You are on page 1of 34

" "Meta-Analysis





" " Meta-Analysis









.

:


.
.


...

...
... .
(Wolf, F.M. 1986 ).

Glass
((1976 primary
analysis
(secondary analysis (re-analysis

.
" " Meta-Analysis
) . (2000
" "
:



.
& Kavale
(Glass (1981:
-1 Narrative Reviews
-2 Box-Score Analysis
-3 (Statistical Integration (Kavale & Glass, 1981




.

" "

:
-1 ,

-2 " "

-3 "
"
-4 " "
-5 " "

-6 "
"


-
1976
- .

1993 .




.

:
:
.1 " "
.
.2 Meta- Analysis
.
.3
.

:

.

" ":Meta-analysis
"" ""Meta-analysis
1930 . 1976
" "Gene Glass
(Garfield, E.,1991) . ) (2000

"
25 "Meta-Analysis at25 ""
" : 25

....
%25
.. 1975
ERIC 1500 ) ." ...
(Glass, G.,2000

) (1 "
" 1990-1975
.(PsycInfo-ERIC-Social ScSearch (Lyons, L. C., 1997

)(1 5 1 5 1 9 9-10 9 7

350
300
250
200

100
50
0

197
5
197
6
197
7
197
8
197
9
198
0
198
1
19 8
2
198
3
198
4
198
5
19 8
6
198
7
198
8
198
9

s y c
In f oE
R
S.Chalmers,IC
o c i a l S cT S
e a r c h
P"
"
" "The wave of the future
(Mann, C., 1990) .
:
" "


) . (2000
" "

.
" "
.
.

150


(Lyons, L. C., 1997) .
- (Drowns & Rudner. Bangert- (1991
:
.

(Bangert-Drowns & Rudner, 1991) .
) (1993
" :
"). (1993
: (Davies & Crombie (2003
. .
.
(Davies & Crombie, 2003).


.




.

(DeCoster (2004 :
-1
.

.
-2
.

.
(DeCoster, J., 2004).
:
: ) Qualitative Summary
(.

: ) Quantitative Summary (.
: .
""
: .
:

-1
.
-2 .
.
-3
.
-4
(Rosenthal (1984
:
) ( 1
combination
comparison
)(3
)(1
probabilities
)(4
)(2
effect sizes

) (1
) (2
) (3
) (4
.
Glass -
) ( .
.

(Schwarzer, R., 1989) .

(Durlak & Lipsey (1991 "


" :
:
-1




:
-2








)(:
-3




:
-4
pooled SD
.




:
-5













:
-6




)
(Durlak & Lipsey, 1991

(Lyons (1997 ": :


" (13)
:
:
-1
- - .
:
-2
- - - .
:
-3
" "d....
:
-4
- -. ....
.
-5

-6
..
.

-7
. : ....
) (3-1
-8
.

-9
.
.
-10 .
2
-
- )- Orwin, 1983) Fail Safe N
-
artificial sources )
( -
-
.
-11 .
-12
.
-13
(Lyons, L. C., 1997) .10

)2004( :
.
-1

-2
.
.
-3
.
-4
(DeCoster, J., 2004) .
-5
" "
-


.

:
Quantitative Review Procedures
: ) "" )Z
scores )Cochran 1937, Fisher
(1932, Person 1933 and Tippett 1931


.
: ""d
. Cohen`s d ""
"".

: Effect Size

.
. "
" Meta-Analysis
. Lipsey & Wilson (1993(

.
) ( ....

:
.
-1

-2
. " effect
(size correlation". (Rosnow, R. L. & Rosenthal, R., 1996



) (2

.SPSS ) (2002 ":


"

) .
(2002

.

