Transcript Support for School Statistics form Statistics NZ

Support for School Statistics
from Statistics NZ
[email protected]
Statistics New Zealand
Auckland Maths Assoc, University of Auckland
Tue 25 Nov 08
1
Achievement objectives for today:
Participants will:
Use Stats NZ resources to
deliver curriculum objectives
Feel more confident and have more fun
with teaching the stats
in Mathematics and Statistics
in the NZ Curriculum
Find out (if time)
what do (some Stats NZ) statisticians
really do!
2
Activities with www.stats.govt.nz:
After an introductory ramble:
Schools corner
StatZing!
SURFs 1, 2, 3
CensusAtSchool (a mention)
Table Builder (= TB); esp Census data
Infoshare:
Time series galore
Hot Off The Presses (= HOTPs):
HOTPs and Statistical Literacy
QuickStats:
about your place etc
Then: what do (some) statisticians really do!
3
Curriculum and Stats NZ Resources 1:
The threads in the Stats and Probability strand:
Statistical investigation
phenomena involving:
multivariate (case) datasets
time-series datasets
Statistical literacy
reports with words, numbers, graphs
risk
Probability
distributions
dependence etc
4
Curriculum and Stats NZ Resources 2:
The threads … and resources for them:
Statistical investigation Schools Corner,StatZing!
phenomena involving:
HOTPs
multivariate (case) datasets: SURFs,TB,CaS
time-series datasets:
Infoshare
Statistical literacy
NZ in Profile
reports with
Quickstats
words, numbers, graphs:
HOTPs
risk, relative risk:
HOTPs, Tables
Probability
distributions:
Tables
dependence etc:
Tables; 2 way
5
Curriculum and Stats NZ Resources 3:
Some are
designed for schools
Some are
(a big one) inadvertently useful for schools!
EG: The HOTPs (Hot Off The Presses):
EG: a rich source of real (we hope) info:
New Zealand Income Survey: June 2008 quarter (a big one)
Highlights | Commentary | Technical notes | Erratum | Tables |
Stat Literacy:
Evaluate stat reports
(L 6,7,8)
Stat investigation:
Methodology:
defining questions,
sampling methods,
errors (samp and non)
etc etc etc etc
Stat investigation:
Story, Data
Time series
Probability:
One-way tables
Two-way tables
Statistical Enquiry Cycle: PPDAC … PPDAC … PPDAC … PPDAC … PPDAC6 …
Why be nice to schools??
Stats NZ:
Dataset
Dataset
Dataset
The World:
Respondents:
People
Businesses
Users:
Public
Professional
Technical
We need our clients to be informed & positive
School stats is a vital way to achieve this
7
Two groups with converging interests:
The
Official Stats
sector
The
Mathematics and Statistics
Education Community
Vision:
an informed society
using statistics.
Curriculum: students will be:
thinking mathematically and statistically;
solving problems, modelling situations.
8
A small problem:
Dataset
Dataset
Dataset
Unit-record multivariate datasets:
Teachers need them!
Official Stats agencies
have lots
but can’t release them! SURF 1
SURF 2
Some smart solutions:
SURF 3
CensusAtSchool (sort-of)
SURFs for Schools: 1, 2, 3
Area Unit
Tables by
Henderson North
geographical Area Henderson South
Tangutu
Woodglen
Glen Eden East
New Lynn North
New Lynn South
Males06 Females06
2,487
2,817
1,956
2,070
1,404
1,554
2,013
2,193
3,237
3,372
1,173
1,233
1,185
1,287
Tot06
5,304
4,023
2,955
4,203
6,609
2,406
9
2,472
Census at school 2009
New dates:
3 March 2009 until 9 April 2009
Register online:
http://www.censusatschool.org.nz/2007/register/
If you have previously registered, OK.
Confirmation in November.
Funded:
X% by Stats NZ
(1-X)% by MoE
Expertise:
Lots of it; from UoA
10
New for 2009
Teachers get their class results back
if they choose.
Early in year so 2009 data
can be used for 2009 teaching.
New questions: from consultations:
Dept of Stats UoA, MoE, Stats NZ, teachers nationwide
Questionnaire critiqued
by StatsNZ Questionnaire design team
www.censusatschool.org.nz
11
www.stats.govt.nz
3 Infoshare
12
StatZing! Latest Sec (Economics)
SURF 2
13
SURFs for Schools
Synthetic Unit Record Files:
Multivariate datasets from Stats NZ surveys
1. Income supplement from the 2004
Household Labour Force Survey
2. 2001 Household Savings Survey
3. Coming soon – 2006 Census
14
15
2001 Household Savings Survey SURF
Based on a survey that collected information
including income, assests, debt,net worth.
