Making Sense of Statistics in Court

I talked about making sense of statistical data in general a while back.  Recently, the latest issue of Nature ( vol.  445, no. 7125, p. 254-255 [electronic version requires subscription]) features a story of how statistics has been used and interpreted in court house.  I guess the author, Mark Buchanan, intended to remind to beware of the use of statistics outside scientific arena.  In social setting, nothing can be controlled as in an experiment room.  Context is crucially needed.  He introduced two particular cases.  The first one is suspected nurse who may have killed her patients.  The second one is a mother who has been suspected of killing her two daughters.  Both cases involved the probability of the chances that suspects could have committed murders.

Two points mentioned in the paper are standing out.  For the nurse’s case, the statistician told that “the chance that her presence was mere coincidence was only 1 to 342 million”.  Well that seems very impossible, right?  But the mathematician said by brining additional independent variable – include more factors – the chance could be as small as 1 in 48 or 1 in 5.  My question is which number seems possible for most of us then.  At what level of chance that you can say that could not be coincidence anymore?

Another point is from the quote below.

[L]awyers have an incentive, and even a duty, to select the evidence that makes their case stronger.  “What the judge ends up hearing often comes from the two extreme ends of the distribution,” he (David Kaye) said.

I think the use of statistics is very outstanding example of how people make sense of “number” is very influential and does affect the way we live.

Note:

Inside this issue of Nature (p.249), there is a article about Apple Inc. has been sued by its shareholder about backdating stock options.  They already confessed and Jobs knows about it.

Freedom of Press in Thailand: Getting Worse?

After reading an opinion piece in today Daily Tarheel about press freedom, I wonder where is Thailand in the press freedom ranking (Too bad, the DTH does not include Thailand in the table). One could easily imagine that the ranking should be worse since the coup.

In the latest ranking done by Reporters Without Borders, French-based association for press freedom, Thailand is ranked 122th place with 33,50 index score. However, the ranking do NOT include the situation after the coup since the data were collected based on the events between September 1, 2005 and September 1, 2006. The method covers 50 criteria including “every kind of violation directly affecting journalists (such as murders, imprisonment, physical attacks and threats) and news media (censorship, confiscation of issues, searches and harassment).”

Well, does the 122th place tell something then?

The number 122 could not tell something without its context, right? Here are some contexts I can grab.

168 countries are ranked in this report. For Thailand’s neighboring countries, Indonesia is ranked 103th, Cambodia 108th, Malaysia 92nd, Philippines 142nd, Vietnam 155th, and Burma (not surprisingly) 164th.

The data is based on the time when the political turmoils were very intense, especially between ousted Prime Minister Thaksin and Sonthi, an opponent leader who have media on hand.

Thaksin has been “accused of using his political and economic power to silence dissenting voices and curbing freedom of speech based on the fact that he has direct authority over the state-owned TV stations while his family controls the other broadcast TV channels.” (Wikipedia on Censorship in Thailand, Nov. 2, 2006)

There were a lot of legal cases between Thaksin government and media, especially those who explicitly were Thaksin’s opponents.

“On 11 October 2005, Thaksin sued Manager newspaper for THB 500 million. (news source) As monks have traditionally been above criticism, Thaksin did not sue Luang Ta Maha Bua. “This is an exercise of an individual’s right to protect his reputation and privacy. The newspaper did not criticise the prime minister fairly as a public official, but rather it took him to task personally, using harsh words, which was damaging to him,” Thana Benjathikul, Thaksin’s lawyer said.” (source: Wikipedia on Thailand political crisis 2005-2006, Nov. 2, 2006)

In terms of longitudinal context, 2006 is the worst year for the press in Thailand.

• 2002 – 65th with score 22.75 (tied with Madagascar)
• 2003 – 82th with score 19.67 (following Thai neighbor, Cambodia placed on 81th rank.)
• 2004 – 59th with score 14,00
• 2005 – 107th with score 28,00
• 2006 – 122th with score 33,50

It seems like 2004 is the best year for Thai press. Comparing only the ranking sounds not quite right. For example, in 2003 the ranking of Thai press was worse than the one in 2002. However, the less score mean the more freedom of the press which mean the freedom score in 2003 is better than the one in 2002. Thus, it seems likely that in 2003 it was a better year for press in overall. However, I am not quite sure about the reliability of these scores over years. Therefore, this is just an observation, not the empirical evidence until I got confirmed by the researcher.

Please note that the data source are Reporters Without Borders’ partner organizations “and its 130 correspondents around the world, as well as to journalists, researchers, jurists and human rights activists“. I do not know the data source for Thailand which would definitely affect the reliability of the index. Anyway, it seems likely that the political turmoil made some media explicitly in the news rather reporting the news.

2006 Questionaire

Note: The list of freedom predators and its black list are also interesting.

