Oh, sorry, to add to that, that's also Brian's later point about
coding the data - which is well worth thinking about. When you say
that the change is positive rather than saying how much change, you've
also thrown away information, so if it's possible, it's even better to
keep as much information as possible.
Jeremy
2008/12/5 Jeremy Miles <[log in to unmask]>:
> The chi-square test would treat your data as categorical. That means
> that you would say that 1 and 2 and 3 are different. But you know
> that they are not just different, they are also ordered, so if you do
> the chi-square test you throw away some information, and therefore you
> have less power.
>
> Jeremy
>
>
>
> 2008/12/5 Davies, Nicola <[log in to unmask]>:
>> Thanks Brian and Jeremy for your replies. Any thoughts on the Chi-Square? I have been reading up on it and it seems another possibility,
>>
>> Best Wishes,
>>
>> Nicola
>>
>>
>> ________________________________________
>> From: Brian K. Saxby [[log in to unmask]]
>> Sent: 05 December 2008 13:43
>> To: Davies, Nicola
>> Cc: [log in to unmask]
>> Subject: Re: Relationships in Changing Variables
>>
>> Hi,
>>
>> Just a quick note (and it may not be possible depending on what your
>> measures are) but the correlations might be more informative if you can
>> use the actual values rather than coding them in the first place.
>>
>> Brian Saxby
>>
>>
>>
>>> Hi All,
>>>
>>> I have four outcome variables that I have coded as 0,1,2 depending on
>>> whether this outcome didn't change, improved, or worsened, in each
>>> participant (from time point 1 to time point 2). I also have 10 variables
>>> that I have coded 0,1,2 depending on whether there has been no change in a
>>> participant, more use of this variable, or less use of this variable. Any
>>> ideas how I would test whether the outcome variables that stay the same
>>> (0), improve (1) or worsen (2) are related to no change (1), more use (1)
>>> or less use (2) of the other variables? Is a correlation appropriate?
>>>
>>> Thanks for your help,
>>>
>>> Nicola
>>
>
>
>
> --
> Jeremy Miles
> Learning statistics blog: www.jeremymiles.co.uk/learningstats
> Psychology Research Methods Wiki: www.researchmethodsinpsychology.com
>
--
Jeremy Miles
Learning statistics blog: www.jeremymiles.co.uk/learningstats
Psychology Research Methods Wiki: www.researchmethodsinpsychology.com
|