Upload
hoangnhan
View
215
Download
0
Embed Size (px)
Citation preview
DOCUMENT RESUME
ED 476 430 TM 034 933
AUTHOR Mumford, Karen R.; Ferron, John M.; Hines, Constance V.;Hogarty, Kristine Y.; Kromrey, Jeffery D.
TITLE Factor Retention in Exploratory Factor Analysis: A Comparisonof Alternative Methods.
PUB DATE 2003-04-00NOTE 70p.; Paper presented at the Annual Meeting of the American
Educational Research Association (Chicago, IL, April 21-25,2003) .
PUB TYPE Reports Research (143) Speeches/Meeting Papers (150)EDRS PRICE EDRS Price MF01/PC03 Plus Postage.DESCRIPTORS Comparative Analysis; *Factor Structure; Monte Carlo Methods;
SimulationIDENTIFIERS *Exploratory Factor Analysis; Parallel Analysis (Horn)
ABSTRACT
This study compared the effectiveness of 10 methods ofdetermining the number of factors to retain in exploratory common factoranalysis. The 10 methods included the Kaiser rule and a modified Kaisercriterion, 3 variations of parallel analysis, 4 regression-based variationsof the scree procedure, and the minimum average partial procedure. Theperformance of these procedures was evaluated based on the average number offactors retained by each method, the proportion of samples retaining the samenumber of factors retained by each method, the proportion of samplesretaining the same number of factors as the true number of factors in thepopulation, and the proportion of samples retaining the same number offactors when a particular rule of thumb is applied to the population. Theperformance of the 10 procedures was investigated using Monte Carlo methodsin which random samples were generated under known and controlled populationconditions. Results clearly suggest that both the choice of method and thedesign of the factor analytic study play crucial roles in retaining thecorrect number of common factors. In terms of overall accuracy across theconditions examined in this study, one of the parallel analysis approaches(Montanelli, and Humphreys, 1976) provided the largest proportion of samplesretaining the correct value. Conditions under which other approaches may workwell are discussed. (Contains 16 tables, 18 figures, 25 charts, and 44references.) (SLD)
Reproductions supplied by EDRS are the best that can be madefrom the on inal document.
Factor Retention in Exploratory Factor Analysis:
A Comparison of Alternative Methods
Karen R. Mumford
John M. Ferron
Constance V. Hines
Kristine Y. Hogarty
Jeffrey D. Kromrey
University of South Florida
U.S. DEPARTMENT OF EDUCATIONOffice of Educational Research and Improvement
EDU ATIONAL RESOURCES INFORMATIONCENTER (ERIC)
This document has been reproduced asreceived from the person or organizationoriginating it.
Minor changes have been made toimprove reproduction quality.
Points of view or opinions stated in thisdocument do not necessarily representofficial OERI position or policy.
1
Factor Retention1
PERMISSION TO REPRODUCE ANDDISSEMINATE THIS MATERIAL HAS
BEEN GRANTED BY
K. Mumford
TO THE EDUCATIONAL RESOURCESINFORMATION CENTER (ERIC)
Paper presented at the annual meeting of the American Educational Research Association, April 21 25,
2003, Chicago
2BEST COPY AVAILABLE
Factor Retention2
Factor Retention in Exploratory Factor Analysis:
A Comparison of Alternative Methods
This study extends previous investigations of the effectiveness of various
procedures for determining the number of factors to retain in exploratory common
factor analysis. In our previous Monte Carlo study on the quality of factor analytic
solutions (Ferron, Kromrey, Hogarty, Hines, & Mumford, 2002), we found that decision
rules used to make a determination of the number of factors to retain, in most cases,
yielded underestimates of the known number of factors in the population. Zwick and
Velicer (1986) suggest that the determination of the number of factors or components to
retain is likely to be the most important decision that the researcher makes in the
conduct of a factor analytic study. Researchers conducting exploratory factor analyses
in most instances do not know the true number of factors that are expected to underlie
the data, and may make decisions resulting in the retention of too few or too many
factors. Since the number-of -factors decision is made prior to the factor rotation stage,
it subsequently impacts the results of the factor analysis, such as rotated factor patterns,
factor score estimates, and the interpretability of the factors. Consequently, several
researchers (e.g., Rummel, 1970; Fava Sr Velicer, 1996; Wood, Tataryn, & Gorsuch, 1996)
have emphasized the importance of extracting the correct number of factors for rotation.
As Turner (1998) and others note, the interpretation of the factors is based on the
assumption that the researcher has extracted the correct number of factors.
The problems of underfactoring (i.e., extracting too few factors) and
overfactoring (i.e., extracting too many factors) have been addressed in the literature
(see for example, Crawford, 1975; Fava Sr Velicer, 1992, 1996; Gorsuch, 1983;
MacCallum, Widaman, Preacher & Hong, 2001; Mosier, 1939; Turner, 1998; Wood,
Tataryn & Gorusch, 1996; Zwick & Velicer, 1986). In the case of underfactoring,
retaining too few factors for inclusion in the rotation phase can result in the loss of
potentially important information (e.g., a substantive factor or factors) in the factor
solution. With underfactoring, it has been argued, the true factors in a data set cannot
be accurately portrayed (Cattell, 1958; Comrey, 1978; Fava & Velicer, 1996). Wood et al.
3
Factor Retention3
(1996) note that when underfactoring occurs, the estimated factors are likely to contain
considerable error. In their investigation of the consequences of underfactoring in both
maximum likelihood factor analysis and principal component analysis, Fava and
Velicer (1996) found "severe degradation of factor score estimates" with underfactoring.
It was noted that the principal component score degraded less rapidly than the factor
score within methods.
Most researchers concur that overfactoring is less a problem than underfactoring.
Fava and Velicer (1992) and others offer two theoretical justifications that support this
point of view. The first, they posit, is based on the fact that each subsequent factor
extracted accounts for less variance than the factor extracted prior to it. The second
relates to the notion that if too many factors are retained, after rotation, it is relatively
easy to discard trivial factors without changing the substantive factors. This
notwithstanding, there is evidence to suggest that overfactoring is none-the-less quite
problematic and should be avoided. With overfactoring, the resultant factor solution
may include factors that are not interpretable or unlikely to replicate (Zwick & Velicer,
1986). It has been suggested that overfactoring may result in factor splitting at the
rotation phase, or when rotating too many oblique factors, high interfactor correlations
may result or the factor space may collapse (Crawford, 1975). Gorsuch (1983) warns
that the extraction of too many factors may cause a common factor to be missed. In
addition, several studies lend support to the notion that overfactoring may result in
change of the overall factor structure (see for example, Keil & Wrigley, 1960; Howard &
Gordon, 1963; Dingman, Miller, & Eyman, 1964).
Wood et al. (1996) found that when overfactoring occurs, the estimated loadings
for true factors usually contain less error than in the case of underfactoring. Based on
the findings of their study, these researchers suggest that overfactoring is preferable to
underfactoring provided that factor splitting is prevented and false factors are
eventually eliminated (these authors advance methods for handling these two
conditions). Fava and Velicer (1996) also concluded from their series of studies, that
underfactoring was a much more severe problem in factor analysis than overfactoring.
4
Pardie0 Andysis
7.0
40
SO
40
3.0
LO
1.0
0.0
- 0 - Expected Elgenvolue
Elgesnotot
.
I 2 3 4 5 6 7 II 9 10
Mother qf Acton
7
IMF MEW AIMRegressionassed RfJethods
Cattell- Nelson- Gorsuch (CNG)
k=>3,p=>6
Uses 6 eigenvalues concurrently
Compares slopes of all possible sets of threeadjacent eigenvalues
Greatest absolute value of difference betweentwo successive slopes
Regressionassed Methods (cont'd)
Multiple Regression (MR)
k=>3,p=>6
Based on same principle as CNG
Utilizes all the eigenvalues in each comparisonof the slopes
Greatest absolute value of difference betweenthe slopes
5
Airirt" '
Regression-Based Methods (cont'd)
t-value index (t)
k=>3,p=>6A variation of the MR procedure
Slopes of the regression lines are comparedusing the usual formula for the t-test
Greatest absolute value of t
10
1111F-Airr'AIIIRegression -Based Methods (cont'd)
Standard Error of Scree (Se,e.)
Errors of estimate are calculated using a sequenceof regression analyses employing a decreasingnumber of eigenvalues.
SEy 1,2,3, ...,
SEy (2+,1 2,3,4, ..., v
SE.v (1.0 3,4,5, ..., v
If SE > 1 v
AIMSiope Comparisons
SO
10
10
LO
110
2 3 4 5 6 7 I 9 JO
Numberqf Fobs12
6
Residual Matrix Analysis
Minimum Average Partial (MAP)
Based on matrices of partial correlations
After each of the factors is pattialed out, theaverage of the squared partial correlation iscalculated
Continued until the residual matrix mostclosely resembles an identity matrix
13
MAP Procedure
I
I
I
maser al ream Pantalk0 14
Method
Number of factors (3, 5, 7)
Number of variables (p) in the correlation
matrices (3k, 5k, 10k)
Sample Size (3p, 5p, 10p, 20p, 40p)
Level of Communality (low, wide, high)
Interfactor Correlation (0, .2, .5, mixed)
7BEST COPY AVAILABLE
Method (continued)
AIM
For each condition (i.e., interfactorcorrelations, communality level, k, p) wegenerated a random sample of 10population R matrices
From each population R matrix generated1,000 samples of each size (N per p)
16
Example of Population Pattern Matrix(k=3 & pr-15)
High Wide Low117 -.01 21 .63 -.01 .07
27 ..08 .16
.88 .18 .18 .74 .49 .13
.76 47 .06 .68 .67 ..05 .44 ..01 .08.45 ..01 .04 .38 .23 .03
.61 -.02 58 .15 .87 .08 .38 .24 .34
.68 .51 .06 .11 .62 ..03-.01 .51 .08 -.02 .58 -.02
-.01 .66 3.15 23 ..04-64 .78 .10 -.03 .11 .70 21 -.63 .60.40 .64 35 .04 .00 54..05 .10 .88 -.04 .59 .67 .10 -.01 .64
.18 -.03 .132 .15 .03 2348 .01 .09 .16 .01 .42 .23 .14 .45
17
Accuracy Criteria
Performance was evaluated based on:
Average number of factors retained by eachmethod
Proportion of samples retaining the same numberof factors as the population
Proportion of samples retaining the same numberof factors as the rule of thumb applied to thepopulation
16
8 BEST COPY AVAILABLE 6
Nuance of Desgrt Factors
Considerable variability was evidencedacross the conditions examined for eachmethod.
To investigate which design factors wereassociated with the most variability, e2was computed and examined.
19
Distributions of RC lean dumber of Factors
Retained, k = 5
28
1 ,
e
;g
,
Fi
2rEFOI1
ao
mums- AIMMean Number of Factors Retained by N per
Variable (N:p) for k = 5
I
14
u
10
2
0CNG 0SE9we0-1 a PA nit 0 PA d...Kr...PA 4 MAP - C. Kw
C5 MOSR
10 15 20 28 30 S 4 4N Po, Variable
21
9
Mean Number of Factors Retained by Number of
Variables Per Factor (p:k) for k = 5
ID
to
I
.3. CMG 0SE Save 0MR0t ..-0...PA RA 0PA d,-6- - PA 11 0MAP - -0 -R512- 41 - M.N5R
3 4 5 0 7 $ 0 10
Number of Vorlobbs Per Fodor22
'7..."--7.7.-7-7311111111
Winn Number of Factors Retained byCommunality Type for k = 5
I
OM 0Eeta SR t M. a Me W ICU HUI23
1014 al Thumb
4111PIPPr .41.1Proportion of Samples Retaining KFactors by True Number of Factors
gu.2
.,
13015017
MG SE 110. RR 1 to PAR PAO PAP RIR 14.1724
Rule of Thumb
BEST COPY AVAILABLE 8
Proportion of Samples Retaining K Factorsby N per Variable (3:p)
Scree
43....PA rah 13PA cl 11
- V5R 43 - WICSR
MAP
6 10 10 20 as 30 16 40 45
25N Vaddis
Proportion of Samples Retaining K Factors byNumber of Variables Per Factor (p:k)
imor- AEIProportion of Samples Retaining K Factors
by Communality TypeID
0.9
0.7
OS
13 03
10.6it 0.3
03
0.0
awm
45119 SE WI 1 PA nil PA d PA 0 NAP KSR SKSRSave
Rule of Thumb
BESTCOPYAVAILABLE
ResuOts
More variables per factor, larger samples,higher communality
Most rules underestimated true number offactors
PAMH is best overall, but MAP is slightlybetter with small samples (N = 3p)
Some methods rarely lead to correctnumber of factors
28
ice"Conclusons
No method should be used in isolation
The 'art of factor analysis'
Some methods should be avoided
What to do about low communality andsmall samples? Need a new and betterthumb
12
BEST COPY AVAILABLE
10
Factor Retention4
Mac Callum et al. (2001) state "given the serious consequences of underfactoring,
...,users of factor analysis should be very rigorous in making the number-of-factors
decision and should err in the direction of overfactoring when the evidence is
ambiguous". Wood, et al. (1996), however, caution about the importance of extracting
the correct number of factors, as they found in their studies that the factor solution with
the least error is the one that extracts the correct number of factors.
Methods for Determining Number of Factors
The factor analysis literature is replete with recommendations regarding the
most appropriate procedures to employ as well as an array of criteria to consider when
addressing the number-of-factors problem (e.g., Gorusch, 1983; Hakstian, Rogers, &
Cattell, 1982; Zwick & Velicer, 1986). According to Horn and Engstrom (1979), after 50
years of work in the field, no fewer than 50 tests had been invented. There is, however,
little consistency in the results obtained from applications of these tests across data sets
and varying conditions. Thus, many applied researchers are often in doubt about the
efficiency of the procedures they use in seeking to determine reasonable estimates of the
number of factors that underlie a given data structure.
Nasser, Benson, and Wisenbaker (2002) note that although researchers
addressing the number-of-factors to retain problem have advanced numerous
approaches based on different theoretical rationales (algebraic, statistical, psychometric,
substantive importance, and interpretability), the most commonly employed methods
for determining the number of factors in applied research are Kaiser's eigenvalue-
greater-than-one rule (Kaiser, 1960) and Cattell's visual scree test (Cattell, 1966). Both
methods are readily available in commonly used statistical software packages, which
may account for their widespread use. However, several researchers have discussed the
limitations of these two procedures and have warned against their use in making
decisions regarding the number of factors to retain.
In the case of the Kaiser-greater-than-one rule (Kaiser, 1960), factor retention is
based upon the number of eigenvalues greater than one. More specifically, the rule
Factor Retention5
calls for as many components or factors to be retained as there are eigenvalues greater
than 1.0. This procedure was initially suggested by Guttman (1954) based on the
consideration that it provided a lower bound for the number of common factors that
underlie a correlation matrix having unities in the main diagonal. Turner (1998) argues
that the Kaiser rule is flawed as it is based on an assumption that a factor is not
psychometrically sound if it accounts for less variance than a single variable does. Cliff
(1988) suggests that the rule lacks a logical basis as one of the rationales for the rule was
based on a misapplication of a formula for estimating the reliability of a multi-item test.
A number of researchers (e.g., Cattell & Jaspers, 1967; Fava & Velicer, 1992; Hakstian,
Rogers, & Cattell, 1982; Lee & Comrey, 1979; Zwick & Velicer, 1986) have reported the
use of the Kaiser rule as problematic as it often leads to an overestimate of the number
of components or factors that underlie the data, thus giving rise to the potential
problems associated with overfactoring. Others report a tendency to either
overestimate or underestimate the number of components or factors depending on the
conditions present in the data, including whether it is applied to population versus
sample matrices (see for example, Cliff, 1988; Hakstian, Rogers & Cattell, 1982).
Cattell's scree test involves identifying the number of factors or components to
retain based on a visual examination of the graph of eigenvalues plotted on the vertical
axis and the factor sequence numbers plotted on the horizontal axis. The process
involves separation of the "scree" of trivial factors from the "cliff" of nontrivial factors
(Cattell, 1966). A straight edge is placed along the bottom part of the graph where the
points form an approximately straight line (Cattell & Vogelman, 1977). The points
above and to the left of the straight line correspond to the nontrivial factors; points on
or close to the straight line constitute the trivial factors. This process of identifying the
nontrivial factors and hence, the number of factors to retain, introduces a considerable
amount of subjectivity on the part of the applied researcher. Several researchers have
noted that issues of reliability often arise in the use of this procedure (e.g., Crawford &
Koopman, 1979; Zwick & Velicer, 1986). The absence of an obvious break or the
presence of multiple breaks in the eigenvalue pattern may make it difficult for one to
14
Factor Retention6
make a decision about the appropriate number of factors to retain (Jurs, Zoski, &
Mueller, 1993; Nasser, Benson & Wisenbaker, 2002). Some researchers have reported
reasonable accuracy of the scree test, however, the range of conditions examined in
their studies were limited (see for example, Tucker, Koopman & Linn, 1969; Cattell &
Jaspers, 1967; Zwick & Velicer, 1986). Zwick and Velicer (1986) indicated in their study
of the number of components to retain that the scree test "was generally accurate but
variable" and recommended against its use as the sole decision method.
In light of the subjective nature of Cattell's visual scree test, researchers have
sought to develop more objective variations of the scree test in order to overcome some
of its limitations. Four regression-based variations have been proposed, namely, the
Cattell-Nelson-Gorsuch (CNG) test (Gorsuch & Nelson, 1981); and three procedures
proposed by Zoski and Jurs (1993, 1996): a multiple regression approach (MR), a t-value
index, and the standard error of scree (SEscree). A recent Monte Carlo study conducted
by Nasser, Benson, and Wisenbaker (2002) compared the performance of these four
regression-based methods under a variety of simulated conditions. The researchers
reported the SEscree procedure to be the most accurate in terms of its performance with
both correlated and uncorrelated factors. The other three methods were found to
consistently underestimate or overestimate the number of factors to be retained. These
researchers noted several limitations of their study, including the range of conditions
examined. They suggested that this investigation should be viewed as an initial but
crucial phase in the investigation of the efficiency of regression-based procedures for
determining the number of factors, as previous work in the field was based solely on a
few correlation matrices from the published literature.
Additionally, other methods considered to be more effective than Kaiser's rule
and Cattell's scree test in estimating the number of factors that underlie a data set have
been proposed. These include parallel analysis (Horn, 1965; Montanelli & Humphreys,
1976; Lautenschlager, Lance, & Flaherty, 1989) and a minimum-average partial (MAP)
correlation procedure proposed by Velicer (1976).
15
Factor Retention7
Given the need for a more comprehensive study and comparison of the
effectiveness of methods proposed for determining the number of factors to retain in
exploratory factor analysis, we undertook an investigation of ten methods that can be
classified under four broad categories of procedures: the Kaiser criterion; parallel
analysis; regression-based variations of the scree test, and residual matrix analysis. The
ten methods examined are briefly described below.
Description of Methods
Kaiser Rule
Eigenvalue Greater Than 1.0 (KSR). This is perhaps the most commonly used
method for determining the number of components to retain in principal component
analysis and is also frequently employed by more than novice researchers in the
conduct of exploratory common factor analysis. Zwick and Velicer (1986) suggest that
the popularity of its use may be due to its availability as the default option in many
statistical packages. The rationale for this method was initially proposed by Guttman
(1954) and later adapted and popularized by Kaiser (1960). Factors or components with
eigenvalues greater than 1.0 (i.e., components which evidence at least as much variance
as one of the original variables) are retained in the analysis and subjected to rotation. As
noted earlier, several researchers have indicated that Kaiser's eigenvalue-greater-than-1
rule is problematic (e.g., Yeomans, & Golden, 1982; Zwick & Velicer, 1986; Wood,
Tataryn, Sr Gorusch, 1996) and have advised against its use in factor analysis. We
include it here because of its common use, and our desire to compare this method with
more recently developed approaches.
Eigenvalue Greater Than Average Eigenvalue in Sample (M-KSR). This modification
of the Kaiser rule was first proposed by Guttman (1954). This method, which may be
more appropriate than the original KSR for common factor analysis, calls for the
retention of factors with eigenvalues greater than the average eigenvalue in the sample.
16
Factor Retention8
Parallel Analysis
This method was developed by Horn (1965). It involves the comparison of
eigenvalues of a correlation matrix of p random uncorrelated variables with those of the
eigenvalues of the correlation matrix of p variables in the actual data set, based on the
same sample size. Factors of the real data set that have eigenvalues greater than the
eigenvalue of the corresponding factor in the random data set are retained and
subjected to rotation.
Parallel Analysis (PAMH). A regression equation for predicting the eigenvalues of
random correlation matrices with squared multiple correlations on the main diagonal of
the random correlation matrix was subsequently developed by Montanelli and
Humphreys (1976):
ln(A, ) = a, + b, ln(N -1) + c, In {(.5) (k(k -1)) (i -1)k}
where N = number of observations,
k = number of variables in the correlation matrix, and
i = eigenvalue sequence number.
The regression weights (ai, bi, 0) were obtained via simulation methods and are
provided by Montanelli and Humphreys (1976).
A decision rule was suggested based on the point at which the plot of sample
eigenvalues crosses the plot of expected eigenvalues from matrices of random data
(Montanelli & Humphreys, 1976). Factors above the point at which the two lines cross,
were retained for rotation. Factors that lie below the point that the plots crossed (i.e.,
those factors whose eigenvalues were less than the eigenvalues of the corresponding
factors of the random data set) were considered trivial factors and were not retained.
Humphreys and Montanelli (1975) found the parallel analysis method to provide
accurate estimates of the number of factors to be retained in common factor analysis.
Zwick and Velicer (1986) also reported that within the context of principal components
analysis, the parallel analysis method yielded consistently accurate estimates of the
number of components to retain. In fact, they indicated that in their study in which
they compared five different rules for determining the number of components to retain,
17
Factor Retention9
the parallel analysis method "was typically the most accurate method at each level of
complexity examined."
Parallel Analysis Close (PAca A modification of Horn's parallel analysis
procedure is to eliminate factors with eigenvalues that are close in magnitude but not
necessarily less than the magnitude of the corresponding eigenvalues of the random
data. Such a modified criterion is expected to retain fewer factors than the original Horn
procedure. For this study, sample eigenvalues within .10 of the random data
eigenvalues were considered close enough to stop retaining factors.
