| Username |
Post |
| ashu |
Posted
on 23-Dec-03 11:44 AM
In the 16-29 Mangsir, 2060 issue of Himal Khabar Patrika magazine, Nepali political scientist Dhruba Kumar has published an interesting article (pages 20-21).
Kumar's argument goes like this:
The median age -- the age that divides a population into two numerically equal groups or half the people are younger than this age, and half are older -- currently stands at 20.1 in Nepal. And due to this particular demographic feature, Nepal's tilt toward conflict and violence is natural. That is, the younger the people, the more prone to conflict they are.
To support his argument, Kumar cites SELECTED examples from other countries.
**********
Intrigued by Kumar's argument, I looked at Nepal's median age over the past 5 decades.
1952-54: 21.1
1961: 20.9
1971: 20.3
1981:19.9
1991:18.9
2001:20.1
(Source: Central Bureau of Statistics, Nepal.)
Based on the above, a question to Kumar: If Nepal's present violence/conflict can be explained from the standpoint of Nepal's current median age, then, how come Nepal didn't experience similar violence/confict in, say, '61, '71, '81, when the median ages were not all that far apart and were closely clustered around one another?
*********
Kumar's international data also seem perplexing.
The much-celebrated African country Botswana, for instance, has a LOWER population median age (19.1 in 2002) than Nepal, and it is doing fine without violence and conflict.
So are present-day Cambodia (19.2), Laos (18.5), Mauritania (16.9) and so many others. In other words, when you look at the median age of populatrion of various countries, you find a number of countries that do NOT fall tidily into Kumar's theory of lower median age means tilt toward violence/conflict. And the median age in conflict-ridden Croatia is 39.0, Serbia, 36.2, and Bosnia, 35.5.
Did Kumar self-select his international sample to fit his theory?
More to the point, wwith so many counter-examples, does Kumar even have much of
a thesis here? How much of a tight fit there really is between a country's median age and its being suspectible to internal conflicts?
oohi
ashu
ktm,nepal
|
| Biswo |
Posted
on 23-Dec-03 01:06 PM
Pls kindly supply the link for the article by Dr Dhruba Kumar.
--
Interestingly, more recently, specially in America's top universities, political science is more about rigorous mathematical analysis. In the past, most of the students of political science could graduate without studying courses like Econometrics, Game Theory and Analysis, but seems like those days are over.
Given the complexity of politics, and demand for more rigorous explanation of politics in America and worldwide, it is natural that analysis are couched not in the vague political terminologies, but in numbers which leaves little room for equivocation.
|
| bhunte |
Posted
on 23-Dec-03 01:17 PM
spurious
|
| sadabichar |
Posted
on 23-Dec-03 02:05 PM
Isn't that the basic problems with nepalese? Trying to be HERO within Nepal by hastily publishing article by hook or by crook. This typical nepalese mentality is deeply rooted among most of the people in Nepal, more so in the case of politicians and so called intelligensia.
Dhruba Kumar may like himself to be called political scientists, but does he understand what statistics is about? Or is he just an outlier within the sample group of political sceintists? His attempt to show off as a political scientist is quite similar to Pushkar Shah who keeps on claiming bogus astronimical kms without any proof.
I would like to see D Kumar's piece of work publsihed in a reputable Journal as much as Pushkar's name being entered in Guinness Records.
|
| chipledhunga |
Posted
on 23-Dec-03 06:56 PM
Ashu, good observation.
What confuses me though is the CBS data you presented. There is a steady decline in the median age from the 1950's to the 90's. I thought it should have been the opposite due to the increased life expectancy. Or did the reduction in infant mortality set this trend? Did better health care and awareness help reduce the mortality rate but fail to control the birth rate? Also why is there a sudden increase in the median age in 2001 (close to what it was three decades ago)? Anyone, any thoughts about it?
Btw sorry for the diversion from the original topic :)
|
| ashu |
Posted
on 23-Dec-03 09:36 PM
Biswo,
The Web site of Nepali Himal Khabar Patrika (www.nepalihimal.com) does NOT have Kumar's article as of this morning. I'll alert you as soon as the article become available.
I agree with your observation about the trend of increasing mathematization that has been happening at top political science departments, especially in the US. Gone are the days when good prose carried the day. Now, regressions, tables, charts and so forth dominate political science journals.
Sadabichar,
Let's not be harsh against Kumar, the person behind the research. It's one thing to criticize his research with reason, but let us agree that it's not fair to criticize him as
a person. Maybe next time, Kumar will not neglect to consult his colleagues at TU's Statistics Department, which I hear, is very good..
At any rate, vigorous but non-personalized debates back and forth can only assist
with the growth of a stronger academia in Nepal.
Chiple,
I also noticed that while in the last 50 years, the world median age increased by 2.8 years to 26.4 (in 2000), in Nepal's case, it actually decreased.
