I'm an A2 student and thought that i'd put up a few pictures of how i learnt about statistics - because when i was asked to do them i wasn't taught or helped by anyone we just had to do it which i hated because i hated maths! and was scared that i could not do it.
So the first thing you have to do is decide the statistical test you are using, which requires three pieces of information (at this level)
1) the type of data (nominal/at least ordinal)
2) the relationship (difference/correlation)
3) the measures you use (independent/repeated-matched)
I've put up a picture as an attachment which is a flow diagram (of sorts) to show you how you use this - but the image is very blurry so I've also attached 3 closer up ones of each section.
Once you have chosen and carried out the test on your research, you then have to decide whether the results you collected were significant or fluke (essentially) to do this - we need a critical values table (which you can actually get online if you just type in mann whitney critical value table - for example)
you will firstly need to decide whether the hypothesis is directional (one-tailed) or non-directional (two-tailed) this will be obvious (you don't just decide) directional = means there will be one outcome e.g. men will do better in the test by ticking more correct answers than women. Non-directional would instead be something like = there will be a difference between men and women's scores on the test.
You then need to choose a value which will be along the top of the critical value table i.e. 0.05, 0.10, 0.01 this literally means 5%, 10%, 1% and is referring to the likelihood of you making an error (type 1- which will come afterwards) IMPORTANT: unless the question/research/teacher has given you this value - always default to 0.05.
Once you have decided these things you also need to consider the amount of participants - (unless you are doing chi squared where it is instead the number of rows and columns) and then add them all up.
For example - if we are using Mann Whitney - and i used participants, had a directional hypothesis and a value of 0.05 and my score was 28 (random number) i would check the critical value table - find the value for 20 participants at a one tailed 0.05 test and find the number - if the critical value is lower than 28 then there is a significant difference - if it is higher then the result is basically a fluke.
NOTE: CHECK THE PICTURES ATTACHED TO FIND IF THE CRITICAL VALUES NEED TO BE HIGHER OR LOWER FOR EACH TEST.
the last part i'm going to mention is type one errors - you may have a significant difference however it could be a fluke because there's still a 5% chance you've made an error. This is a TYPE ONE ERROR If you want to make it less likely that you've made an error you make it 0.01 = which is 1%.
if anything doesn't make sense here feel free to leave a comment/message. Hope it helps a little bit. Attachment 473513473515Attachment 473513473515473540[attach]4.735134735154735e+23[/attach]
x Turn on thread page Beta
You are Here: Home > Forums >< University and university courses >< Science, Technology, Engineering and Maths university courses >< Psychology
Statistics help resource. watch
- Thread Starter
- 30-10-2015 12:23