Modified Parallel Analysis (PALL). Lautenschlager, Lance and Flaherty (1989)
proposed a modification to the general regression equation first developed by
Montanelli and Humphreys (1976) and further refined by Allen and Hubbard (1986) to
predict eigenvalues of factors resulting from a random data correlation matrix. These
researchers noted that the Allen and Hubbard's equation did not yield an accurate
estimate of the first eigenvalue; this in turn directly influenced the accuracy of
subsequent eigenvalues in the series. They modified the equation proposed by Allen
and Hubbard (1986) by adding a variables-to-subjects ratio term:
ln(X) = a, + b,1n(N 1)+c, ln {(.5)(k i-1)(k + 2))+ d, ln(X, )+ e, (k / N)
The regression weights (as, bi, ci, di and ei) are provided by Lautenschlager, Lance
and Flaherty (1989).
The results of a Monte Carlo study using the newly augmented regression
equation showed that it served to significantly improve the prediction of the first, and
consequently subsequent eigenvalues (up to 48) of a random data matrix
(Lautenschlager, et al., 1989) when compared to eigenvalues generated from the Allen
and Hubbard (1986) equation. These researchers recommend use of the modified
regression equation with the added variables-to-subject ratio term in future parallel
analysis applications.
Regression-Based Variations of the Visual Scree
Is
Factor Retention10
Each of the methods described in this category is based on Cattell's (1978)
guidelines for determining the number of common factors to retain.
Cattell- Nelson Gorsuch (CNG). In an attempt to find a more objective approach to
the visual scree test, Gorsuch and Nelson (1983) developed an analytical method using
multiple linear regression to determine the number of factors to retain. In this
procedure, the slopes of all possible sets of three adjacent eigenvalues each are
compared. More specifically, the slope of the first three points (roots) in the eigenvalue
plot (1, 2, and 3) is compared with the slope of the next three points (4, 5,and 6). Then
the slope of points 2, 3, and 4 is compared with the slope of points 5, 6,and 7, and so on
until all points have been incorporated in the comparisons. The number of factors to be
retained is defined by the point at which the difference between the two successive
slopes is greatest.
Multiple Regression (MR). The CNG procedure uses only six eigenvalues (i.e., six
data points) at a time to determine each pair of slopes for comparison. Zoski and Jurs
(1993) suggest that this procedure uses only a limited amount of information to
determine the number of factors to retain. They therefore proposed a multiple linear
regression approach that would include more data points in computing the slope of the
regression lines. The MR approach is based on the same principle as the CNG, but it
utilizes all the eigenvalues in each comparison of pairs of slopes. A series of adjacent
slopes are obtained and all possible successive pairs of slopes are compared. One slope
in each paired comparison is based on an increasing number of eigenvalues and
corresponds with the major factors; the second slope is based on a decreasing number of
eigenvalues and corresponds with the trivial (scree) factors. More specifically, the slope
of the regression line developed from points 1, 2, and 3 in the eigenvalue plot is
compared to the slope of the regression line developed from points 4, 5, 6, ..., p (where
p is the number of eigenvalues in the plot). Then, the slope of the regression line for
points 1, 2, 3, and 4 is compared to the slope for points 5, 6, 7, ..., p, and so on. This
process continues until all adjacent slopes of all possible successive pairs of regression
lines are compared. The decision regarding the number of factors to retain corresponds
19
Factor Retention11
to the point at which the absolute value of the difference between the slopes is the
greatest.
t-value index (t). Another procedure examined introduced a variation of the MR
procedure known as the t-value index. Using the slopes obtained in the MR procedure,
the slopes of the regression lines are compared using the usual formula for the t-test of
the difference between slopes. The number of factors to retain is based on the largest
absolute value of t.
Standard Error of Scree (SEscred. Because the CNG, MR, and t-value index are not
applicable when the number of factors is less than three and/or the number of variables
is less than six, Zoski and Jurs (1996) developed the standard error of scree (SEscr,)
procedure which is based on the standard error of estimate for a set of points in the plot
of the eigenvalues. In this procedure, the errors of estimate are calculated using a
sequence of regression analyses employing a decreasing number of eigenvalues.
Initially, all eigenvalues are regressed onto their ordinal numbers. The procedure
continues with subsequent sets of eigenvalues and concludes when the standard error
of estimate meets the 1/p criterion. Because the error variance tends to be inversely
related to sample size, Zoski and Jurs (1996) set the value of 1/p as the criteria for
determining the number of factors to retain (i.e., the number of standard errors that
exceed 1/p is the number of factors to retain). Zoski and Jurs (1996) and Nasser, et al.
(1996) found SEscree to be more accurate in identifying the number of factors to retain
than the CNG and multiple regression (MR) procedures.
Residual Matrix Analysis
Minimum Average Partial (MAP) Correlation. The Minimum Average Partial
(MAP) method developed by Velicer (1976) is based on a matrix of partial correlations.
After each of the factors has been partialed out, the average of the squared partial
correlations is calculated. When the residual matrix most closely resembles an identity
matrix, no further factors are extracted and rotated. Using this method, at least two
variables will have high loadings on each retained factor. Zwick and Velicer (1986)
20
Factor Retention12
reported the MAP to be one of the two most accurate methods (the other being parallel
analysis) in determining the number of components to retain in their study of five
decision rules in principal components analysis.
Purpose
The purpose of this study was to compare the effectiveness of the ten
aforementioned methods (the Kaiser rule (KSR), a modified Kaiser criterion (M-KSR),
three variations of parallel analysis-- Montanelli and Humphrey's (PAmH), parallel
analysis-close (PAIL), and Lautenschlager et al. modified parallel analysis (PALL), four
regression-based variations of the scree procedure (CNG, MR, t, SEscree), and the
minimum-average partial (MAP) procedure) in determining the number of factors to
retain in exploratory common factor analysis. The performance of these procedures
was evaluated based on the average number of factors retained by each method, the
proportion of samples retaining the same number of factors as the true number of
factors in the population, and the proportion of samples retaining the same number of
factors when a particular rule of thumb is applied to the population.
Method
The performance of the ten procedures were investigated using Monte Carlo
methods, in which random samples were generated under known and controlled
population conditions. The population correlation matrices varied with respect to the
particular aspects of interest. Population correlation matrices were constructed based
on the methods described by Tucker, Koopman and Linn (1969), which have been used
in large-scale simulation studies (Mac Callum, et al., 1999; Tucker et al., 1969). For
correlated factors, a generalization of the method was utilized that was described by
Hong (1999). The true factor loading patterns underlying these matrices were
constructed to exhibit relatively clear simple structure. For these population
correlation matrices, the number of measured variables, the number of common factors,
the level of communality and the correlation between factors were controlled. Sample
21
Factor Retention13
correlation matrices were generated from each of the known population correlation
matrices. Each of the sample matrices was then analyzed using principal axis factor
extraction. The pattern of eigenvalue magnitude was used to determine the number of
factors that should be retained.
Generation of Population Matrices
In the Monte Carlo study, uncorrelated population correlation matrices were
generated using the method presented by Tucker, Koopman, and Linn (1969). These
matrices were generated under the assumption that the common factor model holds
exactly in the population. This method produces population matrices leading to
relatively clear simple structure and has been used in other simulation studies
(Mac Callum et al. 1999; Mundfrom, Shaw & Ke, 2001; Tucker, Koopman and Linn,
1969). The population R is generated based on major, minor, and unique factors,
R = A,A; +A2Al2 + A3A;
where Al is the pxk matrix of actual input factor loadings for the major factors, A2 is the
matrix of actual input factor loadings for the minor factors, and A3 is the pxp diagonal
matrix of actual input factor loadings for the unique factors. The contribution of the
minor factors (A2) was set to zero in this study so that the data generation model
matched a factor analytic model with k common factors. The common factors were
specified as orthogonal.
The process for creating A, starts with the creation of a matrix of conceptual
input factor loadings, A, . To create A, , the loading of a variable on a randomly
selected factor, j=1 thru k, is set to a value randomly chosen between 0 and k-1, (for a 3
factor model a,, could be 0,1, or 2). Next the loading on a randomly selected factor
from those remaining is set to a value randomly chosen between 0 and k-1-ilj . This
process continues until a conceptual input factor loading has been chosen for each
factor, and ensures the sum of the loadings across the factors is k-1. This process is then
repeated for each of the p variables.
Factor Retention14
The matrix of actual input factor loadings, A, , is then created from the matrix of
conceptual input factor loadings, AI , through a series of three steps: (1) normal deviates
are added to introduce error, (2) a skewing function is used to limit negative factor
loadings, and (3) the matrix is scaled to ensure desired levels of communality. The
diagonal matrix, A3, for the unique factors is also scaled to ensure the desired levels of
communality. The levels of communality (h 2) that were simulated were high (h 2 for
each variable drawn randomly from values of .6, .7, and .8), wide (h 2 for each variable
drawn randomly from values of .2, .3, .4, .5, .6, .7, and .8), or low (h 2 for each variable
drawn randomly from values of .2, .3, and .4), which are consistent with the levels
used in other simulation studies. An example matrix of input factor loadings, A, , that
was created using these methods is presented in Table 1 (with k=3 and p=15) for each
level of communality.
To construct population matrices for correlated factors, the above method was
modified following Hong (1999). More specifically,
JBJ' + A321:
where
and
j = [Al A2]
B =[(1) T
r]'where 'D is a matrix of correlations among major factors, r is a matrix of correlations
among minor factors, and I is a matrix of correlations among major and minor factors.
The simulations conducted for this study were somewhat simplified since the
contributions of minor factors (A2, r, r) were set to zero.
It should be noted that this method of generating population correlation matrices
controls the number of measured variables, the number of common factors, and the
level of communality, but that more than one population correlation matrix can be
generated having the desired number of items, factors, and communality. The
Factor Retention15
specifications define a general class of population matrices, but the random selection
involved in the creation of the conceptual input loading matrix, the introduction of
error, and the random selection of specific communality values all influence the
population correlation matrix that is obtained.
Monte Carlo Study Design
The Monte Carlo study included five factors in the design. These factors were (a)
the number of common factors, k, present in the population (populations were
simulated with k = 3, 5 and 7 factors), (b) the number of variables, p, in the correlation
matrices (with p = 3*k, 5*k and 10*k), (c) the level of communality, (high, wide, and low),
(d) the level of interfactor correlation (with ry = 0, .3, .5 and a mixed condition with
interfactor correlations ranging from 0 to .5), and (e) sample size, N (with samples of 3p,
5p, 10p, 20p, and 40p). These N:p ratios represent values that range from those generally
considered insufficient to those considered more than adequate (MacCallum et al.,
1999). The factors in the design were crossed with each other, providing a total of 540
conditions in the Monte Carlo study.
The research was conducted using SAS/IML version 8.2. Conditions for the study
were run under Windows 98 and Windows 2000. For each condition investigated, 10
population matrices were generated. Then for each population matrix, 1,000 samples
were generated. This provided the opportunity to examine the degree to which the
results varied across different population matrices within a condition. The use of 10,000
samples provided adequate precision of estimates of the sampling behavior of the factor
recovery indices. For example, 10,000 samples provide a maximum 95% confidence
interval width around an observed proportion that is ± .0098 (Robey & Barcikowski,
1992).
24
Factor Retention16
Results
Mean Number of Factors Retained
The mean number of factors obtained by each method is presented for all
combinations of k, p, N, and communality in Tables 2-5, where each table contains a
different level of correlation among the factors. A quick perusal of these tables suggests
that the mean number of factors being retained differs across methods, and for each
particular method the closeness between the mean number of factors retained and the
true number of factors, k, varies across the conditions studied, with larger sample sizes
and more variables tending to lead to better results.
To summarize the performance of the methods, a series of box plots was created.
Each box plot shows the distribution of the mean number of factors retained for a
particular method. These box plots are presented in Figures 1-3, for k = 3, 5, and 7
factors, respectively. For the three-factor condition (Figure 1), the CNG method has a
mean number of factors that is very close to 3, the true value, for all conditions. The
other methods show somewhat more variability across conditions. Interestingly, if one
examines the five-factor conditions (Figure 2), the CNG is still tending to suggest 3
factors, and it continues to suggest three factors even when the true number of factors is
seven (Figure 3). Thus its effectiveness drops off considerably as the number of factors
increases.
Two of the parallel analysis methods (PAmil and PAIL) appear to do relatively
well in the sense that the mean number of factors retained is close to the true number of
factors for a relatively large portion of the conditions. Note that when these methods
produce means that differ from k, the means tend to be smaller, suggesting too few
factors are being retained. The tendency to have too few factors is also found and
somewhat more prevalent for the PALL, MAP, and KSR methods. The distributions for
t, SE., and M-KSR show the most variability, and are the only methods that tend to
have an average number of retained factors that exceeds k for many conditions.
Since the mean number of factors retained differs across conditions for each
method, 62 was used to determine which design factors were associated with the most
25
F a c t o r Retention17
variability. These values are presented in Table 6. Note that one would hope that the
majority of the variation in the mean number of factors retained would be associated
with k, the true number of factors. The only method for which over 50% of the
variation was attributable to k was the PAci, method (a2= .61). Figure 4 contains a
graph showing the mean number of factors retained for each method as a function of k.
As k increases from 3 to 5 to 7 the mean number of factors retained also increases for
each method except the CNG, which stays level at 3 factors extracted. Ideally the bars
would reach, but not exceed, values of 3, 5, and 7, and the method that is coming closest
to this is the M-KSR.
The a2analysis also shows that the mean number of factors retained varies with
other design factors. Figure 5 shows variation as a function of sample size per variable
(N/p) for each method. This figure displays results for conditions with k = 5, but the
pattern was consistent across the values of k investigated. With increasing Nip, the
PAS, PAcL, and M-KSR tend to get close to the expected value of 5. With small N/p
ratios M-KSR tends to retain too many factors, while PAS and PAci, tend to retain too
few factors.
The effect of the number of variables per factor (p/k) is shown in Figure 6. The
methods tend to have a larger mean number of retained factors when the number of
variables per factor is greater. When the number of variables per factor is relatively
small (3), all methods tend to underestimate the number of factors. When number of
variables per factor is large (10), four methods have a mean number of retained factors
that is near the true value of 5. These are the PAmH, PAcL, MAP and KSR methods.
The effects of communality are depicted in Figure 7. As communality increases
the mean number of factors retained by the methods tends to get closer to the true value
of 5. For some methods this implies the mean number of retained factors decreases
with communality while for others it increases with communality. Phi also appears to
have some effect on the mean number of factors retained for most methods (Figure 8).
Generally the mean number of retained factors decreases with increases in the
correlations among factors.
'
Factor Retention18
Proportion of Samples Retaining the Same Number of Factors as the Population
In addition to the mean number of factors obtained, the extent to which the
number of factors retained in each sample matrix matched the number of factors
present in the population correlation matrix was examined. The proportion of
agreement between the sample matrices and the population matrices (i.e., the
proportion of samples retaining k factors) is presented for all combinations of k, p, N,
and communality in Tables 7 -10, aggregated among each level of inter-factor
correlation. A review of these tables reveals that the proportion of samples in
agreement with the population differs considerably across the methods and conditions
investigated. Generally, larger sample sizes, a greater number of variables per factor,
high communality levels, and low inter-factor correlations demonstrated better results.
Due to evidence of large variability across the conditions examined for each
method, w 2 was computed to determine which design factors were associated with the
most variability in the proportions. Table 11 displays these results. Contrary to the
expectations about a 2 in relation to the number of factors extracted, it was expected
that little variation in the proportion of agreement would be associated with k, the true
number of factors underlying the population correlation matrix. One method for which
the majority of variation was attributable to k was the CNG method (62=.98). In
addition, a considerable amount of variation that was attributable to k was the MR
procedure (62=.48). As demonstrated in Figure 9, as k increased, the proportion of
samples in agreement with the population decreased for all of the methods. Although
the CNG method demonstrated excellent results for k = 3 matrices, agreement dropped
dramatically as the number of factors increased. Additionally, for the MR procedure,
agreement dropped to zero for populations of k = 5 and k = 7.
The Vanalysis also revealed that the proportion of samples in agreement with
the population varied with other design factors. Figure 10 depicts the variation in
proportions as a function of sample size per variable (N/p) for each method. With
increasing N/p, the majority of the methods evidenced an increase in agreement, with
2%
Factor Retention19
the pAmH method exhibiting superior performance. In addition, the SE scree method
evidenced the greatest improvement in factor retention with increasing N/p.
The influence of the number of variables per factor (p/k) on the proportion of
samples in agreement with the population is shown in Figure 11. For the majority of
the methods, the proportion of samples in agreement with the population increased as
the number of variables per factor increased, with the PArviii and PAci methods
demonstrating exceptional performance. In particular, the MAP procedure showed the
most dramatic improvements in agreement as the number of variables per factor
increased. In contrast, the CNG and t procedures demonstrated the opposite trend,
declining in their agreement with the population as the number of variables per factor
increased. Interestingly, the SEscree method showed significant improvement in the level
of agreement as the number of variables per factor increased until the p/k ratio was 5.
At that point, the proportion of samples in agreement with the population declined.
The effects of communality on the proportion of samples in agreement with the
population were also examined. As shown in Figure 12, the majority of the methods
evidenced that as communality increased, the proportion of samples in agreement with
the population increased. In contrast, the MR and t methods evidenced a decrease in
performance with an increase in communality level. Phi also appeared to affect the
proportion of agreement for almost all methods (Figure 13). With the exception of the
CNG and MR methods, agreement decreased as the interfactor correlation increased.
Negligible differences were observed for the CNG and MR procedures as the
correlation among factors increased.
Proportion of Samples Retaining the Same Number of Factors as the Rule of Thumb
Applied to the Population
Lastly, an examination was made of the extent to which the number of factors
retained by each rule matched the number of factors that would have been retained if
we had used the rule on the population correlation matrix. The agreement with that
which we would have seen in the population is presented for all combinations of k, p, N,
28
Factor Retention20
and communality in Tables 12-15, with each table displaying a different level of inter-
factor correlation. An examination of these tables suggests that the proportion of
samples differs considerably across the methods, and for each method this proportion is
seen to vary across the conditions studied. In general, conditions with larger sample
sizes and more variables per factor evidenced better results.
Because the proportions evidenced variability across the conditions examined for
each method, we computed a series of a2values to investigate which design factors
were associated with the most variability in these proportions. These values are
presented in Table 16. Consistent with the results presented above, with regard to
agreement between the sample matrices and the population matrices, we would not
expect a sizeable portion of the variation in agreement to be associated with k, the true
number of factors underlying the population correlation matrix. The only method for
which a considerable amount of variation was attributable to k was the MR method
(Qv = .46). Figure 14 illustrates the proportion of samples agreeing with the rule applied
to the population R-Matrix as a function of k. As k increases, the proportion of samples
in agreement with the population decreased for most of the methods with the exception
of MR, which decreased with k = 5 but increased substantially with k = 7. Additionally,
the CNG remained zero regardless of the number of factors. Finally, for the t
procedure, agreement dropped quite dramatically as the number of factors increased.
The a2analysis revealed that the proportion of samples in agreement with the
population also varied with other design factors. Figure 15 shows variation in
proportions as a function of sample size per variable (N/p) for each method. With
increasing N/p, the majority of the methods evidenced an increase in agreement, with
the non-scree based methods exhibiting superior performance. However, the SEsave
method evidenced the biggest improvement in factor retention with increasing N/p.
The influence of the number of variables per factor (p/k) is shown in Figure 16. It
was our expectation that a considerable influence would be associated with p/k. Three
methods were observed for which a notable amount of variation was attributable to
p/k; t, PAS and PAct (cti2= .28, 62= .28 and 62= .24 respectively). The non-scree
29
Factor Retention21
based methods tended to fare slightly better or remain constant as the number of
variables per factor increased. The opposite trend was evidenced for the scree-based
methods, all of which decreased in agreement as the number of variables per factor is
increased.
Finally, the effects of communality are depicted in Figure 17. As communality
increases the proportion of samples agreeing with the population tended to increase for
the majority of methods examined, with the exception of the PALL method, which
evidenced a decrease in performance with increased communality. Phi also appeared
to have some effect on the proportion of agreement for most methods (Figure 18).
Generally, agreement decreased as the interfactor correlation increased for the parallel
analysis approaches (PAmH, PAcL and PALL) and the methods based on the Kaiser
criterion (KSR and M-KSR). Negligible differences were observed for the MAP and the
t procedures as the correlation among factors increased, however, the MAP proceddre
outperformed all other methods for the conditions examined.
Discussion
This empirical investigation of methods to determine the number of factors to
retain in common factor analysis clearly suggests that both the choice of method and the
design of the factor analytic study play crucial roles in retaining the correct number of
common factors. For example, with phenomena characterized by low communality of
variables, a study designed with few variables per factor and small sample size is
unlikely to lead to the correct number of factors, regardless of the method selected
(except for the trivial case of the success of the CNG method when the true number of
factors is 3). Investigations in such areas should include larger numbers of variables
(i.e., p = 10k) and a very large number of observations (N = 40p). With such a design, the
SEscree, parallel analysis (both PAmH and PAcL) and the M-KSR methods provided
nearly 100% accuracy across levels of k and inter-factor correlation. Phenomena that are
characterized by higher levels of communality appear to have less stringent data
requirements, especially if the correlation between factors is low.
30
Factor Retention22
The most prevalent type of error made in the number-of-factors problem appears
to be one of underfactoring if the parallel analysis methods (PAmH, PAIL, PALL), the
MAP or the Kaiser methods (KSR, M-KSR) methods are used. Turner (1998) also found
that under some circumstances, parallel analysis may underestimate the number of
factors that are in the data. For the methods based on fitting regression lines to the scree
plot, underfactoring was also characteristic of CNG and MR, while the t procedure
showed a tendency towards overfactoring (especially as the true number of factors
increased). The SEscree method appeared to offer a balance between overfactoring and
underfactoring, although it showed a tendency to estimate an exceptionally large
number of factors in conditions with small sample sizes.
