Thanks to the hard work and dedication of Alastair Lupton, you can not download a copy of a student workbook that details the events of this LUMAT offering. As usual it is a work in progress, but it is looking pretty good from where we sit.
You will recall this activity aimed to assist students in developing a way to think about the process of sampling: the variability that is associated with sampling and how sampling can help us to determine facts about a large population (without looking at each of the population's members).
This document (pdf) provides the basic information about the weight of healthy koalas.
This document is a simple, one page, handout (pdf) that outlines the task.
This spreadsheet provides the data from which the students sample. Simply open it, (instructions are given on the SS) press print and you will get 6 pages of square bears!
This spreadsheet provides the data from which the students sample, but consists of the rogue bears. Recall the dirty trick I played by giving one group a different population to the rest. Simply open it (instructions are given on the SS), press print and you will get 6 pages of square, but rogue, bears!
This sheet of axes (pdf) provides the place where the students can produce their 5 dotplots and which you can then staple together to make the lovely wall art that we made in the sessions - see below.
You will recall we started the second session by wanting to refine what we had learned in session one. Was it possible to find a more refined description for the way samples varied, in particular the mean of each sample? What might effect the amount of wiggling (variation) between the means of the samples? Might the size of the sample effect this?
We investigated this by using the the NORMPOP and REPSAMP programs for the Casio 9860. NORMPOP produced a population of 999 individuals with normally distributed weights with mean and standard deviation of your choosing. Then we used REPSAMP to repeatedly draw sample of size (n) of your choosing. We all used a different value of n so we could see the effect.
We used this sheet (pdf) to record each persons work. Once we had stapled them together we made the wall art you see below. Discussion followed - bigger n, less wiggling!
You will recall we started the second session by wanting to refine what we had learned in session two. Can we predict how much wiggling there is if we know the sample size (n)?
We investigated this by using the the NORMPOP and SEEMEANS programs for the Casio 9860. NORMPOP produced a population of 999 individuals with normally distributed weights with mean and standard deviation of your choosing. Then we used SEEMEANS to compute and store the means of MANY samples of various sizes (n).
We used this sheet (pdf) to record the work. Once completed we then used some modelling to see if we could determine a relationship between 'sigma x bar n' and n. We used a power model you will recall.
So we were nearly there now, but what about populations that were not 'normal'?
We investigated this by using the the SKEWPOP1,2 or 3 and SEEMEANS programs for the Casio 9860. SKEWPOP(X) produced a 'skewed' population of 999 individuals. SKEWPOP3 being the most extreme in skewness. Then we used SEEMEANS to compute and store the means of MANY samples of various sizes (n).
We used this sheet (pdf) to record the work. You will recall finding out that 30 is NOT the magic number!
I hope these materials and the 3 or so hours we spent together are/were useful to you and your students.