SPSS Help

top_drawer_2

Right I have been battling with this all afternoon ....

I have constructed a questionaire, in it I have asked people what risk factors they believe may be relevant to contracting an illness. I have listed possibles including silly ones and myths. At data entry I input them as individual variables ie person didnt tick = 1 person did tick = 2 . Now I want to total them up to work out whether there's a relationship between right answers and age/anxiety (other aspects of my questionaire).

I thought if I totalled them up using the compute function that would do it but as that still a/c for all the wrong answers and the 1's then that doesnt work....

Any ideas??

Jen

GlasweJen

I hate SPSS, never been able to make heads or tails of it but you can post a question on http://talkstats.com/ it's a forum with people who are wizz kids with spss and the like. hope that helps.

blue_ashleigh

Not sure if this is what you're after, but have you tried correlations?

top_drawer_2

I am struggling to even work out how to run a repeated measures with a control condition. I've split the file (tried organise and conpared and no change) there is no obvious option to choose so that I can conpare both the control and the intervention group, both before and and after.

I think the problem Im having is that I can't work out to define my control group - I just have it input as CG (control group) or RI (Received Intervention). How do I make it possible to analyse??


Jen

xara

What are your variables? youre IVs and DVs?

Its probably the way you have inputted/coded it.

melancholly

i don't fully understand exactly what you want to do, but one useful thing when coding data is to always use 0 and 1 rather than 1 and 2. so 0 if they didn't tick it and 1 if they did.

it sounds like some kind of repeatedly mesures ANOVA with two groups where you're looking for an interaction. however that's no help unless you know what your outcome score is composed of..... wasn't the measure you would be using to summarise each person's responses decided on at the start?

it sounds like you have 2 scores for each person? the number of 'sensible' responses they ticked and the number of 'silly' ones? if you can't total them easily in spss then copy and paste the data into excel, code each question as silly or sensible, then sort data by that column and then you can easily sum the total responses.

i think you need to speak to your supervisor asap tbh. or find some online slides on how to do different types of ANOVAs - try googling Andy Field as he is the guy who write the most popular psycholology stats and spss book out there. he has lots of handouts from his own teaching up there.

don't panic - at this point in your degree you can work this out (or you should be able to!) but it sounds like you are too flustered to think straight right now!

top_drawer_2

xara wrote: »

What are your variables? youre IVs and DVs?

Its probably the way you have inputted/coded it.

hi, the ivs are time/intervention no intervention.

The dvs are knowledge and anxiety.

melancholly wrote: »

i don't fully understand exactly what you want to do, but one useful thing when coding data is to always use 0 and 1 rather than 1 and 2. so 0 if they didn't tick it and 1 if they did.

it sounds like some kind of repeatedly mesures ANOVA with two groups where you're looking for an interaction. however that's no help unless you know what your outcome score is composed of..... wasn't the measure you would be using to summarise each person's responses decided on at the start?

it sounds like you have 2 scores for each person? the number of 'sensible' responses they ticked and the number of 'silly' ones? if you can't total them easily in spss then copy and paste the data into excel, code each question as silly or sensible, then sort data by that column and then you can easily sum the total responses.

i think you need to speak to your supervisor asap tbh. or find some online slides on how to do different types of ANOVAs - try googling Andy Field as he is the guy who write the most popular psycholology stats and spss book out there. he has lots of handouts from his own teaching up there.

don't panic - at this point in your degree you can work this out (or you should be able to!) but it sounds like you are too flustered to think straight right now!

hi, yes I have administered a questionaire (to 140 people!) and then 40 have done the intervention condition involving viewing a presentation and answering the same questionaire and another 40 were as used as control and completed the questionaire again without receiving the intervention.

I will definitely replace my coding as you suggest - I'm using 1 and 2 at the moment and its confusing!

Outcome score - its being measured as anxiety/knowledge in response to the information provided.

Yes we agreed on Repeated Measures from the beginning but I've only ever done it on data provided, never actually had to invent/do the experiment.

The questions are not exactly silly but they are based on myths, people often dont even seem to realise how not based on reality their beliefs are.

I have a book by Andy Field and I seem to be running it right but I clearly am not ..........

My tutor is away now for Easter.

Jen

melancholly

right, well you have your score, whatever that may be and then you have 2 columns for 80 people for their scores from the two time points. this score should be the summary score using all their responses - calculate this in spss using compute or in excel and copy and paste it - whatever is easiest. then you have a separate column marking them as control versus intervention (e.g. put 1s for all the controls and 2s for all the interventions). then you do a simple repeated measures ANOVA selecting the variables from the 2 time points. you can also add in the information showing that they are in two groups in a box underneath whether you select the time points. off the top of my head i can't remember what spss calls it but it's in the books.

you may then 'lose' the data from the 60 people who did not do the questionnaire twice. for an undergrad project i would just accept that as a loss, but if you wanted you could do some kind of factor analysis looking for underlying latent traits but that is probably overkill and i'm guessing a bit complicated for you right now! you could devise some kind of path anaylsis incorporating age and anxiety but, you should probably just do a whole load of correlations including all the data from 140 people, with the repeated measures anova on 80 as a follow-up test......

top_drawer_2

I thought about the second part ... my tutor has never really mentioned correlations - she talked a bit about regression but she didnt think that was a priority.

I think the way I am inputting is the problem as I have't provided a option to split the data - it doesnt seem to offer one.

I have all day "booked" to spend on it tommorrow so I will do it tommorrow.

Thank you!!

Jen

melancholly

you define a variable that defines the groups..... you just call it 'group' and then make 1 = control and 2 = intevention (change it in the variable view). then you put the number in for whatever group the person was in..... so in total you only need 3 columns per person for the anova - group allocation, overall first score and overall second score (perhaps an extra column for participant number). you don't need to 'split' the data in any complicated way at all.....

in your position i would have two separate spss files - one with all 140 and one with the 80 for the anova - it's just less confusing.

this isn't very complicated to set up....

top_drawer_2

no I know it wasnt complicated at all to set the data file up when they just handed me the data set on paper but I've become befuddled when I have so much info x2 coming at me. I've entered it all at once as thats what my tutor suggested (code everything regardless of whether its relevant to your hypothesis).

Jen