-1
"Hedges`s
"g " "Cohen`s d :

: ""g


:
g = M1-M2/Spooled
M1 M2
[S=[(X-M)2/N-1
Spooled=MSwithen
" "g
.
=

" "g "" " " " ":


" "g
"":
n1=n2
g=2t/2n

(g=t(n1+n2)/(n1n2
" "g ""
.
= ""

" "z " "t ". "z
" - "f = 1
" "f
" "g "" :
g=2f/n
" "f
" "f
.
" "g " "d :
(g=d/ (n/df

10

" "g " "d


.
" "g " "r :[(g=[r/(1-r2)]/ [df(n1+n2)/(n1n2
" "g
)-1 (
.
: "" "d "g
" "d : (Cohen (1988 " "d M1 M2
. "
"
.
- . " "d
.
d=(M1-M2)/
1996 Rosnow & Rosenhal
pooled
1988
:
2
2
pooled=( 1 + 2 )/2
pooled :
( pooled= Spooled(df/n
df n
" "d "" : (Rosenthal & Rosnow (1991 " "d "" :
(d=2t/(df )]/(df) (n1n2)] d=t(n1+n2
" "df "" " "n
.
" "d "" : " "d (d=2r/(1-r2
" "d " : "g ""d " "g :
(d=g(1-3/4(n1+n2)-9

:
" "g
". "d

-2

size

11

Correlation measures of effect


= .rdv, iv ry
2
""r = 1 ry== : 2/n =
2 .
" "r "" " " [(ry=[t2/(t2+df : = 1
. f=t2
" "r " "U - :
(ry =1-(2U/n1n2
" "r " "g" "d :2
2
"[(g": ry=g n1n2/[g n1n2+(n1+n2)(n1+n2-2
":"d
2
2
(ry= d l(d +4

:
(ry== d2l(d2+1/pq
p=n1/(n1+n2) q=1-p



" "z :
[(Zr=1/2 ln[(1+r)/(1-r
" "r : :
y=b0+b1x1 : y x1

:
(ry,x1=b1(sx1/sy
s .
)& Cohen, 1988; Rosnow
(Rosenthal,1996 ; Rosenthal & Rosnow 1991 ;DeCoster, 2004 ; Schwarzer, 1989
:
Population effect size
-1
.
Variance due to sampling error
-2
2
=])-1 ( [/ .
) ( =Population or residual variance
-3
- .

12

= Residual stander deviation


-4
) (.

:
2
-1
.
2
Zi-) )=2
2 Z
) (i=1,2,3,..k = ).k) -1

-2
.
%75 (Schwarzer, 1989) .
-3

Cluster Method
moderator variables :
... .

:
" "d
d=0.2 d=0.5 .d=0.8

. d=0.0

) % ( d=0.8
%47.7
d=1.7 %75.4
.
" "d
: d=0.0
) (50 d=0.8
) (79
d=1.7
) ( 95.5 .

13

) ( 2


97.7
97.1
96.4
95.5
94.5
93.3
91.9
90
88
86
84
82
79
76
73
69
66
62
58
54
50

2
1.9
1.8
1.7
1.6
1.5
1.4
1.3
1.2
1.1
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0

) ( 3 ""d" ; "r"; "r2


""r
""d
0.707
2
0.689
1.9
0.669
1.8
0.648
1.7
0.625
1.6
0.600
1.5
0.573
1.4
0.545
1.3

14


%
81.1
79.4
77.4
75.4
73.1
70.7
68.1
65.3
62.2
58.9
55.4
51.6
47.4
43
38.2
33
27.4
21.3
14.7
7.7

"*"r2
0.500
0.474
0.448
0.419
0.390
0.360
0.329
0.297

0.514
0.482
0.447
0.410
0.371
0.330
0.287
0.243
0.196
0.148
0.100
0.050
.0

1.2
1.1
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0

0.265
0.232
0.200
0.168
0.138
0.109
0.083
0.059
0.038
0.022
0.010
0.002

" "r2

:
-
0.8 0.2 0.8
0.2

" " :




.
(Kavale & Mattson (1983 1980

.
:


.
(Gersten, R. & Vaughn, S., 2001) .

:



.


. :

15

(Kavale, K. (1982 ":


: " .
.
:
:
-1
Psychological Abstracts, Dissertation Abstracts, Educational Index, bibliographic
.(citations from review articles, and two computer searches (ERIC & MEDLARS

500.
:
-2
-
) (
.
135 500
5300 .
:
-3

.
:
-4
)
( .
:
-5
.
) (1993 ":
".

. :
"
-1
" " :
meta analysis
."...

:
- .
- .
- .
- .


16

- )(13
.

-2
" " d
" : g r Delta
."rho
Orwin's
-3
Fail Safe n .

-4

.
2

-5
.

