(complex) cases to vars problem

[Maguin, Eugene]

I'd like some help to find a more efficient way to work a cases to vars problem. I going to present only a simplified version of the actual dataset because the actual dataset has additional but irrelevant complexity. The data are from an EMA (ecological momentary assessment) measure. Briefly, persons have a piece of software installed on their phones that presents a set of survey items to them at multiple times during the day. This is now the simplified version. The survey consists of 7 items. Item 1 is always presented first and the remaining six items are presented in a ramdomized order for each presentation. 

The data are provided to us in long format and a data record has the following structure:

Person, instance, item_id, item_label, response, item_time, response_time. 

The data for one presentation of the survey items consists of 7 records, one for each of the items presented. Call_time is the time of presentation initiation, item_time is the time at which the item was presented, response_time is the time at which the response was entered. Time is UTC seconds. 

Casestovars works exactly as it should; however, the resulting dataset is not correctly structured because column contents are in call_time order. They need to be in a consistent order across records (person-call_time). I have a plan for this, which is basically a programmed casestovars operation using aggregate which is going to be a lot of work (and error prone) as the number of items is not 7 but rather 45. I'm interested in better ideas.

Here is a sample dataset.

person,call_time,item_id,item_label,response,item_time,response_time,seq

1278,1520541611,1022,Qxtwii,8,1520541619,1520541623

1278,1520541611,1031,NrfghJ,6,1520541641,1520541643

1278,1520541611,1024,AgHmmw,5,1520541648,1520541657

1278,1520541611,1051,BpweBd,5,1520541665,1520541672

1278,1520541611,1059,LkxCwu,6,1520541689,1520541691

1278,1520541611,1029,BooDzz,4,1520541704,1520541712

1278,1520541611,1040,KeioUy,2,1520541719,1520541722

1389,1521539923,1022,Qxtwii,7,1521539951,1521539957

1389,1521539923,1024,AgHmmw,7,1521539972,1521539977

1389,1521539923,1029,BooDzz,4,1521539981,1521539984

1389,1521539923,1051,BpweBd,5,1521540003,1521540010

1389,1521539923,1059,LkxCwu,6,1521540018,1521540022

1389,1521539923,1040,KeioUy,6,1521540026,1521540034

1389,1521539923,1031,NrfghJ,5,1521540047,1521540055

1395,1521783301,1022,Qxtwii,5,1521783306,1521783312

1395,1521783301,1024,AgHmmw,3,1521783323,1521783332

1395,1521783301,1029,BooDzz,6,1521783347,1521783349

1395,1521783301,1040,KeioUy,1,1521783365,1521783374

1395,1521783301,1031,NrfghJ,5,1521783381,1521783388

1395,1521783301,1051,BpweBd,4,1521783398,1521783403

1395,1521783301,1059,LkxCwu,3,1521783408,1521783414

​Thanks, Gene Maguin

​

===================== 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

[Rich Ulrich]

I don't understand where the great complexity comes in.

There seem to be two choices for how to order the responses - by order

of presentation, which you easily obtain but don't want; or by item_label.

I think that you want to a "line number" for order of presentation

to a person, so you can save that information; then sort by Person

and item_id before doing CasesToVars on Person and item_id.

What am I missing?

--

Rich Ulrich

---

From: SPSSX(r) Discussion <[hidden email]> on behalf of Maguin, Eugene <[hidden email]>
Sent: Monday, September 30, 2019 11:21 AM
To: [hidden email] <[hidden email]>
Subject: (complex) cases to vars problem

 

I'd like some help to find a more efficient way to work a cases to vars problem. I going to present only a simplified version of the actual dataset because the actual dataset has additional but irrelevant complexity. The data are from an EMA (ecological momentary assessment) measure. Briefly, persons have a piece of software installed on their phones that presents a set of survey items to them at multiple times during the day. This is now the simplified version. The survey consists of 7 items. Item 1 is always presented first and the remaining six items are presented in a ramdomized order for each presentation. 

The data are provided to us in long format and a data record has the following structure:

Person, instance, item_id, item_label, response, item_time, response_time. 

The data for one presentation of the survey items consists of 7 records, one for each of the items presented. Call_time is the time of presentation initiation, item_time is the time at which the item was presented, response_time is the time at which the response was entered. Time is UTC seconds. 

Casestovars works exactly as it should; however, the resulting dataset is not correctly structured because column contents are in call_time order. They need to be in a consistent order across records (person-call_time). I have a plan for this, which is basically a programmed casestovars operation using aggregate which is going to be a lot of work (and error prone) as the number of items is not 7 but rather 45. I'm interested in better ideas.

