## Friday 16 October 2015

### Definitely not.

At The Institute this year, I've got this rather sharp class of second year sporty-people.  It's good fun because they are able to motor on ahead, they seem to be getting something from the class and they help each other out.  I've been able to find some interesting data to analyse and they come up with interesting ways for processing the information.  The class runs in parallel with a series of more formal lectures on the theory and practicality of statistics, so we were requested and required to sort out the definitions of some statistical terms. This can be a rubbish assignment as it's easy to slip off to wikipedia and copy down the first sentence of the appropriate entry or, because that's plagiarism <shock horror> you try to  paraphrase that first sentence while engaging as little of your brain as will accomplish this task.  Five minutes, let alone five days, later you've forgotten 90% of what you have 'learned'.

A slightly different approach occurred to me when I was driving to work on the morning. a) Ask everyone to choose 3 or 4 out of the dozen or so head-words rather that slugging through the whole list. b) Put the submitted definitions in a long list and ask the rest of the class to decide which of the head-words is being defined. c) Offer a prize to the person who submits the most transparent definition. That worked rather well although it was a bit of work for me to fillet out and discard the head-word from each definition make the font uniform and print out the list of head-words [see immed below] and the list of defns.

 1.Test of hypothesis 4.Statistical test 7.Significance level 10.Normal distribution 2.Control group 5.Blind experiment 8.Placebo effect 11.Degree of freedom 3.Test group 6.Double Blind experiment 9.Student’s t-distribution
First task is for you, dear reader, to have a punt at defining any of these in words that your mother (a redoubtable and intelligent woman with no training in science) could understand. The next is to see if you can work out what is being defined in the following list. In the original exercise, the terminal check box was empty and the students had to enter a number between 1-11 to indicate the concept be defined. The numbers entered here are the number, out of 10, of fellow-students who grasped what was being defined.
1. a statistical test that is used to determine whether probability that a given hypothesis is true  [8]
2. A group isolated from the experiment so that they provide baseline to compare the results of the experiment against [10] >!Prize!<
3. some kind of “fake” treatment, where no effect on the health of the patient it is administered to can be attributed to the treatment itself  [8]
4. Commonly, two statistical data sets are compared, or a data set obtained by sampling is compared against a synthetic data set from an idealized model [3]
5. where neither the tester nor the volunteer knows the outcome of the test until it has been completed  [9]
6. Suppose the test statistic is equal to S.The P-value is the probability of observing a test statistic as extreme as S, assuming the null hypothesis is true.If the P-value is less than the significance level, we reject the null hypothesis.The region of acceptance is a range of values.[2]