Repeated Measures data in spreadsheets

[Darinka Andjelkovic]

Dear colleagues,

 

I have tried the command that Leonid recently made for Joyce Marzan for the purpose of rotating CLAN spreadsheet output on its side, turning columns into row and vice versa:

spreadsheet stat.frq.xls

and it works perfectly.

 

But my question is about  possibility to use this device for the preparation of data file for further statistical analysis.

 

In most studies of language development one would look for one or more measures (let’s say number of utterances or morphemes) at different age samples for the same child.

And, for the statistical reasons we have to treat this as a Repeated Measures design for different age levels. So, working for example in SPSS, we have to sort age samples into different columns like all other variables (utterances, morphemes, etc.), while participants should be sorted in rows  like this:

 

                utter18m    utter19m   utter20m  morph18m   morph19m  morphs20m

Peter

Mary

John

 

I do not see any other way of preparation of Excel data into that kind of sort,  except for manual cutting rows and pasting them into collumns by means of PasteSpecial/Transpose.

It is rather slow, and very susseptable for making errors, especialy for a large quantity of data.

 

How do you handle this?

Maybe a Macro would work, but I do not know how to make it. Is there any available around, that I could use?

 

Any suggestion would be most appreciated.

 

Best wishes,

 

Darinka Andjelkovic

 

 

 

 

 

[Brian MacWhinney]

Dear Darinka,

    This format of subjects in rows and scores in columns is what we are now using for the MEASURES, MORTABLE, and EVAL programs that output Excel-type tables.  However, if you are creating various custom measures through special purpose runs on data, it is difficult for me to see how CLAN can guess the format of what you are creating.  Maybe you are saying that you are creating a collection of separate files for each subject and then trying to cut and paste data from these separate files into a single file.  But, in that case, maybe you should have run your command on *.cha using *CHI as the consistent identifier for your target child rather than "Peter" "Mary" and "John".  

In any case, some of this can be done with what we already have and some cannot.  It all depends on the details of how your data is coded, how subjects appear across files, and so on.  In the end, sometimes it may just be necessary to resort to cut and paste.  However, if there is some consistent and typical method that we can implement, we would try.  But we would need to have it very closely described.  Take a look at MEASURES, MORTABLE, and EVAL to get some sense of how this might work.

Best regards,

-- Brian MacWhinney

-- 
You received this message because you are subscribed to the Google Groups "chibolts" group.
To post to this group, send email to chib...@googlegroups.com.
To unsubscribe from this group, send email to chibolts+u...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/chibolts?hl=en.

[Darinka Andjelkovic]

Dear Brian and colleagues,

 

It seems I was not clear in my letter.

 

This kind of order is necessary when applying General Linerar Model Repeated Measures since same measures are within-subjects repeated.

It is actualy general and typical model of ordering data for any within-subject factors design (like large quantity of data from psychological tests where one participant has data for different items-measures from test and retest, etc...).

 

I am aware of MORTABLE output. It is the same in ordering data as output of FREQ and MLU, and I find it realy usefull.

But if one wants to compare the same measure (like morphems/utterances) for the same child  from different age levels (repeated measures), then this order is required for statistical reasons.

 

I send to Brian one of my output files as an example.

 

The first sheet contains the table generated by CLAN MLU.

The second sheet (named Target) has data that I sorted manualy.

I had to change the cells in column A in order to make clear that data are coming from a set of files (different ages) for the same child.

And I added one row above all in order to make clear that columns contain the same data (measure) coming from different age levels.

 

I know nothing about MEASURES and EVAL and I dont find any information of it on the site, nor in the Manual (I use the last version).

 

I hope the file that I made will make it more clear.

 

Best wishes and thanks for your concern,

 

Darinka

[Kevin Donnelly]

Hi Darinka

::::On Sunday 06 May 2012 Darinka Andjelkovic said::::

> I send to Brian one of my output files as an example.

Could you maybe post this to the list? I'd be interested in seeing it.

--
Pob hwyl / Best wishes

Kevin Donnelly
kevindonnelly.org.uk