Welcome to NBlog, the NoticeBored blog

I may meander but I'm 'exploring', not lost

May 4, 2018

NBlog May 4 - fraud and corruption


I was genuinely surprised to find New Zealand topping the 'corruption perceptions index 2017' from Transparency International. I thought we'd be in the top quartile maybe but didn't expect to lead the field.


New Zealand's 89% score leaves room for improvement but is way above the "average" (the mean score, presumably - or do they mean the median or some other statistic?) of 43%.

The index rates public sector corruption, specifically. According to Transparency International's video promoting the latest findings, high scores are associated with the ability for journalists and activists to speak up about corrupt officials. 

Ah, OK then, so this isn't really about bribery and corruption in general but more specifically about journalism and activism, and repression by the authorities. 

I'm not entirely sure I understand the scale. It is described as a 'scale of 0-100 where 0 equals the highest level of perceived corruption and 100 equals the lowest level of perceived corruption'.  Errr, I'm confused already since the top result in 2017 is patently 89%, not 100%. It gets worse when they say: 
"standardisation is done by subtracting the mean of each source in the baseline year from each country score and then dividing by the standard deviation of that source in the baseline year. This subtraction and division using the baseline year parameters ensures that the CPI scores are comparable year on year since 2012. After this procedure, the standardised scores are transformed to the CPI scale by multiplying with the value of the CPI standard deviation in 2012 (20) and adding the mean of CPI in 2012 (45), so that the data set fits the CPI’s 0-100 scale."
That resembles "think of a number, double it, and take away the number you first thought of" doublespeak, but perhaps it's just the fog in my mathematically challenged brain at 6:30pm on a Friday after a long week, plus 2 rum-n-tonics. The more detailed Technical Methodology Note that ends with "assuming a normal distribution" doesn't clarify so much as reinforce the impression than these are not lies, nor damn lies, but statistics.  

Worse still, it seems the index is not based on raw data obtained each year by a global research team in a scientific and statistically valid manner, but an analysis of data from other studies. Aside from the lack of clear references to those sources, I'm out of energy at this point to 

So, I am left thinking that the people behind Transparency International mean well, but I have some reservations about their methods and motivations.