PWEIGHT= allows for differential weighting of persons. The standard weights are 1 for all persons. To change the weighting of persons, specify IWEIGHT=.
PWEIGHT of 2 has the same effect on item estimation as putting the person and responses into the analysis twice. It does not change a person's responses scored 0-1 into responses scored 0-2. When PWEIGHT is 0 for a person, the measure and fit statistics are reported for the person, but the person does not influence the measures or fit statistics of the other persons or items. PWEIGHT= applies to everything except the dimensionality computations (Tables 23, 24) where PWEIGHT= is set to 1 and IWEIGHT= is set to 1 for all persons and items.
Raw score, count, and standard error of measurement reflect the absolute size of weights as well as their relative sizes. Measure, infit and outfit and correlations are sensitive only to relative weights. If you want the standard error of the final weight-based measure to approximate the S.E. of the unweighted measure, then ratio-adjust case weights so that the total of the weights is equal to the total number of independent observations.
PWEIGHT= file name
|
weights are in a text file of format:
(person entry number) (weight)
|
PWEIGHT = *
(person entry number) (weight)
1 0.5
2 0
*
|
in-line list of person-entry numbers and weights
person entry number 1 is weighted 0.5
person entry number 2 is weighted 0.Person 2 is reported, but it is not used to estimate item measures
|
PWEIGHT = $S2W3
PWEIGHT = $S2E4
|
weights in person labels which start at NAME1= using the column selection rules, e.g., starting in column S... with a width of W... or starting in column S and ending in column E.
This can be expanded, e.g, PWEIGHT = $S23W1+"."+$S25W2 places the columns next to each other (not added to each other)
|
PWEIGHT = $C2W3
PWEIGHT = $C2E4
|
weights in data records using the column selection rules. column1 of the data recode is C1. The format is starting in column C... with a width of W... or starting in column C... and ending in column E. The columns can be anywhere in the data record
|
Example 1:
In a sample of 20 respondents, person 1 is to be given a weight of 2.5, all other persons have a weight of 1.
PWEIGHT=*
1 2.5
2-20 1
*
A better weighting, which would make the reported item standard errors more realistic by maintaining the original total sum of weights at 20 , is:
PWEIGHT=*
1 2.33 ; 2.5 * 0.93
2-20 0.93 ; the sum of all weights is 20.0
*
+---------------------------------------------------------------------------------+
|ENTRY RAW MODEL| INFIT | OUTFIT |PTMEA| | |
|NUMBER SCORE COUNT MEASURE S.E. |MNSQ ZSTD|MNSQ ZSTD|CORR.|WEIGH| KID |
|------------------------------------+----------+----------+-----+-----+----------|
| 1 4 10 -1.20 .85| .65 -.9| .41 .3| .82| 2.33| Richard M|
| 2 7 10 1.70 1.17| .18 -1.2| .10 -.4| .94| .93| Tracie F|
| 3 7 10 1.70 1.17|2.04 1.2|1.32 .7| .84| .93| Walter M|
Example 2:
The data records contain the weights in columns 16-18 of the record. The person label starts in column 15, so this is also column 2 of the person label
NAME1 = 15 ; start of person label
NAMELEN = 20 ; length of person label
PWEIGHT= $S2W3 ; location in person label
&END
...
END NAMES
10110110011001 0.5 Person A
01001100011011 0.7 Person B
......
or
NAME1 = 15 ; start of person label
NAMELEN = 20 ; length of person label
PWEIGHT= $C16W3 ; or $C16E18 ; column in data record also in person label
or
NAME1 = 20 ; start of person label in column 20 excluding the person weight
NAMELEN = 15 ; length of person label
PWEIGHT= $C16W3 ; or $C16E18 ; column in data record, not in person label
Example 3:
Person 4 is a dummy case, to be given weight 0. The measure and fit statistics are reported for person 4, but person 4 does not influence the measures or fit statistics of the other persons or items.
PWEIGHT=*
4 0 ; Person 4 has weight 0, other persons have standard weight of 1.
*
