One important tool that is used for scientific research, particularly in the social sciences where findings cannot be so concretely observed, is data analysis, which in this case is made up mainly of statistics. These are really important because they can provide extra information about our data that we cannot see straight away, but they can also tell us whether are results are what is called ‘statistically significant’, which is today often considered the aim of all research. Many beginner psychologists find statistics difficult to understand and an annoyance more than anything else, but its use has had a massive impact on the accuracy of the discipline and cannot be ignored. Personally it’s something I find quite interesting.
There are two key types of statistic, the first being the ‘descriptive statistic’. This provides further information about the data we already have. The most obvious example of this is the mean, which is an average of all of the different scores from the various different test trials. For example, a study looking at the reaction speeds of 100 people in the morning and in the evening will have 200 different speed results, and when looking at so many, it is hard to notice any patterns. But, all of the reaction speeds can be put through an equation, resulting in an average (most typical) speed for both morning and evening. This makes it much easier to compare the results for each group, leading to useful conclusions. However, the mean does not tell us about the variation of the results in each group and how spread out the speeds for each group are. This can cause a problem, since even if the means for each group are far apart, if there is a lot of variation in the individual values of the two groups, then the results may still not be very strong, and may in fact just be down to chance!
To avoid this error from occurring, the concept of ‘significance’ is used. There are various tests, known as ‘inferential statistics’ (the second type of statistic) which take the mean and a measure of variation known as standard deviation, and use them to work out the likelihood that the results of the study were caused only by chance. If this figure is below a certain value, then the results are declared significant and we assume that there is a real world psychological relationship going on. If the figure is too high, then we must accept that there is either no real relationship going on, or that we simply do not have enough evidence to prove it yet. This is probably the most important part of statistics in research, since it determines whether we accept or reject theories. Another type of inferential statistic is a correlational test. Rather than testing the significance of a difference between two conditions, these simply look at two different variables (things that change), and tells us how related the movements of them are. For example, shoe size and height would correlate highly, since one tends to go up as the other does for an individual.
Though statistics can be hard to get your head around, it is important to recognise their effectiveness at testing psychological ideas and their ability to determine what goes down as psychological ‘truth’ and what is rejected. On the other hand, they still have their problems, and no test ever comes back 100% positive in agreement with a particular hypothesis, since human behaviour is so complex and is affected by so many factors, so we can never be entirely sure that errors are not being made, but as you’ll have learnt if you’ve read my other research methods articles, it is near impossible to design a faultless study and sometimes we just have to do the best we can.
Image from: http://4vector.com/i/free-vector-statistics-form-the-vector_001665_Business_charts1.jpg
0 Comment:
Be the first one to comment on this article.