Hello Raquel Garcia. It will be much easier to give advice if you provide a small dataset showing how your data is set up. Judging from what you said about the members of pairs sharing the same ID number (Nomem_encr), I think you probably have a LONG file structure, with one row per person, so two rows per pair. But even if that is true, there are other unanswered questions. E.g.,

• How many rows per ID number are possible?
• What are the other possible values for the position variable (if any)?

Thank you for clarifying.

To filter out unwanted pairs in SPSS, go to "Select Cases" under the "Data" menu. Use a condition like "(position = 1 AND LAG(position) = 5) OR (position = 5 AND LAG(position) = 1)" to keep only the desired pairs where one member is a household head and the other is a residing child. Apply the filter to retain these pairs while excluding others.

Hello Hewa,

Thank you for your response

I tried this, but now all IDs with 1's were filtered together and IDs with 5's were filtered together, so it didn't check if they had the same ID (household number). Can I perhaps add that, something like if they have the same household number and then add this afterward: (position = 1 AND LAG(position) = 5) OR (position = 5 AND LAG(position) = 1)? But how do type 'same householdnummer' in SPSS language?

Raquel Garcia, you "liked" my earlier response, but you did not answer my questions about the structure of your data file. So what I'm posting here is based on my guess about the structure. Assuming it is a LONG file structure, I would use AGGREGATE to add some new variables to the file, as shown below. HTH.

* Generate some fake data to illustrate.

* Raquel has not yet answered my question

* about how the data file is structured.

* I'll assume it is a LONG file with at

* most 2 rows per ID code, and with position

* values of 1 and 5 only. The general approach

* of using AGGREGATE should work, but the exact

* approach may have to be tweaked if the data

* file structure is different from what I guessed.

NEW FILE.

DATASET CLOSE ALL.

DATA LIST LIST / Nomem_encr position (2F5.0).

BEGIN DATA

1 1

1 1

2 5

2 5

3 1

3 5

4 1

5 5

END DATA.

* Use AGGREGATE to add new variables to the dataset.

SORT CASES BY Nomem_encr position.

AGGREGATE

/OUTFILE=* MODE=ADDVARIABLES

/BREAK=Nomem_encr

/position_min=MIN(position)

/position_max=MAX(position)

/numrecs=NU.

COMPUTE wanted = (position_min EQ 1) AND (position_max EQ 5).

FORMATS wanted(F1).

SELECT IF wanted.

LIST.

Hello Bruce Weaver,

I am very sorry I didn't reply earlier. You mentioned that you wanted a shorter version of my data, and I wasn't sure how to do that. Now, I understand what you meant from your response. I followed your advice, and it worked perfectly. Thank you very much for your help and information.

No worries, Raquel Garcia. Good to hear it worked for you.

Would you perhaps be willing to help me Bruce Weaver ?

Further along in the process, I discovered that I made a mistake. There are multiple pairs possible with the child living at home (5), with positions such as head of household (1), married partner (2), unmarried partner (3), parent-in-law (4), housemate (6), or household member (7). Now, sometimes I also have more than two members of the household. Ideally, I have one member who is 1, 2, 3, 4, 6, or 7, and the other must be a child living at home. So, can I still use the above with adjustments? And how do I ensure that I then have two members of the household left?"

Hello Raquel Garcia. Judging by the scenarios you showed in that small dataset, I think that all you need to do is add some code to keep only the rows that have position = 1 or 5, and then use the approach shown before. For the data file you uploaded, this works:

NEW FILE.

DATASET CLOSE ALL.

GET FILE='C:\Temp\Garcia_exampledatasethousehold.sav'.

FREQUENCIES position.

* Keep only records with position = 1 or 5.

SELECT if ANY(position,1,5).

FREQUENCIES position.

* Now use the AGGREGATE method, as before.

SORT CASES BY nohouse_nem position.

AGGREGATE

/OUTFILE=* MODE=ADDVARIABLES

/BREAK=nohouse_nem

/position_min=MIN(position)

/position_max=MAX(position)

.

* Keep only the pairs with both 1 and 5.

SELECT IF (position_min EQ 1) AND (position_max EQ 5).