Well, I guess we have to wait for the next year then to see the “coup” will make “democracy” or “democrazy”.

Social Networking on Daily Show

Hilarious.. It’s originally uploaded on youtube on May 07, so some of you may have already watched it.

[via Masters of Media]

What does the Internet society look like in 2020?

The latest Pew study report (collaborating with Elon University), entitled The Future of the Internet II, shows the vision of “internet leaders, activists, builders and commentators” on “the future social, political, and economic impact of the internet”. Seven scenarios have been discussed including:

1. the deployment of global network
2. human control over technology
3. transparency vs. privacy
4. luddites, technological “refuseniks,” and violence
5. compelling or “addictive” virtual worlds
6. the fate of language online
7. investment priorities

The full report contains tons of interesting quotes, although at some points I found it’s too overwhelming. Hence, I would think citing this piece should be more cautious, as the introduction already state that the samples are not absolutely representative.

I have to suspect that the visionary does not necessarily need democracy or to be democratized, although innovation will probably do. (see also Hippel’s Democratizing Innovation – I have to admit that I have read only a couple of chapters)

Anyway, it is also a good resource for some folks to learns about internet leaders.

Overall, I am not quite surprised with the results. It seems like there is no a totally undiscovered idea appeared. The results are pretty much based on current tendency. Therefore, the scenarios is quite predictable with just stronger degree of intensiveness. And I think it is reasonable, why?

Evolution or revolution?

In 2020, although the introduction of new innovation may happen as often as weekly or even daily, the Internet (or maybe more specific to WWW), as we have been used for many years, will still play a significant role in the society. We still look forward to the new societal scheme. However, the existence of today Internet society will be merge in tomorrow world. Online society is accumulative where we will never want to delete anything. We will no longer have archives, which we will have to store and retrieve them separately. Instead, our memories will follow us in everyday life in everywhere. We will live with all kinds of tenses within a single moment. Therefore, the future of online society will still be based on evolution not revolution.

Quantifiable responses?

To prove that citing this piece needs some additional attention, here is my example.

Scenario Four: Transparency vs. privacy

Prediction: As sensing, storage and communication technologies get cheaper and better, individuals’ public and private lives will become increasingly ‘transparent’ globally. Everything will be more visible to everyone, with good and bad results. Looking at the big picture – at all of the lives affected on the planet in every way possible – this will make the world a better place by the year 2020. The benefits will outweigh the costs. (p. 30)

49% says disagree. However, I am quite confused about the instrument on how the question is structured. There seems to be three major arguments inside the scenario given.

• more transparency
• transparency makes better place
• outweighing the costs

The first and the last makes more sense to me. But I am not sure if the high transparent atmosphere would make a better place. The word “better” (or good), as we all know, is absolutely subjective and too broad. Therefore, the percentage of response can not tell anything specifically because you can not tell which statement is valid to result. So only valid responses, to me, are those qualitative responses.

Anyway, it is very sparking in overall…

Sense Making of Statistical Information

It is interesting to see how statisticians and researchers make sense of the data they collected and analyzed. From today’s statistics class, I made two major points in my notebook.

Firstly, a part of the discussion is about whether Likert scale should be treated as ordinal or interval/ratio scale. Likert scale is intended to measure people attitude ranging from negative to positive attitudes. The degree of agreement is given numeric value for scale. It can be 5, 7, or 10 point scale. Most of the papers I have gone through so far treat those scale as interval/ratio. The obvious practice is they seek the central tendency by mean (average). Then some of them imply the numeric value back to the meaning. For instance, in 5 point likert scale (from 1-most unsatisfied to 5-most satisfied), the average of the response is 2.5326. I am not sure you can imply that it should fall in to just “satisfied” or “neutral”.

For me, it is hard to say because the distance between degrees is hard to be justified and given the meaning, although there is a choice to add granularity of the scale to 10. I still think there is limitation in that scale.

Also, I think it might be interesting to see the reliability testing of the scale. Although the explanation is given to each degree in the instrument, it is hard to say that my “4 rate” is equal to other people’s “4 rate”. Anyway, the scale range might help a lot in this case.

Another interesting point to me, as it is still the unanswered question to me, is about how people make sense of decimal digits. How people decide how many digits they want? When they will and will not use it, whether intentionally or unintentionally? What criterias (variables) involved to define the context of use? For example, you go to the supermarket and you found the number “$54.99″ on the price tag. The decimal digits would help you to think that it is nearly $55. However, it might be worthless to specify the decimal place when someone asked you the price of the product you bought for a fun talk. I wonder too that, in relation to the fractions of the amount of money, people who strick their financial activities would be more serious with decimal than others. Anyway, that is just my assumption. We might be able to use basic communication model to capture the elements roughly which may include background, time, location and such.