Here is a sample dataset.

person,call_time,item_id,item_label,response,item_time,response_time,seq

1278,1520541611,1022,Qxtwii,8,1520541619,1520541623

1278,1520541611,1031,NrfghJ,6,1520541641,1520541643

1278,1520541611,1024,AgHmmw,5,1520541648,1520541657

1278,1520541611,1051,BpweBd,5,1520541665,1520541672

1278,1520541611,1059,LkxCwu,6,1520541689,1520541691

1278,1520541611,1029,BooDzz,4,1520541704,1520541712

1278,1520541611,1040,KeioUy,2,1520541719,1520541722

1389,1521539923,1022,Qxtwii,7,1521539951,1521539957

1389,1521539923,1024,AgHmmw,7,1521539972,1521539977

1389,1521539923,1029,BooDzz,4,1521539981,1521539984

1389,1521539923,1051,BpweBd,5,1521540003,1521540010

1389,1521539923,1059,LkxCwu,6,1521540018,1521540022

1389,1521539923,1040,KeioUy,6,1521540026,1521540034

1389,1521539923,1031,NrfghJ,5,1521540047,1521540055

1395,1521783301,1022,Qxtwii,5,1521783306,1521783312

1395,1521783301,1024,AgHmmw,3,1521783323,1521783332

1395,1521783301,1029,BooDzz,6,1521783347,1521783349

1395,1521783301,1040,KeioUy,1,1521783365,1521783374

1395,1521783301,1031,NrfghJ,5,1521783381,1521783388

1395,1521783301,1051,BpweBd,4,1521783398,1521783403

1395,1521783301,1059,LkxCwu,3,1521783408,1521783414

​Thanks, Gene Maguin

​

===================== 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

===================== 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

[Maguin, Eugene]

Hi Rich, 

Perhaps i've made the problem more complex than it really is but that is why i'm asking. I hope there's other people on the list who work with EMA data and have encountered this problem. 

I first want to add just two points. A "case" is defined by both person and call_time. Presentation order is recorded by item_time. 

I appreciate your point about the choice between either order by item_id or call_time. It seems like we can have one or the other but not both.

Thank you,

Gene Maguin 

---

From: Rich Ulrich <[hidden email]>
Sent: Monday, September 30, 2019 12:28 PM
To: [hidden email]; Maguin, Eugene
Subject: Re: (complex) cases to vars problem

 

I don't understand where the great complexity comes in.

There seem to be two choices for how to order the responses - by order

of presentation, which you easily obtain but don't want; or by item_label.

I think that you want to a "line number" for order of presentation

to a person, so you can save that information; then sort by Person

and item_id before doing CasesToVars on Person and item_id.

What am I missing?

--

Rich Ulrich

---

From: SPSSX(r) Discussion <[hidden email]> on behalf of Maguin, Eugene <[hidden email]>
Sent: Monday, September 30, 2019 11:21 AM
To: [hidden email] <[hidden email]>
Subject: (complex) cases to vars problem

 

I'd like some help to find a more efficient way to work a cases to vars problem. I going to present only a simplified version of the actual dataset because the actual dataset has additional but irrelevant complexity. The data are from an EMA (ecological momentary assessment) measure. Briefly, persons have a piece of software installed on their phones that presents a set of survey items to them at multiple times during the day. This is now the simplified version. The survey consists of 7 items. Item 1 is always presented first and the remaining six items are presented in a ramdomized order for each presentation. 

The data are provided to us in long format and a data record has the following structure:

Person, instance, item_id, item_label, response, item_time, response_time. 

The data for one presentation of the survey items consists of 7 records, one for each of the items presented. Call_time is the time of presentation initiation, item_time is the time at which the item was presented, response_time is the time at which the response was entered. Time is UTC seconds. 

Casestovars works exactly as it should; however, the resulting dataset is not correctly structured because column contents are in call_time order. They need to be in a consistent order across records (person-call_time). I have a plan for this, which is basically a programmed casestovars operation using aggregate which is going to be a lot of work (and error prone) as the number of items is not 7 but rather 45. I'm interested in better ideas.

Here is a sample dataset.

person,call_time,item_id,item_label,response,item_time,response_time,seq

1278,1520541611,1022,Qxtwii,8,1520541619,1520541623

1278,1520541611,1031,NrfghJ,6,1520541641,1520541643

1278,1520541611,1024,AgHmmw,5,1520541648,1520541657

1278,1520541611,1051,BpweBd,5,1520541665,1520541672

1278,1520541611,1059,LkxCwu,6,1520541689,1520541691

1278,1520541611,1029,BooDzz,4,1520541704,1520541712

1278,1520541611,1040,KeioUy,2,1520541719,1520541722

1389,1521539923,1022,Qxtwii,7,1521539951,1521539957

1389,1521539923,1024,AgHmmw,7,1521539972,1521539977

1389,1521539923,1029,BooDzz,4,1521539981,1521539984

1389,1521539923,1051,BpweBd,5,1521540003,1521540010

1389,1521539923,1059,LkxCwu,6,1521540018,1521540022

1389,1521539923,1040,KeioUy,6,1521540026,1521540034

1389,1521539923,1031,NrfghJ,5,1521540047,1521540055

1395,1521783301,1022,Qxtwii,5,1521783306,1521783312

1395,1521783301,1024,AgHmmw,3,1521783323,1521783332

1395,1521783301,1029,BooDzz,6,1521783347,1521783349

1395,1521783301,1040,KeioUy,1,1521783365,1521783374

1395,1521783301,1031,NrfghJ,5,1521783381,1521783388

1395,1521783301,1051,BpweBd,4,1521783398,1521783403

1395,1521783301,1059,LkxCwu,3,1521783408,1521783414

​Thanks, Gene Maguin

​

===================== 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

===================== 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