Filed under: Uncategorized
Why do we bother to conduct research and statistical analyses? Behaviour just happens anyway, without knowing the what for’s or the why for’s.
Experimenters use inferential statistics to determine whether the results of an experiment are meaningful, and to draw inferences about a population based upon measures taken from a representative sample of that population.
Statistical tests in psychological research, test the probability of obtaining a given set of results. They produce a statement of the probability that an observation represents a true causal relationship and not a chance occurrence. Statistical significance or a statistically significant result is unlikely to have occurred by chance, but is the result of an intervention, of the effect of the independent variable upon the dependent variable indicating a real relationship most likely exists.
Statistical significance testing tells you the probability of a particular result occurring. Psychology often uses a significance level of p < .05. This means that the probability of an event or effect occurring by chance is less than 5%. There is less than 5 chances in 100 that the results are due to chance. Some research, for example medical interventions such as drug trials, requires a more stringent significance level of p < .01. This means the probability of an event occurring by chance is less than 1 %, therefore there is less than 1 chance in 100 that the results are due to chance.
Example of study with p < .05.
http://findarticles.com/p/articles/mi_m0FCG/is_3_34/ai_n27416665/
Example of study with p < .01.
http://www.ncbi.nlm.nih.gov/pubmed/11594943
Human beings have a natural curiosity into understanding the reasons we behave how we do, but research and statistical analyses is not just for the satisfaction of human curiosity. Many decisions in areas like psychotherapy, business and social policy etc (the list is endless) depend on, or are influenced by psychological research and the consequential results. Therefore we want to be as sure as possible that our theories about the mind and behaviour are correct. Psychology aims to describe, explain, predict and control behaviour and much can be gained from the knowledge we gather that can be of benefit. For example the norms of behaviour (relevant to culture, time etc) need to be established before the extremes of behaviour can be identified.
http://news.bbc.co.uk/1/hi/3254074.stm
Significance testing is not perfect. You must look at other things too, such as the effect size, the power, the theoretical underpinnings. Combined, we are better able to interpret data and gain a fully picture of the story they tell.
5 Comments so far
Leave a comment
ellies1mpson 
Firstly, let me say what a great blog this is. You have really grasped the topic very well. You explain everything very well and in great depth.
Comment by bangorlc25 February 7, 2012 @ 5:17 pmThe section on statistical significance, i think, needs further explanation. You’ve explained very well what it is but you haven’t really explained what the benefits of it are or why researchers should conduct.
Furthermore, i would have liked to have seen further discussion in regards to areas such as psychotherapy, business and social policy, which you mention briefly. I thought i was a good idea to bring these areas in as you expand outside of the realms of psychology. Showing that research is not restricted to psychology, or science in that fact, is a great point to make. Unfortunately, its not one people seem to write about a lot :).
Nevertheless, well done. This is an excellent blog.
You mention effect size at the end of your blog and I just wanted to mention what it is and the most common measures. In statistics effect size is essentially a descriptive measure of the two variables in a sample. There are two main measures of effect size: Cohen’s d and eta squared (n2) or partial eta square (np²). A report of the different types of effect size and how commonly they were used in reports was conducted finding interesting results (hopefully this link will work but it took http://www.psych.lancs.ac.uk/people/uploads/petermorris20110624T090210.pdf). The most commonly used measure of effect size was np2 but according to the report this is probably because it is the measure that is used by SPSS. The np2 can only be used across studies to compare effect size if the studies have the same design, and can not be used to compare the size of different effects in a study.
Comment by psychmja1 February 9, 2012 @ 1:52 pm[…] https://ellies1mpson.wordpress.com/2012/02/05/why-do-we-bother-to-conduct-research-and-statistical-an… […]
Pingback by Comments for my TA « psychmja1 February 9, 2012 @ 2:28 pmReally enjoyed reading this, you clearly put a lot of thought into this blog 🙂 I looked at all the links you provided and found them really interesting and relevant to the message you were getting across. I agree with bangoric25, you could have gone into a bit more depth with the statistical significance, but then again you had a lot of good points to make and the blogs aren’t meant to be too long, so overall I thought it was brilliant
Comment by klbpsych February 10, 2012 @ 7:17 pm[…] https://ellies1mpson.wordpress.com/2012/02/05/why-do-we-bother-to-conduct-research-and-statistical-an… […]
Pingback by Homework for Shanti « klbpsych February 10, 2012 @ 7:53 pm