A more likely explanation could be that, all things being equal, "better health care and awareness helped reduce the mortality rate but failed to control the birth rate".
Nepal's population growth rate, at 2.4 per year until a few years ago, stands as one of the highest in the world.
******
oohi
ashu
ktm,nepal
|
| Brook |
Posted
on 23-Dec-03 11:56 PM
Necessary but not sufficient, perhaps?
|
| ashu |
Posted
on 25-Dec-03 10:11 AM
At any rate, I wrote up a short letter to editor, and sent it off to Himal Khabar Patrika for publication.
Several friends in Kathmandu. affiliated with various political science groups, have warned me that Mr. Dhruba Kumar is quite powerful in his own way, and is known to
be vindictive against those who dare question him . . . even nursing grudges for
years.
That's fine.
In Nepal (and, to some extent even on Sajha), I am quite used to dealing with
such people anyway . . .and so, my letter PRAISES Kumar's research, while
asking -- quite respectfully in the name of research -- some of the questions
raised above in this thread.
Let's see what happens.
oohi
ashu
ktm,nepal
|
| sadabichar |
Posted
on 25-Dec-03 11:30 AM
Way to go, Ashu.
"Mr. Dhruba Kumar is quite powerful in his own way, and is known to
be vindictive against those who dare question him . . . even nursing grudges for
years."
It's Nepal, that's why you still have that in your back of the head. That's how we are brought up, aren't we? But such thinkings must be squashed by all of us, to lead the country to progress, not only in a materialistic way, but also in honest thinking and respecting each others view without any grudges. We must learn to accept the facts and truth.
I remember a case in Canada, where a top notch Professor Bill Montelpare concluded with his statistical analysis that allowing body-check in Minor Hockey (junior/kids hockey tourneys/practice) doesn't cause any increase in injuries among the kids. His conclusion led the Hockey federation to allow body-check. Later, CBC (Canadian Broadcasting Corporation) news:' Disclosure' found a serious error in the study's calculations. There were, in fact, more than three times the injuries in the group that was allowed to check compared to the one that was not. That conclusion was based on the data provided to Lakehead University by the hockey teams involved in the study. A statistician consulted about the study by Disclosure told the program that, based on the study's data, the injury rate difference between the two groups is, in fact, "highly statistically significant."
When asked to reconsider his earlier findings Montelpare conceded that there is "an error" in his conclusions. He has since issued a supplemental report to the Canadian Hockey Association rescinding his earlier findings.
For two decades, bodychecking was banned in atom hockey. Study after study said there were too many injuries when children are allowed to hit before they are at least 12 years old. But last year the Canadian Hockey Association (CHA) voted to drop that ban. It based its decision on the results of a study commissioned out of Lakehead University in Thunder Bay, Ont.
Result: The sports body that controls amateur hockey in Canada has banned bodychecking for the game's younger players.
The Canadian Hockey Association ruled Monday that checking would be allowed only among players 11 years of age and older.
That's Statistics!!!!
|
| ashu |
Posted
on 01-Jan-04 11:14 AM
Biswo,
Here's the link to Dhruba Kumar's article.
http://www.nepalihimal.com/2060/mangshir16-30-2060/dristikorne_drinda.htm
*******
Sadabichar, enjoyed reading your parable about statistics and intellectual honesty. Thanks for sharing it.
*****
BTW, my letter to editor, making pretty much the points made above, has been published. It's on page 8, Himal Khabar Patrika, Poush 16-30, 2060 or Jan 1-15, 2003. The letter's NOT up on the Web yet.
Thanks everyone for weighing in with your own views.
oohi
"happy new year, everyone!!
ashu
ktm,nepal
|
| madhav |
Posted
on 05-Jan-04 11:32 PM
Ashu, Biswo, Bhunte and Sadbichar ji,
I did not find Prof. Kumarýs statistical argument uncommon among academic community even here in the United States. Today, I discussed with my professors regarding the statistical validity if someone claims possibility of revolution or civil war given the median age as an explanatory variable. Several research works have been done on age factor and its correlation with political change. However, the crucial shortcoming with Prof. Kumarýs claim is that he has not explained why median age could be an independent variable to explain conflict. For instance, the explanation could be that the availability of young people for political mobilization. If young people are convinced that they have least to lose from their participation and immediate incentives and long term benefits are high, i.e. subsistence security, they could be easily mobilized. Think about present situation in Nepal. What will young people lose taking part in insurgency? What they have achieved that barricades them from taking part in such insurgency?
I would like to debate and explore more on this topic.
Madhav
|
| sadabichar |
Posted
on 06-Jan-04 01:30 AM
I agree with you madhavjee and your prof. Just considering mean/median and try to lead public forcefully to one's view or try to be HERO in addressing current problem in Nepal in order to publish an article and fill up ones resume is what we are talking about.
