Hans Rosling's Gapminder shows world's health trends over 200 years

It’s early days but it is encouraging to see ABC use the crowd sourcing platform Ushahidi to map the Queensland floods from the perspective of its audience. Ushahidi means testimony in Kiswahili and works best when there are lots of people witnessing the same large event. It was developed to allow people to map incidents when ethnic violence erupted in Kenya in late 2007 and proved influential in exposing fraud in the 2009 Namibian election.

It is great to see innovative tools used here and it reminds me of my favourite thing on the Internet right now. It is a four minute video by Swedish doctor and professor of statistics Hans Rosling produced by the BBC. Rosling has also developed remarkable statistics software called Gapminder which has a dazzlingly brilliant way of interpreting statistics in a way that is informative and compelling.

In this BBC video he shovels 120,000 sets of numbers through his program from world census surveys for two hundred years. He plots the data by countries of the world since 1810 on a graph where the x-axis is income per person and the y-axis is life expectancy in years. Near coordinates 0,0 are the sick and poor, and near n,n are the very healthy and wealthy. In fast forward, we can see 200 years of trends flashing in front of our eyes as two centuries of data is plotted on the graph.

In 1810 all the countries of the world are clustered in the lowest quadrant. The UK and the Netherlands were clearly better than every other country on both indicators, though they were still low with life expectancy around 40 years and average per capita income less than $3,000. By 1860 the Nordic countries Norway, Sweden and Denmark were leading the way with remarkable improvements in life expectancy by up to ten years. The UK was still the wealthiest in the world as it was about to enter Pax Britannica and its new colonies in Australia and New Zealand weren’t far behind though life expectancy was low. The US was also catching up fast.

Fast forward another 50 years and Scandinavia was still the healthiest part of the world with average life expectancy pushing 60 years. New Zealand and Australia were finally seeing the benefits of their remarkable riches (second and third wealthiest in the world behind the US) to push life expectancy above 50. With the exception of colonial countries Canada and Argentina, the European countries were next highest on both indicators, though Japan was rising quickly. At the bottom, average life expectancy was just 22 years in the area now called Bangladesh and 23 in India.

By 1960, the discrepancy between rich and poor were quite pronounced. Most of Europe, North America, the colonial countries and Japan were achieving life expectancy of up to 70 years. The US and Switzerland were pulling away with average incomes up to $20,000. Small oil-rich states such as Brunei and Qatar were averaging over $40,000 though life expectancy was lower. China had slumped to the bottom as it suffered through the famine trauma of the Great Leap Forward. Yet the Chinese were still living ten years longer than they did in 1910. African countries were the poorest but surprisingly healthy with Lesotho people living to 47 years on just $365 (literally a dollar a day).

In 1985 Brunei and Qatar were still the wealthiest countries in the world and their citizens were living longer too. The Japanese were living an average 78 years making them the healthiest in the world. All the First World countries were clustered close behind. The developing nations were catching up quickly. Countries (or soon to be countries) such as Mexico, Latvia, Ukraine, Albania and even North Korea were averaging over 70 year lifespans. The five biggest Asian nations (China, India, Pakistan, Indonesia and Bangladesh) were still poor but beginning to make a charge. Post-colonial Africa was bringing up the rear. Yet even in the poorest country, Mozambique on just $366, the average lifespan was three years higher than Britain in 1810.

In 2009, Japan is still the long-living nation in the world, now averaging 83 years. Its ageing population is presenting new challenges for economists as well as demographers. But most of the West is now averaging 80 years of lifespan with an average wage clustered around $30,000. Qatar remains the richest country in the world (a clue to why a country of 1.1 million has won the right to stage the World Cup in 2022). Most of Middle East, North Africa, East Asia and the Pacific were also living longer than 70 years. There were improvements at the bottom end too. War-torn Afghanistan was the least healthiest country in the world, living just 44 years even there both indicators have remarkably been improving since 1994. The equally war-torn DRC (Congo) is now the poorest country with an average salary of just $359 (less than the poorest in 1960) but the Congolese still live to 48 years.

