[求助]怎样理解因子得分?

是 否 +2 论坛币

k人 参与回答

经管之家送您一份

应届毕业生专属福利!

求职就业群

赵安豆老师微信：zhaoandou666

经管之家联合CDA

送您一个全额奖学金名额~ !

立即领取

感谢您参与论坛问题回答

经管之家送您两个论坛币！

+2 论坛币

请教各位达人,在进行因子分析时,通过自动保存所得到的因子得分具体表示什么含义?有哪些分析用途?等待赐教!

谢谢!!

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

分享0 收藏0 回帖

关键词：因子得分 自动保存 因子分析 得分

The estimated values of the factors, called the factor scores, may also be useful in the interpretation as well as in the
diagnostic analysis. To be more precise, the factor scores are estimates of the unobserved random vectors Fl, l = 1, . . . , k, for each individual xi, i = 1, . . . , n. Johnson and Wichern (1998) describe three methods which in practice yield very similar results

[此贴子已经被作者于2008-2-14 4:58:32编辑过]

Factor Scores Aren't Sacred: Comments on "Abuses of Factor Scores"
Robert F. Schweiker
American Educational Research Journal, Vol. 4, No. 2 (Mar., 1967), pp. 168-170
doi:10.2307/1162125
This article consists of 3 page(s).

The relationships among principal component, image component, three types of factor
score estimates, and a scale score method were compared over different levels of variables
(p), saturations (a_ij), sample sizes (N), variable to component ratios (p/m), and pattern
rotations. Scores were compared on both same (convergent) components and different
(divergent) components. There were virtually no overall differences among score methods.
The average correlation between matched scores across all conditions was .98. Comparisons
within methods, that is, among component scores or among factor scores, generally were
slightly higher than comparisons between methods. The scale score method, while
correlating slightly lower (overall .96), was still highly correlated with both component and
factor scores. When a score did depart from another, it usually occurred in one of the
conditions of low component saturation, low sample size, low p/m ratio, or a combination
of these conditions.

Factor analysis and component analysis, which includes principal
component analysis ( Hotelling, 1933) and image component analysis ( Guttman,
1953; Harris, 1962), are competing methods used in data reduction. Both
factor analysis and component analysis serve the same two broad purposes.
The first purpose involves pattern interpretation to determine which variables
are related to each other. The pattern relates the original p variables to m new
variables (m < p). The m new variables are called components or factors
depending on the method. Previous research ( Velicer, 1977; Velicer & Fava,
1987, 1992; Velicer & Jackson, 1990; Velicer, Peacock, & Jackson, 1982) has
found very little practical difference between patterns derived by the different
methods. The second purpose concerns a question of parsimony. In this case,
the p observed scores are replaced by m new scores.

A problem involved in the calculation of scores is selecting among various
alternative score methods. These score methods include: (a) two types of
component scores, that is, principal component and image component scores,
(b) as many as five types of factor scores ( Harris, 1967), and (c) scale scores,

感谢各位达人指点,但你们的英文方式回答,我读起来确实有点困难(不好意思,英语水平不高),

不知能否用中文回答?再谢各位!

[em04][em04][em04][em04]

也要用到，关注

One strength of the factor analytic measurement model is the ability to "purify" the measures by distinguishing between variance related to the common factor(s) and variance due to measurement error. The question then arises, can the researcher use the measurement model to assign scores to individual cases or respondents on the factors? For example, suppose a school headmaster develops a model of "teaching effectiveness,"based on data collected from the school's teachers. Can the headmaster then use the model's parameter estimates, and the manifest variable scores, to derive "factor scores" for the purpose of grading teaching performance?

Bollen indicates that one way to approach the problem (although other approaches exist) is to set up a hypothetical regression of the common factors () on the manifest variables:

 

One may then estimate the "B" regression parameter in the usual way, as the ratio of the covariance of and X over the variance of X. This leads to the formula:

 

where is the covariance matrix of the common factors, is the matrix of loadings, is the model-implied covariance matrix of the manifest variables, and ^ indicates that the matrices are based on estimated parameters.

Factor scores are problematic for several reasons. Besides the existence of competing methods for deriving the scores, the principle of factor indeterminacy suggests that different rotations of the factor solution could lead to different factor scores. While the different sets of factor scores that result may be highly correlated, there is no guarantee that the relative positions of the cases or individuals on the factor score continuum will be identical.

In addition, the measurement model indicates that the manifest variables are functions of both the common factors and the measurement error terms. The factor score equation constructs the factor scores as weighted composites of the manifest variables. Thus, the factor scores are also influenced by measurement error. For this reason, a factor score should not be considered a perfect measure of the factor itself.

This dilemma highlights one of the relative strengths of partial least squares (PLS). In PLS, the "latent variables" are weighted composites of the manifest variables. So the PLS approach leads directly to explicit factor scores. Some researchers and practitioners find this to be one of the most attractive features of PLS.

[此贴子已经被作者于2008-2-16 22:52:34编辑过]

Encyclopedia of Biostatistics
Factor Scores
Standard Article
Ralph B. D'AGOSTINO SR1 and Heidy K. Russell1
1Boston University, Boston, MA, USA
Copyright © 2005 John Wiley & Sons, Ltd. All rights reserved.
 
Abstract
After a set of interpretable factor loadings is determined, a next possible step is to compute factor scores. Exact factor scores can be obtained for a principal components analysis model. Methods that compute factor scores in the common factor analysis model are also presented. In general, there is limited value in the use of factor scores, for even though the factors are interpretable, the estimated factor scores are not well determined.

 

[此贴子已经被作者于2008-2-16 22:49:29编辑过]

Hi

Does anyone know how to quickly calculate factor scores for cases?

I've used principal components analysis in JMP to identify factors (which
are meaningful) as well as finding the rotated components.  JMP kindly
provides the Std Score Coefficients, (they are the same as Factor Score
coefficients in SPSS I believe), which are then used to calculate factor
scores for cases.  Is there an easy way to do this?

Right now I'm creating new columns and writing the formula for each factor
score.  There surely is an easier way?

(One other question which I'm not certain of:  Do I need to standardize the
values of the original cases before calculating the factor scores?  If so,
is this done automatically in some way or do I need to do that?)

Appreciate your help...

1.如果是主成分分析，综合得分是自己算的，即factor做完之后（得选在factor anaylsis界面选中scores中的display factor score……才会出来这个矩阵）因子载荷矩阵下面那个带score的的表格就是计算主成分得分的系数矩阵，然后将原始数据标准（用描述性统计分析就能直接得到）化后的结果带入方程式（方程系数就是系数矩阵，这个过程得自己算），得到各个主因子的综合的得分，若要计算综合得分，则需要在写一个方程式，Y=Y1*a+y2*b……，y1,y2……为各个主成分得分，a,b……为各个主成分的发差贡献率，在特征值那表里头了。最后得到Y即为综合得分。
2.要是因子分析算因子得分就简单了，直接在scores中选中save as variable，那么在表格中直接就会多出来一列得分变量了，即为因子得分。
    在说说他俩区别吧，主成分就一个用途，那就是排序，比如算哪个城市发展的好可以用，主成分没有含义。因子分析可以分析出来各个因子代表什么，比如算影响各个城市发展的主要因素是什么。共同点就是在SPSS中操作的过程是一样。