Date: Fri, 18 Apr 2008 11:33:25 -0300 Hector Maletta "SPSSX(r) Discussion" Hector Maletta Re: weight by syntax To: "Oliver, Richard" text/plain; charset="us-ascii"

Of course weighting in one variant and not weighting in the other variant will affect the results. The same would happen with any procedure: a weighted crosstabs will be different that an unweighted one. But weighting per se (applied in both variants) would not affect anything regarding the specific problem posed in this thread. Hector

-----Original Message----- From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU] On Behalf Of Oliver, Richard Sent: 17 April 2008 21:13 To: SPSSX-L@LISTSERV.UGA.EDU Subject: Re: weight by syntax

Weight will definitely affect your results unless the value of the weight variable is a constant. Consider the following very simple example:

data list list /x y z. begin data 1 2 1 2 4 1 3 1 5 4 5 1 end data. REGRESSION /DEPENDENT x /METHOD=ENTER y. WEIGHT BY Z. REGRESSION /DEPENDENT x /METHOD=ENTER y.

The WEIGHT command basically says: Treat every case as if there were N cases with these values, where N is the value of the WEIGHT variable. In this case, once you weight the data, there are five cases with a values of 3 and 1 for for x and y respectively instead of just one.

________________________________

From: SPSSX(r) Discussion on behalf of jimjohn Sent: Thu 4/17/2008 4:43 PM To: SPSSX-L@LISTSERV.UGA.EDU Subject: Re: weight by syntax

thanks guys let me try to explain this problem better.

so when i implement the syntax written in my original comment: my output gives me nine different regression outputs, for example: for balbucket 1, i get the following 3 regression outputs: redrate_m vs saving3, redrate_m vs saving3 and saving3_sq, and redrate_m vs saving3, saving3_sq, age

and so i get those same three regressino models for balbucket2 and for balbucket3.

now without using any syntax, i want to get the same regression output result that i got for balbucket 1: redrate_m vs saving3, saving3_sq, and age so i go select cases -> balbucket = 1. then i go to the regression and i should be getting the same output but i dont. here is the syntax that shows up from what i do:

USE ALL. COMPUTE filter_\$=(balbucket = 1). VARIABLE LABEL filter_\$ 'balbucket = 1 (FILTER)'. VALUE LABELS filter_\$ 0 'Not Selected' 1 'Selected'. FORMAT filter_\$ (f1.0). FILTER BY filter_\$. EXECUTE. REGRESSION /MISSING LISTWISE /STATISTICS COEFF OUTS R ANOVA /CRITERIA=PIN(.05) POUT(.10) /NOORIGIN /DEPENDENT redrate /METHOD=ENTER saving3 saving3_sq age.

so im not understanding why i get a different regression output for balbucket1 regression. the one thing i dont really understand in the syntax i used in the beginning was WEIGHT BY so im thinking this must have something to do with why the results are different. any ideas? thanks.

Hector Maletta wrote: > > No time to go into the problem right now, but just an offhand comment: > Weight, I believe, has nothing to do with it. Each case has a weight, and > the weight of a case does not change when you filter cases or split the > file. Filter executes a binary decision (selects a case if the condition > if > the filter variable equals 1, and leaves it out if the filter variable > does > not equal 1). Since BALBUCKET has three possible values, this may be a > source of differences with SPLIT FILE. > Hector > > > -----Original Message----- > From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU] On Behalf Of > jimjohn > Sent: 17 April 2008 17:36 > To: SPSSX-L@LISTSERV.UGA.EDU > Subject: weight by syntax > > can someone plz help me with this, im trying to understand this syntax: i > thought this is just a multiple regression model with redrate_m vs > (saving3, > saving3_sq, age) for different values of balbucket (either 1, 2, or 3). > but > if i just filter balbucket and run the regression model i get different > output. so im just wondering what does this WEIGHT BY rembal do to change > my > regression output? thanks so much. > > WEIGHT > BY rembal . > SORT CASES BY balbucket . > SPLIT FILE > LAYERED BY balbucket . > REGRESSION > /MISSING LISTWISE > /STATISTICS COEFF OUTS R ANOVA CHANGE > /CRITERIA=PIN(.05) POUT(.10) > /NOORIGIN > /DEPENDENT redrate_m > /METHOD=ENTER saving3 /enter saving3_sq /enter age . > -- > View this message in context: > http://www.nabble.com/weight-by-syntax-tp16754311p16754311.html > Sent from the SPSSX Discussion mailing list archive at Nabble.com. > > ===================== > To manage your subscription to SPSSX-L, send a message to > LISTSERV@LISTSERV.UGA.EDU (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@LISTSERV.UGA.EDU (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://www.nabble.com/weight-by-syntax-tp16754311p16756378.html Sent from the SPSSX Discussion mailing list archive at Nabble.com.

===================== To manage your subscription to SPSSX-L, send a message to LISTSERV@LISTSERV.UGA.EDU (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@LISTSERV.UGA.EDU (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@LISTSERV.UGA.EDU (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

Back to: Top of message | Previous page | Main SPSSX-L page