Passing Skills

http://2013stats.wikispaces.com/3.10+Lesson+Log
http://2013stats.wikispaces.com/3.10+Introduction
http://2013stats.wikispaces.com/3.10+HELP+I+might+FAIL (if this page is just too much information).


PPDAC
produced a report that shows they have used each component of the statistical enquiry cycle to make a formal inference
PROBLEM
posed a comparison investigative question using a given multivariate data set
PLAN
use iNZight to analyse data and produce formal confidence intervals using bootstrapping
DATA
give details about the data set
ANALYSIS
Inzight:
Boxplot with confidence intervals
Bootstrap comparison confidence intervals
Sample statistics: Mean, Median, Range, Standard Deviation, IQR


ANALYSIS
Write Up: State the obvious! Describe and compare your data. Lots of writing!

Compare the statistics of the two box plots, central tendency and spread.

  • Language: Shift, overlap.
  • central tendency - mean median mode
  • spread - standard deviation, range and IQR inter-quatile-range
  • Know about Box Plots





discussed sample distributions - shape - skew - symmetry/asymmetry - outliers - gaps - tail size
The new TARSOG is TOGSSS Tail, Outliers, Gaps, Shape, Skew, Symmetry



discussed sampling variability, including variability of estimates
EG: It is important to note that the calculations are on one sample, which I can not be sure is representiative of the population. The results WOULD be different with another sample. Sampling error will be present. There could also be non-sampling errors within my sample. For example: the data is from school students who may not be very accurate at measuring!
Note VERY well: Yes, please state a larger sample would give a more accurate result, BUT, saying that does NOT show that you understand sampling VARIABILITY!
http://www.stat.auckland.ac.nz/~wild/WPRH/


Sampling Error: make sure you understand this


http://www.censusatschool.org.nz/2012/data-viewer/
CONCLUSION
Made an appropriate formal statistical inference:

Example of a Statistical inference:
It is a very safe bet that for the population of athletes in AIS, the median %Bfat of ball sport players will be between 5.45% and 11.54% higher than the median %Bfat of track/field athletes.
Therefore I can make the call that the median %Bfat of ball sport players is higher than the median %Bfat of track/field athletes.

Possible writing frames:
Looking at my bootstrap confidence interval I can fairly state that there is a difference in the female and male median % of variable in the database. This is clearly evident because my bootstrap confidence interval does not include 0.
OR
Looking at my bootstrap confidence interval I can not state there is difference in the female and male median % of variable in the database. My bootstrap confidence interval does include 0, which means I do not have enough information to make a call.