Investigate right!

Reliability

If I were to collect a fair, random sample of Editor's letters found in both male and female aimed magazines I would simply split the two via their intended gender of audience and then select half of my intended total from each pile without reading them. Therefore they should be a fair sample and any anomalies would be very insignificant.

For Brian Cox's tweets I would follow a more strategic path. First I would divide the total of his tweets by the number of tweets I wish for my sample. For an example lets say I wish to find 20 tweets out of a total of 500. 500 divided by 20 equals 25. I would then choose a number at random between 1 and 25, lets say 16 for example. From 16 I would then at 25 and take the tweet represented by that number and continue adding 25 and taking the corresponding tweet until I run out of tweets. I will then end up with a sample of 20 tweets taken purely at random.

The same technique would be used for sampling Michael Gove's comments on education.


Comparability
How far does the language of advertising vary according to the gender of the target audience?

For this investigation I would opt to look at the difference in carbonated drinks. It has been publicized that drinks like Diet Pepsi and Pepsi Max or Diet Coke and Coke Zero are advertised purely for different genders and are both, actually, almost identical. As evidence I would use the product, the adverts and the contents of the drink itself. I believe this would show how language, in multiple formats, is used to appeal to the different genders on carbonated drinks.



How does a 'live' commentary on radio differ from one on television?

This is an investigation that I would breaking into two. Firstly I would investigate the difference in news reports between the two formats from numerous channel/stations. I would, secondly, then look at the statistics between the two so that I would have both linguistic evidence and popular it evidence to compare.


Ethicality

The sample data that I had planned to collect would have been ethical as they would have all have been published as videos, in books or online. It would be reliable as I would have followed the method of fair collection as stated above. In terms of comparability it would lack strength. My data would range from dictionary denotations to the assumed adding of a suffix to a word when trying to speak in an outdated tongue. However, if I could find at least two of each forms then I could compare them together easily.