OCR Psychology G544 June 18th 2012

We do!

We don't need to know any tests for interval/ratio data, because these tests are parametric and we only need to know non-parametric, so Chi squared, Sign test, Mann Whitney, Wilcoxon and Spearman's Rho.

Do we actually need to know the test and how to do it?

Or do we just need to know why we use that particular test? Ie, nominal/ordinal data and independent/repeated/correlation

I'm still really confused about how to conduct question A. If someone could reply I'd be eternally grateful!

So, say the question was: You've been asked to design an observational study that would collect a) nominal b)interval c) ordinal data

What would you do for that?! eeek.

Also, How would you go about conducting a correlation study? could someone kindly write a brief example of what i should write?

big thanks! xxxxxxxxxxxxxxxxxxxxxxxxxx

For observation:
nominal: do like a frequency grid, so number of times a behaviour occurs
Interval: recording the amount of time a behaviour occurs
ordinal: rate the severity of the behaviour on a rating scale
When you are designing an observation:
1. what type of sampling method (opportunity, random, snowball etc)/what is your sample
2. Where you are doing it, for what and for how long
3. ethics: gaining consent from somewhere?
4. participant or non participant
5. covert or overt
6. event sampling or time sampling (or both)
7. how you are going to get data: either structured in frequency grid, or unstructured writing everything
8. give a few examples of your categories and what you aim to record
9. inter rater reliability by having second observer? Maybe say you would have done pilot study etc
For correlation:
1. sampling method(opportunity, random, snowball)/what is your sample
2. Gaining consent
3. what are your variables and how are you going to measure them
4. exact instructions on what you would tell participants to do e.g with correlation between memory and iq, you would tell them how long they have etc
5. give them the right the withdraw and remain anonymous
6. Thank them and put results up somewhere e.g college website?

Hope that helped!

If I could hug you right now I would! Thank you so so sososoos much!

What would I do if it asked for nominal, interval and ordinal in an experimental design, or a correlation?

Im really really sorry for all the questions, if theres any way I can help you, let me know! x

Lol aww its fine! It helps me to remember it too its similar really for correlation and experiment
nominal: Is categorical data such as the number of people that answered red and the number of people that answered blue.
Ordinal: Is things such as rating scales or something that doesn't have equal intervals such as in a correlation one variable could be how happy the person is on a scale of 1-10, as this is subjective data it is ordinal.
Interval: Is data that has equal intervals, this is probably the one you'd use the most in say experiments for example testing memory the data may be the number of words each participant remembered

Any other questions just ask i'm hoping that i'm right aha

Thank you!!

I'm alittle bit confused about interval data, would you mind expanding a little bit? xxx

Its something that is like a fact, such as temperature, gcse scores, words remembered and they have equal intervals between such as if someone remembered 50 words you can say that they did twice as well as someone who remembered 25 words.
You cant do that with ordinal data because for example if its happiness and someone rated them self 10/10 another 5/10 you cant say one person is double the happiness of the other person because they have both interpreted the scale different if that makes sense!

Its something that is like a fact, such as temperature, gcse scores, words remembered and they have equal intervals between such as if someone remembered 50 words you can say that they did twice as well as someone who remembered 25 words.
You cant do that with ordinal data because for example if its happiness and someone rated them self 10/10 another 5/10 you cant say one person is double the happiness of the other person because they have both interpreted the scale different if that makes sense!

I'm really worried i'm going to get ordinal and interview data mixed. Because GCSE scores could be ordinal, couldn't they? if you put the scores in order?

Also, for example say they were testing how many words a particpant could recall

id say that would be nominal data as its put in categories numbers that were and were not recalled? But people say thats ordinal, would you mind explaining why?

Ordinal isnt necessarily putting scores in order, maybe easier if you remember it as:
Ordinal= data that doesn't have equal intervals and is subective, so gcse results are internationally recognised and there is usually the same difference between a A, B, C and D etc so you can say that someone that got an A did better than someone that got a B whereas with ordinal data it would be rating scales and you cant really say that one person is happier than the other based on ordinal data as it is subjective.
Nominal= it is categories that you use and then count a frequency, so number of people that finished the race and number of people that didn't, so if you imagine a bar chart you could put finish the race, not finish the race at bottom then the number of people vertically. However you can't do this with ordinal or interval data as there isnt 2 clear categories, as the data you are getting is how many words they remembered.