In terms of overall accuracy across the conditions examined in this study, the
PAmH approach to parallel analysis provided the largest proportion of samples
retaining the correct value of k, an advantage seen across levels of communality, inter-
factor correlation and values of k (except for k = 3, in which the CNG approach was the
most accurate). Across sample sizes, the PAmH method was the most accurate except for
the smallest samples examined (N = 3p), in which the MAP evidenced a slight
advantage. Similarly, for values of p/k, the PAME was the most accurate except for the
smallest value (p/k = 3) in which the CNG provided the highest average accuracy.
Interestingly, the more recent, modified prediction equations provided by
Lautenschlager, Lance and Flaherty (1989) substantially reduced the accuracy in
number-of-factors determination. These results, of course, must be considered in the
light of the limitations of this research. Although the simulation approach we have
followed has examined a range of values for p, k and N, and a variety of communality
levels and levels of inter-factor correlation, further research is recommended to extend
these findings. Additional work is also needed in the development of methods for
accurately determining the number of factors in the more challenging conditions
revealed here (e.g., low communality, few variables per factor and small sample sizes),
conditions in which none of these rules of thumb were successful.
3i
Factor Retention23
Such additional work is important because exploratory factor analysis is a
frequently used multivariate technique in educational research. In the conduct of actual
exploratory factor analyses, researchers face critical decisions regarding the number of
factors to extract and retain. The interpretations gleaned from factor analytic solutions
depend in large part upon the appropriate use of factor retention strategies. As
exploratory factor analysis continues to enjoy a prominent position among the currently
available multivariate methods, researchers must remain mindful of the limitations of
certain procedures and methods. The results of this study underscore the need to
exercise caution in factor retention decisions and highlight the need to consider, in the
planning stages, those aspects of the factor solution that are important in the research
application. This research furnishes valuable information about the sensitivity of
commonly employed methods and provides guidance regarding the choice of
alternative strategies.
32
Factor Retention24
ReferencesAllen, S. J., & Hubbard, R. (1986). Regression equations for latent roots of random data
correlations with unities in the diagonals. Multivariate Behavioral Research, 21, 393-398.
Cattell, R. B. (1958). Extracting the correct number of factors in factor analysis.Educational and Psychological Measurement, 18, 791-838.
Cattell, R. B. (1966). The scree test for the number of factors. Multivariate BehavioralResearch, 1, 245-276.
Cattell, R. B. (1978). The scientific use of factor analysis in behavior science and life sciences.New York: Plennum.
Cattell, R. B., & Jaspers, J. (1967). A general plasmode for factor analytic exercises andresearch. Multivariate Behavioral Research Monographs, 3, 1-212.
Cattell, R. B., & Vogelman, S. (1977). A comparison trial of the scree and KG criteria fordetermining the number of factors. Multivariate Behavioral Research, 12, 289-325.
Cliff, N. (1988). The eigenvalue-greater-than-one rule and the reliability of components.Psychological Bulletin, 103(2), 276-279.
Comrey, A. L. (1978). Common methodological problems in factor analytic studies.Journal of Consulting and Clinical Psychology, 46, 648-659.
Crawford, C.B. (1975). Determining the number of interpretable factors. PsychologicalBulletin, 82(2), 226-237.
Dingman, H. F., Miller, C. R., & Eyman, R. K. (1964). A comparison between twoanalytic rotational solutions where the number of factors is indeterminate.Behavioral Science, 11, 400-404.
Fava, J. L. & Velicer, W. F. (1992). The effects of overextraction on factor and componentanalysis. Multivariate Behavioral Research, 27, 387-415.
Fava, J. L., & Velicer, W. F. (1996). The effects of underextraction in factor andcomponent analyses. Educational and Psychological Measurement, 56, 907-929.
Ferron, J. M., Kromrey, J. D., Hogarty, K. Y., Hines, C. V. & Mumford, K. R. (2002,April). Alternative Perspectives on Sample Size in Exploratory Factor Analysis: AnEmpirical Investigation of the Influence of Communality and Overdetermination. Paperpresented at the annual meeting of the American Educational ResearchAssociation, New Orleans.
Ford, K. J., MacCallum, R. C., & Tait, M. (1986). The application of exploratory factoranalysis in applied psychology: A critical review and analysis. PersonnelPsychology, 39, 291-314.
33
Factor Retention25
Gorsuch, R. L. & Nelson, J. (1981, October). CNG scree test: An objective procedure fordetermining the number of factors. Paper presented at the annual meeting of theSociety for Multivariate Experimental Psychology.
Gorsuch, R.L. (1983). Factor analysis (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum.
Guttman, L. (1954). Some necessary conditions for common factor analysis.Psychometrika, 19, 149-161.
Hakstian, A. R., Rogers, W. T., & Cattell, R. B. (1982). The behavior of number-of-factorsrules with simulated data. Multivariate Behavioral Research, 17, 193-219.
Hong, S. (1999). Generating correlation matrices with model error for simulation studiesin factor analysis: A combination of the Tucker-Koopman-Linn model andWijsman's algorithm. Behavior Research Methods, Instruments, & Computers, 31,727-730.
Horn, J. L. (1965). A rationale and test for the number of factors in factor analysis.Psychometrika, 30 (2), 179-185.
Horn, J. L. & Engstrom, R. (1979). Cattell's scree test in relation to Bartlett's chi-squaretest and other observations on the number of factors problem. MultivariateBehavioral Research, 14, 283-300.
Howard, K. J., & Gordon, R.A. (1963). Empirical note on the "number of factors"problem in factor analysis. Psychological Reports, 12, 247-250.
Humphreys, L. G., & Montanelli, R. G. (1975). An investigation of the parallel analysiscriterion for determining the number of common factors. Multivariate BehavioralResearch, 10, 193-206.
Jurs, S., Zoski, K., & Mueller, R. (1993, April). Using linear regression to determine thenumber of factors to retain in factor analysis and the number of issues to retain in delphistudies and other surveys. Paper presented at the annual meeting of the AmericanEducational Research Association, Atlanta, GA.
Kaiser, H. F. (1960). The application of electronic computers to factor analysis.Educational and Psychological Measurement, 20, 141-151.
Keil, D. F., & Wrigley, C. (1960). Effects upon the factorial solution of rotating varyingnumbers of factors. American Psychologist, 15, 487-488.
Lambert, Z. V., Wildt, A. R., & Durand, R. M. (1990). Assessing sampling variationrelative to number-of-factors criteria. Educational and Psychological Measurement,50, 33-48.
Lautenschlager, G. F., Lance, C. E., & Flaherty, V.L. (1989). Parallel analysis criteria:Revised equations for estimating the latent roots of random data correlationmatrices. Educational and Psychological Measurement, 49, 339-345.
3
Factor Retention26
Lee, H. B., & Comrey, A. L. (1979). Distortions in commonly used factor analyticprocedures. Multivariate Behavioral Research, 14, 301-321.
MacCallum, R. C., Widaman, K. F., Zhang, S., & Hong, S. (1999). Sample size in factoranalysis. Psychological Methods, 4, 84-99.
MacCallum, R. C., Widaman, K. F., Preacher, K. J. & Hong, S. (2001). Sample size infactor analysis: The role of model error. Multivariate Behavioral Research, 36(4),611- 637.
Montanelli, R. G., Jr., & Humphreys, L. G. (1976). Latent roots of random datacorrelation matrices with squared multiple correlations on the diagonal: A MonteCarlo study. Psychometrika, 41, 341-348.
Mosier, C. I. (1939). Influence of chance error on simple structure: An empiricalinvestigation of the effect of chance error and estimated communalities on simplestructure in factorial analysis. Psychometrika, 4, 33-44.
Nasser, F., Benson, J., & Wisenbaker, J. (2002). The performance of regression-basedvariations of the visual scree for determining the number of common factors.Educational and Psychological Measurement, 62, 397-419.
Robey, R. R., & Barcikowski, R. S. (1992). Type I error and the number of iterations inMonte Carlo studies of robustness. British Journal of Mathematical and StatisticalPsychology, 45, 283-288.
Rummel, R. J. (1970). Applied factor analysis. Evanston, IL: Northwestern UniversityPress.
Tucker, L. R., Koopman, R. F., & Linn, R. L. (1969). Evaluation of factor analytic researchprocedures by means of simulated correlation matrices. Psychometrika, 34, 421-459.
Turner, N. E. (1998). The effect of common variance and structure pattern on randomdata eigenvalues: Implications for the accuracy of parallel analysis. Educationaland Psychological Measurement, 58(4), 541-568.
Velicer, W.F. (1976). Determining the number of components from the matrix of partialcorrelations. Psychometrika, 41, 321-327.
Wood, J. M., Tataryn, D. J., & Gorsuch, R. L. (1996). Effects of under- and overextractionon principal axis factor analysis with varimax rotation. Psychological Methods, 1,354-365.
Yeomans, K. A., & Golden, P. A. (1982). The Guttman-Kaiser criterion as predictor ofthe number of common factors. The Statistician, 31(3), 221-229.
Zoski, K., & Jurs, S. (1993). Using multiple regression to determine the number offactors to retain in factor analysis. Multiple Linear Regression Viewpoints, 20, 5-9.
35
Factor Retention27
Zoski, K., & Jurs, S. (1996). An objective counterpart to visual test for factor analysis:The standard error scree. Educational and Psychological Measurement, 56, 443-451.
Zwick, W. R., & Velicer, W. F. (1986). Comparison of five rules for determining thenumber of components to retain. Psychological Bulletin, 99 (3), 432-442.
36
Factor Retention28
Table 1Example Population Pattern Matrices as a Function of Communality when k=3 and p=15.
High Communality Wide Communality Low Communality.87 -.01 .21 .89 -.03 -.03 .63 -.01 .07
.87 -.05 .19 .83 .04 -.04 .55 -.02 -.03
.86 .18 .18 .74 .49 .13 .45 -.02 -.02
.76 .47 .06 .69 .57 -.05 .44 -.01 .09
.77 -.03 -.03 .45 -.01 .04 .38 .23 .03
.61 -.02 .58 .18 .87 .08 .36 .24 .34
.58 .51 -.06 .16 .69 -.01 .11 .62 -.03-.01 .89 .08 .15 .69 -.02 -.02 .55 -.02-.04 .84 -.03 -.01 .56 .3 -.02 .45 -.02.16 .82 -.04 -.03 -.04 .84 .24 .38 -.03
-.04 .78 .30 -.03 .11 .70 .21 -.03 .60.40 .64 .36 .48 -.06 .68 .04 .00 .55
-.06 .10 .89 -.04 .59 .67 .10 -.01 .54
-.03 .03 .77 .18 -.03 .52 .15 .03 .53
.48 -.01 .69 .16 -.01 .42 .23 .14 .48
Note. Largest loading for each variable is bolded.
3 7
Tab
le 2
Mea
n N
umbe
r of
Fac
tors
Ret
aine
d, P
hi--
0.0
0
Fact
or R
eten
tion 29
kp
Size
Low
Com
mun
ality
Wid
e C
omm
unal
ityH
igh
Com
mun
ality
SEC
NG
Scr
eeM
RT
M-
PAN
T},
PA
L-L
PA
LL
MA
P K
SR K
SRSE
CN
G S
cree
MR
TSE
PAS
PA 1
. PA
LL
MA
P K
SR M
-KSR
CN
G S
cree
MR
TM
-PA
S PA
IL P
AL
L M
AP
KSR
KSR
39
3n3.
32.
30.
00.
02.
11.
60.
51.
11.
73.
33.
12.
50.
00.
02.
32.
31.
01.
51.
91.
93.
12.
60.
00.
02.
42.
32.
42.
02.
22.
45p
3.2
1.8
0.0
0.0
2.2
1.8
0.5
1.1
1.3
3.1
3.1
2.3
0.0
0.0
2.5
2.5
1.1
1.5
1.7
1.7
3.1
2.6
0.0
0.0
2.5
2.5
2.5
2.1
2.1
2.4
10p
3.1
1.7
0.0
0.0
2.3
1.9
0.6
1.1
1.1
2.8
3.0
2.3
0.0
0.0
2.7
2.7
1.2
1.5
1.5
1.5
3.1
2.6
0.0
0.0
2.6
2.6
2.6
2.2
2.1
2.4
20p
3.1
1.7
0.0
0.0
2.4
2.0
0.7
1.0
1.1
2.5
3.0
2.4
0.0
0.0
2.8
2.8
1.2
1.4
1.4
1.4
3.1
2.6
0.0
0.0
2.7
2.6
2.7
2.3
2.1
2.4
40p
3.1
1.8
0.0
0.0
2.4
2.1
0.8
1.0
1.0
2.3
3.0
2.4
0.0
0.0
2.9
2.9
1.3
1.4
1.4
1.4
3.1
2.6
0.0
0.0
2.7
2.6
2.7
2.4
2.1
2.4
315
3p3.
14.
43.
14.
33.
22.
61.
21.
62.
75.
13.
14.
03.
14.
23.
03.
02.
02.
72.
92.
93.
13.
13.
24.
03.
02.
93.
02.
92.
93.
05p
3.1
3.3
3.1
4.0
3.2
2.7
1.2
1.7
2.2
4.7
3.1
3.2
3.1
4.1
3.0
3.0
2.2
2.8
2.8
2.8
3.1
3.0
3.2
3.9
3.0
3.0
3.0
3.0
2.9
3.0
10p
3.1
2.9
3.0
3.7
3.2
2.9
1.2
1.7
2.0
4.0
3.1
3.0
3.0
3.9
3.0
3.0
2.4
2.9
2.8
2.8
3.1
3.0
2.6
3.1
3.0
3.0
3.0
3.0
3.0
3.0
20D
3.1
2.9
2.1
2.5
3.1
3.0
1.3
1.7
1.8
3.2
3.1
3.0
1.7
2.2
3.0
3.0
2.5
2.9
2.8
2.8
3.1
3.0
0.3
0.4
3.0
3.0
3.0
3.0
3.0
3.0
40p
3.0
3.0
0.2
0.2
3.0
3.0
1.3
1.7
1.8
3.0
3.1
3.0
0.2
0.3
3.0
3.0
2.6
2.9
2.8
2.8
3.1
3.0
0.0
0.0
3.0
3.0
3.0
3.0
3.0
3.0
330
3p3.
210
.23.
24.
83.
23.
02.
83.
03.
68.
83.
18.
13.
14.
43.
03.
03.
03.
03.
03.
03.
14.
73.
14.
03.
03.
03.
03.
03.
03.
05p
3.2
6.6
3.3
4.1
3.1
3.0
2.9
3.0
3.0
7.6
3.1
5.7
3.1
4.1
3.0
3.0
3.0
3.0
3.0
3.0
3.1
3.2
3.2
4.0
3.0
3.0
3.0
3.0
3.0
3.0
10p
3.2
3.6
3.3
4.1
3.0
3.0
2.9
3.0
3.0
5.5
3.1
3.5
3.2
4.1
3.0
3.0
3.0
3.0
3.0
3.0
3.1
3.0
3.2
4.1
3.0
3.0
3.0
3.0
3.0
3.0
20p
3.2
3.0
3.4
4.1
3.0
3.0
3.0
3.0
3.0
3.4
3.1
3.0
3.2
4.0
3.0
3.0
3.0
3.0
3.0
3.0
3.1
3.0
3.2
4.1
3.0
3.0
3.0
3.0
3.0
3.0
40p
3.2
3.0
3.4
4.1
3.0
3.0
3.0
3.0
3.0
3.0
3.1
3.0
3.2
4.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.2
4.1
3.0
3.0
3.0
3.0
3.0
3.0
515
3n3.
34.
83.
24.
63.
82.
80.
81.
32.
85.
53.
24.
83.
24.
43.
83.
81.
72.
23.
13.
13.
54.
83.
35.
04.
24.
04.
23.
03.
74.
25D
3.2
3.8
3.1
4.3
4.0
3.0
0.9
1.3
2.1
5.3
3.2
4.2
3.1
4.2
4.1
4.1
1.9
2.2
2.7
2.7
3.6
4.8
3.3
4.5
4.6
4.3
4.6
3.2
3.6
4.2
10p
3.1
3.4
3.1
4.0
4.4
3.4
1.0
1.2
1.7
4.9
3.2
4.0
3.1
3.8
4.5
4.5
2.0
2.3
2.5
2.5
3.7
4.8
2.2
2.6
4.8
4.6
4.8
3.4
3.6
4.2
20p
3.1
3.5
2.4
2.9
4.6
3.8
1.0
1.1
1.4
4.5
3.2
4.0
2.1
2.5
4.7
4.7
2.1
2.3
2.4
2.4
3.9
4.8
0.2
0.2
4.9
4.8
4.9
3.4
3.6
4.2
40p
3.1
3.6
0.5
0.6
4.6
4.1
1.0
1.1
1.2
4.4
3.3
4.1
0.5
0.6
4.7
4.7
2.2
2.3
2.4
2.4
3.9
4.9
0.0
0.0
5.0
4.9
5.0
3.4
3.5
4.2
525
3p3.
29.
03.
17.
25.
14.
11.
92.
54.
38.
53.
37.
93.
16.
64.
74.
73.
24.
24.
54.
53.
95.
63.
16.
25.
04.
95.
04.
74.
95.
05p
3.2
6.6
3.1
6.5
5.2
4.4
1.9
2.6
3.6
7.8
3.3
6.2
3.0
6.2
4.9
4.9
3.4
4.4
4.3
4.3
4.0
5.0
3.1
6.4
5.0
5.0
5.0
4.8
4.9
5.0
10p
3.2
5.1
3.1
6.1
5.2
4.7
2.0
2.7
3.1
6.5
3.4
5.1
3.0
6.1
5.0
5.0
3.5
4.5
4.2
4.2
4.2
5.0
3.1
6.6
5.0
5.0
5.0
4.8
4.9
5.0
20p
3.2
4.9
3.1
6.0
5.1
4.9
2.1
2.8
2.9
5.2
3.5
5.0
3.0
6.1
5.0
5.0
3.6
4.6
4.1
4.1
4.3
5.0
3.1
6.5
5.0
5.0
5.0
4.9
4.9
5.0
40p
3.3
4.9
3.1
5.8
5.0
5.0
2.1
2.8
2.8
5.0
3.6
5.0
3.1
6.0
5.0
5.0
3.6
4.6
4.1
4.1
4.4
5.0
3.2
4.7
5.0
5.0
5.0
4.9
4.9
5.0
550
313
3.3
21.2
3.0
11.3
5.1
5.0
4.3
5.0
5.9
14.7
3.8
16.9
3.0
7.3
5.0
5.0
5.0
5.0
5.0
5.0
4.6
11.5
3.0
7.0
5.0
5.0
5.0
4.9
5.0
5.0
513
3.4
14.7
3.0
8.0
5.0
5.0
4.4
5.0
5.0
12.9
3.9
12.7
3.0
7.2
5.0
5.0
5.0
5.0
5.0
5.0
4.9
7.2
3.0
7.2
5.0
5.0
5.0
5.0
5.0
5.0
10p
3.6
8.3
3.0
7.4
5.0
5.0
4.5
5.0
5.0
9.4
4.1
8.3
3.0
7.3
5.0
5.0
5.0
5.0
5.0
5.0
5.2
5.0
3.0
7.3
5.0
5.0
5.0
4.9
5.0
5.0
20p
3.9
5.2
3.0
7.4
5.0
5.0
4.5
5.0
5.0
5.8
4.3
5.5
3.0
7.4
5.0
5.0
5.0
5.0
5.0
5.0
5.3
5.0
3.0
7.3
5.0
5.0
5.0
4.9
5.0
5.0
40p
4.1
5.0
3.0
7.4
5.0
5.0
4.4
5.0
5.0
5.0
4.5
5.0
3.0
7.4
5.0
5.0
5.0
5.0
5.0
5.0
5.4
5.0
3.0
7.3
5.0
5.0
5.0
5.0
5.0
5.0
721
3n3.
47.
83.
26.
55.
64.
01.
21.
64.
07.
73.
47.
63.
26.
95.
75.
72.
73.
04.
54.
53.
36.
73.
17.
25.
85.
45.
84.
05.
05.
75p
3.3
6.2
3.2
6.2
6.0
4.5
1.2
1.6
3.1
7.4
3.5
6.7
3.2
6.9
6.2
6.2
3.0
3.2
4.2
4.2
3.2
6.5
3.1
7.0
6.2
5.9
6.2
4.4
4.9
5.7
10D
3.3
5.4
3.1
5.9
6.3
5.0
1.2
1.5
2.4
6.8
3.5
6.3
3.2
6.7
6.5
6.5
3.1
3.3
3.9
3.9
3.1
6.5
3.1
5.9
6.5
6.3
6.5
4.7
4.8
5.7
20p
3.3
5.4
3.1
5.3
6.5
5.4
1.3
1.5
2.0
6.4
3.6
6.2
3.3
5.9
6.6
6.6
3.2
3.3
3.8
3.8
3.1
6.5
3.0
4.1
6.6
6.4
6.6
4.8
4.8
5.7
40p
3.3
5.6
3.0
4.2
6.6
5.8
1.3
1.5
1.9
6.2
3.7
6.2
3.3
5.0
6.7
6.7
3.3
3.3
3.7
3.7
3.0
6.5
3.0
3.5
6.8
6.5
6.8
4.9
4.8
5.7
735
31)
3.5
14.4
3.1
10.7
6.9
5.8
2.9
3.8
6.1
11.8
3.2
12.2
3.1
9.2
6.6
6.6
4.8
6.1
6.4
6.4
3.5
8.9
3.1
9.2
6.9
6.8
6.9
6.2
6.8
7.0
5D3.
510
.63.
110
.17.
16.
23.
04.
15.
210
.93.
29.
83.
08.
86.
86.
85.
06.
36.
26.
23.
57.
23.
09.
67.
07.
07.
06.
36.
87.
010
p3.
67.
63.
09.
77.
26.
63.
14.
24.
79.
13.
27.
63.
08.
96.
96.
95.
16.
46.
06.
03.
57.
03.
09.
87.
07.
07.
06.
46.
77.
020
p3.
76.
83.
09.
57.
06.
83.
14.
34.
57.
23.
27.
03.
09.
06.
96.
95.
16.
35.
95.
93.
57.
03.
09.
87.
07.
07.
06.
36.
77.
040
p3.
