Tag Archives: gfi

Big Week o’ Blog Stats Celebration, Day 5: Overall Trends: Word Count and Gunning Fog Index

Today is mega trends day. I’ll be looking at blog-wide stuff like the overall changes in word count and the overall changes in the Gunning Fog Index. Woohoo!

Hypotheses:
A. The Word Count per blog has increased as time has gone on. That is, my blogs today are longer than my blogs when I first started.
B. The GFI per blog has increased as time has gone on.
C. There is no significant correlation between Word Count and the GFI.

Analyses!
I performed a regression (aka a glorified correlation in this case) between Word Count and Blog Number to determine if the number of words per blog has increased as time has gone on. Which indeed it has; predicting Word Count by Blog Number, the regression equation can be written as Word Count = 0.0613*Blog Number. Blog Number predicts a significant proportion of variance in the Word Count variable, F(1,2190) = 14.15, p < 0.001. Here is a plot. The red line is the regression line. As always, click on those bad boy plots to see them more clearly.

Same procedure for GFI vs. Word Count. The GFI, or Gunning Fog Index, remember, is a measurement of the readability of English writing and its values correspond to the number of years of formal education a person must achieve in order to fully understand the written passage. For example, a GFI of 10 suggests that an individual must have completed 10th grade in order to understand the material. To achieve near universal understanding, Wiki recommends that the GFI of a bit of text hover around an eight.

Anyway. The regression equation here is Word Count = 0.0008639(GFI). GFI predicts a significant proportion of variance in the Word Count variable, F(1, 2190) = 51.86, p < 0.0001. Here is another plot with another regression line.

Finally, I tested the correlation between Word Count and GFI. The correlation was -0.0028 but was not significant with t = -.0.1287, p = 0.8976.

Results:
A. Supported! The regression line isn’t very steep, but it’s significant still.
B. Supported! That’s actually a pretty impressive regression line, in my opinion.
C. Supported! There’s practically no correlation at all between the length of my blogs and the level of comprehension. I blame the surveys.

Woo!