300 synthetic people
representing the 5000+ people
who responded to the survey.
16
2001 Household Savings Survey SURF
Variables include:
–Gender
–Employment
–Qualification
–Ethnicity
–Partnered
–Age
–Age of Partner
–Total income
–Wages/Salary
income
–Total debt
–Total networth
17
18
19
Using the SURF
20
Teacher page for each activity
•Curriculum links
•Possible answers
•Available as a PDF document
21
What can we improve?
For teachers
For students
22
Census: SURF 3
Under development; final checking
Based on
2006 Census of Population and Dwellings
Contains unit record datasets for each of New
Zealand’s 16 main Regional Authorities
300 synthetic people who represent everyone
that responded for each region
23
Census: SURF 3
Variables included
– Sex
– Work and Labour force status
– Qualification
– Ethnicity
– Income
– Age Group
– Mode of transport to work
– Hours worked
– Cigarette smoking behaviour
– Access to a cellphone/mobile phone
– Access to the internet
24
Access to internet
Age
x
Cigarette smoking behaviour
Ethnicity
Hours worked in employment
per week
x
x
Main means of travel to work
Qualification – highest
Sex
Work and labour force status
x
x
x
x
x
x
x
x
x
x
x
x
x
x
Access to a
Cellphone/Mobile
Phone
Access to internet
Age
Cigarette smoking
behaviour
Ethnicity
Hours worked in
employment per
week
Main means of
travel to work
Sex
x
x
x
x
x
x
Qualification –
highest
Work and labour
force status
Variable
Access to a Cellphone/Mobile
Phone
Personal Income
Preserved relationships
x
x
x
x
25
Limitations: SURF 3: Census
Synthetic data
Not all relationships and patterns are preserved
Joining tables together
does not represent the whole of New Zealand
However, you can compare regions!
26
Battle for the ‘greener suburb’:
an example using case data
Compare the ‘traveling to work’ habits of
geographic areas.
Which area has the ‘greener’ workers?
– Walking / Running / Cycling
– Public transport
– Carpooling???
– Working at home?
(Graphic from CensusAtSchool
)
27
Battle for the ‘greener suburb’:
where to find the data
We want a data source that contains
information about modes of travel to work by
area units.
Luckily, we have the 2006 Census of
Population and Dwellings on Table Builder!
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
Travel to
work
in the
four
Auckland
cities
Counts: People by
City of usual residence,
Main Means of Travel to Work:
2006 Census
Not Elsew here Included
Other
Walked or Jogged
Bicycle
Motor Cycle or Pow er
Cycle
Train
Manukau City
Public Bus
Auckland City
Passenger in a Car, Truck,
Van or Company Bus
Waitakere City
North Shore City
Drove a Company Car,
Truck or Van
Drove a Private Car, Truck
or Van
Did Not Go To Work Today
Worked at Home
0
50,000
100,000
44
Travel to work in Kapiti and Wellington
50%
Kapiti Coast District
Wellington City
40%
30%
20%
10%
0%
Worked at
Home
Did Not Go
Drove a
Drove a
Passenger Public Bus
To Work Private Car, Company
in a Car,
Today Truck or Van Car, Truck Truck, Van
or Van
or Company
Bus
Train
Motor Cycle
or Power
Cycle
Bicycle
Walked or
Jogged
Other
Not
Elsewhere
Included
45
Table Builder: Datasets on Area Units:
At this point
the screen-shots
stop.