LIST nohouse_nem position.

OUTPUT from the LIST command:

nohouse_nem position

1 1

1 5

2 1

2 5

3 1

3 5

Number of cases read: 6 Number of cases listed: 6

Hello Bruce Weaver,

I actually mean that the pairs 1-5, 2-5, 3-5, 4-5, 6-5, 7-5 are possible, and not only 1-5, how can I apply that?

Can i do this:

COMPUTE wanted =(position_min EQ 1) AND (position_max EQ 5) OR (position_min EQ 2) AND (position_max EQ 5) OR (position_min EQ 3) AND (position_max EQ 5) OR (position_min EQ 5) AND (position_max EQ 6) OR (position_min EQ 5) AND (position_max 7).

Hello Raquel Garcia. In your sample data, some ID numbers can have 3 or more of those position values. E.g., if I run your latest computation of "wanted", I get this:

nohouse_nem position wanted

1 1 1

Dear Bruce Weaver

Really great that you're helping me with this!

I wanted two records of the same household ID. I'd like pairs of parents and their children living at home, preferably pairs 1-5 as indicated by the arrows.

However, it can also be with 2 and 3 like this:

nohouse_nem position wanted

1 2

Hmm. Having households with 2 or more children, and therefore with 2 or more pairs complicates things. If 2 children and 2 pairs is the max, I think the following code gets you most of the way to what you want. I added another household to the data file you uploaded earlier with two children and 2 pairs.

NEW FILE.

DATASET CLOSE ALL.

GET FILE='C:\Temp\Garcia_exampledatasethousehold.sav'.

LIST nohouse_nem position wanted.

FREQUENCIES position.

SORT CASES BY nohouse_nem position.

AGGREGATE

/OUTFILE=* MODE=ADDVARIABLES

/BREAK=nohouse_nem

/p1=CIN(position 1 1)

/p2=CIN(position 2 2)

/p3=CIN(position 3 3)

/p4=CIN(position 4 4)

/p5=CIN(position 5 5)

/p6=CIN(position 6 6)

/p7=CIN(position 7 7)

.

FORMATS p1 to p7 (F1).

NUMERIC wanted (F1).

* First, get rid of households where there are no position=5 records.

SELECT IF p5 GT 0.

LIST.

DO REPEAT

flag = flag15 flag25 flag35 flag56 flag57 /

c1 = p1 p2 p3 p5 p5 /

c2 = p5 p5 p5 p6 p7 /

a = 1 2 3 5 5 /

b = 5 5 5 6 7.

DO IF missing(wanted).

*COMPUTE flag = (c1 GT 0) AND (c2 GT 0).

COMPUTE flag = (c1 EQ 1) AND (c2 EQ 1).

IF flag wanted = ANY(position,a,b).

END IF.

END REPEAT.

FORMATS flag15 to flag57 (F1).

LIST nohouse_nem position wanted.

DO IF missing(wanted) and p5 EQ 2.

IF ANY(2,SUM(p1,p2),SUM(p1,p3),SUM(p1,p5),SUM(p1,p7)) wanted = 1.

END IF.

LIST nohouse_nem position wanted.

SELECT if wanted.

LIST nohouse_nem position wanted.

Here is the output for that final LIST command:

nohouse_nem position wanted

1 1 1

1 5 1

2 1 1

2 5 1

3 1 1

3 5 1

4 5 1

4 6 1

6 3 1

6 5 1

7 1 1

7 2 1

7 5 1

7 5 1

Number of cases read: 14 Number of cases listed: 14

QUESTION: How do you intend to generate the two pairs for household 7 here? Which 5 gets paired with 1, and which with 2?

it worked! Thank you!

I was thinking of randomly pairing those two. What do you think? Or paired sequentially, where the first 5 in the series are matched with 1 and the next 5 with 2.

Do you happen to know how I can actually match the pairs in my dataset, so that I can see that the household leader (position=1) has reported making debts (finances=4), and then observe that the child living at home has a low score in self-esteem. Because my research question is: 'To what extent is the association between financial stress and self-esteem mediated by the sense of connectedness with others? So, I want to understand the financial situation of the family to be able to draw comparisons with self-esteem.