International data: developing QM social science capacity John MacInnes 1
Dec 30, 2015
International data: developing QM social science capacity
John MacInnes
1
Training/teaching QM: some challenges
•Low confidence in maths or statistics ability
•Low motivation: doubts about worth of QM
•Low expectation of achievement or experience
•Low reinforcement elsewhere in curriculum
•Little curriculum space
•Real, relevant data are most convincing, but rarely yield simple, clear patterns
2
Training/teaching QM: some resources
•More, better, easier to access data
•Better GUIs, range of software and IT infrastructure
•Better visualisation resources e.g gapminder
3
Training/teaching QM: special relevance of international data
•All social sciences consider ‘globalisation’. Study of host society in isolation increasingly seen as parochial
•Cosmopolitan student bodye.g. of Edinburgh CQDA course majority non-UK based students
•Comparison is core of social science and QM•Country level data is typically at interval level•It addresses engaging cross-disciplinary issues•It is suitable for both transversal and time series approaches
4
The CQDA ‘blended learning’ course
5
Using World Bank and UNHDI data
6
New challengesOld model: pay for a data set and analyse with SPSS, SAS etc
New model: data transparency / ‘open data’New skills in data location, manipulation and retrieval which complicate core task of learning e.g. OLS regression analysis
Temporary solution‘Teaching’ datasets
The WDI/HDI dataset
Data from latest available year to minimise missing cases
Only countries with > 3m pop
100 variables: manageable for new learners
Online access to meta data, but sufficient var label description to facilitate simple analyses
Deliberate inclusion of non-interval variables
7
The WDIHDI teaching dataset
8
The CQDA ‘blended learning’ course
9
The strong association between GDP and fertility
The CQDA ‘blended learning’ course
10
The spurious correlation betweenMobile phone subscriptions and Infant mortality
Data checking procedures
3000 tractors per 100 sq. km
= 30 tractors per sq km
= 1 tractor per 3 hectares?
11
ConclusionsAdvantages:Very useful teaching tool
Combines relevance with clarity, but also complexity for more advanced learners
Drawbacks
Resource intensive to produce
Less flexible that original data sources
What facilitates QM T&L may not teach students data complexity management skills
12