-6

.

-7
" ."3 2

.
(Todd, J. t. (2003" :
: " .
:
:
.
38 51231 .
(Barling & Gallagher (1996
(Cooper & Rosenthal (1980



. .


:
PsycInfo, ABI/Info and :
-1
Dissertation Abstracts part-time, work :
.status, work schedule, and employment status

38

17


51231 . ":
" :

"
".

-2
: .....

-3
: ) /( )
( ) (
) (.

-4
""
"" d .

-5
(Huffcutt & Arthur (1995 sample-
(adjusted meta-analytic deviancy statistic (SAMD
. (Gadel (1953
SAMD .
)
-6
(.
:
-7
_ )(d=-0.05
) (d=0.02
).(d=0.1
- ). (d=0.39
- )
(
. d .0.01- 0.06-
(Ellington, A. J. (2003 ":

" .


.
:
:
-1
Educational Resources Information Center

18

((ERIC) & The Dissertation Abstracts International (DAI


Journal of Research in Mathematics Education (JRME), School Science and
Mathematics, and Educational studies in Mathematics 1983
2002
:
1983 2002




= 86.
-2




.

-3
Hedge`s Q
Statistic 2 =
1- (Huffcutt & Arthur (1995
54
.
:
-4


.

(Kavale, K. (1981 ":


: "
.

106 723

.
.

.
(Kavale, K. (1982 ":
"

19


.
:

-1


.
161 1571 8
6 .

-2


.

-3
) (
(Glass (1976
.... .
:
-4
.
(Beck & Scott (2001 ":
: " .

.
:
1998-1950
-1
" "cognitive specificity" "cognitive content-specificity
""PsycINFO

-2
: Beck`s
content-specificity hypothesis
zero-order correlation :

13.
" "Z
-3
" "r
.
:
-4

.

20

(Twenge, J. M. (2002":
: " .

)
(" .
446
312940.

60 .

.
.
) (2000 ":

: " .


) ( . :

-1
:
1232
""
. -
Meta Analysis Program
(Schwarzer, R. (1989 :
= 0.32 = 0.007 =
0.004 = 0.054
) .(0.08 10.44 = 2 4
.

[0.4266-0.2155] 0.05
. .

-2
:
16 3869
:
= 0.19 = 0.019 =
0.004 = 0.122
) .(0.048 78.55=2 15
.0.001

21

.
Cluster Method
: 11 10 12
= 645
9 15 5 6 2 1 4 3 16 7 13 8 14
= .3224
:
: = 0.1358 0.007
= 0.004 =
) (0.034 12.22 =2
12 .

0.05
%100
.
: = 0.47 0.0009
=0.0028 =
) .(0.118 0.920= 2
2 .
. .

. .

-3
:
19 = 3057
:
= 0.127 = 0.0086
= 0.006 = 0.051
) .(0.032 27.086=2 18
" " .


. 2
. .

:


.
)(2000

22

-





.
) (1982) (1981


...
.
:



" : " .

. :
-1

Full Text )) :
(ERIC, Science Direct, EBSCO-Publishing, Blackwell-synergy
Dissertation Abstract International

: "" " " " "
-
Academic self-concept, self- :
concepts, learning disability, learning disorders, learning difficulties
34
) (
....
.
-2 :



....

23

1330
)(.
: )
-3
( ) (
)
( ) (...
.... ) (
....
"Hedges`s "g
:
) ( 4

Hedges`s
*1

2
""g

1.0
0.587
100
100
1
0.71
1.22
44
86
2
0.74
1.15
44
86
3
0.87
0.655
44
86
3
0.77
0.698
44
86
3
1.0
1.065
45
45
4
0.86
1.021
9
18
5
0.86
2.68
9
18
5
0.86
1.38
9
18
5
0.75
0.16
41
567
6
1.0
0.39
38
26
7
1.0
0.47
25
25
8
* ) 1 2 (

" "g " "d " "g



) (

"
"


