organizing data in SPSS

[jasmine]

Hi,
I have been trying to figure this out for days, please help!
I have several months of hourly temperature measurements for numerous subjects (several thousand data points for each participant).
I need to include each and every one of these hourly measurements, date, and time along with the subject id, their location, species etc in the data sheet in such a way that allows for more diverse analyses than just linear regression.
I would really appreciate any help you can give me. Even if it's to tell me it's not possible and why, so I don't keep running in circles...

Thanks so much

Jasmine

[jasmine]

Hi,
I have been trying to figure this out for days, please help!
I have several months of hourly temperature measurements for numerous subjects (several thousand data points for each participant).
I need to include each and every one of these hourly measurements, date, and time along with the subject id, their location, species etc in the data sheet in such a way that allows for more diverse analyses than just linear regression.
I would really appreciate any help you can give me. Even if it's to tell me it's not possible and why, so I don't keep running in circles...

Thanks so much

Jasmine

[Bruce Weaver]

You've answered your own question, as far as I can tell.  I.e., set up a LONG file with the following variables:

ID
Species
Location
Date_and_Time
Temperature
Etc.

Use a DATETIME format for the Date_and_Time variable (http://www-01.ibm.com/support/knowledgecenter/SSLVMB_20.0.0/com.ibm.spss.statistics.help/syn_date_and_time_date_time_formats.htm?lang=en).

It may be that you need to restructure the data for certain analyses you want to do later on.  If so, commands like CASESTOVARS and AGGREGATE will probably be useful.  Look them up (or Google them) and study the examples.

One other thing:  Given the large number of observations per ID, I doubt that the data will be entered by hand.  How are the data being recorded initially?  And how are you getting it into SPSS?

HTH.

jasmine wrote

Hi,
I have been trying to figure this out for days, please help!
I have several months of hourly temperature measurements for numerous subjects (several thousand data points for each participant).
I need to include each and every one of these hourly measurements, date, and time along with the subject id, their location, species etc in the data sheet in such a way that allows for more diverse analyses than just linear regression.
I would really appreciate any help you can give me. Even if it's to tell me it's not possible and why, so I don't keep running in circles...

Thanks so much

Jasmine

[Rich Ulrich]

More detail, consistent with Bruce's advice -

Keep in mind that ordinary "linear regression" is never going to use, as separate
scores, several thousand hourly-temperature data points. (What hypothesis would
that be?) 

If it is ordinary linear regression, you will have one record for each ID.
It will have each of the one-time ("species"?  location?) measures, which should
be entered into one file, one time.  Then you will use Match Files in order to
add to that file a few factor scores/ composite scores/ derived measures. 

The "long form" that Bruce mentions is a pretty natural way to obtain data from
devices - each day, time, and measurement in a separate row.  If the rows are
organized, say, with 24 hours on a row for one day, it *might* be more convenient
for further processing if you use CasesToVars to write the 24 rows per day.  On
the other hand, if you only use the Average for one day, or some within-day contrast
computations, then it could, conceivably, be handy to re-write the long form (if that
is what you have) to make a row per day.... but that seems less likely.
 
Aggregate with Break variables is robust, starting from the long form, if that will
handle what scores you want to create. 

Data ENTRY is often best chosen for what is most natural, non-redundant, and
error free, even if that means creating a number of separate files.  Do your data
reduction using the separate files (long-form with 1000 hours/rows/scores gets
reduced to one or a few Scores for each ID).  Then you MATCH to create a file for
analysis.

--
Rich Ulrich

> Date: Tue, 21 Jul 2015 18:29:59 -0700

> From: [hidden email]
> Subject: organizing data in SPSS
> To: [hidden email]
>
> Hi,
> I have been trying to figure this out for days, please help!
> I have several months of hourly temperature measurements for numerous
> subjects (several thousand data points for each participant).
> I need to include each and every one of these hourly measurements, date, and
> time along with the subject id, their location, species etc in the data
> sheet in such a way that allows for more diverse analyses than just linear
> regression.
> I would really appreciate any help you can give me. Even if it's to tell me
> it's not possible and why, so I don't keep running in circles...
>

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

[David Marso]

"more diverse analyses than just linear regression."
And we are supposed to psychically discern what you have in mind and
suggest how to arrange your data?  
In a nutshell it really doesn't matter how the data are placed into SPSS since it can be readily rederranged into any number of usable formats with a couple of lines of syntax.

Perhaps present what sorts of analyses you have in mind and some hints will be provided.

Rich Ulrich wrote

More detail, consistent with Bruce's advice -

Keep in mind that ordinary "linear regression" is never going to use, as separate
scores, several thousand hourly-temperature data points. (What hypothesis would
that be?)  

If it is ordinary linear regression, you will have one record for each ID.
It will have each of the one-time ("species"?  location?) measures, which should
be entered into one file, one time.  Then you will use Match Files in order to
add to that file a few factor scores/ composite scores/ derived measures.  

The "long form" that Bruce mentions is a pretty natural way to obtain data from
devices - each day, time, and measurement in a separate row.  If the rows are
organized, say, with 24 hours on a row for one day, it *might* be more convenient
for further processing if you use CasesToVars to write the 24 rows per day.  On
the other hand, if you only use the Average for one day, or some within-day contrast
computations, then it could, conceivably, be handy to re-write the long form (if that
is what you have) to make a row per day.... but that seems less likely.
 
Aggregate with Break variables is robust, starting from the long form, if that will
handle what scores you want to create.  

Data ENTRY is often best chosen for what is most natural, non-redundant, and
error free, even if that means creating a number of separate files.  Do your data
reduction using the separate files (long-form with 1000 hours/rows/scores gets
reduced to one or a few Scores for each ID).  Then you MATCH to create a file for
analysis.

--
Rich Ulrich

> Date: Tue, 21 Jul 2015 18:29:59 -0700
> From: [hidden email]
> Subject: organizing data in SPSS
> To: [hidden email]
>
> Hi,
> I have been trying to figure this out for days, please help!
> I have several months of hourly temperature measurements for numerous
> subjects (several thousand data points for each participant).
> I need to include each and every one of these hourly measurements, date, and
> time along with the subject id, their location, species etc in the data
> sheet in such a way that allows for more diverse analyses than just linear
> regression.
> I would really appreciate any help you can give me. Even if it's to tell me
> it's not possible and why, so I don't keep running in circles...
>
     
=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD