Dates in Factor Analysis?
Locked
 
|
I'm using some survey data about the housing decision making process. One key variable I'd like to use is the date household's started looking for their property (as housing markets are not static). Is it possible to use dates in factor analysis?
Thanks, Rich |
|
Please give us more detail about what you are using
the factor analysis for.
Art Kendall Social Research Consultants On 5/29/2012 6:24 AM, RJDunning wrote: ===================== 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 REFCARDI'm using some survey data about the housing decision making process. One key variable I'd like to use is the date household's started looking for their property (as housing markets are not static). Is it possible to use dates in factor analysis? Thanks, Rich -- View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Dates-in-Factor-Analysis-tp5713414.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
Art Kendall
Social Research Consultants |
Administrator
|
I am puzzled as to what using a date variable in FA is going to contribute. If you think it will somehow provide any sort of insight into the dynamic nature of housing market by simply adding it into the factor analysis is likely to prove disappointing at best and probably misleading. FA is used to reduce the dimensionality of high dimensional data and assumes that subsets of variables entering into such analysis are linearly related to one another to some extent. What 'dimension(s)' of your data have to do with the key date variable? What sort of theoretical relationships can you posit to tell a good story?
Please reply to the list and not to my personal email.
Those desiring my consulting or training services please feel free to email me. --- "Nolite dare sanctum canibus neque mittatis margaritas vestras ante porcos ne forte conculcent eas pedibus suis." Cum es damnatorum possederunt porcos iens ut salire off sanguinum cliff in abyssum?" |
|
|
In reply to this post by RJDunning
I cannot see how dates can be used as they are, unless there is a steady trend in the housing market, which seems likely only with very small timespans. However, you might be able to find some other data connected to dates like: populations expectation to their future level of funds, average housing price, months since housing prices were significantly reduced (or number of continuous months of rising housing prices), percent increase in average housing prices during last 12 months, level of mortgage rate, unemployment rate or something else, and add these macro data to your individual level datafile, and then use factor analysis to find out if and how these economic considerations enter into your respondents' decisionmaking.
Best, Eero Olli -----Opprinnelig melding----- Fra: David Marso [mailto:[hidden email]] Sendt: 29. mai 2012 17:29 Emne: Re: Dates in Factor Analysis? I am puzzled as to what using a date variable in FA is going to contribute. If you think it will somehow provide any sort of insight into the dynamic nature of housing market by simply adding it into the factor analysis is likely to prove disappointing at best and probably misleading. FA is used to reduce the dimensionality of high dimensional data and assumes that subsets of variables entering into such analysis are linearly related to one another to some extent. What 'dimension(s)' of your data have to do with the key date variable? What sort of theoretical relationships can you posit to tell a good story? Art Kendall wrote > > Please give us more detail about what you are using > the factor analysis for. > Art Kendall > Social Research Consultants > > On 5/29/2012 6:24 AM, RJDunning wrote: > > I'm using some survey data about the housing decision making > process. One key variable I'd like to use is the date household's > started looking for their property (as housing markets are not > static). Is it possible to use dates in factor analysis? > > Thanks, > Rich > > -- > View this message in context: > http://spssx-discussion.1045642.n5.nabble.com/Dates-in-Factor-Analysis > -tp5713414.html Sent from the SPSSX Discussion mailing list archive at > Nabble.com. > > ===================== > To manage your subscription to SPSSX-L, send a message to > LISTSERV@.UGA (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 > LISTSERV@.UGA (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 > -- View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Dates-in-Factor-Analysis-tp5713414p5713421.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 |
|
|
In reply to this post by RJDunning
It strikes me as being a little bit similar to adding in
a variable like a person's Age, to a personality inventory - but more complicated. For Age, the result is that factors do more to describe Age, than Age does to describe a factor. Housing markets are not static, but a linear trend in change over time is not an "effect" that is readily interpretable. I would suggest that, using the insight that markets are not stable, you would do better to insert "Present Interest Rate", or other indexes of changes. Or use a dummy-indicator variable, Yes/No, to show that a specific condition existed. -- Rich Ulrich > Date: Tue, 29 May 2012 03:24:20 -0700 > From: [hidden email] > Subject: Dates in Factor Analysis? > To: [hidden email] > > I'm using some survey data about the housing decision making process. One key > variable I'd like to use is the date household's started looking for their > property (as housing markets are not static). Is it possible to use dates in > factor analysis? ... |
|
|
In reply to this post by RJDunning
What about using discriminant analysis where the groups to be explained
are aggregations of dates (week1, week2, ...). But a caveat : your survey data are not statistically independent : a housing market value in week5 correlates more or less to its value in week6. And the housing values in area 5.1 which is contiguous to area 5.2 autocorrelate, too. So you should look into spatial and time auto correlation models before starting to find substantial structures. Cheers, F. Thomas On 29/05/2012 12:24, RJDunning wrote: > I'm using some survey data about the housing decision making process. One key > variable I'd like to use is the date household's started looking for their > property (as housing markets are not static). Is it possible to use dates in > factor analysis? > > Thanks, > Rich > > -- > View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Dates-in-Factor-Analysis-tp5713414.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 |
«
Return to SPSSX Discussion
|
1 view|%1 views
| Free forum by Nabble | Edit this page |