1460 - " "g
. 1.36
-
) (

- :
***************************************************************
RESULTS OF META-ANALYSIS FOR EFFECT SIZES d
*

24

***************************************************************
File name: academic.d
Number of effect sizes: 10
Total sample size: 1460
Unweighted mean of
Observed variance of

effect sizes g =
effect sizes g =

0.77700
0.16005

SE =
SD =

0.12651
0.40006

Unweighted mean of adjusted effect sizes d =


Observed variance of adj.
effect sizes d =

0.76809
0.15428

SE =
SD =

0.12421
0.39278

-----------------------"Weighted Integration Method"-------------------Mean effect size d+


= 0.70721
SE = 0.06670
Significance
Z = 10.49572
p = 0.00000
Variance
= 0.00445
SD = 0.06670
Confidence interval from 0.5751 to 0.8393 95%
Homogeneity Q
= 29.7472
df =
9 p = 0.00048
--------------------------"Random Effects Model"-----------------------Mean effect size DELTA
= 0.73868
SE = 0.12012
Confidence interval from 0.5032 to 0.9741 95%
Significance
Z = 6.14956
p = 0.00000
Observed variance
= 0.15428
Error variance
= 0.06269
Population variance
= 0.09159
Homogeneity Q
= 29.56945 df =
9 p = 0.00052
% Amount of variance explained by sampling error:
40.64
******************* Kraemer (1983) method *************
Mean effect size
d = 0.54798
Confidence interval from 0.4412 to 0.6562 95%
Population effect size Rho = 0.26425
Variance of rho
= 0.00069
Confidence interval from 0.2154 to 0.3118 95%
Homogeneity
Chi-square = 54.29504
** Orwin"s Fail Safe n based on "random effects model" DELTA ***
Fail safe for critical d of .20
Fail safe for critical d of .50
Fail safe for critical d of .80

= 26.93405
= 4.77362
= -0.76649

:
:

."g" ; "d"
:
. Z " d"
0.05
.
:
Glass`s Delta
.
""

25

Z .

.
: Kraemer method
" "g " "r Fisher`s Z
Z -
) (rho " ."d
" "d 0.05
" "rho 0.05
.2
:
.
" "Delta
" = 0.20 = 0.50 ."0.80
27
0.20 0.50
0.80 .
) (0.80 0.50 0.20


0.50
15 . -
) (.
2
Cluster Method :
File name: academic.d
CLUSTERS AT 1 % LEVEL OF SIGNIFICANCE
CLUSTER 1
=StudyID
7
=Effect Size
1.3600
=StudyID
3
=Effect Size
1.2200
=StudyID
4
=Effect Size
1.1500
=StudyID
2
=Effect Size
1.0700
0.7000
0.6600
0.5900
0.4700
0.3900
0.1600

=Size
=Size
=Size
=Size
=Size
=Size

Effect
Effect
Effect
Effect
Effect
Effect

CLUSTER 2
=StudyID
6
=StudyID
5
=StudyID
8
=StudyID
10
=StudyID
9
=StudyID
1

Additional Information
Average Sample Size = 146.0000
Sample Size Std. Dev= 158.0285
Correlation
= -0.5188
between Sample and Effect Sizes

26


7 4 3 2
10 9 8 6 5 1
:
:
***************************************************************
RESULTS OF META-ANALYSIS FOR EFFECT SIZES d
*
***************************************************************
File name: cluster1.d
Number of effect sizes:
4
Total sample size: 377
*

Unweighted mean of
Observed variance of

effect sizes g =
effect sizes g =

1.20000
0.01513

SE =
SD =

0.06151
0.12302

Unweighted mean of adjusted effect sizes d =


Observed variance of adj.
effect sizes d =

1.18393
0.01194

SE =
SD =

0.05464
0.10928

-----------------------"Weighted Integration Method"-------------------Mean effect size d+


= 1.15618
SE = 0.11537
Significance
Z = 10.02202
p = 0.00000
Variance
= 0.01331
SD = 0.11537
Confidence interval from 0.9301 to 1.3823 95%
Homogeneity Q
=
0.3967
df =
3 p =

0.94093

--------------------------"Random Effects Model"-----------------------Mean effect size DELTA


