Leading newspapers in Bangladesh on July 10, 2013 sensationally headlined the survey findings from Transparency International (TI)'s Global Corruption Barometer 2013. Approximately 1,000 people from each of 107 countries were surveyed between September 2012 and March 2013. In Bangladesh, 1822 people participated in the survey conducted from February 10 - March 15, 2013. Of the total sample, 61 % were from rural and 39 % from urban areas. Based on Transparency International Bangladesh (TIB) presentation of the TI report to the media on July 9, the media coverage gave the clear impression that most of the important institutional pillars of Bangladesh are perceived to be extremely corrupt. Corruption undoubtedly is a major problem here. However, the way TIB constructed the survey results led to predictably excessive perception bias in favor of corruption.
Let me illustrate using TIB construction of the survey response. One key question in the survey was the extent to which respondents saw institutions such as the political parties, police, judiciary and so on affected by corruption. They were asked to answer on a scale from 1 to 5, where 1 mean not at all corrupt and 5 means extremely corrupt. The TI report presented the results in the form of percentages who thought they are corrupt or extremely corrupt counting only those who rated the institutions as 4 or 5. TIB, however, chose to deviate from this method and counted not just 4 and 5 but also 2 and 3 leaving out only the "not at all corrupt" response.
As may be expected, the results are dramatically different. According to TIB, 93% said political parties are corrupt whereas according to TI 45% said political parties are corrupt; 93% said police are corrupt according to TIB while 64% according to TI; 60% said media is corrupt according to TIB while only 9% according to TI. I could go on, but you get the point. While TIB did present the TI numbers on the same question in a separate slide describing them as the percentage of respondents who thought the extent of corruption is of "serious concern" (ratings of 4 and 5), none of the media paid any attention to this slide in their coverage. Surprise, surprise!
You may be familiar with Framing Effect, a cognitive bias , in which people react differently to a particular question depending on whether it is presented as a positive or as a negative. People tend to avoid risk when a positive frame is presented but seek risks when a negative frame is presented. The concept helps to develop an understanding of the formation of political opinion  where spin  plays a large role.
Imagine what the response would have been had the same respondents been asked to answer on a scale from 1 to 5, where 1 mean totally honest and 5 means extremely dishonest. Let’s assume for the sake of argument that the distribution of responses is the same in both the frames. In this case 2 would be interpreted as honest rather than corrupt because it is closer to 1 relative to 4 or 5. A rating of 3 would then be seen as those who think the glass is exactly half full or half empty. How then would TIB have treated 2 and 3? Will they have lumped together somewhat honest (2 rating) and mildly honest/dishonest (3 rating) with 4 (dishonest) and 5 (extremely dishonest)? I doubt it because they would have reasoned against treating the same somewhat honest and mixture of honest and dishonest with unambiguously dishonest. They probably would have only taken 4 and 5 just as TI did in its report to characterize the extent of corruption. The TIB deviation from TI compilation of responses is thus vulnerable to the framing effect.
Now, it is totally unreasonable to assume that the distribution of responses across the ratings would have been the same under the two frames. For example, studies show beef described as “75% lean” are given higher ratings than beef described as “25% fat”. Similarly, it could be argued, a judiciary rated on a scale of honesty to dishonesty will get better ratings than a judiciary rated on a scale of incorruptible to corruptible.
Let me conclude with the classic question that has relevance in understanding how perceptions on corruption are shaped. Is the glass half full or half empty? Imagine you are thirsty and you ask how much water there is in the glass. You are told there is plenty. The glass is half full. You take a drink. Alternatively, after looking you are told with a worried expression that the glass is half empty. You decide to wait a little longer for your drink. In either case there was exactly the same amount of water in the glass. However, the reaction of how much water is in the glass frames the situation. It gives a context to a glass half full and half empty. That context alters one’s judgment about the glass, the water inside, as well as behavior when it comes to taking a drink.
How does seeing it one way frame everything else we later decide about both glass and water? Does seeing it the other way change our view? Seeing it half empty has a more emotional impact than seeing it half full. Similarly seeing institutions or individuals as more corrupt is more debilitating than seeing them as less honest. Political and legal institutions perceived as fair, just and (reasonably) efficient, increase the likelihood that citizens will overcome social dilemmas. If you think you live in a society where everybody steals, you probably will also choose to steal because you think the probability that you will be caught is low and, even if you do get caught, the chances of your being punished severely for a crime that is so common are low. By contrast, if you live in a society where theft is perceived to be rare, the perceived chances of your being caught and punished are high, so you choose not to steal.