selective weighting
Locked
 
|
Hi everyone,
I browsed the archive section and couldn't find an answer to my question (unless I missed it). I would like to selectively apply weights to certain variables and not others. Can anyone suggest as to how it can be done? Thanks, Greg |
|
|
Can you be more specific or provide some examples?
-----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of grigoris Sent: Sunday, September 21, 2008 10:39 AM To: [hidden email] Subject: selective weighting Hi everyone, I browsed the archive section and couldn't find an answer to my question (unless I missed it). I would like to selectively apply weights to certain variables and not others. Can anyone suggest as to how it can be done? Thanks, Greg -- View this message in context: http://www.nabble.com/selective-weighting-tp19596012p19596012.html Sent from the SPSSX Discussion mailing list archive at Nabble.com. ===================== 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 |
|
In reply to this post by Greg
Please explain in more detail what you would like to do.
Are you creating summative scales? Do you want to weight cases while doing analysis on some variables, but not when doing analysis on other? Art Kendall Social Research Consultants grigoris wrote: > Hi everyone, > > I browsed the archive section and couldn't find an answer to my question > (unless I missed it). > > I would like to selectively apply weights to certain variables and not > others. Can anyone suggest as to how it can be done? > > Thanks, > Greg > -- > View this message in context: http://www.nabble.com/selective-weighting-tp19596012p19596012.html > Sent from the SPSSX Discussion mailing list archive at Nabble.com. > > ===================== > 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
Art Kendall
Social Research Consultants |
|
|
In reply to this post by Greg
I apologize for the broad question.
I have two datasets from the census, one is a 5% sample with individual-level data, and the other is from the 100% summary files, aggegrate level data. I merged some variables from the 5% to the aggegrate one. However, i want to weight the variables from the sample so their totals can match thosefrom the 100% file. Specifically, the 100% has the total pop for metro area, and by merging the # whites from the 5% sample, I can compute, let's say, white's proportion within each metro area. However, I want to weight the pop of whites-I have the weights-in order to calculate their proportion. I tried weighting and saving the variables from the 5% sample before merging, but when I merged the numbers went back to pre-weighting state. Therefore, what I was hoping in doing is to slectively weight the white pop in the merged file, but cannot make it to work. (I'm not sure if this is related to the link I used to merge both files. Each file has a metro area code and used it to merge them.) Any advice would be greatly appreciated! greg
|
|
|
First be sure that you have sample counts and not expansion estimates
from the 5% sample. If you need just the point estimates (totals, percents, means, etc.) and not the interval estimates (SDs, SEs, significance, etc.) do the weighting and aggregation on the 5% file and then do the merge. For percentages/proportions, the SD is determined by what the percentage/proportion is. Art Kendall Social Research Consultants grigoris wrote: > I apologize for the broad question. > > I have two datasets from the census, one is a 5% sample with > individual-level data, and the other is from the 100% summary files, > aggegrate level data. > > I merged some variables from the 5% to the aggegrate one. However, i want to > weight the variables from the sample so their totals can match thosefrom the > 100% file. > > Specifically, the 100% has the total pop for metro area, and by merging the > # whites from the 5% sample, I can compute, let's say, white's proportion > within each metro area. > > However, I want to weight the pop of whites-I have the weights-in order to > calculate their proportion. I tried weighting and saving the variables from > the 5% sample before merging, but when I merged the numbers went back to > pre-weighting state. Therefore, what I was hoping in doing is to slectively > weight the white pop in the merged file, but cannot make it to work. > > (I'm not sure if this is related to the link I used to merge both files. > Each file has a metro area code and used it to merge them.) > > Any advice would be greatly appreciated! > > greg > > > > grigoris wrote: > >> Hi everyone, >> >> I browsed the archive section and couldn't find an answer to my question >> (unless I missed it). >> >> I would like to selectively apply weights to certain variables and not >> others. Can anyone suggest as to how it can be done? >> >> Thanks, >> Greg >> >> > > -- > View this message in context: http://www.nabble.com/selective-weighting-tp19596012p19614371.html > Sent from the SPSSX Discussion mailing list archive at Nabble.com. > > ===================== > 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
Art Kendall
Social Research Consultants |
|
|
Thought more about it. you probably want to have a file with expanded
values before doing the match In which case you can use AGGREGATE on the weighted file or you can use something like this DO REPEAT X = x01 x02 x13 x15 x19 to x30/expandedx = newx01 newx02 newx13 newx15 x19 to newx30. compute expandedx= wgt* X. end repeat. Art Art Kendall wrote: > First be sure that you have sample counts and not expansion estimates > from the 5% sample. > > If you need just the point estimates (totals, percents, means, etc.) and > not the interval estimates (SDs, SEs, significance, etc.) do the > weighting and aggregation on the 5% file > and then do the merge. > > For percentages/proportions, the SD is determined by what the > percentage/proportion is. > > Art Kendall > Social Research Consultants > > grigoris wrote: >> I apologize for the broad question. >> >> I have two datasets from the census, one is a 5% sample with >> individual-level data, and the other is from the 100% summary files, >> aggegrate level data. >> >> I merged some variables from the 5% to the aggegrate one. However, i >> want to >> weight the variables from the sample so their totals can match >> thosefrom the >> 100% file. >> >> Specifically, the 100% has the total pop for metro area, and by >> merging the >> # whites from the 5% sample, I can compute, let's say, white's >> proportion >> within each metro area. >> >> However, I want to weight the pop of whites-I have the weights-in >> order to >> calculate their proportion. I tried weighting and saving the >> variables from >> the 5% sample before merging, but when I merged the numbers went back to >> pre-weighting state. Therefore, what I was hoping in doing is to >> slectively >> weight the white pop in the merged file, but cannot make it to work. >> >> (I'm not sure if this is related to the link I used to merge both files. >> Each file has a metro area code and used it to merge them.) >> >> Any advice would be greatly appreciated! >> >> greg >> >> >> >> grigoris wrote: >> >>> Hi everyone, >>> >>> I browsed the archive section and couldn't find an answer to my >>> question >>> (unless I missed it). >>> >>> I would like to selectively apply weights to certain variables and not >>> others. Can anyone suggest as to how it can be done? >>> >>> Thanks, >>> Greg >>> >>> >> >> -- >> View this message in context: >> http://www.nabble.com/selective-weighting-tp19596012p19614371.html >> Sent from the SPSSX Discussion mailing list archive at Nabble.com. >> >> ===================== >> 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 > > ===================== 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
Art Kendall
Social Research Consultants |
|
|
I appreciate your help.
This is what I did: I weighted the 5% file, then used aggregate, as you mentioned. Afterwards, I merged it with the 100% file. Greg
|
«
Return to SPSSX Discussion
|
1 view|%1 views
| Free forum by Nabble | Edit this page |