total readers owned by the population 1 million

year 2009 – readers sold 2 million – readers owned 3 million*

year 2010 – readers sold 4 million – readers owned 7 million*

year 2011 – readers sold 8 million – readers owned 15 million*

year 2012 – readers sold 7 million – readers owned 22 million*

*This, of course, ignores the fact that some of the e-readers sold in previous years broke and were thrown out, so the total, cumulative number of readers would be smaller than the sum of all units sold.

Even though “peak sales” were 8 million units in 2011, the sale of another 7 million units in 2012 still increases the total number of units in the hands of the population by a significant amount.

And, of course, the 10% and 19% numbers for e-reader ownership came from survey data, and these numbers are not perfectly accurate, since every survey has a margin of error.

————————

I think I really agree with the title you put on this story. Its mostly bullshit, or perhaps looking at the tea leaves in the bottom of the teacup. There is hard data out there. Amazon.com could tell you exactly how many kindles they have sold in each month of the last 5 years. They could also tell you how many ebooks they have sold in each month of the last 5 years. However, Amazon WON’T release this data. Instead, we are reduced to scrutinizing the tidbits of information contained in press releases about various surveys.

Perhaps in another 20 or 30 years there will be scholarly articles about the development of ereading that are based on real data that has been ‘declassified’ by hardware vendors, ebook retailers, and publishers. This might make a wonderful PHD thesis for someone who has not yet been born. Until then, all we can do is guess at what is really happening.

]]>“Overall, the number of book readers in late 2012 was 75% of the population ages 16 and older, a small and **statistically insignificant** [emphasis added] decline from 78% in late 2011.”

You say:

“But if the overall number of book readers—which, for the purposes of this study, seems to include readers of both e-books and print books—has declined from 78 percent to 75 percent over the past year, wouldn’t that suggest that the act of reading has, in fact, declined slightly?”

———————-

The problem here is that every survey has a margin of error. I don’t know what the margin of error is in the Pew survey, but **imagine** (that is, assume for the sake of this discussion) that the error is +/- 5 percent, 19 times out of 20. That means that for one poll out of 20 the error will be MORE than +/- 5%.

Therefore, we can be pretty sure (but not absolutely certain) that the number of readers in the general population in 2011 was between 73 and 83 percent. (71 or 72, or 84, or 85 percent are still ‘possible’ true numbers for the percentage of readers in 2011.)

We can also be pretty sure (but not absolutely certain) that the number of readers in the general population in 2012 was between 70 and 80 percent. (with a possibility that the ‘true’ percentage of readers was outside of this range).

From the data provided, therefore, it is entirely possible that the percentage of readers in the general population increased from 72 percent to 81 percent over the last year. It is also possible that the percentage of readers dropped from 84 percent to 69 percent last year. It is also possible that the percentage of readers was 77 percent in 2011 and was 77 percent in 2012, so there was no change at all.

That’s what “statistically insignificant” means. The supposed drop of 3 percent is small when compared to the known accuracy of the polls. Therefore, you can not determine the real change in the number of readers from the data provided.

]]>