So the great big project I’ve been working up to all semester, the one that is a major hunk of my grade in two separate classes. I finally got the data sets together to attempt an answer to my research question. And none of my variables had significant associations. NONE OF THEM!!!!!!!!
I was so sure of the right-ness of my hypothesis. So sure, based on having lived in the skin of this hypothesis for the past six years. SO DAMN SURE!!! But the numbers said no. No. NO. NOOOO!!!!
So I ran them again. In reverse. NO! Then I looked again at the scatter-plots of my variables and there they were. Nine observations which were outside of all realistic experience. Nine out of over 1100. So I took them out. And guess what. “What?” My predictor variable had a statistically significant association with 5 out of 8 of my outcome variables.
So I can climb down from the top of the building, drive away from the edge of the cliff, cancel that imaginary appointment with the registrar to drop the program NOW! I can breath. And I can finish my paper. Both papers. And probably even keep my grades high enough to keep my funding.
No animals were harmed in the writing of this blog.
How I Haven't Killed Myself 
Oh holy crap. When I was reading about the non-relationships of your variables I thought you’d have to spend your whole paper explaining the lack of correlation. Thank goodness you threw out the outliers. I’m surprised statistical analysis gave them as much weight as it did when included in the set. They must’ve been waaaay out! And I’m glad you didn’t harm any animals. Love you!
LikeLike
Thanks. I was on the brink for a while. But in some places the ratio was like 12,000 to one. So totally ridiculous. Then I was scared that taking them out was like cheating. But I talked to a couple of the students ahead of me in the program, and their advice was to pull them out, but be sure to explain that in the paper. So I guess it’s not all that uncommon.
LikeLike