Professional Documents
Culture Documents
Statistic Introductroduction
Statistic Introductroduction
Statistic Introductroduction
Statistical Methods
1. Introduction
Peter
Samuels
Birmingham
City
University
Summary
q What is statistics?
q What is a mean?
q Data types
q The research study process
q The statistical analysis process
q Some basic statistical concepts
q Benefits of good study design
q Comparison of two study designs
www.statstutor.ac.uk
Peter
Samuels
Birmingham
City
University
Activity:
What is statistics?
1 minute:
q Write down your own definition
2 minutes:
q Discuss it with your neighbour and agree on
a definition
www.statstutor.ac.uk
Peter
Samuels
Birmingham
City
University
What is statistics?
The word statistics is used in 3 main ways:
1. Common meaning: factual information involving
numbers. A better word for this is data.
2. Precise meaning: quantities which have been
derived from sample data, e.g. the mean (or
average) of a data set
3. Common meaning: an academic subject which
involves reasoning about statistical quantities
In order to use statistics properly you need to be
able to think about statistics in the right way
www.statstutor.ac.uk
Peter
Samuels
Birmingham
City
University
Peter
Samuels
Birmingham
City
University
What is a mean?
The mean of a data
Country
Number
Country
Number
set is a measure of
Canada
22
Spain
10
its middle value.
France
52
Sweden
12
Example: The
Japan
43
UK
41
number of nuclear
South Korea 9
USA
119
power stations in
West Germany 23
various countries in Soviet Union 73
1989.
To calculate the mean, add all the data values together and
divide by the number of values.
22 + 52 + 43 + 9 + 73 + 10 + 12 + 41 + 119 + 23 384
X=
=
= 38.4
10
10
www.statstutor.ac.uk
Peter
Samuels
Birmingham
City
University
Data types
In statistics it is vital to understand what types of data
you are working with.
There are three main types:
q Nominal categories that do not have a natural
order, e.g. gender, eye colour, types of building
q Ordinal categories which have a natural order but
are not numerical, e.g. Likert scales
q Scale/continuous numerical data ordered against
a constant scale, e.g. date, temperature, length, weight,
frequency
www.statstutor.ac.uk
Peter
Samuels
Birmingham
City
University
Activity:
CensusAtSchool
Phase 6
Questionnaire
Available from:
http://
www.censusatschool.org.uk/
images/phases/phase6questionnaire.pdf
Discuss with your neighbour
the data type of each question
www.statstutor.ac.uk
Peter
Samuels
Birmingham
City
University
Answers
Qn. No.
1
2
3
4
5
6
7
Type
Nominal
Nominal
Scale
Each: Scale
Nominal
Scale
Scale
8
9
10
Nominal
18
Each: Ordinal 19
Scale
20
www.statstutor.ac.uk
Qn. No.
11
12
13
14
15
16
17
Type
Each: Scale
Nominal + free text
Ordinal (unsure in middle)
Nominal (multi-answer)
Nominal + free text
Nominal
Nominal
Nominal
Each: Scale
Each: Scale
Peter
Samuels
Birmingham
City
University
Define
objectives
and draft
research
question(s)
Process
data
Design
study
and plan
statistical
analysis
Statistical
analysis
Conduct
survey,
study or
experiment
Report
results
End
Peter
Samuels
Birmingham
City
University
www.statstutor.ac.uk
Peter
Samuels
Birmingham
City
University
Peter
Samuels
Birmingham
City
University
Peter
Samuels
Birmingham
City
University
www.statstutor.ac.uk
Peter
Samuels
Birmingham
City
University
Peter
Samuels
Birmingham
City
University
Imprecise
Biased
Unbiased
www.statstutor.ac.uk
Peter
Samuels
Birmingham
City
University
Data richness
q You should always use the richest
(most detailed) data available
because it will give more accurate
results
q Here, the Age data is richer than the
Age Category data
q However, there might be ethical
issues in obtaining detailed data
q Here, the respondents might feel
embarrassed to give their exact age
www.statstutor.ac.uk
Peter
Samuels
Birmingham
City
University
25-29
50
40+
27
25-29
27
25-29
31
30-30
24
18-24
31
30-30
32
30-30
34
30-30
17
18-24
Sample:
Population:
May be too big /
expensive to study
www.statstutor.ac.uk
Estimates
Sample
?
