In section 16.9 in the book it says “a sample size of 1 has one final horizontal pixel for each horizontal pixel in the original file, and a sample size of 2 has one horizontal pixel for every two horizontal pixels in the original file.” The next line has me a little puzzled.
“So when inSampleSize is 2, the image has a quarter of the number of pixels on the original.”

Maybe I’m not getting something but wouldn’t that mean if you have 1 pixel for every 2 pixels in the original file, you would now have an image HALF the size of the original? Am I reading this wrong?

This is geometry, my friend.
If you halve the length and width of a rectangle or square, you reduce it’s area by four times.
Imagine a 4x4 square of area 16 (a pixel is equivalent to a unit of area). If you reduce it’s area to 2x2, it will now have an area of 4.
Hope that helps.

You are ONLY thinking about the horiz axis (x). What about the vertical axis (y)? The inSampleSize subsamples on both the axis. So, when the inSampleSize is 2, then 1 out of every 2 pixels on the x-axis and 1 out of every 2 pixels on the y-axis is decoded. As a result, the decoded image is 1/2 the width of the original image and 1/2 the height of the original image. Consequently, this implies that the decoded image has (1/2 * 1/2) 1/4 the number of pixels in comparison to the original image.

In other words, if inSampleSize = x, then the x number of pixels on both dimensions width, height map to a single pixel in the decoded bitmap.