86.
83.
09.
26.
96.
83.
14.
34.
46.
83.
16.
93.
08.
97.
07.
05.
16.
35.
85.
83.
57.
03.
08.
37.
07.
07.
06.
36.
67.
07
703D
3.0
33.1
3.0
20.4
7.0
7.0
6.0
7.0
8.2
20.5
3.1
26.3
3.0
11.1
7.0
7.0
6.9
7.0
7.0
7.0
3.3
19.4
3.0
10.2
7.0
7.0
7.0
6.9
7.0
7.0
5D3.
024
.13.
013
.07.
07.
06.
17.
07.
017
.93.
120
.73.
010
.77.
07.
07.
07.
07.
07.
03.
212
.93.
010
.57.
07.
07.
07.
07.
07.
010
03.
014
.73.
011
.27.
07.
06.
07.
07.
013
.13.
114
.63.
010
.87.
07.
07.
07.
07.
07.
03.
27.
53.
010
.67.
07.
07.
07.
07.
07.
020
o3.
08.
73.
011
.17.
07.
05.
97.
07.
07.
93.
19.
83.
010
.97.
07.
06.
97.
07.
07.
03.
17.
03.
010
.67.
07.
07.
07.
07.
07.
040
o3.
07.
03.
011
.07.
07.
05.
87.
07.
07.
03.
17.
13.
011
.07.
07.
06.
97.
07.
07.
03.
07.
03.
010
.77.
07.
07.
07.
07.
07.
0
Tab
le 3
Mea
n N
umbe
r of
Fac
tors
Ret
aine
d, P
hi =
0.3
0
Fact
or R
eten
tion 30
kp
Size
Low
Com
mun
ality
Wid
e C
omm
unal
ityH
igh
Com
mun
ality
SEC
NG
Scr
eeM
RT
M-
PAS
PAL
, PA
LL
MA
P K
SR K
SRSE
CN
G S
cree
MR
TSE
PAS
PAL
. PA
u, M
AP
KSR
M-K
SR C
NG
Scr
ee M
RT
M-
PAS
PAL
I, P
AL
L M
AP
KSR
KSR
39
3n3.
12.
10.
00.
01.
81.
50.
71.
11.
53.
23.
02.
20.
00.
01.
81.
81.
01.
21.
51.
53.
02.
50.
00.
02.
12.
02.
11.
91.
92.
25o
3.0
1.6
0.0
0.0
1.9
1.5
0.8
1.0
1.1
3.0
3.0
1.9
0.0
0.0
2.0
2.0
1.0
1.2
1.3
1.3
3.0
2.5
0.0
0.0
2.4
2.3
2.4
2.0
1.8
2.2
100
3.0
1.3
0.0
0.0
2.0
1.5
0.9
1.0
1.0
2.6
3.0
1.8
0.0
0.0
2.4
2.4
1.0
1.2
1.1
1.1
3.0
2.5
0.0
0.0
2.7
2.5
2.7
2.1
1.8
2.2
20p
3.0
1.3
0.0
0.0
2.2
1.7
1.0
1.0
1.0
2.3
3.0
1.8
0.0
0.0
2.7
2.7
1.0
1.2
1.1
1.1
3.0
2.6
0.0
0.0
2.8
2.6
2.8
2.1
1.7
2.3
40D
3.0
1.3
0.0
0.0
2.3
1.9
1.0
1.0
1.0
2.2
3.0
1.8
0.0
0.0
2.8
2.8
1.0
1.2
1.0
1.0
3.0
2.6
0.0
0.0
2.9
2.7
2.9
2.2
1.7
2.2
315
3p3.
04.
13.
04.
82.
52.
01.
01.
42.
34.
93.
03.
83.
04.
62.
52.
51.
32.
22.
42.
43.
03.
13.
04.
32.
82.
72.
82.
92.
72.
95o
3.0
3.0
3.0
4.3
2.7
2.1
1.0
1.4
1.7
4.4
3.0
3.2
3.0
4.3
2.8
2.8
1.4
2.3
2.2
2.2
3.0
3.0
3.0
4.2
2.9
2.9
2.9
3.0
2.7
2.9
101,
3.0
2.5
3.0
3.9
2.8
2.4
1.0
1.3
1.4
3.7
3.0
2.9
2.9
3.9
3.0
3.0
1.5
2.4
2.1
2.1
3.0
3.0
2.5
3.4
3.0
3.0
3.0
3.0
2.7
2.9
20D
3.0
2.5
2.3
2.9
2.9
2.6
1.0
1.3
1.2
3.0
3.0
3.0
2.0
2.6
3.0
3.0
1.7
2.4
2.0
2.0
3.0
3.0
0.5
0.6
3.0
3.0
3.0
3.0
2.7
2.9
40o
3.0
2.6
0.4
0.6
3.0
2.7
1.0
1.3
1.2
2.9
3.0
3.0
0.4
0.5
3.0
3.0
1.7
2.4
2.0
2.0
3.0
3.0
0.0
0.0
3.0
3.0
3.0
3.0
2.7
2.9
330
3o3.
010
.23.
011
.73.
12.
81.
72.
63.
68.
93.
08.
13.
09.
23.
03.
02.
73.
03.
03.
03.
04.
83.
04.
53.
03.
03.
03.
03.
03.
05p
3.0
6.5
3.0
9.2
3.1
3.0
1.8
2.7
2.9
7.8
3.0
5.6
3.0
6.2
3.0
3.0
2.8
3.0
3.0
3.0
3.0
3.2
3.0
4.4
3.0
3.0
3.0
3.0
3.0
3.0
10p
3.0
3.6
3.0
6.2
3.1
3.0
1.9
2.8
2.8
5.8
3.0
3.4
3.0
4.8
3.0
3.0
2.9
3.0
3.0
3.0
3.0
3.0
3.0
4.4
3.0
3.0
3.0
3.0
3.0
3.0
20p
3.0
3.0
3.0
4.8
3.0
3.0
2.0
2.9
2.6
3.6
3.0
3.0
3.0
4.7
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
4.4
3.0
3.0
3.0
3.0
3.0
3.0
40o
3.0
3.0
3.0
4.5
3.0
3.0
2.0
2.9
2.5
3.0
3.0
3.0
3.0
4.7
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
4.4
3.0
3.0
3.0
3.0
3.0
3.0
515
3n3.
04.
23.
05.
12.
41.
71.
01.
12.
25.
23.
04.
33.
05.
22.
62.
61.
11.
72.
42.
43.
04.
23.
05.
52.
92.
62.
92.
72.
73.
35p
3.0
2.8
3.0
4.7
2.5
1.7
1.0
1.1
1.4
4.9
3.0
3.5
3.0
4.8
3.1
3.1
1.0
1.7
1.9
1.9
3.0
4.1
3.0
4.9
3.6
3.2
3.6
2.9
2.5
3.3
10D
3.0
2.0
3.0
4.2
2.9
1.9
1.0
1.0
1.0
4.3
3.0
3.2
3.0
4.2
3.7
3.7
1.0
1.8
1.6
1.6
3.0
4.2
2.6
3.5
4.2
3.8
4.2
3.1
2.4
3.3
20p
3.0
1.9
2.7
3.6
3.4
2.3
1.0
1.0
1.0
3.7
3.0
3.3
2.4
3.1
4.1
4.1
1.0
1.8
1.4
1.4
3.0
4.3
0.5
0.7
4.5
4.2
4.5
3.1
2.3
3.2
400
3.0
2.0
1.2
1.5
3.8
2.7
1.0
1.0
1.0
3.3
3.0
3.3
1.0
1.3
4.3
4.3
1.0
1.8
1.3
1.3
3.0
4.3
0.0
0.0
4.7
4.4
4.7
3.2
2.2
3.2
525
3p3.
08.
63.
011
.23.
42.
41.
01.
53.
68.
23.
07.
53.
011
.33.
63.
61.
63.
33.
73.
73.
05.
53.
010
.94.
13.
84.
14.
74.
04.
65o
3.0
6.0
3.0
10.1
3.9
2.8
1.0
1.5
2.4
7.5
3.0
6.0
3.0
10.3
4.3
4.3
1.7
3.5
3.2
3.2
3.0
5.0
3.0
10.0
4.7
4.5
4.7
4.8
3.9
4.6
10D
3.0
4.5
3.0
9.2
4.7
3.5
1.0
1.4
1.7
6.3
3.0
5.0
3.0
9.3
4.9
4.9
1.9
3.7
3.0
3.0
3.0
4.9
3.0
8.8
4.9
4.9
4.9
4.9
3.8
4.6
200
3.0
4.4
3.0
8.4
5.0
4.2
1.0
1.4
1.4
5.2
3.0
5.0
3.0
8.2
5.0
5.0
2.0
3.8
2.8
2.8
3.0
4.9
3.0
7.3
4.9
4.9
4.9
4.9
3.8
4.7
40p
3.0
4.5
3.0
7.4
5.1
4.7
1.0
1.4
1.3
4.9
3.0
5.0
3.0
7.0
5.0
5.0
2.1
3.8
2.7
2.7
3.0
4.9
3.0
5.4
5.0
4.9
5.0
4.9
3.8
4.6
550
3p3.
020
.83.
026
.94.
54.
02.
14.
15.
714
.43.
016
.83.
027
.05.
05.
03.
95.
05.
05.
03.
011
.43.
023
.85.
04.
95.
05.
05.
05.
05p
3.0
14.4
3.0
24.7
4.8
4.6
2.2
4.3
4.6
12.6
3.0
12.7
3.0
24.6
5.0
5.0
4.2
5.0
5.0
5.0
3.0
7.2
3.0
19.5
5.0
5.0
5.0
5.0
5.0
5.0
10p
3.0
8.1
3.0
22.3
5.0
4.9
2.3
4.4
4.2
9.2
3.0
8.4
3.0
21.2
5.0
5.0
4.4
5.0
5.0
5.0
3.0
5.0
3.0
15.2
5.0
5.0
5.0
5.0
5.0
5.0
20p
3.0
5.2
3.0
19.9
5.0
5.0
2.4
4.5
4.0
5.6
3.0
5.6
3.0
17.6
5.0
5.0
4.4
5.0
5.0
5.0
3.0
5.0
3.0
13.3
5.0
5.0
5.0
5.0
5.0
5.0
40p
3.0
5.0
3.0
17.3
5.0
5.0
2.5
4.5
3.9
5.0
3.0
5.0
3.0
14.7
5.0
5.0
4.4
5.0
5.0
5.0
3.0
5.0
3.0
12.4
5.0
5.0
5.0
5.0
5.0
5.0
721
3n3.
07.
03.
08.
73.
22.
11.
01.
23.
17.
33.
06.
73.
09.
03.
43.
41.
32.
33.
33.
33.
06.
33.
09.
44.
23.
64.
23.
43.
74.
751
33.
04.
93.
08.
13.
62.
21.
01.
11.
96.
93.
05.
73.
08.
34.
44.
41.
32.
42.
72.
73.
06.
13.
08.
45.
24.
65.
23.
73.
64.
710
p3.
03.
63.
07.
34.
52.
81.
01.
01.
16.
13.
05.
23.
07.
55.
55.
51.
42.
42.
22.
23.
06.
13.
06.
96.
05.
46.
03.
93.
44.
720
p3.
03.
43.
06.
75.
43.
51.
01.
01.
05.
43.
05.
23.
06.
76.
06.
01.
52.
52.
02.
03.
06.
23.
05.
16.
45.
96.
44.
03.
34.
640
p3.
03.
63.
05.
75.
84.
21.
01.
01.
05.
03.
05.
43.
05.
46.
26.
21.
62.
51.
91.
93.
06.
33.
03.
96.
76.
26.
74.
23.
34.
77
3530
3.0
13.6
3.0
17.7
4.0
3.0
1.1
2.1
4.8
11.0
3.0
12.1
3.0
17.9
5.1
5.1
2.1
4.5
5.2
5.2
3.0
8.7
3.0
17.7
5.7
5.3
5.7
6.4
5.6
6.4
5p3.
09.
73.
016
.25.
03.
71.
12.
134
10.0
3.0
9.7
3.0
16.4
6.0
6.0
2.3
4.8
4.6
4.6
3.0
7.2
3.0
16.0
6.5
6.2
6.5
6.6
5.4
6.4
10v
3.0
6.8
3.0
14.7
6.0
4.8
1.1
2.1
2.5
8.2
3.0
7.6
3.0
14.6
6.8
6.8
2.4
5.0
4.2
4.2
3.0
7.0
3.0
13.9
6.8
6.7
6.8
6.7
5.3
6.5
20D
3.0
6.1
3.0
13.2
6.6
5.5
1.1
2.1
2.1
6.6
3.0
7.0
3.0
13.1
7.0
7.0
2.5
5.2
4.0
4.0
3.0
7.0
3.0
11.5
7.0
6.9
7.0
6.8
5.3
6.5
400
3.0
6.1
3.0
11.6
6.8
6.0
1.1
2.1
1.9
6.3
3.0
7.0
3.0
11.4
7.0
7.0
2.6
5.3
3.9
3.9
3.0
7.0
3.0
8.9
7.0
7.0
7.0
6.8
5.2
6.5
770
3p3.
032
.73.
039
.36.
15.
52.
96.
08.
020
.13.
026
.33.
039
.46.
96.
95.
27.
07.
07.
03.
019
.63.
039
.16.
96.
96.
97.
07.
07.
05D
3.0
23.6
3.0
36.2
6.8
6.5
3.0
6.3
6.6
17.6
3.0
20.7
3.0
36.2
7.0
7.0
5.6
7.0
7.0
7.0
3.0
13.0
3.0
35.7
7.0
7.0
7.0
7.0
7.0
7.0
10p
3.0
14.3
3.0
33.0
7.0
6.9
3.0
6.4
6.2
12.8
3.0
14.5
3.0
32.7
7.0
7.0
5.7
7.0
6.9
6.9
3.0
7.6
3.0
31.9
7.0
7.0
7.0
7.0
7.0
7.0
20D
3.0
8.6
3029
.97.
07.
03.
06.
55.
97.
83.
09.
730
29.4
7.0
7.0
5.6
7.0
6.9
6.9
3.0
7.0
3.0
28.4
7.0
7.0
7.0
7.0
7.0
7.0
40p
3.0
7.0
3.0
26.7
7.0
7.0
3.1
6.5
5.7
7.0
3.0
7.1
3.0
26.2
7.0
7.0
5.5
7.0
6.9
6.9
3.0
7.0
3.0
24.7
7.0
7.0
7.0
7.0
7.0
7.0
Fact
or R
eten
tion 31
Tab
le 4
Mea
n N
umbe
r of
Fac
tors
Ret
aine
d, P
hi =
0.5
0
kp
Low
Com
mun
ality
Wid
e C
omm
unal
ityH
igh
Com
mun
ality
SESi
ze C
NG
Scr
eeM
RT
M-
PAM
H P
AIL
PA
LL
MA
P K
SR K
SRSE
CN
G S
cree
MR
TSE
PAm
H P
AIL
PA
u, M
AP
KSR
M-K
SR C
NG
Scr
ee M
R -
TM
-PA
mH
PA
cL P
AL
L M
AP
KSR
KSR
39
3n3.
12.
00.
00.
01.
71.
40.
81.
11.
43.
03.
01.
90.
00.
01.
51.
51.
01.
21.
31.
33.
02.
10.
00.
01.
61.
41.
61.
61.
41.
9
5v3.
01.
40.
00.
01.
61.
30.
91.
01.
12.
83.
01.
60.
00.
01.
61.
61.
01.
21.
11.
13.
02.
00.
00.
01.
81.
61.
81.
61.
31.
8
10D
3.0
1.2
0.0
0.0
1.7
1.3
1.0
1.0
1.0
2.4
3.0
1.4
0.0
0.0
1.9
1.9
1.0
1.2
1.0
1.0
3.0
2.0
0.0
0.0
2.2
2.0
2.2
1.7
1.2
1.7
20p
3.0
1.1
0.0
0.0
1.8
1.4
1.0
1.0
1.0
2.0
3.0
1.4
0.0
0.0
2.2
2.2
1.0
1.1
1.0
1.0
3.0
2.1
0.0
0.0
2.5
2.3
2.5
1.7
1.1
1.7
400
3.0
1.1
0.0
0.0
1.9
1.5
1.0
1.0
1.0
1.8
3.0
1.4
0.0
0.0
2.5
2.5
1.0
1.1
1.0
1.0
3.0
2.1
0.0
0.0
2.7
2.4
2.7
1.7
1.1
1.7
315
3D3.
03.
93.
05.
22.
01.
51.
01.
12.
04.
83.
03.
63.
05.
11.
91.
91.
11.
82.
02.
03.
03.
03.
04.
92.
11.
92.
12.
72.
12.
65p
3.0
2.5
3.0
4.7
2.0
1.5
1.0
1.1
1.2
4.4
3.0
2.9
3.0
4.6
2.3
2.3
1.1
1.9
1.7
1.7
3.0
3.0
3.0
4.5
2.6
2.4
2.6
2.8
2.0
2.6
10D
3.0
1.9
3.0
4.3
2.4
1.7
1.0
1.0
1.0
3.7
3.0
2.6
3.0
4.1
2.7
2.7
1.1
1.9
1.4
1.4
3.0
3.0
2.7
3.7
2.9
2.8
2.9
2.8
1.9
2.6
20D
3.0
1.8
2.6
3.5
2.7
2.0
1.0
1.0
1.0
3.0
3.0
2.7
2.3
2.9
2.9
2.9
1.1
2.0
1.4
1.4
3.0
3.0
0.7
0.9
3.0
3.0
3.0
2.9
1.9
2.6
40p
3.0
1.9
1.0
1.3
2.9
2.4
1.0
1.0
1.0
2.7
3.0
2.7
0.7
0.9
3.0
3.0
1.1
2.0
1.3
1.3
3.0
3.0
0.0
0.0
3.0
3.0
3.0
2.9
1.9
2.6
330
3D3.
010
.03.
014
.32.
52.
11.
12.
03.
38.
53.
08.
13.
013
.52.
82.
81.
83.
03.
03.
03.
04.
73.
08.
13.
02.
93.
03.
03.
03.
05D
3.0
6.3
3.0
12.6
2.8
2.5
1.1
2.1
2.5
7.4
3.0
5.6
3.0
10.9
2.9
2.9
2.0
3.0
2.9
2.9
3.0
3.2
3.0
5.7
3.0
3.0
3.0
3.0
3.0
3.0
10D
3.0
3.5
3.0
10.4
3.0
2.9
1.1
2.2
2.1
5.4
3.0
3.5
3.0
7.1
3.0
3.0
2.2
3.0
2.8
2.8
3.0
3.0
3.0
5.0
3.0
3.0
3.0
3.0
3.0
3.0
20D
3.0
3.0
3.0
8.0
3.0
3.0
1.1
2.2
1.9
3.3
3.0
3.0
3.0
5.3
3.0
3.0
2.3
3.0
2.8
2.8
3.0
3.0
3.0
4.9
3.0
3.0
3.0
3.0
3.0
3.0
40D
3.0
3.0
3.0
6.1
3.0
3.0
1.1
2.2
1.7
3.0
3.0
3.0
3.0
4.8
3.0
3.0
2.4
3.0
2.7
2.7
3.0
3.0
3.0
4.9
3.0
3.0
3.0
3.0
3.0
3.0
515
3n3.
03.
93.
05.
22.
01.
51.
01.
12.
05.
03.
03.
73.
05.
51.
61.
61.
01.
41.
81.
83.
03.
63.
05.
52.
01.
72.
02.
32.
02.
75D
3.0
2.5
3.0
4.7
2.0
1.4
1.0
1.0
1.2
4.6
3.0
3.0
3.0
5.0
2.0
2.0
1.0
1.3
1.3
1.3
3.0
3.5
3.0
4.9
2.7
2.2
2.7
2.5
1.8
2.6
10D
3.0
1.6
3.0
4.2
2.2
1.4
1.0
1.0
1.0
4.0
3.0
2.6
3.0
4.3
2.9
2.9
1.0
1.3
1.0
1.0
3.0
3.5
2.8
3.8
3.4
3.0
3.4
2.6
1.6
2.5
20D
3.0
1.4
2.8
3.6
2.7
1.6
1.0
1.0
1.0
3.2
3.0
2.6
2.5
3.3
3.6
3.6
1.0
1.2
1.0
1.0
3.0
3.6
1.0
1.2
3.8
3.5
3.8
2.7
1.5
2.5
40p
3.0
1.3
1.5
1.9
3.3
1.9
1.0
1.0
1.0
2.7
3.0
2.7
1.3
1.6
4.0
4.0
1.0
1.2
1.0
1.0
3.0
3.6
0.0
0.1
4.1
3.7
4.1
2.7
1.4
2.5
525
3D3.
07.
93.
011
.61.
91.
41.
01.
22.
97.
63.
07.
13.
011
.92.
42.
41.
12.
63.
03.
03.
05.
43.
011
.82.
62.
22.
64.
13.
03.
8
5p3.
05.
23.
010
.52.
21.
51.
01.
11.
66.
83.
05.
63.
010
.83.
33.
31.
12.
72.
42.
43.
04.
93.
010
.63.
73.
13.
74.
32.
73.
7
10p
3.0
3.3
3.0
9.3
3.0
2.0
1.0
1.1
1.1
5.4
3.0
4.7
3.0
9.6
4.3
4.3
1.1
2.9
2.1
2.1
3.0
4.9
3.0
9.2
4.6
4.2
4.6
4.5
2.5
3.7
20D
3.0
3.0
3.0
8.3
3.8
2.6
1.0
1.1
1.0
4.2
3.0
4.7
3.0
8.5
4.8
4.8
1.1
2.9
1.9
1.9
3.0
5.0
3.0
7.6
4.9
4.7
4.9
4.5
2.4
3.7
40D
3.0
3.1
3.0
7.5
4.4
3.2
1.0
1.0
1.0
3.7
3.0
4.7
3.0
7.2
4.9
4.9
1.1
3.0
1.7
1.7
3.0
5.0
3.0
5.8
5.0
4.9
5.0
4.5
2.4
3.7
550
3v3.
020
.73.
027
.12.
92.
21.
12.
75.
414
.23.
016
.33.
027
.24.
04.
