3 ms·

I guess then I'm not the only one to think the interpretation of the data in these slides is consistently poor. e.g. The slide with the correlation scatter plo

by sago 11y ago

I guess then I'm not the only one to think the interpretation of the data in these slides is consistently poor. e.g. The slide with the correlation scatter plots - who devised the model to draw those trend lines? That's just awful statistics. e.g.2. The slide that judges the important pages in the deck based on the length of time they're looked at (important? why? why not 'complex' or 'difficult to grok'?) e.g.3. Observed slide order actually says 'team is never in the middle', but then averages the team to be in the middle of the deck. Hmmm. What an awful piece of data analysis on potentially interesting data.

beat 11y ago
Zero is an even number. Two is an even number. The average of two even numbers is an even number. Therefore, one is an even number.
- EliRivers 11y ago
  The average of two even numbers is an even number In which universe is that true?
  - lgas 11y ago
    One where all odd numbers are even numbers.
  - beat 11y ago
    It isn't. That's the point. Fallacy in the middle of the argument.
    - EliRivers 11y ago
      At risk of outing myself as massively non hipster, I truly don't see the point. Stating obviously incorrect statements. What's the point of this game? If you're trying to suggest that the original poster made a mistake, grow up and just say so. Sometimes, there is a benefit to demonstrating a point obliquely, when doing so causes people to think about things in a new way and thus reach a burst of enlightenment. That's not the case here. This is just smug hipsterness.
minimaxir 11y ago
> e.g. The slide with the correlation scatter plots - who devised the model to draw those trend lines? That's just awful statistics. Since the tool appears to be ggplot2, the line (generated via geom_smooth() ) is likely LOESS: http://en.wikipedia.org/wiki/Local_regression http://en.wikipedia.org/wiki/Local_regression However, they deliberately removed the confidence interval provided, which is misleading.

by sago 11y ago

beat 11y ago

Zero is an even number. Two is an even number. The average of two even numbers is an even number. Therefore, one is an even number.

EliRivers 11y ago
The average of two even numbers is an even number In which universe is that true?
- lgas 11y ago
  One where all odd numbers are even numbers.
- beat 11y ago
  It isn't. That's the point. Fallacy in the middle of the argument.
  - EliRivers 11y ago
    At risk of outing myself as massively non hipster, I truly don't see the point. Stating obviously incorrect statements. What's the point of this game? If you're trying to suggest that the original poster made a mistake, grow up and just say so. Sometimes, there is a benefit to demonstrating a point obliquely, when doing so causes people to think about things in a new way and thus reach a burst of enlightenment. That's not the case here. This is just smug hipsterness.

minimaxir 11y ago

> e.g. The slide with the correlation scatter plots - who devised the model to draw those trend lines? That's just awful statistics. Since the tool appears to be ggplot2, the line (generated via geom_smooth() ) is likely LOESS: http://en.wikipedia.org/wiki/Local_regression http://en.wikipedia.org/wiki/Local_regression However, they deliberately removed the confidence interval provided, which is misleading.