Even I might agree that lower median age of the population might help in civil war, but median age alone cannot. Now, along with median age, how many other factors should be considered to explain the insurgency or civil war in a particular country could be of great interest. But, what's the use simply by understanding how the median age could cause upheaval in political arena? Are people gonna shift this median age to cuase or not to cause upheaval inthier favor?
|
| Biswo |
Posted
on 06-Jan-04 02:21 AM
Ashu,
Obviously, the article has reductionist approach. In fact, any student of statistics or econometrics or political science of present days can easily calculate the the correlation between the median age and violence:it may be as simple as using just the celebrated F-test.I don't have the data on median ages of the countries and violences there, but my hunch is this hypothesis is not easy to reject(that professor Dhruba Kumar may actually be true despite the examples in the contrary).However, looking at the professor's article does give the impression that the professor selectively culled out some countries and presented his case.
I think this is the pitfall of writing in a popular magazine. You can't present the methodology and calculation there. You have to write in a simple words for masses. But you will then leave plenty of your peers unsatisfied. So, I think it will be wise for the professor to provide some kind of reference to the work based on which his results are presented.I hope he will provide a curious reader like you with the link to his work now that you have sent a letter raising the doubt.
My intuition is: lower median age means higher number of children, and probably an increasing number of work force in the labor market. If the economy is not healthy enough to create proportionate number of jobs, some of these children are likely to be disgruntled rebels.Taking Nepal as an example: with the population growth rate more than 2%, and depending upon which report you believe, number of workers seeking job each year being from 2,50,000 to 3,00,000,and with the economy unable to produce that much job, we are forcing more and more young people to the lands that their fathers worked on.Thus, we are having more people living off the same land. It likely means our peasants are getting poorer and more numerous. This may also explain why relatively rich but agricultural Rukum/Rolpa, not absolutely poor Khotang/Humla etc are epicenters of the rebellion. Again, this is just my intuition. Causes of revolutions/rebellions are not easy to explain, anyway.
Madhav, I hope you will find my reply above helpful to answer your question.
|
| Sadabichar |
Posted
on 07-Jan-04 04:31 PM
Let's consider, for an example, 2 factors - Median Age [MA] and Per Capita Income [PCI]).
Now set the range between -1 and +1 as the lowest and the highest of the factor, respectively. Which means, the lowest median age will get scoring of -1, and highest median age gets +1. Between ý1 and +1, rest of the median ages of the countries under the study will fall. Same coding for the Per Capita Income [PCI] follows.
So the statistician would propose his/her model as following, to explain if [MA] and [PCI] cause the [VUL]:
[VUL] = a. [MA] + b. [PCI] + c. [MA]x[PCI]+.............. (a, b, c being coefficients)
which means, the vulnerability [VUL] of any country to experience civil war/upheaval/conflict/violence may be dependent on [MA], [PCI]. There is a term [MA]x[PCI], which is the interaction effect of the two factors on [VUL]. Well, you might wanna add other terms, such as quadratic effect of these factors [MA]^2 and [PCI]^2. And you might even consider other factors in the equation, such as how much religious people of that country are, whether the country is in the northern or southern hemisphere, climates, and what not. But after proposing such a model with numerous factors, statisticians will run Multiple Regression analysis to determine the coefficients (a, b, c, etc) and ANOVA to determine if the factors (such as ý [MA], [PCI], [MAxPCI]) are statistically significant or not by looking at the associated p-values. If the factors are statistically insignificant, then such factors can be dropped out from the model to come up with the more refined version of model and then run the Multiple Regression again.
Simply considering a single factor such as Median Age [MA] in the study to come up with any sort of conclusion (as done by Dhruba Kumar) could be disastrous. In the model above, if we analyze the data, perhaps [MA] or [PCI] alone might not cause [VUL] to be high or low. Rather the interaction effect between these two factors, Median Age and Per Capita Income, might exert considerable effect on determining the value of [VUL].
Letýs consider the interaction effect portion of the model.
[VUL] = c. [MA]x[PCI]
[MA] [PCI] [MA]x[PCI]=[VUL]
-1 -1 +1 (Example: Country Nepal, +1 means higher probability of violence/conflicts)
+1 -1 -1 (Example: Country with high MA and low PCI, lower probability of violence)
-1 +1 -1 (Example Rich Countries with lower median age, -1 means low probability of VUL)
+1 +1 +1 (Example: Rich countries with higher MA, higher probability of violence????)
In the four scenarios presented above, first three seems to be obvious. Country like Nepal, with lower MA and lower PCI, there is more chance of violence/conflicts (first scenario 1). And in Scenario 2, although the country is poor with low PCI, it may not experience violence/conflicts as the majority of the population is mature (high median age). In third Scenario, the rich countries with high PCI but low median age may not be prone to violence and conflicts. But the fourth scenario, rich country with high MA might experience conflicts is little bit hard to be true. There must be other parameters (factors) that might explain this.