02.
14.
84.
84.
83.
011
.23.
027
.04.
64.
44.
65.
04.
95.
05o
3.0
14.2
3.0
25.0
3.9
3.1
1.0
2.9
3.7
12.5
3.0
12.2
3.0
25.0
4.6
4.6
2.4
4.9
4.4
4.4
3.0
7.0
3.0
24.6
5.0
4.9
5.0
5.0
4.9
5.0
10D
3.0
8.0
3.0
22.8
4.8
4.3
1.0
3.0
2.8
9.1
3.0
8.0
3.0
22.5
5.0
5.0
2.7
5.0
4.1
4.1
3.0
5.0
3.0
21.9
5.0
5.0
5.0
5.0
4.9
5.0
20D
3.0
5.2
3.0
20.7
4.9
4.8
1.0
3.1
2.3
5.6
3.0
5.4
3.0
20.2
5.0
5.0
2.8
5.0
4.0
4.0
3.0
5.0
3.0
19.5
5.0
5.0
5.0
5.0
4.9
5.0
40D
3.0
5.0
3.0
18.6
5.0
4.9
1.0
3.1
2.1
5.0
3.0
5.0
3.0
18.0
5.0
5.0
2.9
5.0
3.9
193.
05.
03.
017
.05.
05.
05.
05.
04.
95.
07
213n
3.0
6.3
3.0
9.0
1.8
1.3
1.0
1.1
2.5
6.6
3.0
6.1
3.0
9.4
1.8
1.8
1.0
1.5
2.4
2.4
3.0
5.4
3.0
9.5
1.9
1.6
1.9
2.6
2.4
3.5
5p3.
04.
03.
08.
21.
91.
31.
01.
01.
36.
03.
04.
93.
08.
62.
52.
51.
01.
41.
61.
63.
05.
23.
08.
63.
12.
43.
12.
72.
03.
3
10D
3.0
2.5
3.0
7.4
2.6
1.5
1.0
1.0
1.0
5.0
3.0
4.2
3.0
7.7
4.1
4.1
1.0
1.3
1.1
1.1
3.0
5.3
3.0
7.3
4.8
3.8
4.8
2.9
1.7
3.2
20p
3.0
2.1
3.0
6.7
3.7
2.0
1.0
1.0
1.0
4.0
3.0
4.2
3.0
6.9
5.3
5.3
1.0
1.3
1.0
1.0
3.0
5.4
3.0
5.7
5.7
4.8
5.7
3.0
1.4
3.1
40D
3.0
2.2
3.0
5.9
4.7
2.6
1.0
1.0
1.0
3.4
3.0
4.4
3.0
6.0
6.0
6.0
1.0
1.2
1.0
1.0
3.0
5.5
3.0
4.4
6.2
5.4
6.2
3.0
1.3
3.1
735
313
3.0
13.0
3.0
17.9
1.8
1.3
1.0
1.3
4.1
10.6
3.0
11.7
3.0
18.1
2.7
2.7
1.0
3.0
4.1
4.1
3.0
8.4
3.0
17.9
3.5
3.0
3.5
5.5
4.2
5.2
5v3.
09.
03.
016
.52.
51.
51.
01.
22.
19.
63.
09.
23.
016
.64.
44.
41.
03.
23.
13.
13.
07.
03.
016
.24.
94.
34.
95.
83.
95.
1
10p
3.0
5.9
3.0
15.0
4.4
2.4
1.0
1.1
1.1
7.8
3.0
7.1
3.0
14.9
6.0
6.0
1.0
3.3
2.4
2.4
3.0
6.8
3.0
14.2
6.2
5.6
6.2
5.9
3.7
5.1
20D
3.0
5.2
3.0
13.7
6.0
4.0
1.0
1.1
1.1
6.2
3.0
6.5
3.0
13.4
6.5
6.5
1.0
3.3
1.9
1.9
3.0
6.8
3.0
12.0
6.8
6.4
6.8
6.0
3.6
5.1
40p
3.0
5.4
3.0
12.3
6.6
5.0
1.0
1.1
1.0
5.6
3.0
6.5
3.0
11.9
6.8
6.8
1.0
3.4
1.8
1.8
3.0
6.8
3.0
9.5
6.9
6.8
6.9
6.0
3.5
5.1
770
3D3.
032
.43.
039
.43.
22.
41.
13.
77.
519
.83.
025
.93.
039
.65.
25.
22.
66.
66.
66.
63.
019
.43.
039
.36.
25.
96.
27.
06.
87.
051
33.
023
.43.
036
.55.
03.
91.
13.
85.
117
.33.
020
.43.
036
.56.
36.
33.
06.
76.
16.
13.
012
.93.
036
.16.
96.
76.
97.
06.
87.
010
D3.
014
.13.
033
.56.
75.
91.
14.
03.
812
.73.
014
.33.
033
.26.
96.
93.
26.
85.
85.
83.
07.
53.
032
.57.
07.
07.
07.
06.
77.
020
D3.
08.
53.
030
.77.
06.
81.
24.
03.
17.
73.
09.
63.
030
.27.
07.
03.
36.
85.
75.
73.
07.
03.
029
.17.
07.
07.
07.
06.
77.
040
o3.
07.
03.
027
.97.
07.
01.
34.
02.
87.
03.
07.
13.
027
.17.
07.
03.
46.
85.
75.
73.
07.
03.
025
.57.
07.
07.
07.
06.
77.
0
Fact
or R
eten
tion 32
Tab
le 5
Mea
n N
umbe
r of
Fac
tors
Ret
aine
d, P
hi =
Mix
ed
kp
Low
Com
mun
ality
Wid
e C
omm
unal
ityH
igh
Com
mun
ality
SESi
ze C
NG
Scr
eeM
RT
M-
PAS
PAa.
PA
LL
MA
P K
SR K
SRSE
CN
G S
cree
MR
TSE
PAS
P&L
PA
LL
MA
P K
SR M
-KSR
CN
G S
cree
MR
TM
-PA
mH
PA
LL
PA
LL
MA
P K
SR K
SR
39
3n3.
12.
10.
00.
0L
81.
40.
71.
11.
53.
23.
02.
30.
00.
01.
91.
91.
11.
41.
71.
73.
02.
30.
00.
01.
91.
81.
91.
91.
82.
1
5o3.
01.
50.
00.
01.
81.
40.
81.
01.
13.
03.
02.
00.
00.
02.
12.
11.
11.
51.
41.
43.
02.
20.
00.
02.
22.
02.
22.
01.
72.
0
10p
3.0
1.3
0.0
0.0
1.9
1.4
0.9
1.0
1.0
2.6
3.0
1.9
0.0
0.0
2.4
2.4
1.1
1.4
1.3
1.3
3.0
2.2
0.0
0.0
2.4
2.2
2.4
2.0
1.7
2.0
20p
3.0
1.2
0.0
0.0
2.1
1.6
1.0
1.0
1.0
2.2
3.0
1.9
0.0
0.0
2.6
2.6
1.2
1.4
1.2
1.2
3.0
2.3
0.0
0.0
2.6
2.4
2.6
2.0
1.7
2.0
40o
3.0
1.2
0.0
0.0
2.2
1.7
1.0
1.0
1.0
2.0
3.0
1.9
0.0
0.0
2.7
2.7
1.2
1.4
1.2
1.2
3.0
2.3
0.0
0.0
2.8
2.5
2.8
2:0
1.7
2.0
315
3p3.
14.
33.
14.
52.
92.
31.
11.
52.
55.
03.
03.
63.
04.
32.
22.
21.
32.
02.
22.
23.
03.
13.
04.
12.
52.
42.
52.
92.
42.
8
5D3.
03.
23.
04.
13.
02.
51.
11.
52.
04.
63.
02.
93.
03.
82.
42.
41.
42.
12.
02.
03.
03.
03.
03.
92.
92.
72.
92.
92.
32.
8
10p
3.0
2.7
3.0
3.8
3.0
2.7
1.1
1.6
1.7
3.9
3.0
2.7
2.9
3.5
2.7
2.7
1.5
2.1
1.9
1.9
3.0
3.0
2.6
3.3
3.0
3.0
3.0
3.0
2.3
2.8
20p
3.0
2.7
2.2
2.7
3.0
2.8
1.1
1.6
1.6
3.1
3.0
2.7
1.9
2.2
2.9
2.9
1.6
2.1
1.8
1.8
3.0
3.0
0.6
0.7
3.0
3.0
3.0
3.0
2.2
2.8
40p
3.0
2.8
0.3
0.4
3.0
2.9
1.2
1.6
1.5
3.0
3.0
2.8
0.3
0.3
3.0
3.0
1.7
2.1
1.8
1.8
3.0
3.0
0.0
0.0
3.0
3.0
3.0
3.0
2.2
2.8
330
3p3.
010
.23.
07.
02.
92.
61.
92.
53.
58.
83.
08.
13.
05.
02.
92.
92.
33.
03.
03.
03.
04.
73.
04.
73.
03.
03.
03.
03.
03.
0
5p3.
06.
53.
04.
93.
02.
82.
02.
52.
87.
73.
05.
73.
03.
93.
03.
02.
43.
02.
92.
93.
03.
23.
04.
33.
03.
03.
03.
03.
03.
0
10v
3.0
3.6
3.0
3.7
3.0
3.0
2.0
2.5
2.4
5.6
3.0
3.5
3.0
3.7
3.0
3.0
2.5
3.0
2.9
2.9
3.0
3.0
3.0
4.3
3.0
3.0
3.0
3.0
3.0
3.0
20p
3.0
3.0
3.0
3.6
3.0
3.0
2.0
2.6
2.3
3.5
3.0
3.0
3.0
3.7
3.0
3.0
2.6
3.0
2.9
2.9
3.0
3.0
3.0
4.3
3.0
3.0
3.0
3.0
3.0
3.0
40p
3.0
3.0
3.0
3.6
3.0
3.0
2.0
2.6
2.2
3.0
3.0
3.0
3.0
3.7
3.0
3.0
2.6
3.0
2.9
2.9
3.0
3.0
3.0
4.2
3.0
3.0
3.0
3.0
3.0
3.0
515
3n3.
04.
23.
05.
02.
41.
81.
01.
22.
25.
13.
04.
23.
04.
92.
52.
51.
21.
82.
32.
33.
04.
03.
05.
02.
92.
62.
92.
72.
73.
2
5v3.
02.
93.
04.
62.
51.
81.
01.
11.
54.
73.
03.
43.
04.
42.
82.
81.
21.
81.
91.
93.
03.
83.
04.
63.
33.
03.
32.
82.
53.
1
10p
3.0
2.2
3.0
4.1
2.8
2.1
1.0
1.1
1.1
4.1
3.0
3.0
3.0
4.0
3.4
3.4
1.2
1.9
1.7
1.7
3.0
3.8
2.7
3.6
3.8
3.5
3.8
2.9
2.5
3.1
20p
3.0
2.1
2.6
3.3
3.2
2.4
1.0
1.1
1.1
3.4
3.0
3.0
2.5
3.1
3.9
3.9
1.3
1.9
1.6
1.6
3.0
3.9
0.7
0.9
4.2
3.8
4.2
3.0
2.4
3.1
40p
3.0
2.1
1.1
1.3
3.4
2.7
1.0
1.1
1.0
3.1
3.0
3.1
1.0
1.2
4.3
4.3
1.3
1.9
1.5
1.5
3.0
3.9
0.0
0.0
4.4
4.1
4.4
3.0
2.4
3.0
525
3p3.
08.
43.
010
.33.
32.
51.
21.
93.
57.
83.
07.
93.
011
.04.
24.
21.
93.
54.
04.
03.
05.
63.
010
.04.
03.
84.
04.
54.
04.
2
5p3.
05.
83.
09.
13.
62.
91.
21.
92.
57.
13.
06.
23.
010
.24.
74.
72.
13.
83.
63.
63.
05.
03.
09.
24.
24.
14.
24.
63.
94.
1
10p
3.0
4.1
3.0
7.9
4.1
3.4
1.2
1.9
2.0
5.8
3.0
5.1
3.0
9.2
4.9
4.9
2.3
4.0
3.3
3.3
3.0
4.9
3.0
8.4
4.7
4.3
4.7
4.7
4.0
4.1
20p
3.0
3.8
3.0
7.2
4.3
3.7
1.2
1.9
1.8
4.5
3.0
4.9
3.0
8.2
5.0
5.0
2.4
4.1
3.1
3.1
3.0
4.9
3.0
7.2
4.9
4.8
4.9
4.8
4.0
4.1
4013
3.0
3.9
3.0
6.5
4.5
3.9
1.2
1.9
1.7
4.0
3.0
4.9
3.0
7.1
5.0
5.0
2.5
4.1
3.0
3.0
3.0
5.0
3.0
5.5
5.0
4.9
5.0
4.8
4.0
4.0
550
3P3.
020
.83.
027
.03.
52.
91.
53.
25.
514
.43.
016
.33.
027
.14.
74.
73.
55.
05.
05.
03.
011
.53.
015
.04.
94.
84.
95.
05.
05.
0
5p3.
014
.33.
024
.94.
33.
71.
63.
33.
912
.63.
012
.33.
024
.75.
05.
03.
75.
04.
94.
93.
07.
23.
011
.15.
05.
05.
05.
05.
05.
0
1013
3.0
8.1
3.0
22.5
4.9
4.5
1.7
3.3
3.1
9.3
3.0
8.1
3.0
21.9
5.0
5.0
3.8
5.0
4.8
4.8
3.0
5.0
3.0
9.5
5.0
5.0
5.0
5.0
5.0
5.0
20p
3.0
5.2
3.0
20.3
5.0
4.9
1.8
3.4
2.8
5.7
3.0
5.4
3.0
18.9
5.0
5.0
3.9
5.0
4.8
4.8
3.0
5.0
3.0
9.2
5.0
5.0
5.0
5.0
5.0
5.0
40p
3.0
5.0
3.0
17.9
5.0
5.0
1.9
3.3
2.7
5.0
3.0
5.0
3.0
15.9
5.0
5.0
3.9
5.0
4.8
4.8
3.0
5.0
3.0
9.2
5.0
5.0
5.0
5.0
5.0
5.0
721
3n3.
06.
83.
08.
33.
12.
21.
01.
43.
17.
03.
06.
43.
08.
43.
33.
31.
42.
53.
23.
23.
05.
83.
09.
43.
32.
83.
33.
23.
24.
1
5D3.
04.
73.
07.
53.
42.
41.
01.
42.
06.
53.
05.
33.
07.
73.
93.
91.
52.
62.
72.
73.
05.
63.
08.
44.
33.
74.
33.
43.
04.
010
D3.
03.
53.
06.
74.
02.
81.
01.
31.
55.
63.
04.
73.
06.
94.
74.
71.
52.
72.
42.
43.
05.
63.
07.
15.
34.
65.
33.
52.
94.
0
20D
3.0
3.3
3.0
6.1
4,6
3.3
1.0
1.3
1.3
4.7
3.0
4.6
3.0
6.2
5.3
5.3
1.6
2.8
2.3
2.3
3.0
5.8
3.0
5.7
6.1
5.2
6.1
3.5
2.8
4.0
40D
3.0
3.4
3.0
5.4
4.9
3.7
1.0
1.3
1.2
4.2
3.0
4.6
3.0
5.4
5.7
5.7
1.7
2.8
2.3
2.3
3.0
5.9
3.0
4.3
6.3
5.8
6.3
3.5
2.8
4.0
735
3D3.
013
.63.
017
.64.
33.
41.
32.
44.
911
.13.
011
.73.
017
.74.
64.
62.
44.
74.
94.
93.
08.
43.
017
.44.
44.
14.
45.
94.
75.
4
513
3.0
9.6
3.0
16.1
5.0
4.0
1.2
2.5
3.6
10.1
3.0
9.3
3.0
16.1
5.5
5.5
2.6
5.0
4.3
4.3
3.0
6.9
3.0
15.8
5.3
4.9
5.3
6.0
4.5
5.4
10D
3.0
6.5
3.0
14.6
5.8
4.7
1.2
2.5
2.9
8.2
3.0
7.3
3.0
14.3
6.2
6.2
2.8
5.2
4.0
4.0
3.0
6.6
3.0
13.8
6.0
5.7
6.0
6.0
4.3
5.3
20p
3.0
5.8
3.0
13.0
6.1
5.3
1.2
2.5
2.5
6.3
3.0
6.7
3.0
12.7
6.7
6.7
2.9
5.4
3.9
3.9
3.0
6.6
3.0
11.9
6.3
6.0
6.3
6.0
4.2
5.3
4013
3.0
5.9
3.0
11.4
6.2
5.8
1.2
2.6
2.3
6.0
3.0
6.7
3.0
11.1
6.9
6.9
3.0
5.4
3.8
3.8
3.0
6.7
3.0
9.5
6.8
6.2
6.8
6.0
4.1
5.3
770
3o3.
032
.53.
039
.34.
74.
22.
94.
97.
619
.93.
026
.03.
039
.15.
95.
95.
16.
56.
46.
43.
019
.43.
033
.95.
75.
55.
77.
06.
46.
95D
3.0
23.5
3.0
36.2
5.7
5.0
2.9
5.0
5.6
17.3
3.0
20.2
3.0
35.2
6.1
6.1
5.2
6.7
6.0
6.0
3.0
13.0
3.0
30.1
6.3
6.1
6.3
7.0
6.2
6.8
1013
3.0
14.2
3.0
32.9
6.7
6.2
3.0
5.1
4.7
12.6
3.0
14.0
3.0
29.2
6.7
6.7
5.4
6.8
6.0
6.0
3.0
7.6
3.0
24.1
6.9
6.8
6.9
7.0
6.1
6.8
20D
3.0
8.5
3.0
29.6
7.0
6.8
3.0
5.2
4.3
7.7
3.0
9.4
3.0
24.6
6.9
6.9
5.3
6.9
5.9
5.9
3.0
7.0
3.0
20.5
7.0
6.9
7.0
7.0
6.0
6.8
40D
3.0
7.0
3.0
25.9
7.0
7.0
3.0
5.2
4.1
7.0
3.0
7.0
3.0
21.1
6.9
6.9
5.3
6.9
5.9
5.9
3.0
7.0
3.0
18.4
7.0
7.0
7.0
7.0
6.0
6.8
Tab
le 6
Om
ega
Squa
red
for
Mea
n N
umbe
r of
Fac
tors
Ret
aine
d.
Eff
ect
k p/k
N Com
mun
ality
Phi
k X
p/k
k X
Nk
X C
omm
k X
Phi
p/k
X N
p/k
X C
omm
p/k
X P
hiN
X C
omm
N X
Phi
Com
m X
Phi
k X
p/k
X N
k X
p/k
X C
omm
k X
N X
Com
mp/
k X
N X
Com
mk
X p
/k X
Phi
k X
N X
Phi
p/k
X N
X P
hiC
omm
X p
/k X
Phi
Com
m X
k X
Phi
Com
m X
n X
Phi
k X
p/k
X N
X C
omm
k X
p/k
X N
X P
hiC
omm
X k
X p
/k X
Phi
Com
m X
p/k
X n
X P
hiC
omm
X k
X n
X P
hik
X p
/k X
N X
Com
m X
phi
CN
GSE
AM
Rt
Para
llel A
naly
sis
MA
PK
SRM
-KSR
PAS
PAIL
PAL
L
0.02
70.
257
0.25
90.
320
0.61
00.
456
0.11
90.
258
0.27
10.
305
0.00
30.
230
0.23
90.
351
0.10
40.
180
0.34
00.
361
0.38
00.
183
0.00
00.
154
0.04
20.
034
0.06
20.
066
0.00
40.
001
0.03
20.
082
0.00
60.
011
0.00
30.
005
0.01
30.
045
0.14
20.
162
0.05
60.
128
0.16
00.
003
0.00
00.
060
0.05
50.
077
0.17
00.
032
0.08
70.
011
0.01
80.
055
0.25
90.
074
0.01
10.
021
0.03
30.
055
0.04
70.
022
0.00
10.
040
0.02
10.
005
0.01
80.
016
0.00
00.
000
0.00
50.
008
0.01
60.
004
0.00
20.
001
0.00
30.
007
0.01
80.
022
0.00
70.
014
0.08
70.
000
0.00
00.
024
0.01
50.
020
0.04
20.
008
0.02
00.
003
0.00
00.
126
0.02
20.
006
0.01
00.
006
0.00
00.
000
0.00
00.
034
0.00
20.
034
0.00
20.
004
0.00
20.
007
0.01
70.
030
0.00
50.
048
0.01
00.
001
0.00
00.
054
0.01
20.
013
0.02
10.
004
0.01
20.
004
0.00
00.
026
0.00
40.
000
0.00
10.
001
0.00
10.
000
0.01
70.
080
0.00
10.
000
0.00
10.
004
0.01
90.
012
0.00
00.
000
0.00
20.
000
0.01
80.
000
0.00
00.
001
0.00
50.
005
0.00
50.
006
0.00
10.
000
0.00
00.
030
0.10
60.
003
0.00
10.
001
0.00
00.
000
0.00
00.
004
0.00
20.
005
0.00
80.
001
0.00
00.
001
0.00
30.
002
0.00
00.
004
0.00
00.
004
0.00
20.
000
0.00
10.
000
0.00
00.
000
0.00
20.
008
0.00
00.
012
0.00
20.
001
0.00
30.
005
0.00
00.
001
0.00
30.
040
0.05
20.
000
0.00
10.
020
0.00
30.
003
0.00
30.
001
0.00
30.
001
0.00
30.
000
0.00
00.
001
0.00
60.
004
0.00
00.
000
0.00
00.
000
0.00
00.
000
0.00
00.
003
0.00
40.
004
0.00
10.
000
0.00
10.
000
0.00
60.
000
0.00
00.
002
0.00
10.
003
0.03
50.
016
0.01
30.
000
0.04
60.
000
0.00
00.
002
0.00
20.
002
0.00
40.
002
0.00
10.
000
0.00
00.
000
0.00
00.
000
0.00
10.
000
0.00
00.
000
0.00
00.
000
0.00
00.
002
0.01
10.
000
0.00
10.
001
0.00
00.
000
0.00
00.
004
0.00
00.
000
0.00
10.
001
0.00
10.
001
0.00
00.
