Psychology research: Confused on design and variables, can anyone help?

Tigersilly

Psychology research: Confused on design and variables, can anyone help?

I think the design is repeated measures and that the level of measurements is ordinal level.

I have to write a report based on a questionnaire (Likert Scale) about levels of environmental concern vs egoistic concern.
The questionnaire was distributed amongst a group of 1st year psych students. Below the questionnaire were further questions enquiring about gender, first language, employment, voluntary work etc. I am confused as to whether these are ALL considered independent variables, whilst the mean (catergorised- egoistic vs environmental) concern scores are considered dependent variables.

The question we have been given to answer is: In the sample is there a difference between the egoistic and biosphric concern scores, and if so what's the direction of the difference?

I'm confused! Can anyone help me?

Zziggi

Tigersilly wrote: »

Below the questionnaire were further questions enquiring about gender, first language, employment, voluntary work etc.

I would have said these were simply pieces of information you have collected about the background of your participants. Therefore i would list these in the participants section of the Method. (i.e. 18 of the p's were male, 12 were female, all 40 p's had english as a first language etc etc etc). These aren't I.V.'s as far as i can see.

Was the same questionnaire given to the same participants after a time delay? If not, why is it you think it is a repeated measures design?

Tigersilly

I'm a first year student so bear with me

So, it can be the case that there are no IVs and levels of IV because nothing here is being manipulated?

It is considered to be a survey? 'a passive collection of info that already exists'

Dependent variables do exist, right? They are the concern scores?

Is the level of measurement ordinal and the design repeated measures?

Edit: on second thoughts no time delay was involved... I guess it's not repeated measures. My problem is that I have been given a series of stock questions to answer e.g. group design and I can't work out what the group design is.

the_devil_made_me_do_it

IV is the thing you are manipulating, ie if you were doing an experiment on memory, the IV maybe whether the participants were asked to free recall a list of words or whether thay had prompts (catagory titles), the DV would be how many words were recalled.

Gender ect.. can't be IVs because they cannot be manipulated. Those questions are being asked either cause they are relevant in some way, or are needed to describe the participants used.

There is also something called Extraneous Variables which are things which can effect your study, for instance: lighting in the room ect...

Did the questionaire ask the same questions of all the participants, or did each participant get allocated to a condition and given a questionairre according to the condition they were in?

If it was the first option, then I'd say the design was within-participants, because all participants took part in all conditions. You don't (or shouldn't) be required to say why you chose this design, unless you are doing A-level psychology, and then you may get asked to say why the design was chosen.

It's difficult to offer a decent answer because you haven't explained your situation very well.

melancholly

as far as i can see, you have a repeated measures design. you have asked a group of people a set of questions - and you have two outcome variables: their score on the environmental concern questions and their score on the egotistic concern questions.

i assume you are looking to see if there is an association between how peopel think about themselves versus think about the environment.

the next key point is what your actual outcome measure is.... did you add up scores from a number of questions, or average them, or something more complicated? and how many points were there on the likert scale and how many questions there were for each DV?

personally, i would take averaged data from a scale and just stick it through a paired t test in your situation. it may not be strictly continuous data as it clearly starts from an ordinal scale, but when you have averaged lots of them together, you can get values that are inbetween each point on the scale. whether that's you can 'officially' do according your stats lecturer is a different matter. i've published papers like that, so it can be done! otherwise i guess you're in the realm of non-parametric stats to look for a significant difference.

with the data you collected at the bottom, you can then look to see whether the pattern in performance changes across different groups (assuming you have sufficient numbers of people in each category). you could look for gender differences in a mixed ANOVA (i think a non-parametric version is the krusall-wallis?), or see if people who do voluntary work have different levels of egotistical concern. but these are all extra tests, not the primary aim!

anewman

the_devil_made_me_do_it wrote: »

Gender ect.. can't be IVs because they cannot be manipulated. Those questions are being asked either cause they are relevant in some way, or are needed to describe the participants used.

Although if using multiple regression you could call Gender and Age etc "predictor variables", and the one you're interested in the "outcome variable".

melancholly wrote: »

personally, i would take averaged data from a scale and just stick it through a paired t test in your situation. it may not be strictly continuous data as it clearly starts from an ordinal scale, but when you have averaged lots of them together, you can get values that are inbetween each point on the scale. whether that's you can 'officially' do according your stats lecturer is a different matter. i've published papers like that, so it can be done!

The stats books and lecturers will point out many papers use incorrect statistical methods. But this is the reason for reporting the tests used and probability, t statistic values, degrees of freedom etc etc.

To OP a good book is Andy Field - Discovering Statistics using SPSS second edition. There's one by Darren Langdridge which I haven't read but might not be too bad (he was my methods lecturer in my first year at University and was the sort always cracking jokes). The one by Hugh Coolican I always thought was rather poor and a waste of money myself.

the_devil_made_me_do_it

anewman wrote: »

Although if using multiple regression you could call Gender and Age etc "predictor variables", and the one you're interested in the "outcome variable".

Don't think 1st yr students would be expected to go that much into detail. But hey ya never know these days

anewman

the_devil_made_me_do_it wrote: »

Don't think 1st yr students would be expected to go that much into detail. But hey ya never know these days

Actually yes, sorry lol. Focus on the task in hand.

If the course is like mine was and the first year did not contribute to the degree classification, try not to worry too much about it and read feedback on marked work thoroughly. They're probably expecting you to make errors, so you don't make them when the work in year two goes towards your classification.

Things are definitely changing in Universities. Leeds University has an online submission thing which cross-references your work with loads of books and internet websites, including past websites, and if it detects any hint of plagiarism it generates a plagiarism report.