But there’s a
2-slide
summary …
8000
Population: Census 2006 vs
Population: Census 2001
for Area Units of Waitakere City
6000
Sturges North
4000
Here's data for 10 Area Units:
Area Unit
Tot01 Tot06
Sturges North
2283 5772
Kingdale
3480 3537
Fairdene
4410 4554 2000
Whenuapai West
1836 1842
Herald
1656 1698
Hobsonville
3342 3378
Westgate
705 1092
Royal Road West 2424 2664
0
West Harbour
4569 4932
0
Lucken Point
4656 5238
2000
4000
6000
46
8000
www.stats.govt.nz for schools: short guide: p1
Schools Corner
SURF (No. 2)
About the data source | The dataset | Activities
(copy the dataset and paste into your spreadsheet)
StatZing! (the latest Activities)
Find by …(find old StatZing!s etc)
Table Builder
2006 Population Census
Selected tables
Travel to Work
Expand (find the Areas you want)
Tick (use the ticks above and to left)
Click the Table icon
Actions, download to XL format
(then copy and paste into your spreadsheet package)
Age by Sex for 1996, 2001, 2006 (then as above)
47
www.stats.govt.nz for schools: short guide: p2
Infoshare
Browse
Work, Income and Spending
Linked Employer-Employee Dataset
Age and ANZIC96 (ANZ Industry Classification 1996)
Select a few items, and for Time, Select All
Go
Pivot clockwise, to get data into a column
Save as xls
(copy the dataset and paste into your spreadsheet)
Releases by Title (Takes you to Hot Off The Presses)
NZ Income Survey
NZ Income Survey; June 2008
(then explore these:)
Highlights|Commentary|Technical notes|Erratum|Tables
QuickStats about a Place
(and also see QuickStats about a Subject, and NZ in Profile)
Place List (and find your suburb)
(and use the 12 tabs).
48
What Statisticians do all day: an eg:
The new Immigration Survey:
Pop:
36,620 approved immigrants in 2004
Sample:
7,125 of them
We find Estimates (via ‘resampling’)
with Sample Errors (= half the confidence interval)
Immigration survey: Labour Force Activity
Labour force status:
Employed
Looking for work
Immigration approval category
Number Samp err
Number Samp err
Skilled Primary Applicant
11,630
510
220
90
Skilled Secondary Applicant
5,130
350
410
130
Business
1,210
100
40
30
Family partner
4,770
150
240
80
Pacific
1,120
160
70
10
Notes:
This is an example:
values are not necessarily the actual ones.
table is incomplete
Samp Err is sample error:
approximately 2 times the standard error found by jackknife.
Number is the estimate, from the sample, o
of the total number of people in the cell, for the population studied.
The confidence interval for the estimate is:
Number ± Sample error.
49
Sample Error vs Estimate:
300
Sample Error vs Estimate
for cells from a LISNZ table
jackknife
200
100
0
0
1000
2000
Hmmmmmm: what does that show?
50
Sample Error (up) vs Estimate (across):
300
For cells from the
Immigration Survey NZ
Sample Error vs Estimate
for cells from a LISNZ table
jackknife, binomial 1
Sample Error from
the data by Jackknife
(ie resampling)
200
100
Standard Deviation
from binomial model
= √ (p (1-p) N)
0
0
1000
2000
Sample error has lots of variation:
Can we explain some of it? How?
What function might it fit?
For the lower (blue) points, what did we forget?
51
Some consultation:
We asked Pat:
You need to multiply by the 2 value:
Sample Error
= z * Standard Error
= 1.96 * Standard Error
They forgot to multiply by 2
(or 2 ish)
Why are they so dumb in
Wellington??
52
Sample Error vs Estimate:
300
Sample Error vs Estimate
for cells from a LISNZ table
Jackknife, binomial 2
Sample Error from
the data by Jackknife
(ie resampling)
200
Sample Error
from binomial model
= z * √ (p (1-p) N)
100
0
0
1000
2000
Hmmmmmm: how does that look?
53
To confidentialise, we added noise: ± 4
300
Sample Error vs Estimate
for cells from a LISNZ table
jackknife
Noise from
confidentialising is
about this big:
200
100
Noise from sampling
varies, but is this big
0
0
1000
2000
Does the noise from confidentialising matter?
54
What do we do all day?
In a Stats office
Find a problem that matters
Find some Data (Evidence)
Talk, scratch heads
Do some graphs
Try models using Maths
Make mistakes
Consult with wise heads
Do more graphs
Make decisions
Communicate results in:
words, numbers, graphs
In a Stats classroom
Find a problem that matters
Find some Data (Evidence)
Talk, scratch heads
Do some graphs
Try models using Maths
Make mistakes
Consult with wise heads
Do more graphs
Make decisions
Communicate results in:
words, numbers, graphs
We hope you enjoy statistical discovery as we do!!
55
Gender Balance in Waitakere:
Females vs Males
for Area Units of Waitakere Cit
2006 Census
3000
Y+X line
2000
1000
Herald
0
0
1000
2000
3000
56
The Maths and Stats teacher’s vital role:
If the next cohorts of adults
can handle statistical evidence and thinking:
That’ll be nice for Statistics NZ!
which produces:
social, economic and environmental stats
That’s utterly essential for solutions to
NZ’s and the Earth’s challenges.
Enjoy!!
57