000
0.00
00.
000
0.00
80.
000
0.00
00.
003
0.00
10.
001
0.00
40.
004
0.00
20.
001
0.00
00.
000
0.00
00.
000
0.00
30.
005
0.00
10.
000
0.00
20.
000
0.00
00.
000
0.00
00.
001
0.00
00.
000
0.00
00.
000
0.00
00.
000
0.00
00.
000
0.00
00.
001
0.00
00.
001
0.00
00.
000
0.00
00.
000
Fact
or R
eten
tion 34
Tab
le 7
Prop
ortio
n of
Sam
ples
Ret
aini
ng K
Fac
tors
, Phi
= 0
.00
kp
Low
Com
mun
ality
Wid
e C
omm
unal
ityH
igh
Com
mun
ality
SESi
ze C
NG
Scr
eeM
RT
M-
PAM
H P
AIL
PA
LL
MA
P K
SR K
SRSE
CN
G S
cree
MR
TSE
PAS
Ma,
PA
LL
MA
P K
SR M
-KSR
CN
G S
cree
MR
TM
-PA
S PA
ct P
AL
L M
AP
KSR
KSR
39
3nill
271
022
120
02
6193
400
029
290
26
692
560
042
3542
2421
42
5088
140
025
120
00
6796
370
042
420
21
193
560
054
4754
3020
43
10D
917
00
3214
00
065
9735
00
6262
01
00
9458
00
6157
6136
2143
20D
926
00
3817
00
048
9738
00
7777
01
00
9459
00
6560
6540
2343
40D
928
00
4422
00
034
9740
00
8686
00
00
9460
00
7062
7044
2442
315
3v90
2192
3949
460
763
088
3389
2683
8317
7186
8687
8985
2296
9396
9594
99
Si,
9257
9341
6363
06
232
8975
8719
8989
2980
8080
8910
085
1810
099
100
9894
100
10D
9485
9237
8185
06
721
9095
839
9494
4387
7979
9110
069
1210
010
010
099
9610
0
20p
9693
6321
9297
06
281
9196
483
9898
5489
7979
9110
09
110
010
010
099
9710
0
40D
9698
62
9810
00
71
100
9197
70
9999
6089
7979
9210
00
010
010
010
010
098
100
330
3D82
078
1785
9778
9846
089
087
910
010
099
9997
9793
990
1510
010
010
010
010
010
0
Si,
811
7312
9299
8910
010
00
901
867
100
100
100
100
100
100
9284
8411
100
100
100
100
100
100
10D
8056
699
9710
094
100
100
091
5883
710
010
010
010
010
010
094
100
8210
100
100
100
100
100
100
20D
8010
064
899
100
9810
010
062
9110
081
910
010
010
010
010
010
095
100
8210
100
100
100
100
100
100
40p
8010
062
710
010
010
010
010
010
090
100
7910
100
100
100
100
100
100
9610
081
1010
010
010
010
010
010
0
515
3n5
303
1914
40
00
434
393
2014
140
00
011
739
3236
2236
144
24
Si,
417
328
205
00
055
527
232
2222
00
00
1175
1053
5940
5920
324
10D
37
132
336
00
066
715
I25
4040
00
00
1178
17
8162
8123
225
20D
26
07
5112
00
053
912
04
6060
00
00
1282
00
9377
9324
125
4013
18
00
6321
00
040
1214
00
7070
00
00
1386
00
9888
9825
025
525
3D4
11
1036
270
035
06
11
1063
631
4448
4810
511
197
9397
8993
99
513
519
111
5042
00
20
721
17
8282
254
3030
896
00
100
9910
092
9310
0
10D
577
111
7169
00
05
785
03
9393
361
2020
610
00
010
010
010
094
9210
0
2013
890
19
8987
00
076
796
01
9696
364
1616
310
01
110
010
010
095
9110
0
4013
1094
18
9896
00
010
07
961
299
992
6514
141
100
1159
100
100
100
9590
100
550
3p4
00
195
9737
9523
02
00
010
010
010
010
098
985
00
010
010
010
098
100
100
5D4
00
098
100
4599
100
01
00
010
010
010
010
010
010
03
30
010
010
010
099
100
100
10D
31
00
9910
047
100
100
00
00
010
010
010
010
010
010
01
990
010
010
010
099
100
100
20D
278
00
100
100
4610
010
038
057
00
100
100
100
100
100
100
010
00
010
010
010
099
100
100
40D
110
00
010
010
044
100
100
100
010
00
010
010
010
010
010
010
00
100
00
100
100
100
9910
010
0
721
3n1
240
811
20
00
331
350
1114
140
00
02
591
1212
312
40
3
Si,
124
110
173
00
050
244
016
2727
00
00
252
119
3016
307
03
10D
111
122
305
00
061
336
144
4040
00
00
150
148
4734
4710
02
20D
111
133
4710
00
043
433
039
5252
00
00
151
01
6044
6013
01
40D
115
0I
6318
00
036
632
03
6363
00
00
051
00
7850
7814
00
735
3D2
00
833
160
021
02
00
661
610
4343
435
50
093
8293
8482
99
5v3
20
1045
320
00
03
10
380
800
5325
253
790
010
099
100
8677
100
10D
548
08
6164
00
01
253
01
8989
056
1717
210
00
010
010
010
087
7110
0
20D
982
04
7679
00
071
293
00
9292
057
1616
110
00
010
010
010
085
6610
0
40D
1181
01
8880
00
082
291
00
9797
057
1616
010
00
910
010
010
085
6310
0
770
3o0
00
099
9514
9710
00
00
010
010
091
100
9999
10
00
100
100
100
9910
010
0
513
00
00
100
100
1799
100
00
00
010
010
095
100
100
100
10
00
100
100
100
9910
010
0
1013
00
00
100
100
1410
010
00
00
00
100
100
9510
010
010
01
570
010
010
010
010
010
010
0
20D
011
00
100
100
810
010
027
01
00
100
100
9310
010
010
00
100
00
100
100
100
100
100
100
40p
099
00
100
100
110
010
010
00
890
010
010
090
100
100
100
010
00
010
010
010
010
010
010
0
Fact
or R
eten
tion 35
Tab
le 8
Prop
ortio
n of
Sam
ples
Ret
aini
ng K
Fac
tors
, Phi
= 0
.30
kp
Low
Com
mun
ality
Wid
e C
omm
unal
ityH
igh
Com
mun
ality
SESi
ze C
NG
Scr
eeM
RT
M-
PAS
PAL
LL
, PA
LL
MA
P K
SR K
SRSE
CN
G S
cree
MR
TSE
PAN
ili P
Aci
. PA
LL
MA
P K
SR M
-KSR
CN
G S
cree
MR
TM
-PA
mi.,
PA
S PA
LL
MA
P K
SR K
SR3
93n
9221
10
157
00
164
9827
10
1414
00
11
100
490
025
1525
155
275p
988
00
165
00
064
9917
00
2222
00
00
100
500
045
3045
182
2410
o10
01
00
183
00
053
100
100
041
410
00
010
054
00
6552
6522
124
20D
100
00
024
30
00
3010
08
00
6767
00
00
100
580
078
6378
240
2540
p10
00
00
315
00
018
100
70
082
820
00
010
060
00
9069
9025
024
315
3v10
027
100
3333
200
233
110
038
100
3351
511
3343
4310
091
100
1978
6878
9072
91
5D10
047
100
3743
250
13
610
075
100
3173
731
3824
2410
099
100
1393
8993
9672
9210
D10
048
9938
6343
00
040
100
9298
2393
931
3917
1710
010
085
810
097
100
9971
9220
D10
051
7727
8061
00
086
100
9667
1298
982
3915
1510
010
015
110
010
010
010
070
91
40D
100
5615
593
720
00
8610
099
123
9999
236
1616
100
100
00
100
100
100
100
7090
330
3D10
00
100
1475
746
6347
010
00
100
399
9967
9997
9710
08
100
210
010
010
010
010
010
0-5D
100
110
010
9094
774
940
100
210
01
100
100
8410
010
010
010
083
100
010
010
010
010
010
010
01
OD
100
5610
04
9510
08
8277
010
062
100
010
010
094
100
100
100
100
100
100
010
010
010
010
010
010
020
p10
010
010
01
9810
010
8862
4810
010
010
00
100
100
9810
010
010
010
010
010
00
100
100
100
100
100
100
4013
100
100
100
099
100
1292
5110
010
010
010
00
100
100
100
100
100
100
100
100
100
010
010
010
010
010
010
05
153n
021
016
40
00
057
027
017
33
00
00
026
024
10
11
00
5to
05
029
40
00
058
010
037
55
00
00
021
061
61
60
00
10p
00
040
50
00
035
03
041
1414
00
00
023
018
287
280
00
2013
00
018
90
00
08
02
012
2727
00
00
028
00
5323
530
00
40D
00
01
160
00
01
03
02
3939
00
00
032
00
7337
730
00
525
3D0
30
111
20
04
00
30
012
120
74
40
550
031
1731
7918
595D
026
01
222
00
00
031
01
3838
010
00
093
00
7252
7287
1261
10D
043
01
4810
00
011
087
00
8484
013
00
093
00
9188
9190
963
20D
040
00
7832
00
079
097
00
9898
015
00
092
00
9291
9290
765
40:1
054
00
9467
00
094
099
00
9999
018
00
092
049
9991
9990
764
550
3v0
00
049
230
2837
00
00
095
9516
9998
980
00
096
9496
100
100
100
5to
00
00
8262
037
640
00
00
100
100
3210
010
010
00
30
010
099
100
100
9910
0
10p
02
00
9590
044
240
00
00
100
100
4710
099
990
990
010
010
010
010
099
100
20to
083
00
100
960
4810
490
530
010
010
051
100
9898
010
00
010
010
010
010
098
100
40D
010
00
010
010
00
507
100
010
00
010
010
051
100
9898
010
00
010
010
010
010
098
100
721
3n0
230
52
00
00
520
330
30
00
00
00
310
10
00
00
05D
08
08
20
00
055
014
06
11
00
00
023
05
30
30
00
10D
00
020
40
00
027
01
030
55
00
00
023
062
194
190
00
20D
00
051
80
00
02
00
054
1212
00
00
027
05
4115
410
00
40D
00
019
150
00
00
00
014
1818
00
00
031
00
6528
650
00
735
3D0
00
03
00
00
00
00
04
40
20
00
110
019
719
688
485D
08
00
70
00
00
02
00
2727
04
00
084
00
5944
5975
552
10D
039
00
240
00
018
050
00
7474
06
00
010
00
084
7084
804
5520
D0
250
047
60
00
540
960
095
950
80
00
100
00
100
8810
084
457
40D
028
00
6820
00
042
097
00
9999
09
00
010
00
210
010
010
084
358
770
3D0
00
035
120
2922
00
00
086
863
100
9999
00
00
9490
9410
010
010
0SD
00
00
8255
043
640
00
00
9999
710
098
980
00
010
010
010
010
010
010
010
D0
00
010
094
052
300
00
00
100
100
910
094
940
530
010
010
010
010
010
010
020
D0
150
010
010
00
5420
420
10
010
010
09
100
9191
010
00
010
010
010
010
010
010
040
D0
100
00
100
100
055
1610
00
920
010
010
07
100
9090
010
00
010
010
010
010
010
010
0
Fact
or R
eten
tion 36
Tab
le 9
Prop
ortio
n of
Sam
ples
Ret
aini
ng K
Fac
tors
, Phi
= 0
.50
kp
Low
Com
mun
ality
Wid
e C
omm
unal
ityH
igh
Com
mun
ality
SESi
ze C
NG
Scr
eeM
RT
M-
PAN
TH
Ma,
PA
LL
MA
P K
SR K
SRSE
CN
G S
cree
MR
TSE
PAS
PAct
, PA
M. M
AP
KSR
M-K
SR C
NG
Scr
ee M
RT
M-
PAL
m P
Act
PA
LL
MA
P K
SR K
SR
39
3n96
191
011
50
01
6410
018
10
77
00
00
100
231
06
36
41
9
5v99
50
010
30
00
6110
07
00
88
00
00
100
180
014
614
20
5
10p
100
00
010
10
00
3910
01
00
1414
00
00
100
170
033
1533
20
3
20p
100
00
09
00
00
1010
00
00
3030
00
00
100
200
057
3157
10
2
400
100
00
010
00
00
210
00
00
5151
00
00
100
210
071
4971
10
2
315
3p10
026
100
2118
70
017
210
040
100
2419
190
1314
1410
085
100
2026
1526
6719
585p
100
3310
024
226
00
07
100
5910
027
3737
015
22
100
9510
014
6040
6078
1156
101,
100
1799
2637
110
00
3910
063
9825
6666
017
00
100
9891
894
8194
844
5520
p10
014
8821
6325
00
080
100
6576
1782
820
200
010
010
024
110
098
100
852
57
40p
100
1934
685
450
00
7310
066
233
9595
023
00
100
100
00
100
100
100
851
583
3031
310
00
100
345
270
2361
010
00
100
578
789
9294
9410
010
100
096
9296
9910
010
051
310
01
100
572
510
2549
010
01
100
494
9417
9788
8810
084
100
010
010
010
010
010
010
0
10p
100
6110
05
9588
026
150
100
5910
03
100
100
2699
8181
100
100
100
010
010
010
010
010
010
0
2013
100
100
100
298
990
2710
6910
010
010
01
100
100
3410
077
7710
010
010
00
100
100
100
100
100
100
40p
100
100
100
099
100
025
910
010
010
010
00
100
100
3910
074
7410
010
010
00
100
100
100
100
100
100
515
3n-
018
016
20
00
057
017
016
00
00
00
010
021
10
10
00
5p0
40
291
00
00
520
40
381
10
00
00
70
553
13
00
010
p0
00
392
00
00
180
10
454
40
00
00
80
2510
510
00
0
20p
00
019
20
00
01
01
014
1515
00
00
010
01
1310
130
00
40-p
00
02
50
00
00
00
02
3232
00
00
010
00
2810
280
00
525
3p0
60
01
00
00
00
60
01
10
10
00
570
01
01
370
12
5p0
290
11
00
00
50
390
06
60
10
00
880
014
314
480
8
10p
07
01
50
00
045
065
00
3838
00
00
091
00
6337
6357
05
20p
01
01
210
00
027
067
00
7979
00
00
097
00
9469
9460
04
40p
02
02
474
00
08
073
00
9090
00
00
010
00
2810
091
100
610
45
503p
00
00
40
01
520
00
00
2121
078
7878
00
00
6040
6010
093
100
5p0
00
023
40
15
00
00
066
660
8946
460
50
096
9096
100
9010
0
10p
02
00
7738
01
00
01
00
9999
096
2727
099
00
100
100
100
100
8910
0
200
081
00
9383
01
050
068
00
100
100
098
2121
010
00
010
010
010
010
089
100
40p
095
00
9892
00
095
010
00
010
010
00
9916
160
100
00
100
100
100
100
9010
07
213n
020
05
00
00
044
023
03
00
00
00
09
01
00
()0
00
5p0
30
80
00
00
230
50
60
00
00
00
20
40
00
00
010
p0
00
180
00
00
10
00
251
10
00
00
10
581
01
00
020
00
00
440
00
00
00
00
615
50
00
00
20
1313
013
00
0
401)
00
029
10
00
00
00
021
1515
00
00
04
00
331
330
00
735
3p0
00
00
00
00
00
00
00
00
00
00
170
00
00
180
2
500
140
00
00
00
00
60
04
40
10
00
790
03
03
230
1
10t)
019
00
30
00
032
048
00
2929
01
00
082
00
3811
3829
00
200
04
00
230
00
028
052
00
5353
02
00
082
00
7851
7829
00
40v
04
00
522
00
05
050
00
7373
02
00
083
00
9177
9129
00
770
3p0
00
00
00
047
00
00
010
100
6159
590
00
039
2439
100
8499
Sp
00
00
60
00
I0
00
00
4141
070
3333
00
00
8574
8510
078
9910
p0
00
070
260
00
00
00
087
870
7619
190
540
010
099
100
100
7499
20p
017
00
9878
00
043
02
00
9696
080
1313
010
00
010
010
010
010
070
9940
P0
100
00
100
960
00
100
093
00
100
100
081
1010
010
00
010
010
010
010
070
99
Fact
or R
eten
tion 37
Tab
le 1
0Pr
opor
tion
of S
ampl
es R
etai
ning
K F
acto
rs, P
hi =
Mix
ed
kp
Low
Com
mun
ality
Wid
e C
omm
unal
ityH
igh
Com
mun
ality
SESi
ze C
NG
Scr
eeM
RT
M-
PAm
H P
AcL
PA
LL
MA
P K
SR K
SRSE
CN
G S
cree
MR
TSE
PAM
H P
Act
PA
LL
, MA
P K
SR M
-KSR
CN
G S
cree
MR
TM
-PA
mH
PA
IL P
AL
L M
AP
KSR
KSR
39
3n93
221
015
70
01
6499
271
013
130
1I
199
291
010
510
101
10
5D98
70
014
40
00
6710
013
00
1818
00
00
100
250
019
919
90
610
D10
01
00
162
00
055
100
60
033
330
00
010
024
00
3921
398
03
20p
100
00
018
20
00
2410
03
00
5252
00
00
100
250
062
3562
60
1
40D
100
00
023
10
00
710
02
00
6969
00
00
100
260
077
5277
50
03
153p
9523
9638
4336
05
490
100
4110
050
2525
015
1818
100
9110
037
5338
5384
4480
5p97
5297
4254
460
513
310
064
100
5841
410
153
310
099
100
3185
7085
9234
8010
p99
6897
3772
670
54
2710
067
9858
7171
015
'0
010
010
087
2110
097
100
9726
82
20D
100
7271
2287
790
61
8510
069
6435
9393
014
00
100
100
195
100
100
100
9822
8240
p10
076
103
9787
07
097
100
779
499
990
130
010
010
00
010
010
010
099
2182
330
3p10
00
100
5565
552
4651
010
00
100
5388
8829
9395
9510
010
100
910
099
100
100
100
100
5D10
01
100
5884
762
5175
010
01
100
5096
9640
9692
9210
084
100
610
010
010
010
010
010
010
p10
057
100
5296
961
5443
010
055
100
4710
010
048
9789
8910
010
010
04
100
100
100
100
100
100
20D
100
100
100
4698
100
056
2656
100
9910
044
100
100
5599
8989
100
100
100
210
010
010
010
010
010
0
40p
100
100
100
4399
100
057
2110
010
010
010
041
100
100
6010
089
8910
010
010
01
100
100
100
100
100
100
515
3n0
210
153
00
00
560
240
142
20
00
00
150
230
00
00
05D
05
028
30
00
053
08
027
33
00
00
08
049
20
20
00
10D
00
036
40
00
025
01
032
77
00
00
08
019
102
100
00
20p
00
014
50
00
04
00
010
2121
00
00
010
00
238
230
00
40o
00
01
50
00
00
00
01
4141
00
00
013
00
3917
390
00
525
3o0
30
19
10
03
00
10
132
320
1615
150
510
14
14
531
21
5p0
290
213
10
00
20
230
162
620
244
40
910
118
618
650
13
10D
019
04
201
00
034
082
00
8787
032
11
094
00
7031
7074
06
20D
03
06
301
00
043
093
00
9696
036
11
094
00
9379
9381
05
40D
00
05
501
00
09
094
00
9999
037
00
096
046
100
9110
086
04
550
3D0
00
08
10
346
00
00
074
742
9697
970
00
087
7887
100
9910
05D
00
00
348
02
100
00
00
9797
510
091
910
30
010
098
100
100
9810
0
10p
02
00
8550
02
00
00
00
100
100
610
084
840
990
010
010
010
010
098
100
20D
083
00
9988
02
042
064
00
100
100
610
082
820
100
00
100
100
100
100
9710
0
40p
010
00
010
010
00
20
100
010
00
010
010
04
100
8181
010
00
010
010
010
010
097
100
721
3n0
230
41
00
00
530
280
30
00
00
00
160
10
00
00
05p
06
08
10
00
042
08
07
00
00
00
07
05
00
00
00
10p
00
019
10
00
07
00
029
11
00
00
07
062
40
40
00
20p
00
041
10
00
00
00
049
22
00
00
010
09
271
270
00
40p
00
010
10
00
00
00
09
22
00
00
014
00
4111
410
00
735
300
00
03
00
00
00
00
01
10
10
00
170
00
00
110
05p
08
00
50
00
00
05
00
55
01
00
066
00
00
011
00
10D
032
00
110
00
015
053
00
3636
01
00
055
00
20
211
00
20p
02
00
140
00
030
071
00
6969
01
00
058
00
322
3210
00
40D
00
00
160
00
00
070
00
8888
02
00
066
00
7720
7711
00
770
313
00
00
10
02
390
00
00
00
050
3737
00
00
62
698
3886
5p0
00
015
20
35
00
00
06
60
641
10
00
034
1634
9921
83
10D
00
00
7133
04
00
00
00
7373
078
00
051
00
9282
9210
013
8020
D0
170
099
770
60
480
20
090
900
850
00
100
00
100
9210
010
07
8040
130
100
00
100
970
80
100
096
00
9292
088
00
010
00
010
010
010
010
04
80
Tab
le 1
1
Om
ega
Squa
red
for
Prop
ortio
n of
Sam
ples
Ret
aini
ng K
Fac
tors
.
Eff
ect
k p/k
N Com
mun
ality
Phi
k X
p/k
k X
Nk
X C
omm
k X
Phi
p/k
X N
p/k
X C
omm
p/k
X P
hiN
X C
omm
N X
Phi
Com
m X
Phi
k X
p/k
X N
k X
p/k
X C
omm
k X
N X
Com
mp/
k X
N X
Com
mk
X p
/k X
Phi
k X
N X
Phi
p/k
X N
X P
hiC
omm
X p
/k X
Phi
Com
m X
k X
Phi
Com
m X
n X
Phi
k X
p/k
X N
X C
omm
k X
p/k
X N
X P
hiC
omm
X k
X p
/k X
Phi
Com
m X
p/k
X n
X P
hiC
omm
X k
X n
X P
hik
X p
/k X
N X
Com
m X
phi
Fact
or R
eten
tion 38
Para
llel A
naly
sis
CN
GSE
scre
eM
Rt
PAm
HPA
cLPA
LL
MA
PK
SRM
-KSR
0.97
50.