Well, the point here is, the model could be complicated. So care must be practiced before concluding anything. Statistics have been well appreciated in the manufacturing industry, where one can experiment very well with the controlled factors to improve the quality of the products. We have yet to see the utility of statistics in political sciences.
|
| Biswo |
Posted
on 08-Jan-04 06:31 PM
Sadabicharji,
Interesting perspective.
However, it is not true that we have yet to see the utility of statistics in political science. Now, I am not a political scientist, and what I am writing here is borne out of my belief in mathematics and my interaction with some political scientists here in my school.
Let's put it this way: what is statistics? Just a way to represent data. Statisticians try to put in words what the data thrown in front of us are trying to tell. In politics, in economics, in sciences,in law, there are data everywhere, and statistics merely try to suggest what those data are trying to say. Because of its unequivocal nature, we embrace statistics and mathematics in our study.Papers in demographic voting patterns use statistical analysis, that was something I remember running into sometimes ago.I've also heard political scientists -who used statistics to explain their survey- were called in to testify in Florida election fiasco in 2000 and in recent gerimandering suit on Texas redistricting.
|
| Bhunte |
Posted
on 08-Jan-04 07:09 PM
The use of statistics in political analysis should be kept minimal. Otherwise the results will be merely a statistical artifact. I would rather rely more on theory, logical arguments, history, psychology, law, belief and culture, name of game, etc when it is a political analysis. There are other areas where statistics has a crucial decision in policy matters or so...
|
| Xanthipee |
Posted
on 08-Jan-04 07:47 PM
Purpose of stat is to lie. Guys, read:
"How to Lie With Statistics" by Darrell Huff, Irving Geis
|
| Biswo |
Posted
on 08-Jan-04 08:08 PM
I knew a famous professor in his field who casually told me, "What is econometrics? Garbage in, garbage out." In deed, there is a vocal minority which is disgusted by statistics. But not me:-)
Think about this. If English were to be enough, no one would ever have invented mathematics. We resort to new language only when new language facilitates us in expressing what we want. [This is one reason why we insert so many English words while conversing with each other in Nepali. We don't say Radio to show off, we say Radio because it is more easy for us to say Radio than Akashbaani.]The reason why Keynes'"General Theory of Employment, Interest and Money" is so difficult to decipher is because he used English in stead of plain mathematics, according to the economists. The economists of later era felt economics could not be explained by english language alone, and see, where is Economics now in terms of using math.
Ditto about Political science. Let's think about Voting Rules. How can one say these voting rules work? How can one say some voting rules are strategyproof? My limited exposure to May's Theorem, Condorcet, Borda, Simpson, Moulin etc convinced me that these rules are best tested by mathematics, and their efficacy could only be explained by the time honored techiques of mathematics. Why? Simply, because mathematics is least susceptible to equivocation.
|
| Xanthipee |
Posted
on 08-Jan-04 08:44 PM
You can prove what you want to prove by data snooping, changing your models, not explicitly stating your assumptions. Stat is not pure math. It is polluted by bias and subjectivity. However, if you ask me whether it is interesting, I would say- absolutely! Who says Guinness beer is not rich?
|
| ashu |
Posted
on 16-Jan-04 02:13 AM
Sadabichar,
Your model-building work reminded me of those stat classes from the days of yore.
I agree with Biswo math is least susceptible to equivocation.
BTW, for those who are interested in this issue, here's my letter:
http://www.nepalihimal.com/2060/magh1-15-2060/pratikriya.htm
***********
Here's Dhruba Kumar's original article.
http://www.nepalihimal.com/2060/mangshir16-30-2060/dristikorne_drinda.htm
oohi
ashu
ktm,nepal
|
| Biswo |
Posted
on 17-Jan-04 03:41 PM
Ashu,
I still have problem reading your reaction. It seems a lot of words are omitted. And you seem uncharacteristically polite there:-)
|
| ashu |
Posted
on 17-Jan-04 07:24 PM
Biswo,
I have emailed a request to the Himal Media's Web master to make the letter come out clearly, in full. Let's see.
As for seeming to be "uncharacteristically polite there", well, I am always learning and
re-learning the fine art of being polite yet firm in Nepal.
I mean, if you are too polite here, people will jerk you around, thinking that you be taken advantage of easly.
If you are too firm, then people will avoid you, thinking that you are a jerk.
And so, learning to maintain a fine, if tricky, balance between being jerked around
and being a jerk seems to be a winning emotional skill to still be able to eat lunch with different people in Nepal. :-)
oohi
ashu
ktm,nepal
|
