How Free Data Can Drive Some of the Monkey Business Out of Political Journalism and Science

Post on 18-Jul-2015

118 views 0 download

Transcript of How Free Data Can Drive Some of the Monkey Business Out of Political Journalism and Science

How Free Data Can Drive Some of the Monkey

Business Out of Political Journalism and Science

1

Recommended Resources

2

Wikipediahttp://en.wikipedia.org/wiki/Main_Page

IFES Election Guidehttp://www.electionguide.org/

European Election Database (EED)http://www.nsd.uib.no/european_election_database/

Electographhttp://www.electograph.com/

Poll of Pollshttp://www.pollofpolls.no/

TV2 Partibarometerethttp://www.tv2.no/politikk/partibarometeret/#/ http://www.slideshare.net/filipvanlaenen/

Why Bother?

3https://xkcd.com/386/

Why Bother?

Actually:

Lots of people are wrong,not only on the internet,but also on television, radio and in the newspapers…

4https://xkcd.com/386/

Disclaimer

5

I'm an engineer, not a statistician

https://www.flickr.com/photos/polob818/3110877065/

Disclaimer

6

I'm an engineer, not a statistician

● Terminology is most certainly wrong

● I may be doing some “monkey business” too…

https://www.flickr.com/photos/polob818/3110877065/

7https://www.flickr.com/photos/azizul/11428653043/

Define “Monkey Business”

8https://www.flickr.com/photos/azizul/11428653043/

Common Pitfalls

● Global margins of error

● Small differences blown out of proportions

● Scores around thresholds

● Seat distributions without margins of error

● Majority in votes or seats?

● Opinion poll outliers

Example

9

Client: Vårt Land

Pollster: Norstat

Fieldwork: 23–29 March 2015

Published: 1 April 2015

Sample size: 961 respondents

Margin of error: N/A

http://www.vl.no/meninger/t%C3%B8ffe-hareide-holder-koken-1.348682

Above or Below the Threshold?

10

Venstre: 3.8 %

⇒ CI(95%) = 2.7 – 5.2 %

⇒ P(V < 4 %) = 61 %

SV: 2.6 %

⇒ CI(95%) = 1.7 – 3.8 %

⇒ P(SV < 4 %) = 98.7 %

MDG: 4.3 %

⇒ CI(95%) = 3.2 – 5.8 %

⇒ P(MDG ≥ 4 %) = 71 %

http://www.vl.no/meninger/t%C3%B8ffe-hareide-holder-koken-1.348682

Who's Larger?

11

MDG: 4.3 %

Venstre: 3.8 %

⇒ P(MDG > V) ≈ 71 %

SV: 2.6 %

⇒ P(MDG > SV) ≈ 97 %

⇒ P(V > SV) ≈ 93 %

http://www.vl.no/meninger/t%C3%B8ffe-hareide-holder-koken-1.348682

Who's Larger – in Seats?

12

MDG: 4.3 %

⇒ CI(95%) = 3.2 – 5.8 %

⇒ CI(95%) = 1 – 10 seats

Venstre: 3.8 %

⇒ CI(95%) = 2.7 – 5.2 %

⇒ CI(95%) = 1 – 9 seats

SV: 2.6 %

⇒ CI(95%) = 1.7 – 3.8 %

⇒ CI(95%) = 0 – 2 seats

http://www.vl.no/meninger/t%C3%B8ffe-hareide-holder-koken-1.348682

Distribution of Seats

13

Ap: 41.7 % 71 – 82 seats⇒

H: 22.3 % 35 – 46 seats⇒

Frp: 10.0 % 15 – 22 seats⇒

KrF: 6.5 % 9 – 15 seats⇒

Sp: 5.9 % 8 – 14 seats⇒

MDG: 4.3 % 1 – 10 seats⇒

V: 3.8 % 1 – 9 seats⇒

SV: 2.6 % 0 – 2 seats⇒

Rødt: 1.8 % 0 – 2 seats⇒

http://www.vl.no/meninger/t%C3%B8ffe-hareide-holder-koken-1.348682

14https://www.flickr.com/photos/jdhancock/4816476075/

Free Data Sources

Upcoming and Past Elections

Wikipedia

http://en.wikipedia.org/wiki/Main_Page

15https://www.flickr.com/photos/adamkr/4592708889/

Upcoming and Past Elections

IFES Election Guide

http://www.electionguide.org/

16https://www.flickr.com/photos/adamkr/4592708889/

17

18

19

20

Upcoming and Past Elections

European Election Database (EED)

http://www.nsd.uib.no/ european_election_database/

21https://www.flickr.com/photos/adamkr/4592708889/

22

23

24

25

26

Opinion Poll Results

27

Wikipedia

http://en.wikipedia.org/wiki/Main_Page

http://en.wikipedia.org/wiki/Opinion_polling_for_the_2015_United_Kingdom_general_election

28

Opinion Poll Results

29

Electograph

http://www.electograph.com/

Norwegian Opinion Poll Results

TV2 Partibarometeret

http://www.tv2.no/politikk/partibarometeret/#/

30

31

32

33

Norwegian Opinion Poll Results

34

Poll of Polls

http://www.pollofpolls.no/

35

36

37

38

39

40

41

42

43

Election Results Forecasts

Electoral Calculus

http://www.electoralcalculus.co.uk/

44

Election Results Forecasts

Elections Etc

http://electionsetc.com/

45

46https://www.flickr.com/photos/azizul/11428653043/

Introducing “SAPoR”

Why SAPoR?

47

● More fun than using R

● Seat distributions

https://github.com/filipvanlaenen/sapor

https://www.flickr.com/photos/azizul/11428653043/

Recommended Resources

48

Wikipediahttp://en.wikipedia.org/wiki/Main_Page

IFES Election Guidehttp://www.electionguide.org/

European Election Database (EED)http://www.nsd.uib.no/european_election_database/

Electographhttp://www.electograph.com/

Poll of Pollshttp://www.pollofpolls.no/

TV2 Partibarometerethttp://www.tv2.no/politikk/partibarometeret/#/ http://www.slideshare.net/filipvanlaenen/