043
0.47
60.
011
0.04
30.
043
0.03
60.
014
0.02
80.
033
0.00
00.
158
0.12
60.
041
0.36
60.
403
0.25
50.
473
0.46
70.
192
0.00
00.
154
0.01
10.
022
0.13
30.
082
0.00
30.
001
0.00
40.
033
0.00
00.
118
0.00
00.
005
0.04
40.
049
0.07
20.
156
0.06
20.
047
0.00
00.
023
0.00
10.
011
0.08
90.
094
0.13
00.
044
0.08
20.
044
0.00
00.
012
0.25
90.
192
0.00
30.
004
0.03
60.
004
0.01
30.
011
0.00
00.
006
0.02
20.
035
0.00
20.
001
0.00
10.
000
0.00
00.
000
0.00
00.
000
0.00
10.
015
0.00
10.
001
0.00
50.
002
0.00
20.
001
0.00
30.
001
0.00
40.
031
0.01
30.
008
0.00
20.
002
0.00
50.
002
0.00
00.
174
0.02
00.
032
0.01
40.
014
0.00
10.
000
0.00
30.
090
0.00
00.
020
0.00
10.
002
0.00
40.
005
0.06
50.
076
0.03
90.
146
0.00
00.
014
0.00
20.
015
0.00
70.
015
0.10
90.
011
0.04
00.
029
0.00
00.
013
0.00
10.
010
0.00
00.
000
0.00
10.
000
0.00
10.
034
0.00
00.
003
0.00
00.
002
0.00
70.
004
0.00
00.
000
0.00
20.
004
0.00
00.
001
0.00
00.
004
0.00
10.
002
0.01
80.
002
0.00
40.
008
0.00
00.
021
0.04
10.
110
0.00
70.
007
0.00
00.
000
0.00
10.
003
0.00
00.
002
0.00
30.
012
0.00
30.
004
0.00
30.
012
0.00
50.
003
0.00
00.
009
0.00
10.
021
0.00
00.
000
0.00
00.
000
0.00
10.
001
0.00
00.
027
0.00
10.
027
0.01
40.
011
0.00
00.
001
0.00
20.
093
0.00
00.
001
0.00
20.
022
0.00
50.
007
0.00
70.
003
0.02
30.
008
0.00
00.
000
0.00
00.
003
0.00
00.
000
0.00
00.
000
0.00
00.
000
0.00
00.
006
0.00
00.
001
0.02
00.
018
0.00
20.
000
0.01
10.
005
0.00
00.
015
0.00
00.
009
0.00
40.
009
0.03
50.
057
0.04
20.
013
0.00
00.
002
0.00
00.
010
0.00
30.
004
0.01
10.
002
0.00
40.
003
0.00
00.
001
0.00
00.
000
0.00
10.
000
0.00
00.
000
0.00
20.
002
0.00
00.
028
0.00
30.
037
0.00
30.
003
0.00
00.
000
0.00
00.
002
0.00
00.
000
0.00
00.
003
0.00
90.
006
0.00
00.
000
0.00
10.
001
0.00
00.
004
0.00
00.
014
0.00
40.
006
0.04
90.
009
0.01
60.
008
0.00
00.
004
0.00
00.
001
0.00
90.
013
0.00
10.
000
0.00
90.
008
0.00
00.
000
0.00
00.
000
0.00
00.
000
0.00
00.
000
0.00
00.
001
0.00
00.
000
0.00
00.
000
0.00
30.
005
0.00
10.
000
0.00
00.
003
Fact
or R
eten
tion 39
Tab
le 1
2Pr
opor
tion
of S
ampl
es A
gree
ing
with
Rul
e A
pplie
d to
Pop
ulat
ion
R-
Mat
rix,
Phi
= 0
.00
kp
Size
Low
Com
mun
ality
Wid
e C
omm
unal
itySE
CN
G S
cree
MR
Hig
h C
omm
unal
ityM
-T
PAm
s PA
u, P
Au,
MA
P K
SR K
SRSE
CN
G S
cree
MR
TM
-PA
S PA
u, P
AL
L M
AP
KSR
KSR
SEC
NG
Scr
eeM
RT
PAS
PAL
L P
A,,,
MA
P K
SR M
-KSR
39
3n0
4099
9927
3242
8735
220
5299
9931
3162
6749
490
R7
100
100
5264
5262
7584
51,
057
100
100
3543
4890
6729
062
100
100
4747
7173
6464
093
100
100
6477
6471
7988
101,
074
100
100
4557
5794
8447
072
100
100
6868
7881
7979
097
100
100
7187
7179
8592
20p
084
100
100
5670
6696
9074
080
100
100
8585
8487
8989
099
100
100
7590
7585
8995
40p
091
100
100
6582
7398
9391
088
100
100
9595
8992
9595
010
010
010
080
9280
9093
973
1531
30
210
049
4643
5325
00
330
083
8333
7985
850
890
096
9396
9594
995p
057
00
6363
4760
542
075
00
8989
4690
8787
010
00
010
099
100
9894
100
101,
085
11
8185
5071
6821
095
44
9494
6297
9090
010
019
1910
010
010
099
9610
020
p0
9332
3292
9755
8077
810
9646
4698
9874
9993
930
100
9090
100
100
100
9997
100
40p
098
9494
9810
062
8783
100
097
9393
9999
8099
9696
010
010
010
010
010
010
010
098
100
330
3p0
00
085
9778
9846
00
00
010
010
099
9997
970
90
010
010
010
010
010
010
05p
01
00
9299
8910
010
00
01
00
100
100
100
100
100
100
084
00
100
100
100
100
100
100
100
056
00
9710
094
100
100
00
580
010
010
010
010
010
010
00
100
00
100
100
100
100
100
100
20p
010
00
099
100
9810
010
062
010
00
010
010
010
010
010
010
00
100
00
100
100
100
100
100
100
40p
010
00
010
010
010
010
010
010
00
100
00
100
100
100
100
100
100
010
00
010
010
010
010
010
010
05
153n
02R
00
158
6469
214
036
00
2222
4846
3939
075
00
3622
3635
7277
51,
033
00
2311
7373
1321
054
00
3636
5851
6363
081
00
5940
5946
7883
10p
031
11
3917
8279
4341
066
22
6060
6758
8282
086
3131
8162
8153
8589
20p
034
2222
6234
8885
6866
074
3232
8383
7765
9191
091
9595
9377
9360
8993
40p
044
8585
8157
9190
8482
082
8383
9595
8572
9494
096
100
100
9888
9865
9394
525
3p0
10
036
2749
.40
20
01
00
6363
5559
5959
051
00
9793
9789
9399
5p0
190
050
4256
4829
00
200
082
8270
7376
760
960
010
099
100
9295
100
10t
077
00
7169
6457
605
086
00
9393
8183
8585
010
00
010
010
010
094
9810
020
1,0
900
089
8774
6373
760
990
096
9686
8990
900
100
00
100
100
100
9599
100
400
094
00
9896
8168
7910
00
100
00
9999
9191
9494
010
01
110
010
010
095
100
100
550
3p0
00
095
9772
9523
00
00
010
010
010
010
098
980
00
010
010
010
098
100
100
5p0
00
098
100
8099
100
00
00
010
010
010
010
010
010
00
30
010
010
010
099
100
100
100
01
00
9910
087
100
100
00
00
010
010
010
010
010
010
00
990
010
010
010
099
100
100
20p
078
00
100
100
9210
010
038
057
00
100
100
100
100
100
100
010
00
010
010
010
099
100
100
40o
010
00
010
010
095
100
100
100
010
00
010
010
010
010
010
010
00
100
00
100
100
100
9910
010
07
213n
017
7017
146
5256
0li
022
8111
1717
4030
3030
068
9316
1219
1227
6575
5p0
3173
1822
957
648
140
5282
830
3053
3748
480
8596
1430
3830
3771
8010
p0
3176
1839
1761
7533
340
7883
645
45-
6647
6767
093
9817
4762
4750
7788
200
034
7821
6329
6583
5856
086
8210
5858
7457
8080
096
9952
6074
6063
8494
4013
045
8034
8644
7089
7468
087
8124
7171
8069
8888
099
100
8578
8078
7689
967
3531
30
092
1133
2248
358
00
095
1361
6159
5945
450
597
393
8293
7673
99Si
,0
195
845
4255
4539
00
196
1180
8065
6956
560
7999
110
099
100
7879
100
100
048
985
6181
6457
630
048
989
8989
6975
7272
010
010
00
100
100
100
8087
100
20p
090
993
7699
7262
7864
095
100
792
9275
8082
820
100
100
010
010
010
080
9210
040
p0
9110
03
8810
080
6784
980
9910
06
9797
8284
8989
010
099
010
010
010
081
9610
07
703p
00
100
299
9557
9710
00
099
110
010
094
100
9999
00
100
210
010
010
099
100
100
51,
00
100
110
010
055
9910
00
00
991
100
100
9410
010
010
00
010
01
100
100
100
9910
010
010
p0
010
00
100
100
5810
010
00
00
100
010
010
095
100
100
100
057
100
010
010
010
010
010
010
020
p0
1110
00
100
100
6710
010
027
01
100
010
010
097
100
100
100
010
010
00
100
100
100
100
100
100
40p
099
100
010
010
077
100
100
100
089
100
010
010
010
010
010
010
00
100
100
010
010
010
010
010
010
0
Fact
or R
eten
tion 40
Tab
le 1
3Pr
opor
tion
of S
ampl
es A
gree
ing
with
Rul
e A
pplie
d to
Pop
ulat
ion
R-M
atri
x, P
hi =
0.3
0
kp
Low
Com
mun
ality
Wid
e C
omm
unal
ityH
igh
Com
mun
ality
SESi
ze C
NG
Scr
eeM
RT
M-
PAN
LH
PA
ct P
AL
L M
AP
KSR
KSR
SEC
NG
Scr
eeM
RT
SEPA
S PA
L, P
AL
L M
AP
KSR
M-K
SR C
NG
Scr
ee M
RT
PAL
L, P
A.
M-
PAL
L M
AP
KSR
KSR
39
3n0
3799
9925
2766
9251
170
4899
9914
1487
7549
490
7110
010
025
2325
5659
72
5o0
5410
010
030
3277
9688
250
5610
010
022
2289
8175
750
8010
010
045
4045
6365
78
10D
068
100
100
3940
8999
9948
063
100
100
4141
9089
9090
090
100
100
6562
6571
7082
20p
074
100
100
5254
9710
010
072
070
100
100
6767
9093
9595
095
100
100
7873
7877
7587
4013
081
100
100
6571
100
100
100
860
7910
010
082
8290
9596
960
9910
010
090
7990
8280
91
315
300
210
033
2388
6911
10
380
051
5143
6654
540
910
078
6878
9075
92
5o0
480
043
2990
7639
40
750
073
7351
7569
690
990
093
8993
9679
9610
D0
661
163
5091
8571
260
922
293
9363
8378
780
100
1515
100
9710
099
8198
20D
076
2323
8070
9192
8570
096
3333
9898
7588
8484
010
085
8510
010
010
010
085
9940
130
8485
8593
8191
9690
830
9988
8899
9982
9388
880
100
100
100
100
100
100
100
8810
0
330
3D0
00
075
7444
6319
00
00
099
9967
9997
970
80
010
010
010
010
010
010
0
Si,
01
00
9094
5174
460
02
00
100
100
8410
010
010
00
830
010
010
010
010
010
010
0
1013
056
00
9510
059
8262
00
620
010
010
094
100
100
100
010
00
010
010
010
010
010
010
0
20D
010
00
098
100
6788
7848
010
00
010
010
098
100
100
100
010
00
010
010
010
010
010
010
0
40p
010
00
099
100
7492
8810
00
100
00
100
100
100
100
100
100
010
00
010
010
010
010
010
010
0
515
3n0
180
06
R97
RR
81
031
00
66
8051
1111
059
00
53
534
4063
Si,
034
00
68
100
9564
20
440
012
1281
5729
290
660
015
1115
4450
71
101,
039
00
1010
100
9998
110
521
126
2682
6557
570
7614
1438
2838
5461
82
20p
039
99
1720
100
100
100
400
5821
2142
4283
7376
760
.85
8383
6352
6362
7190
400
043
5959
2838
100
100
100
660
6466
6657
5784
8286
860
9010
010
083
6783
6981
96at
to5
253p
03
00
112
9658
00
03
00
1212
3539
2222
055
00
3117
3187
5774
tb5p
026
00
222
9866
80
031
00
3838
4547
4343
093
00
7252
7296
6880
100
043
00
4810
9976
419
087
00
8484
5558
6161
093
00
9188
9199
8087
20D
040
00
7832
100
8469
650
970
098
9866
6773
730
920
092
9192
9989
92
40D
054
00
9467
100
9281
840
990
099
9976
7383
830
920
099
9199
9993
97
550
31,
00
00
4923
2954
40
00
00
9595
5199
9898
00
00
9694
9610
010
010
0
5p0
00
082
6236
6831
00
00
010
010
063
100
100
100
03'
00
100
9910
010
099
100
lOp
02
00
9590
4379
580
00
00
100
100
7610
099
990
990
010
010
010
010
099
100
20D
083
00
100
9652
8875
490
530
010
010
084
100
9898
010
00
010
010
010
010
098
100
40D
010
00
010
010
061
9385
100
010
00
010
010
090
100
9898
010
00
010
010
010
010
098
100
117
213n
011
603
32
99R
I0
10
2570
42
259
459
90
6090
05
25
3142
55
m coSi
,10
p0 0
29 2360 60
3 3
5 102 5
100
100
91 9723 87
2 100 0
43 4970 70
3 3
6 286 28
66 7753 66
32 6732 67
0 067 75
90 900 1
12 3111 28
12 3141 52
53 6563 74
20p
017
603
2316
100
9999
330
5670
255
5588
7383
830
8590
1360
5360
6376
82
0 07
3540
03p 5D
0 0 0
210 6
60 100
100
6 0 0
42 3 9
38 0 1
100 70 72
100
45 54
100 0 3
550 0
0 0 0
66 0 2
70 100
100
10 0 0
68 4 27
68 4 27
97 40 53
82 33 43
91 9 29
919
29
0 0 0
92 11 84
91 100
100
49 0 0
85 19 59
807 44
85 19 59
72 73 81
85 58 70
89 73 82
lOp
043
100
030
1074
6431
20
5010
00
7474
6754
5656
010
010
00
8470
8486
7890
20p
047
100
054
3175
7565
420
9610
00
9595
7463
7171
010
010
00
100
8810
090
8294
40p
049
100
075
5177
8283
700
9710
00
9999
8170
8282
010
010
00
100
100
100
9385
977
703p
00
100
035
1251
504
00
010
00
8686
6210
089
890
010
00
9490
9410
010
010
0
Si,
00
100
082
5559
6924
00
010
00
9999
5110
092
920
010
00
100
100
100
100
100
100
101,
00
100
010
094
6982
460
00
100
010
010
042
100
9696
053
100
010
010
010
010
010
010
0
w r20
p40
1,
0 015
100
100
100
0 010
010
010
010
080 89
87 9065 75
42 100
0 01
9210
010
00 0
100
100
100
100
46 6210
010
099 10
099 10
00 0
100
100
100
100
0 010
010
010
010
010
010
010
010
010
010
010
010
0
PAC
CZ
)
Fact
or R
eten
tion 41
Tab
le 1
4Pr
opor
tion
of S
ampl
es A
gree
ing
with
Rul
e A
pplie
d to
Pop
ulat
ion
R-
Mat
rix,
Phi
= 0
.50
kp
Size
Low
Com
mun
ality
Wid
e C
omm
unal
ityH
igh
Com
mun
ality
SEC
NG
Scr
eeM
RT
Ni-
PAm
H P
ApL
PA
LL
MA
P K
SR K
SRSE
CN
G S
cree
MR
TSE
PAM
H P
Acr
, PA
LL
MA
P K
SR M
-KSR
CN
G S
cree
MR
TM
-PA
NH
, PA
IL P
AL
L M
AP
KSR
KSR
39
3n0
3899
9925
3177
9462
110
4699
999
999
7768
680
6199
996
136
6462
6650
062
100
100
2932
8998
9520
058
100
100
1111
100
8192
920
6710
010
014
2114
7276
73
10D
077
100
100
3636
9810
010
040
068
100
100
1919
100
8598
980
7510
010
034
4034
8287
81
20D
081
100
100
4846
100
100
100
670
7410
010
038
3810
089
100
100
081
100
100
5759
5789
9487
40D
083
100
100
6261
100
100
100
820
8010
010
059
5910
093
100
100
086
100
100
7177
7192
9691
315
3D0
190
018
1010
089
161
035
00
1919
8351
3030
085
00
2615
2671
5667
5p0
410
022
910
095
775
065
00
3737
8459
5454
095
00
6040
6083
6473
1Cto
041
11
3715
100
9999
250
792
266
6686
6974
740
989
994
8194
9072
8020
D0
4012
1263
3110
010
010
060
084
2424
8282
8978
8282
010
076
7610
098
100
9379
86
40D
049
6666
8554
100
100
100
770
8677
7795
9591
8688
880
100
100
100
100
100
100
9582
903
3031
50
00
045
2775
617
00
00
078
7838
9272
720
100
096
9296
9910
010
0
5D0
10
072
5177
6731
00
10
094
9453
9781
810
840
010
010
010
010
010
010
0
10D
061
00
9588
7777
530
059
00
100
100
6799
8787
010
00
010
010
010
010
010
010
0
20D
010
00
098
9980
8468
690
100
00
100
100
7710
092
920
100
00
100
100
100
100
100
100
40D
010
00
099
100
8491
7810
00
100
00
100
100
8610
095
950
100
00
100
100
100
100
100
100
515
3n0
90
06
1099
9115
00
280
01
I10
063
2828
058
00
42
447
3961
5p0
320
05
810
097
831
040
00
11
100
6872
720
670
011
711
5553
6810
D0
560
07
810
099
994
051
11
55
100
7497
970
798
820
2220
6567
75
2013
064
77
1312
100
100
100
180
6016
1618
1810
081
100
100
089
6868
2850
2873
7581
40D
069
4949
2719
100
100
100
420
7158
5839
3910
086
100
100
094
9999
4869
4879
7984
525
3D0
20
01
110
077
10
05
00
11
7237
77
057
00
10
152
3355
Si,
022
00
21
100
8641
00
340
06
673
4721
210
880
014
314
6746
63
10p
037
00
74
100
9192
60
750
038
3875
5837
370
910
063
3763
7961
7220
p0
280
025
1410
095
9937
085
00
7979
7768
5151
097
00
9469
9487
7279
40p
033
00
5538
100
9810
069
093
00
9090
8175
6464
010
00
010
091
100
9280
855
503D
00
00
40
9041
00
00
00
2121
2778
2525
00
00
6040
6010
093
100
5p0
00
023
491
492
00
00
066
6644
8950
500
50
096
9096
100
9510
0
10D
02
00
7742
9260
270
01
00
9999
6796
7878
099
00
100
100
100
100
9710
0
20D
081
00
9391
9268
6046
068
00
100
100
7998
9090
010
00
010
010
010
010
098
100
40D
095
00
9810
092
7582
980
100
00
100
100
8599
9595
010
00
010
010
010
010
099
100
721
3n0
510
11
110
090
30
022
800
00
100
585
50
4390
00
00
3215
445D
025
100
11
100
9771
00
3880
10
010
064
4242
045
900
00
036
3054
lfto
031
101
21
100
100
100
10
3680
13
310
073
8888
052
900
53
544
5366
20D
022
101
83
100
100
100
130
3880
119
1910
081
9999
062
906
2316
2353
7376
40D
024
101
2912
100
100
100
420
4980
449
4910
088
100
100
072
9036
4531
4561
8583
735
3D0
010
00
00
100
760
00
010
00
00
9027
00
017
100
00
00
4441
645D
09
100
00
010
089
130
03
100
04
490
358
80
7910
00
30
353
5572
10p
031
100
04
010
097
881
049
100
032
3291
4639
390
8210
00
3811
3862
7280
20D
013
100
030
010
099
9424
083
100
060
6092
5766
660
8210
00
7851
7871
8187
40D
013
100
064
810
010
097
560
8610
00
8282
9366
8383
083
100
091
7791
7986
91
770
3v0
010
00
00
5933
00
00
100
010
1035
7425
250
010
00
3924
3910
085
995p
00
100
06
061
380
00
010
00
4141
5184
5050
00
100
085
7485
100
9199
10D
00
100
070
2663
4711
00
010
00
8787
6491
7878
054
100
010
099
100
100
9599
20p
017
100
098
7867
5636
430
210
00
9696
7595
8787
010
010
00
100
100
100
100
9899
40D
010
010
00
100
9679
6251
100
093
100
010
010
080
9790
900
100
100
010
010
010
010
010
099
Fact
or R
eten
tion 42
Tab
le 1
5Pr
opor
tion
of S
ampl
es A
gree
ing
with
Rul
e A
pplie
d to
Pop
ulat
ion
R-M
atri
x, P
hi =
Mix
ed
kp
Size
Low
Com
mun
ality
Wid
e C
omm
unal
ityH
igh
Com
mun
ality
SEC
NG
Scr
eeM
RT
M-
PAS
PAc
PAL
L M
AP
KSR
KSR
SEC
NG
Scr
eeM
RT
SEPA
S Pi
ta, P
AL
L M
AP
KSR
M-K
SR C
NG
Scr
ee M
RT
M-
PAm
ii PA
ct P
AL
L M
AP
KSR
KSR
39
3n0
3899
9925
3068
9353
90
5399
9920
2084
6953
530
7099
991R
2618
7267
855D
054
100
100
2932
8098
9116
068
100
100
2626
8874
7474
077
100
100
2733
2781
7390
10D
062
100
100
3640
9310
010
037
078
100
100
4141
9278
9191
083
100
100
4746
4788
7995
20D
066
100
100
4754
9910
010
067
085
100
100
6161
9582
9797
087
100
100
7060
7092
8498
40D
068
100
100
5870
100
100
100
840
8910
010
079
7997
8599
990
9010
010
085
7685
9487
100
315
3D0
210
043
3670
5314
00
380
025
2549
7066
660
910
053
3853
8467
795D
052
00
5446
7560
343
064
00
4141
5776
8383
099
00
8570
8592
7782
10D
073
11
7267
7966
5624
072
22
7171
6881
9191
010
013
1310
097
100
9787
8720
130
8028
2887
7982
7173
780
7536
3693
9377
8396
960
100
8181
100
100
100
9894
9240
p0
8590
9097
8787
7784
920
8491
9199
9988
8699
990
100
100
100
100
100
100
9998
963
303D
00
00
6555
8864
60
00
00
8888
5793
8989
010
00
100
9910
010
010
010
05D
01
00
8476
9371
340
01
00
9696
6896
9797
084
00
100
100
100
100
100
100
1 O
D0
570
096
9697
7867
00
550
010
010
077
9799
990
100
00
100
100
100
100
100
100
20D
010
00
098
100
9983
8356
099
00
100
100
8399
9999
010
00
010
010
010
010
010
010
040
130
100
00
9910
010
088
8910
00
100
00
100
100
9010
099
990
100
00
100
100
100
100
100
100
515
3n0
190
012
1797
827
20
290
03
361
5733
330
580
03
33
5058
745D
039
00
1519
9989
564
045
00
66
6665
5858
066
00
1110
1159
6781
lOp
042
00
2329
100
9389
190
591
113
1371
7276
760
7310
1029
3529
7076
8920
130
4113
1334
4510
095
9557
067
1616
2929
7679
8686
080
7676
4363
4379
8394
40D
044
6565
4869
100
9598
830
7567
6750
5081
8590
900
8610
010
059
8659
8488
975
253D
00
00
1110
5457
20
01
00
3232
4241
2121
051
00
41
461
9379
513
013
00
1619
5562
280
023
00
6262
5655
3636
091
00
186
1874
9487
CA
IOD
067
00
2748
5672
602
082
00
8787
7369
5353
094
00
7031
7084
9594
1""
20D
087
00
4080
5880
8048
093
00
9696
8779
6868
094
00
9379
9391
9795
40D
096
00
6296
6487
9090
094
00
9999
9583
8181
096
00
100
9110
096
9996
550
313
00
00
81
4547
00
00
00
7474
5796
8181
00
00
8778
8710
099
100
5D0
00
034
851
5312
00
00
097
9771
100
8888
03
00
100
9810
010
098
100
10p
02
00
8550
6061
630
00
00
100
100
8110
094
940
990
010
010
010
010
098
100
20D
083
00
9988
7071
8242
064
00
100
100
8810
097
970
100
00
100
100
100
100
9710
0
40D
010
00
010
010
082
8091
100
010
00
010
010
093
100
9898
010
00
010
010
010
010
097
100
721
3n0
710
28
796
611
00
1710
01
I32
4921
210
4990
10
00
3938
64Si
,0
3010
210
997
6618
10
3810
04
440
6147
470
4990
12
02
5046
71
10D
038
102
2013
9973
598
052
100
1717
4772
7070
054
900
223
2264
5379
20D
035
101
3820
100
7979
310
6010
040
4052
8283
830
6590
464
2164
7759
8440
D0
4110
257
3310
083
8559
068
100
6767
6289
8989
076
9036
8152
8187
6487
735
3p0
010
01
71
3839
00
00
100
01
139
3914
140
1310
01
00
068
3662
5D0
210
00
162
3845
20
05
100
05
551
5548
480
7310
00
30
378
5368
10D
047
100
037
936
5227
10
5210
00
3636
6371
7676
075
100
012
1312
8666
7620
D0
7910
00
5740
3759
5964
080
100
069
6973
8386
860
7810
00
4222
4291
7581
40D
087
100
065
7740
6879
900
8010
00
8888
8289
9292
086
100
087
4087
9381
877
703D
00
100
01
076
450
00
010
00
00
7559
5656
00
100
06
26
9862
845D
00
100
015
282
522
00
010
00
66
7574
9191
00
100
034
1634
9972
87IO
D0
010
00
7133
8863
470
00
100
073
7374
8893
930
5110
00
9282
9210
081
8720
D0
1710
00
9977
9371
8348
02
100
090
9075
9597
970
100
100
010
092
100
100
8888
40D
010
010
00
100
9797
7895
100
096
100
092
9280
9899
990
100
100
010
010
010
010
093
89
Tab
le 1
6
Om
ega
Squa
red
for
Prop
ortio
n of
Sam
ples
Agr
eein
g w
ith R
ule
App
lied
to P
opul
atio
n R
-Mat
rix.