Rosling notes that within countries there are massive discrepancies. If Shanghai was removed from China it would be in the top 1 percentile. Similarly Australians live for 82 years and enjoy an average wage of $34,327 but if the Indigenous population was measured alone, it would be much worse. Yet despite the wider discrepancies across the world that now exist, Rosling’s imaginative graph makes a powerful point: the trend is worldwide for higher earnings and longer lifespans. Projecting out to 2060, it is clear all countries are pushing towards the mythical “n,n”. Feeding all these long-living people in a world of catastrophic climate change is beyond the powers of this data, but it is certainly engrossing food for thought.

Some stats on Australian media's use of Twitter

With the Media140 conference coming up in Sydney next month, I thought it was time for a bit of local research on Twitter usage. I used Dave Earley’s comprehensive list of Australian journalists on twitter as a starting point for some research on local media usage. I spent a painstaking few hours getting the details of 313 people from Earley's list (plus a few others I though should be there). My list is not as exhaustive and I ran out of time to include the photographers, radio and some miscellaneous names.

But for the ones I did look at I simply captured the number of people they follow, the number of people that follow them and how many times they had tweeted (to Sunday 11 Oct 2009). I then sorted the results into various criteria.

Firstly, the data is sorted by how many people they follow. The higher numbers here are the people who are interested in everybody.

Frombecca leads the way following 24857 people on Twitter. After her comes Sarahprout who follows 22296 people. There is considerable distance to third place:

3. cameronreilly 10015
4. mumbrella 5749
5. mediahunter 4736
6. katekendall 4433
7. ccoaster2002 4351
8. earleyedition 3323
9. Joe_Hildebrand 3111
10. debsm 3105
11. landeryou 3060
12. firstdogonmoon 2865
13. tabloid terror 2678
14. JohnBirmingham 2576
15. grahamy 2556
16. renailemay 2429
17. PaulWiggins 2000
18. barrysaunders 1977
19. Warlach 1925
20. LeonGettler 1850

Looking at the data in a different way is a Top 20 judged by how many followers they have. These are the influencers, the people that everyone is reading. Once again the same top two is well out in front - Frombecca at 22864, about a hundred ahead of Sarahprout on 22754. Again there is a large distance to number 3. The rest of the top 20 is considerably different than the previous list (though there are overlaps). It has several well known people who tend not to follow many people themselves.

3. triplejdools 11540
4. MyfWarhurst 11477
5. doctorkarl 10968
6. miafreedman 9932
7. cameronreilly 9650
8. MariekeHardy 7358
9. mumbrella 6780
10. duncanriley 5958
11. adambspencer 5418
12. katekendall 4860
13. tabloid terror 4815
14. AnnabelCrabb 4746
15. mediahunter 4604
16. triplejtvdoctor 4058
17. LeighSales 3995
18. firstdogonmoon 3914
19. ccoaster2002 3906
20. Joe_Hildebrand 3592

Next is a look at who is doing all the Tweeting. This time a very different Top 20 emerges with Stilgherrian well out in front heading towards 30,000 tweets. Just 3 of the people in previous lists make this one. They line up as follows:

1. stilgherrian 29382
2. johnlacey 25347
3. cameronreilly 22669
4. Warlach 21145
5. djackmanson 15049
6. bronwen 14180
7. kateedwards 14072
8. frombecca 12865
9. bengrubb 12629
10. mediahunter 12470
11. duncanriley 12374
12. ssharwood 11599
13. barrysaunders 11270
14. tomatom 10837
15. firstdogonmoon 10367
16. renailemay 10133
17. jason_a_w 9593
18. earleyedition 8775
19. tabloid terror 8289
20. wolfie_rankin 7814

Next, a look at the ratio of following to followers.

Here are the top 20 of those who follow a lot more people than follow them. Poppy Masselos leads the way following 2.193 times as many people as follow her. It could be argued that these people are the media listeners of Australian Twitter (though there are any number of other explanations). This list has very little overlap with any of the previous lists.

