Wow, it's been a month already - how time flies when you're having fun.
In today's Straits Times, I came across an article on Page 2 of the Home Section that raised my eyebrows - 1.44 Million Read ST screamed the headline (note: this link will be dead if you're reading this more than a week after this entry was made - so you can read the PDF snapshot of the article here). I thought to myself, so 1.44 million readers actually bothered to contact ST to tell them so, not bad. I was further intrigued to find out how this feat of contacting 1.44 million people was done - especially since I was not contacted myself about this, and duly read the article.
Well, what actually happened was this - the Nielson Media Index 2008 was released yesterday or the day before. This Index uses a 'face-to-face survey' of 4,700 persons over the age of 15 to gather information and subsequently reports this finding (ie. how many feel this, how many agree with that etc.) through the Index. And in this Index (which I can't find and suspect must buy to read), the ST was quoted as being read by 39% of those surveyed - ie. 1,833 of the 4,700 persons said they read the ST.
Apparently, in last year's report (Nielson Media Index 2007), this percentage was quoted as 37.9%. So, using 39% - 37.9% = 1.1%, the real difference between the 2 years would be more or less 52 persons (1.1% of 4,700). However, this percentage could have also been reached if there had simply been more total respondents to begin with - ie. 1,833 of 4,836 = 37.9%. So, if there had been just 136 more persons surveyed last year as compared to this year, this 'increase' would not be an increase after all.
Interesting right? It gets better. The 1.44 million figure is 39% of 3.69 million 'consumers' (roughly 80% of the population since the survey was only for those aged 15 and older). The actual corresponding numerical figure for this 39% is 1,833 persons, which means each person in the survey acted on behalf of 784 other 'consumers'. Therefore, the '105,000 more readers' translates to just 133 real survey respondents.
Hey, the actual numbers don't seem as impressive, I hear you say? Well, that's the idea, isn't it? Instead of reporting the facts outright, the suggestive figures were reported glamourously. Any survey is only indicative with nothing to laud about, and even though 4,700 persons may very well be a large sample size, it is only representative of 0.12% of the intended target (that's LESS THAN ONE PERCENT mind you).
What would have been a better approach (better for the independent understanding of the reader that is) would be stating the respondent size (4,700) followed by just the percentage and/or respondent figures for each category (eg. 'ST readers comprised 39% or 1,833 of the respondents' and so on). Extrapolating the figures is for dramatic effect only and therefore should have been limited to a one-liner somewhere in the middle of the article.
Well, this concludes the lesson from ST. Remember, now you can just find any 100 strangers at Orchard Road, give them a sweet or freebie, then ask them to answer the question "Do you think I am a nice person?", collate the figure (which would be a 'yes' for almost all), get an official figure on shoppers at Orchard Road from somewhere (which should be in the millions), and finally proclaim, 'XXX is Liked by Millions of Shoppers at Orchard Road!' - ie. 90+% extrapolated to cover that official figure on shoppers.
I sincerely hope to find the Nielsen Media Index and have a look at it and also study the methodology used. But I am pretty sure that no one in the 4,700 respondents was a homeless person, an old aunty or uncle wiping tables, a person serving jail time, or a person at IMH. But of course the 1.44 million figure includes such persons by default.
Well, if anyone can point me to the source of these findings, please inform me so that I can update this entry with that information, thank you. In the meanwhile, do have fun manipulating statistical figures to make yourself look good.
In today's Straits Times, I came across an article on Page 2 of the Home Section that raised my eyebrows - 1.44 Million Read ST screamed the headline (note: this link will be dead if you're reading this more than a week after this entry was made - so you can read the PDF snapshot of the article here). I thought to myself, so 1.44 million readers actually bothered to contact ST to tell them so, not bad. I was further intrigued to find out how this feat of contacting 1.44 million people was done - especially since I was not contacted myself about this, and duly read the article.
Well, what actually happened was this - the Nielson Media Index 2008 was released yesterday or the day before. This Index uses a 'face-to-face survey' of 4,700 persons over the age of 15 to gather information and subsequently reports this finding (ie. how many feel this, how many agree with that etc.) through the Index. And in this Index (which I can't find and suspect must buy to read), the ST was quoted as being read by 39% of those surveyed - ie. 1,833 of the 4,700 persons said they read the ST.
Apparently, in last year's report (Nielson Media Index 2007), this percentage was quoted as 37.9%. So, using 39% - 37.9% = 1.1%, the real difference between the 2 years would be more or less 52 persons (1.1% of 4,700). However, this percentage could have also been reached if there had simply been more total respondents to begin with - ie. 1,833 of 4,836 = 37.9%. So, if there had been just 136 more persons surveyed last year as compared to this year, this 'increase' would not be an increase after all.
Interesting right? It gets better. The 1.44 million figure is 39% of 3.69 million 'consumers' (roughly 80% of the population since the survey was only for those aged 15 and older). The actual corresponding numerical figure for this 39% is 1,833 persons, which means each person in the survey acted on behalf of 784 other 'consumers'. Therefore, the '105,000 more readers' translates to just 133 real survey respondents.
Hey, the actual numbers don't seem as impressive, I hear you say? Well, that's the idea, isn't it? Instead of reporting the facts outright, the suggestive figures were reported glamourously. Any survey is only indicative with nothing to laud about, and even though 4,700 persons may very well be a large sample size, it is only representative of 0.12% of the intended target (that's LESS THAN ONE PERCENT mind you).
What would have been a better approach (better for the independent understanding of the reader that is) would be stating the respondent size (4,700) followed by just the percentage and/or respondent figures for each category (eg. 'ST readers comprised 39% or 1,833 of the respondents' and so on). Extrapolating the figures is for dramatic effect only and therefore should have been limited to a one-liner somewhere in the middle of the article.
Well, this concludes the lesson from ST. Remember, now you can just find any 100 strangers at Orchard Road, give them a sweet or freebie, then ask them to answer the question "Do you think I am a nice person?", collate the figure (which would be a 'yes' for almost all), get an official figure on shoppers at Orchard Road from somewhere (which should be in the millions), and finally proclaim, 'XXX is Liked by Millions of Shoppers at Orchard Road!' - ie. 90+% extrapolated to cover that official figure on shoppers.
I sincerely hope to find the Nielsen Media Index and have a look at it and also study the methodology used. But I am pretty sure that no one in the 4,700 respondents was a homeless person, an old aunty or uncle wiping tables, a person serving jail time, or a person at IMH. But of course the 1.44 million figure includes such persons by default.
Well, if anyone can point me to the source of these findings, please inform me so that I can update this entry with that information, thank you. In the meanwhile, do have fun manipulating statistical figures to make yourself look good.