= 1.15618
SE = 0.11591
Confidence interval from 0.9290 to 1.3834 95%
Significance
Z = 9.97497
p = 0.00000
Observed variance
= 0.01194
Error variance
= 0.08460
Population variance
= 0.00000
Homogeneity Q
= 0.38910 df =
3 p = 0.94249
% Amount of variance explained by sampling error: 100
******************* Kraemer (1983) method *************
Mean effect size
d = 1.16847
Confidence interval from 0.9366 to 1.4127 95%
Population effect size Rho = 0.50445
Variance of rho
= 0.00271
Confidence interval from 0.4241 to 0.5769 95%
Homogeneity
Chi-square = 0.40600
** Orwin"s
Fail safe
Fail safe
Fail safe

Fail Safe n based


for critical d of
for critical d of
for critical d of

on "random effects model" DELTA ***


.20
= 19.12366
.50
= 5.24946
.80
= 1.78091

27



:
. 2
-1
. -2
. %75 -3


.
:
***************************************************************
RESULTS OF META-ANALYSIS FOR EFFECT SIZES d
*
***************************************************************
File name: cluster2.d
Number of effect sizes:
6
Total sample size: 1083
*

Unweighted mean of
Observed variance of

effect sizes g =
effect sizes g =

0.49500
0.04043

SE =
SD =

0.08209
0.20107

Unweighted mean of adjusted effect sizes d =


Observed variance of adj.
effect sizes d =

0.49087
0.03997

SE =
SD =

0.08162
0.19993

---------------------"Weighted Integration Method"---------------------Mean effect size d+


= 0.47651
SE = 0.08263
Significance
Z = 5.74023
p = 0.00000
Variance
= 0.00683
SD = 0.08263
Confidence interval from 0.3138 to 0.6392 95%
Homogeneity Q
=
6.4729
df =
5 p =

0.26288

------------------------"Random Effects Model"-------------------------Mean effect size DELTA


= 0.47651
SE = 0.08309
Confidence interval from 0.3137 to 0.6394 95%
Significance
Z = 5.73504
p = 0.00000
Observed variance
= 0.03997
Error variance
= 0.04808
Population variance
= 0.00000
Homogeneity Q
= 6.46658 df =
5 p = 0.26343
% Amount of variance explained by sampling error: 100
******************* Kraemer (1983) method *************
Mean effect size
d = 0.34801
Confidence interval from 0.2266 to 0.4706 95%
Population effect size Rho = 0.17143
Variance of rho
= 0.00093
Confidence interval from 0.1126 to 0.2291 95%
Homogeneity
Chi-square = 13.54216

28

*** ** Orwin"s Fail Safe n based on "random effects model" DELTA


= 8.29524
= -0.28191
= -2.42619

Fail safe for critical d of .20


Fail safe for critical d of .50
Fail safe for critical d of .80






.

.

:

-1

. :

.

-2
.

.

-3
.

-4
-
.

:

:

-1


.

-2

.
.
-3

29


.

-4
.

-5
.

.



.

:

-1

.

-2
.

-3

-
...
.

.

-4
" "

.

) :(2000
-1
:

.

30

:(1993)
-2

.
:(2002)
-3
:
. Bangert-Drowns, R. L. & Rudner, L. M. (1991). Meta- -4
.analysis in
Educational Research. ERIC ED339748
Barbara, H. J. (1996). School-Related Stress: Children -5
with and
without Disabilities. Paper presented at the Annual
Conference
of the American Educational Research Association
,(New York
.(NY, April (8-12
Beck, R. & Perkins, S. T. (2001). Cognitive content- -6
Specificity for
Anxiety and Depression: A Meta-Analysis. Cognitive
Therapy
.and research, 25 (6), 651-663
Carlisle, J. F. (1996). Evaluation of Academic -7
Capabilities in
Science by Students with and without Learning
Disabilities and
their Teachers. The Journal of Special Education, 3(1),
.18-34
Cohen, J. (1988). Statistical Power Analysis for the -8
Behavioral
Sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum
.Associates
.pp. 21-23, 25, 44
Davies, H. T. & Crombie, l. K. (2003). What is Meta- -9
.?Analysis
.www.evidence-based-medicine.co.uk
DeCoster, J. (2004). Meta Analysis Nots. Partially from -10
a
Course in Meta-Analysis Taught by Alice Eagly at
Northwestern
University. Handbook References refer to
Cooper& Hedges
.eds), The Handbook of Research Synthesis)