Eff
ect
k p/k
N Com
mun
ality
Phi
k X
p/k
k X
Nk
X C
omm
k X
Phi
p/k
X N
p/k
X C
omm
p/k
X P
hiN
X C
omm
N X
Phi
Com
m X
Phi
k X
p/k
X N
k X
p/k
X C
omm
k X
N X
Com
mp/
k X
N X
Com
mk
X p
/k X
Phi
k X
N X
Phi
p/k
X N
X P
hiC
omm
X p
/k X
Phi
Com
m X
k X
Phi
Com
m X
n X
Phi
k X
p/k
X N
X C
omm
k X
p/k
X N
X P
hiC
omm
X k
X p
/k X
Phi
Com
m X
p/k
X n
X P
hiC
omm
X k
X n
X P
hik
X p
/k X
N X
Com
m X
phi
Fact
or R
eten
tion 43
CN
GSE
s,,
MR
t
Para
llel A
naly
sis
MA
PK
SRM
-KSR
PAm
HPA
ci,
PAL
L
0.00
00.
076
0.45
80.
238
0.04
00.
055
0.00
30.
058
0.04
00.
013
0.00
00.
040
0.06
40.
281
0.27
60.
239
0.01
50.
095
0.03
60.
005
0.00
00.
359
0.02
30.
051
0.19
70.
168
0.07
10.
102
0.25
60.
255
0.00
00.
135
0.00
90.
005
0.03
30.
036
0.01
70.
006
0.07
80.
361
0.00
00.
006
0.00
20.
003
0.08
60.
099
0.01
00.
002
0.01
60.
011
0.00
00.
001
0.24
60.
213
0.00
30.
001
0.00
20.
010
0.00
40.
004
0.00
00.
009
0.01
10.
013
0.00
70.
004
0.00
00.
003
0.01
10.
000
0.00
00.
001
0.00
20.
001
0.00
10.
001
0.00
40.
000
0.00
40.
001
0.00
00.
001
0.00
30.
001
0.00
90.
013
0.00
50.
001
0.00
30.
001
0.00
00.
086
0.01
20.
026
0.01
70.
009
0.00
10.
010
0.01
30.
005
0.00
00.
009
0.00
90.
004
0.00
40.
004
0.05
50.
170
0.04
90.
008
0.00
00.
004
0.00
40.
002
0.00
80.
015
0.05
30.
030
0.02
40.
007
0.00
00.
013
0.00
20.
006
0.00
10.
001
0.00
50.
001
0.06
30.
123
0.00
00.
001
0.00
00.
001
0.01
30.
009
0.00
00.
001
0.00
70.
000
0.00
00.
001
0.00
10.
000
0.00
20.
004
0.03
50.
015
0.00
60.
000
0.00
00.
026
0.05
70.
103
0.00
40.
005
0.00
00.
000
0.00
10.
000
0.00
00.
003
0.01
20.
008
0.00
10.
002
0.00
90.
008
0.00
20.
001
0.00
00.
012
0.00
10.
002
0.00
00.
000
0.00
00.
000
0.00
20.
003
0.00
00.
031
0.00
10.
005
0.01
00.
009
0.00
30.
014
0.00
60.
030
0.00
00.
001
0.00
70.
002
0.00
60.
004
0.01
00.
005
0.00
10.
001
0.00
00.
001
0.00
00.
000
0.00
20.
001
0.00
00.
000
0.00
00.
000
0.00
00.
003
0.00
00.
001
0.01
90.
017
0.00
90.
002
0.00
60.
000
0.00
00.
008
0.00
30.
000
0.00
50.
011
0.06
80.
077
0.04
60.
002
0.00
00.
001
0.00
30.
000
0.00
20.
001
0.01
00.
004
0.00
50.
001
0.00
00.
001
0.00
00.
000
0.00
10.
002
0.00
00.
000
0.00
30.
002
0.00
00.
032
0.00
50.
011
0.00
10.
001
0.00
00.
000
0.00
00.
000
0.00
00.
001
0.00
00.
001
0.00
70.
005
0.00
00.
000
0.00
10.
000
0.00
00.
001
0.00
50.
001
0.00
50.
005
0.01
70.
011
0.00
60.
002
0.00
00.
004
0.00
00.
000
0.00
90.
012
0.00
50.
003
0.01
00.
002
0.00
00.
000
0.00
00.
000
0.00
00.
000
0.00
00.
000
0.00
00.
000
0.00
00.
000
0.00
00.
001
0.00
20.
003
0.00
00.
000
0.00
10.
000
0 0 m
Fact
or R
eten
tion 44
16 14 12 10
0
CN
G S
E S
cree
MR
O O O O t'P
A m
h PA
cl
PA 1
1M
AP
KSR
M-K
SR
Met
hod
Figu
re 1
. Dis
trib
utio
ns o
f M
ean
Num
ber
of F
acto
rs R
etai
ned,
k =
3
Fact
or R
eten
tion 45
30 20 10
6 a 6
I1
1.
I
T
0 Pti
CN
G S
E S
cree
MR
PA m
h PA
cl
PA 1
1M
AP
KSR
M-K
SR
Met
hod
Figu
re 2
. Dis
trib
utio
ns o
f M
ean
Num
ber
of F
acto
rs R
etai
ned,
k =
5
Fact
or R
eten
tion 46
40 30 20 10
-o
ti a 8 I
It
II
I
I
'IIS
I-N
.T
, I
I I
e3
CN
G S
E S
cree
MR
tPA
mh
PA c
lPA
11
MA
PK
SRM
-KSR
Met
hod
Figu
re 3
. Dis
trib
utio
ns o
f M
ean
Num
ber
of F
acto
rs R
etai
ned,
k =
7
Figu
re 4
. Mea
n N
umbe
r of
Fac
tors
Ret
aine
d by
Tru
e N
umbe
r of
Fac
tors
16-
14-
11.
Elk
= 3
Elk
= 5
Elk
= 7
CN
GSE
Scr
eeM
Rt
PA m
hPA
cl
PA 1
1M
AP
KSR
M-K
SR
Rul
e of
Thu
mb
Fact
or R
eten
tion 47
Figu
re 5
. Mea
n N
umbe
r of
Fac
tors
Ret
aine
d by
N p
er V
aria
ble
(N:p
) fo
r k
= 5
05
1015
2025
N P
er V
aria
ble
3035
4045
Figu
re 6
. Mea
n N
umbe
r of
Fac
tors
Ret
aine
d by
Num
ber
of V
aria
bles
Per
Fac
tor
(p:k
) fo
r k
= 5
20 -
187
16-
Ti
14 -
m g 4),
12-
.F.J cz 44
100 ct .0
CA
58-
0oo
z6- 4 2
(=30
221°
CN
G0S
E S
cree
6 M
R0P
A tr
ill 0
PA c
lPA
11
0 -
KSR
°-
M-K
SR
Fact
or R
eten
tion 49
0,2
34
56
78
Num
ber
of V
aria
bles
Per
Fac
tor
910
11
Figu
re 7
. Mea
n N
umbe
r of
Fac
tors
Ret
aine
d by
Com
mun
ality
Typ
e fo
r k
= 5
12 10
Fact
or R
eten
tion 50
tt
Low
Wid
e
ID H
igh
CN
GSE
Scr
eeM
Rt
PA m
hPA
cl
PA 1
1M
AP
KSR
M-K
SR
Rul
e of
Thu
mb
Figu
re 8
. Mea
n N
umbe
r of
Fac
tors
Ret
aine
d by
Phi
for
k =
5
14-
12 10 4
Fact
or R
eten
tion 51
r =
.00
r =
.30
r =
.50
0 r
= M
ixed
CN
GSE
Scr
eeM
Rt
PA m
hPA
cl
PA 1
1M
AP
KSR
M-K
SR
Rul
e of
Thu
mb
Figu
re 9
. Pro
port
ion
of S
ampl
es R
etai
ning
K F
acto
rs b
y T
rue
Num
ber
of F
acto
rs
1.0
0.9
0.8
0.7
t 0.6
*JZ ;.
0.5
00.
40 6.1
0.3
0.2
0.1
0.0
Fact
or R
eten
tion 52
-
k=3
k=5
k=7
CN
GSE
Scr
eeM
R
111.
1.0.
tPA
mh
PA c
lPA
11
MA
PK
SRM
-KSR
Rul
e of
Thu
mb
Figu
re 1
0. P
ropo
rtio
n of
Sam
ples
Ret
aini
ng K
Fac
tors
by
N p
er V
aria
ble
(N:p
)
1.0
0.9
-
0.8
-
0.7
-
0.5
0 0 o 0.
4o.
0.3
-
0.2
0.1
-
0.0
Fact
or R
eten
tion 53
05
1015
2025
N P
er V
aria
ble
3035
4045
Figu
re 1
1. P
ropo
rtio
n of
Sam
ples
Ret
aini
ng K
Fac
tors
by
Num
ber
of V
aria
bles
Per
Fac
tor
(p:k
)
1.0
0.9
0.8
0.7
0.3
0.2
0.1
0.0
Fact
or R
eten
tion 54
CN
G0S
E S
cree
-.°9
6F M
R0
t0
PA m
h.D
PA c
l°=
PA
11
MA
P-
0 -
KSR
° M
-KSR
23
45
67
8
Num
ber
of V
aria
bles
Per
Fac
tor
910
11
Figu
re 1
2. P
ropo
rtio
n of
Sam
ples
Ret
aini
ng K
Fac
tors
by
Com
mun
ality
Typ
e
1.0
0.9
0.8
0.7
`cf.
3 E 0
.6
v-1
0.5
0 a, 0
.40 4."
0.3
-
0.2
-
0.1
0.0
Fact
or R
eten
tion 55
NIO
NM
INJ
*IV
CN
GSE
Scr
eeM
R
Low
Wid
e
Hig
h
tPA
mh
PA c
lPA
11
MA
PK
SRM
-KSR
Rul
e of
Thu
mb
Figu
re 1
3. P
ropo
rtio
n of
Sam
ples
Ret
aini
ng K
Fac
tors
by
Phi
1.0
0.9
0.8
0.7
Fact
or R
eten
tion 56
0 r
= .0
0
0 r
= .3
0
0 r
.50
0 r
= M
ixed
0.3
0.2
0.1
0.0
CN
GSE
Scr
eeM
Rt
PA m
hPA
cl
PA 1
1M
AP
KSR
M-K
SR
Rul
e of
Thu
mb
Fact
or R
eten
tion 57
Figu
re 1
4. P
ropo
rtio
n of
Sam
ples
Agr
eein
g w
ith R
ule
App
lied
to P
opul
atio
n R
-Mat
rix
by T
rue
Num
ber
of F
acto
rs
1.0
0.9
0.8
-
0.7
5 0.
6
0.5
0.4
41,4
0.3
0.2
0.1
k=3
k=5
k=7
0.0
CN
GSE
Scr
eeM
Rt
PA m
hPA
cl
Rul
e of
Thu
mb
PA 1
1M
AP
c:
KSR
M-K
SR
Fact
or R
eten
tion 58
Figu
re 1
5. P
ropo
rtio
n of
Sam
ples
Agr
eein
g w
ith R
ule
App
lied
to P
opul
atio
nR
-Mat
rix
by N
per
Var
iabl
e (N
:p)
1.0
0.9
5
N P
er V
aria
ble
Fact
or R
eten
tion 59
Figu
re 1
6. P
ropo
rtio
n of
Sam
ples
Agr
eein
g w
ith R
ule
App
lied
to P
opul
atio
nR
-Mat
rix
by N
umbe
r of
Var
iabl
es P
er F
acto
r (p
:k)
a.,
..30.
6 -
E
1.0
0.9
-
0.8
-
=0=
CN
G°=
4:1S
E S
cree
==
t1=
MR
0t=
O=
PA m
h =
2PA
ci
='P
A 1
1M
AP
° 0
° K
SR°
0 °
M-K
SR
0.7
-O
°.
(:-,
s o 0
5z 1 0
0.4-
0.3
-
0.2
-
0.1
0.0
23
45
67
Num
ber
of V
aria
bles
Per
Fac
tor
89
1011
Fact
or R
eten
tion 60
Figu
re 1
7. P
ropo
rtio
n of
Sam
ples
Agr
eein
g w
ith R
ule
App
lied
to P
opul
atio
nR
-Mat
rix
by C
omm
unal
ity T
ype
1.0
0.9
0.8
0.7
cow t0.
6
4.6
0.5
0
cx)
ow 0
.441
4
0.3
0.2
0.1
0.0
Low
Wid
e
Hig
h
CN
GSE
Scr
eeM
Rt
PA m
hPA
cl
PA 1
1M
AP
KSR
M-K
SR
Rul
e of
Thu
mb
Figu
re 1
8. P
ropo
rtio
n of
Sam
ples
Agr
eein
gw
ith R
ule
App
lied
to P
opul
atio
n R
-Mat
rix
by P
hi
1.0
0.9
0.8
0.7
cu cu Sa.
§0.
6
c.n
0.5
0.4
CZ
)0.
3
0.2
0.1
0.0
Or
= .0
0
Or
= .3
0
Or
= .5
0
r =
Mix
ed
CN
GSE
Scr
eeM
RPA
mh
PA c
lPA
11
MA
PK
SRM
-KSR
Rul
e of
Thu
mb
Fact
or R
eten
tion 61
1.
U.S. Department of EducationOffice of Educational Research and Improvement (OERI)
National Library of Education (NLE)Educational Resources Information Center (ERIC)
REPRODUCTION RELEASE
DOCUMENT IDENTIFICATION:
(Specific Document)
ERICEMU! Bisons Dna
TM034933
Title: Factor Retention in Exploratory Factor Analysis: A Comparison ofAlternative Methods
Author(s): Karen R Mumford; John M. Ferron; Constance V. Hines; Kristine Y. Hogarty; Jeffrey D. Kromrey
Corporate Source: University of South FloridaPublication Date:
04/25/2003
II. REPRODUCTION RELEASE:
In order to disseminate as widely as possible timely and significant materials of interest to the educational community, documentsannounced in the monthly abstract joumal of the ERIC system, Resources in Education (RIE), are usually made available to users in microfiche,reproduced paper copy, and electronic media, and sold through the ERIC Document Reproduction Service (EDRS). Credit is given to thesource of each document, and, if reproduction release is granted, one of the following notices is affixed to the document.
If permission is granted to reproduce and disseminate the identified document, please CHECK ONE of the following three options and signat the bottom of the page.
The sample sticker shown below will beaffixed to all Level 1 documents
PERMISSION TO REPRODUCE ANDDISSEMINATE THIS MATERIAL HAS
BEEN GRANTED BY
OleTO THE EDUCATIONAL RESOURCES
INFORMATION CENTER (ERIC)
Level 1
Check here for Level 1 release, permittingreproduction and dissemination in microfiche orother ERIC archival media (e.g., electronic) and
paper copy.
The sample sticker shown below will beaffixed to all Level 2A documents
2A
PERMISSION TO REPRODUCE ANDDISSEMINATE THIS MATERIAL IN
MICROFICHE, AND IN ELECTRONIC MEDIAFOR ERIC COLLECTION SUBSCRIBERS ONLY,
HAS BEEN GRANTED BY
TO THE EI UCATIONAL RESOURCESINFORMATION CENTER (ERIC)
Ole
Level 2A
Check here for Level 2A release, permitting reproductionand dissemination in microfiche and in electronic media for
ERIC archival collection subscribers only
The sample sticker shown below will beaffixed to all Level 28 documents
PERMISSION TO REPRODUCE ANDDISSEMINATE THIS MATERIAL IN
MICROFICHE ONLY HAS BEEN GRANTED BY
2B
SaleTO THE EDUCATIONAL RESOURCES
INFORMATION CENTER (ERIC)
Level 2B
Check here for Level 2B release, permitting reproductionand dissemination in microfiche only
Documents will be processed as indicated provided reproduction quality permits.If permission to reproduce is granted, but no box is checked, documents will be processed at Level 1.
I hereby grant to the Educational Resources Information Center (ERIC) nonexclusive permission to reproduce and disseminate this documentas indicated above. Reproduction from the ERIC microfiche or electronic media by persons other than ERIC employees and its systemcontractors requires permission from the copyright holder. Exception is made for non-profit reproduction by libraries and other service agenciesto satisfy information needs of educators in response to discrete inquiries.Signature; ir
L It'a 4- 1/1 7 I-,:49 I) 7iIM 1 adPrinted Name/Position/Title:
karen Mufnkrd Poore/ hialtr 0,Organization/Address:
0 E. ieennedy Blvd. R0.602 3.103 kuntx, FL 33601.3goe
Telephone: z., _(613)27,Z-0'34 F AY 6 7 3) 272 -i
E-Mail Address:
Karen. iviunitard PDate: h103
scthe . 2 . as
Emit/417'o,,
III. DOCUMENT AVAILABILITY INFORMATION (FROM NON-ERIC SOURCE):
If permission to reproduce is not granted to ERIC, or, if you wish ERIC to cite the availability of the document from another source,please provide the following information regarding the availability of the document. (ERIC will not announce a document unless it ispublicly available, and a dependable source can be specified. Contributors should also be aware that ERIC selection criteria aresignificantly more stringent for documents that cannot be made available through EDRS.)
Publisher/Distributor:
Address:
Price:
IV. REFERRAL OF ERIC TO COPYRIGHT/REPRODUCTION RIGHTS HOLDER:
If the right to grant this reproduction release is held by someone other than the addressee, please provide the appropriate name andaddress:
Name:
Address:
V. WHERE TO SEND THIS FORM:
Send this form to the following ERIC Clearinghouse:
ERIC Clearinghouse on Assessment and EvaluationUniversity of Maryland, College Park1129 Shriver LabCollege Park, MD 20742
EFF-088 (Rev. 4/2003)-TM-04-03-2003