Hope that helped a little bitt

Thanks for taking the time to explain this to me, i really appreciate it! x

I still can't get my head around it though! Why am I so special haha. If you wouldnt mind could you answer this for me, perhaps then I might understand a bit more?

Say they asked you to collect nominal data for an experiment. How would you do this? questionnaire, interview?

same for correlation, and observation?

How would you do the same for ordinal and interval data for each research method?

Really really really really really appreciate this!

xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

haha its fine, aw its so annoying when you cant understand something but dw it will come to you

Nominal- Number of people that e.g get it right or wrong, say yes or no, put red or blue, finish the race don't finish the race, say the persons attractive, say the persons unnattractive. For self report this would be closed questions and for observation this would be event sampling.
Ordinal- Ordered data e.g the order people come in a race, the order of people's happiness, Order of the most scared to least scared, any data that can be ordered but the difference between 1 and 2 2 and 3 isn't always the same. someone that came first in a race may be just in front of the 2nd person whereas those that came 3rd and 4th were half an hour later, so not equal gaps. For self report and observation this would be rating scales.
Interval- measured using publicly recognised measurement, so instead of ordering where they came in race, you would time them and then this would provide you with equal intervals, in experiments this could also be timing, in self reports this would be something like for how long do you revise a day, or what did you get in your GCSE's etc and for observation this could also be timing the behaviour

haha its fine, aw its so annoying when you cant understand something but dw it will come to you

Nominal- Number of people that e.g get it right or wrong, say yes or no, put red or blue, finish the race don't finish the race, say the persons attractive, say the persons unnattractive. For self report this would be closed questions and for observation this would be event sampling.
Ordinal- Ordered data e.g the order people come in a race, the order of people's happiness, Order of the most scared to least scared, any data that can be ordered but the difference between 1 and 2 2 and 3 isn't always the same. someone that came first in a race may be just in front of the 2nd person whereas those that came 3rd and 4th were half an hour later, so not equal gaps. For self report and observation this would be rating scales.
Interval- measured using publicly recognised measurement, so instead of ordering where they came in race, you would time them and then this would provide you with equal intervals, in experiments this could also be timing, in self reports this would be something like for how long do you revise a day, or what did you get in your GCSE's etc and for observation this could also be timing the behaviour

I think ive figured out why im confused... Im getting confused because i cant identify what measure is what data level. For example.

Happiness scales from 1-5 is ordinal, right? But how is that not also nominal because you could argue that people fall under 2/5 happy and 5/5 happy.

Same with grades in school. People could fall under the A category, the B category etc etc .

Also say with student ID numbers, which is nominal but It could be ordinal too, because they could be ranked in order?

and also, an experiment investigating audience effects in which 10 participants all had to see how many basketball shots they could get through the net from a particular position when no one was watching as opposed to when an audience of 12 people were watching. This is ordinal but why isnt this nominal?


I think im okay with interval data, because thats just thinks like time, temperature things that have standardised scales


So just to double check that Im kind of on the right lines:

Say the question in the exam was: 1) Correlation study that plans to collect nominal data using independent measures design. I chose the option happiness and popularity

I would administer a closed ended questionnare asking the question : Would you consider yourself happy. yes/no would you consider yourself popular yes/no to males and to females.

2) you must use independent measures and collect ordinal data. whether people remember names of food when theyre hungry or not

how would you measure that ^ id have no idea?! Just as a guess I would put them in two conditions, hungry or not hungry and then just conduct a standardised memory test, and then place the scores in order. Is that right?!

A way in which my psychology teacher has got us to remember the tests is
Nominal, Independent measures, Chi squared test
Nick Is Cool
Nominal Repeated measures, Sign Test
Nick Rides Scooter
Nominal, Correlation, Phi Test
Nick Can't Punch

Ordinal, Independent Measures, Mann Whitney U Test
Ollie Is Mad
Ordinal, Repeated Measures, Wilcoxon T test
Ollie, Runs Wild
Ordinal, Correlation, Spearmans Rank Test
Ollie Can Swim

And for interval/ ratio data as long as you use a small sample use the test which would be used for ordinal level data as assumptions of parametric test cannot be met:
Interval/ ratio + independent measures = Mann Whiteney U test
interval/ ratio + repeated measures = Wilcoxon T test
interval/ ratio + correlation = Spearmans Rank test

I'm making some booklet thing for G544, will post it on here when I've done it!!! Covers everything on the spec xx