Analysis of Survey results

You can find the raw results in this spreadsheet:

choose a format

gnumeric format.
[I slightly prefer gnumeric to Open Office.] ods format
[Open Office Calc is very similar to Microsoft Excel.]

If you click on this link, it's likely the spreadsheet will open in gnumeric; if not, save the file, and open it with gnumeric.) (Alternative instructions below.) Save this file then open it with Open Office Calc.

choose a format
gnumeric format. [I slightly prefer gnumeric to Open Office.]	ods format [Open Office Calc is very similar to Microsoft Excel.]
If you click on this link, it's likely the spreadsheet will open in gnumeric; if not, save the file, and open it with `gnumeric`.) (Alternative instructions below.)	Save this file then open it with Open Office Calc.

Useful things to try in gnumeric are: Insert → chart and Tools → Statistical Analysis → Histogram

Exploratory analysis: question 2

Raw responses 'P' to question 2 (vertical axis) versus the value of the random number 'X' (horizontal axis). Does X have an effect on P?

Histogram of responses 'P' to question 2.

Cumulative frequency of responses 'P' to question 2.

Exploratory analysis: questions 3 and 4

1: Are Aunts different from Mothers?

Except for the two mega-aunts, does it look as if mothers have bigger families than aunts?

Notice that more points are below the line (y=x) than above. Is the difference between the numbers below and above the line significant? (There are 16 above the line and 29 below. 4 are on the line.)

2: Are Aunts' family sizes correlated with Mothers'?

We might expect so, since we know that some countries have higher average numbers of children per woman than others. Statistics has many ways to measure correlation. A correlation coefficient is a standard quantity to compute. A rank-rank scatter plot is also a good idea, especially in this problem, since our data contains outliers, and a correlation coefficient is not robust to outliers. Ranks, in contrast, are robust to outliers. I haven't found an easy way to make a rank-rank scatter plot yet in gnumeric, though it is certainly possible.

Back to the main survey page

Listen to a 'Yes Minister' play, that explains how opinion pollsters can get whatever answer they want to a question.

(Here are instructions for copying the gnumeric file and opening it using a Terminal.)
mkdir survey cp ~mackay/probability/survey/Responses.gnumeric survey cd survey gnumeric Responses.gnumeric