Population of students at
Birmingham City University
www.statstutor.ac.uk
Population mean
(unknown) Parameter
Peter
Samuels
Birmingham
City
University
Random selection
q Most research study designs require a sample to be
randomly selected from a population
q Research1 suggests humans cannot generate random
numbers and thus cannot make random selections
q Suggested methods:
Select numbered balls out of a bag (as in the National
Lottery)
Use an online random number generator, such as
www.random.org/integers
Use the RAND or RANDBETWEEN functions in Excel
Peter
Samuels
Birmingham
City
University
Robustness
q Parameterbased
statistical tests
make certain
assumptions
in their
underlying
models
q However, they
often work
well in other
situations when these assumptions are violated
q This is known as robustness
www.statstutor.ac.uk
Peter
Samuels
Birmingham
City
University
www.statstutor.ac.uk
Peter
Samuels
Birmingham
City
University
Peter
Samuels
Birmingham
City
University
Activity: In-car
control panel design
A new type of car control panel has been developed to control
various functions within a vehicle, e.g. air conditioning, heater,
radio/CD etc.
Two studies were undertaken where subjects used a driving
simulator, so that their mean distraction time could be measured
using eye-tracking technology, whilst driving and using various
control panel functions.
The idea behind the studies was to ask subjects to use the new
design in the driving simulator and then repeat this using a
standard design of control (i.e. one found in a large number of
cars currently on the road).
The aim was to assess the research hypothesis that the new
control reduces distraction times.
www.statstutor.ac.uk
Peter
Samuels
Birmingham
City
University
Study 1
Plot of Individual Data Values from Vehicle Control Design Study 1
Sample
means
1.75
1.50
Data
Ten subjects
used the new
design in a
driving simulator,
whilst ten
different
subjects used the
standard design.
A plot of their
distraction times
is shown on the
right.
1.25
1.00
0.75
0.50
Standard Design
New Design
Discuss with your neighbour whether you believe this supports the
research hypothesis that the new control reduces distraction times.
www.statstutor.ac.uk
Peter
Samuels
Birmingham
City
University
Study 2
As for Study 1
Plot of Individual Data Values from Vehicle Control Design Study 2
but the same ten
subjects used
1.75
each design.
1.50
A plot of their
distraction times
1.25
is again shown
on the right.
1.00
Again, discuss
0.75
with your
neighbour
0.50
whether you
7 out of 10
Standard Design
New Design
believe this
lower with
supports the research hypothesis that the new
new design
control reduces distraction times.
Data
Subject
1
2
3
4
5
6
7
8
9
10
www.statstutor.ac.uk
Peter
Samuels
Birmingham
City
University
Recap
We have considered:
q What is statistics?
q The mean of a data series
q Data types
q The research study process
q The data analysis process
q Some basic statistical concepts
q Benefits of good study design
q Two study designs
www.statstutor.ac.uk
Peter
Samuels
Birmingham
City
University
Bibliography
CensusAtSchool (2014) 2005/2006 CensusAtSchool Questionnaire. [pdf]
Available at:
http://www.censusatschool.org.uk/images/phases/phase6-questionnaire.pdf
[Accessed 6/01/14].
Coolican, H. (2009) Research Methods and Statistics in Psychology, 5th ed.,
London: Hodder and Stoughton.
Easton, V. J. and McColl, J. H. (n. d.) Online statistics glossary, version 1.1.
Available at: http://www.stats.gla.ac.uk/steps/glossary/alphabet.html
[Accessed 6/01/14].
Gonick, L. and Smith, W. (1993) The Cartoon Guide to Statistics, New York:
HarperCollins.
Hayslett, H. T. (1991) Statistics Made Simple, 3rd ed., London: Made Simple
Books.
Phillips, J. L. (1999) How to think about statistics, 6th ed., New York: Henry Holt.
Rowntree, D. (2000) Statistics without Tears: An introduction for nonmathematicians, New ed., London: Penguin.
Stirling, W. D. (2013) Textbooks for Learning Statistics: Public CAST e-books.
Available at: http://cast.massey.ac.nz/collection_public.html [Accessed
6/01/14].
Reviewer:
Ellen
Marshall
Peter
Samuels
www.statstutor.ac.uk
University of Sheeld