31

Durlak, J. A. & Lipsey, M. Y. (1991). A practitioner's -11


Guide to
Meta-Analysis. American Journal Communication
.Psychology,
19 (3), 291-332
Ellington, A. J. (2003). A Meta-analysis of The Effects of -12
calculators on Students` Achievement and attitude
Levels in
Precollege Mathematics Classes. Journal for research
in
.Mathematics Education, 34 (5), 433-463
Garfield, E. (1991). Meta-Analysis and the -13
Metamorphosis of
the Scientific Literature Review. Science Reviews,
-14(43), 170
.176
Gerstein, R. & Vaughn, S. (2001). Meta-Analysis in -14
Learning
Disabilities: introduction to the Special Issue. The
Elementary
.School journal, 101(3),247-272
.Glass, G. V. (2000). Meta-Analysis at 25 -15
Glass.edu.asu.edu\gene\papers\meta25.html
Kaval, k. A. (1981). The relationship Between Auditory -16
Perceptual Skills and reading Ability: A Meta-Analysis.
Journal
.of Learning Disability, 14 (9), 539-546
Kaval, K. A. (1982). Meta-Analysis of the Relationship -17
Between
Visual Perceptual Skills and Reading Achievement.
Journal of
.Learning Disabilities, 15 (1), 42-51
Kaval, K. A. (1982). The Efficacy of Stimulant Drug -18
treatment
for hyperactivity: a Meta-analysis. Journal of Learning
.Disability, 15 (5), 280-289
Kavale, K. A. & Forness, S. R. (1983). Hyperactivity -19
and Diet
.Treatment: A Meta-Analysis of the Feingold Hypothesis
.Journal of Learning Disabilities, 16 (6), 324-330

32

Kavale, K. A. & Glass, G. V. (1981). Meta-Analysis and -20


the
Integration of Research in special Education. Journal of
.learning Disabilities, 14 (9), 531-538
Lipsey, M. W. & Wilson, D. B. (1993). The Efficacy of -21
:Psychological, Educational, and Behavioral treatment
Confirmation from Meta-Analysis. American
,Psychologist, 48
.1181-1209
Lyons, L. C. (1997). Meta-Analysis: Methods of -22
Accumulating
:Result Across Research Domains. Home Page
.www.mnsinc.com/solomon/MetaAnalysis.html
Mann, C. (1990). Meta-Analysis in the Breech. Science, -23
.249, 80
Orwin, R. G. (1983). A fail Safe N for Effect Size in -24
-Meta
.Analysis. Journal for Educational Statistics, 8, 157-159
Rogers, H.; Saklofske, D. H.(1985). Self-Concepts, -25
Locus of
Control and Performance Expectations of Learning
Disabled
Children. Journal of Learning Disabilities, 16(5), 273.278
Rosenthal, R. & Rosnow, R. L. (1991). Essentials of -26
Behavioral
Research: Methods and data analysis (2nd ed). New
:York
.McGraw Hill
Rosnow, R. L. & Rosenthal, R. (1996).Computing -27
,Contrasts
Effect Sizes, and Counter nulls on other People's
Published
.Data: General Procedures for Research Consumers
.Psychological Methods, 1, 331-340
Schwarzer, R. (1989).Meta-Analysis Programs. Berlin: -28
Institute
.of Psychology
Stone, C. & Alison, L. M. (2002).The Accuracy of -29
Academic

33

Self-Evaluations in Adolescents with Learning


.Disabilities
.Journal of Learning disabilities, 35(4), 370-383
Tabassam, W. & Grainger, J. (2002). Self-Concept, -30
Attributional
Style and Self-efficacy Beliefs of Students with
Learning
Disabilities with and without Attention Deficit
Hyperactivity
.Disorder. Learning Disability Quarterly, 25, 141-151
Todd, J. T. (2003). Job Attitudes of Par-time vs. full- -31
time
workers: A Meta-Analytic review. Journal of
& Occupational
.Organizational Psychology, 76 (2), 151-167
Twenge, J. M. (2002). Self-Esteem and Socioeconomic -32
Status: A
Meta-Analysis. Personality & Social Psychology
,(Review, 6 (1
.59-71
Wolf, F. M. (1986). Meta-Analysis Quantitative Methods -33
for
.Research Synthesis. California: SAGE Publication, Inc

34

You might also like