Wednesday 28 March 2012

lies, dam lies and statistics!

At the end of last week i read a blog post by tilly from tilly and the buttons. People are doing a challenge to use one pattern and wear clothes made form it for one week. She had collated the data into various charts showing which make of patterns was most popular and what item people were making. Being sad i asked to be sent the data and set to, to investigate any stats i could! 
When i usually do statistics I'm working with numbers and so the software package i use is number related! i surfed about for a more qualitative based one but came up with nothing. so i set to with what i had a performed a correlation between pattern type and pattern manufacturer to see if there was any relation ship between what the participants are making and who designed/sells the pattern.

As you can see there is no relationship and the R square value is very low. (the closer to 1 it is the better the relationship, also i gave each manufacture and item a number so the graph would plot, i can give you the details if you are desperate!)

I also collated all the data into one graph, showing easily that tops by collette are the most popular! i can not, however, proved statistical significance (this is where i waffle on about n numbers and power calculations!)
Ok, enough with the data mining! i have science data to sort!

2 comments:

  1. Replies
    1. I use instat. Minitab is too hard for me, have to remember what you do every time where as instat is designed for idiots like me.

      Delete

Your comments are alway welcome, i know they take you time, i appreciate each one because, the smile they bring is mine.