1. PoppyMasselos 2.193906
2. sam_ikin 2.149254
3. ebaral 1.946903
4. jodieobrien 1.918605
5. loopymedia 1.752525
6. nadine_lee 1.672065
7. UrsulaHeger 1.492877
8. PaulWiggins 1.456664
9. Ericdot 1.444444
10. tanwatt 1.442623
11. princessfluffy 1.422535
12. bjj4me 1.406926
13. claireorourke 1.402778
14. andrewjcrook 1.382653
15. SarahElks 1.378125
16. barrysaunders 1.329523
17. LeonGettler 1.3019
18. MWikramanayake 1.293839
19. rossprowd 1.280822

At the other end of the scale are the people whose followers vastly outnumber the amount of people they follow. These are the people who presumably are there to be heard but not necessarily to listen (but again many other interpretations are possible). Leading the way is the ABC’s Dr Karl who follows just three people but is followed by 10, 968 for a ratio of 0.000274. The top 20 is:

1. doctorkarl 0.000274
2. adambspencer 0.006645
3. MyfWarhurst 0.006709
4. cmrosemary 0.009901
5. triplejtheking 0.010442
6. MariekeHardy 0.012639
7. miafreedman 0.021043
8. AClennell 0.023256
9. paulajoye 0.023335
10. LeighSales 0.023529
11. therealrocky123 0.027778
12. newslab 0.027778
13. triplejtvdoctor 0.028093
14. Jeanti 0.040359
15. liamphillips 0.045455
16. jessmcguire 0.04797
17. triplejdools 0.05156
18. bernardkeane 0.051931
19. brandstand 0.05852
20. JoannaSavill 0.059499


Next, a look at the ratio of followers to tweets. A high number here means that the user is not Tweeting much despite having a large audience. I’m not sure if this metric is in any way important though it is interesting that 7 out of the top 20 are from the ABC.

The ABC’s Misha Ketchell leads the way with a following of 168 despite having tweeted just once. The data in this category needs to be divided into two segments. Firstly there are those like Ketchell with a smaller amount of followers who essentially seem to use Twitter, if at all, to lurk. The second category is those who have a very large following but their Twitter posting is low in comparison. Adam Spencer is an example here with 5,418 followers but just 53 posts giving him roughly 102 followers per tweet.

The full top 20 of high followers to tweets (with category number in brackets) is:
1. mishaketch 168 (1)
2. adambspencer 102.2264 (2)
3. liamphillips 66 (1)
4. Djfcm 52.2 (1)
5. triplejtheking 28.5614 (2)
6. Heidi75 20.33333 (1)
7. Jeanti 20.27273 (1)
8. MyfWarhurst 16.87794 (2)
9. marshallheald 16.66667 (1)
10. tburton 13.44444 (1)
11. zarabee 12 (1)
12. Joe_Hildebrand 11.40317 (2)
13. ScottPape 10.89051 (2)
14. paulajoye 10.10266 (2)
15. valens quinn 9.6 (1)
16. GeorgiaWaters 9.380952 (1)
17. abcmarkscott 8.715909 (2)
18. heraldsunscoop 8.062201 (2)
19. triplejtvdoctor 7.84913 (2)
20. triplejdools 7.734584 (2)

At the other end of the scale are those who are generous with their tweets compared to the number of followers they attract.

John Lacey leads the way with his 1,451 followers enjoying over 25,000 tweets for a ratio of 0.057 followers per tweet. The top 20 is as follows:

1. johnlacey 0.057245
2. Ericdot 0.068381
3. wolfie_rankin 0.069235
4. s_bridges 0.081579
5. ssharwood 0.092508
6. djackmanson 0.093428
7. dellvink 0.098243
8. kateedwards 0.099204
9. jason_a_w 0.101115
10. stilgherrian 0.111735
11. ChrisRhyss 0.11488
12. mcjp 0.124486
13. cate3221 0.12573
14. sonya_gee 0.128358
15. alexkidman 0.129713
16. barrysaunders 0.131943
17. foraggio 0.133285
18. Warlach 0.138
19. bengrubb 0.141104
20. JenBennett 0.150236

There were no real surprising conclusions from the data apart from the two ladies Frombecca and Sarahprout that dominate the following/followers count. I don't have a ready explanation why they are so far in front in both categories.

The next step I'd like to do is measure influence by Retweets over a week to see whether there are any insights from that data.

Any thoughts?