ML20235D821

Applicant Exhibit A-7,consisting of Excerpts of Publication Entitled, Belief,Attitude,Intention & Behavior:Introduction to Theory & Research. Copyrighted Matl

ML20235D821
Person / Time
Site:	Shoreham File:Long Island Lighting Company icon.png
Issue date:	07/08/1987
From:	AFFILIATION NOT ASSIGNED
To:
References
OL-3-A-007, OL-3-A-7, NUDOCS 8709250364
Download: ML20235D821 (10)

Text

..' ~.

.....-. ~.

.m m..

.,4 _..

.... -. -. ~..

.y m*..

.., r' r*.

a >... &j'gy v:, y.'i

.p..

~.,. 6 ' ' * - '

. G),$' ry <},7

t4 gs.u'.-~..

, - q{

..Q ;]yje,,',]d,.f a.,.

o

' l, l. 41

. ' ~ j, k,(:y,.-M.W.; $ c. d.

9. :. l ~ A M. ?:..,. f 4,..d.,, -

;7c;.'7.

..,:; y ;* -

n Q

4

.. s. tt"

%,.. %-iy*,%(.. u.s

{x. _ h.',,/, 'r. "

. +*

= '

f,'?-

,. n.,-

"r.

....g

..c.. -

.1,.

MI+

so-z22-n 3 uico y#37 scP 17 a11:a 7sg;7-A-7 BELIEFi /TTITUDE, INTENTION AND BEKMOR:

An Introduction to Theory and Research

,1 -

f.< n cy/

t#'f

%.- r ;f : ' ' l ; 'l i

(f. j'

.s

(

y O~d

~. ~

a' W

..)

i

%f,,j E

d 4

eb 8709250364 870709 PDR ADOCK 05000322 0

PDR

m_.,,,.

__g-,,

__,.._c

_7.-_-_______,.,_-_-

-__,v,

-,._m_,m_,_-,-

77,

,._7 i

l l

[

i e

R-v T

N\\

b' ~

\\#

4 q%1

.s

• t, 9

\\s g,V g,

w

-s1w+

w n.

,.yy 3, y.y, v.,

3.,-.-

w ;,

.s.

.-y 1

< #* *..,3 ; '. --

.. g

.y

_. p. :

._,. ;., n. (.. ;. 3 ig;.

6,p: th c ye s

- +

' y.;ty y J. f )jg1;ffj;4 g.fgj;jGg[

g.g j, _

'. y ;'

!~. '

Y s

This book is in the As Go ADDISON WESLE'i SERIES IN SOCIAL PSYCHOLOGY the me psycho cal lite trality portanc many c i

by the explain

-&r researc t? little at

,,, - role, if

{

' clearly i

a Cour5 i

tude-re ture th radicaR more t!

as no s tude cc Copyright @ 1975 by Addison-Wesley Publishing Company, Inc. Philippines copyright 1975 the ficic by Addison-Wesley Publishing Company, Inc.

g,-

All rights reserved. No part of this publication may be reproduced, stored in a retrieval sys.

The pt -

tem, or transmitted, in any torm or by any mear electronic, mechanical, photocopying, framew recording, or otherwise, without the prior written prmission of the publisher. Printed an the United States of America. Published simultaneously in Canada. Library of Congress Catalog show ti l Card No. 74 21455.

first ap,

ise= a nn emeo l

CDEFCHIJJWA 790 l

^

.I L

Yt b+

c.t

((

4 lt f

l I,*

356 Chapter 8. Prediction of Itcharlor j

t V

l l

Once the attitude scale co'nstructor has selected his set of opinion statements, Th he, like the behavioral observer, goes out and observes whether the individuals in criteric his sample agree or disagree with his statements. The re pondent may be asked to tude at i

simply agree or disagree (i.e:, be forced to make a dichotomous choice), or he indeed may be asked to indicate the degree of his agreement (e.g., to respond on a five-attitudt or seven-place agree-disagree scale). At this stage, then, the attitude scale con-scaling structor,like the behavioral observer, has before him a set of numbers assumed to w

imply something about an individual': religiosity. Unfortunately, it is here that the Table l

similarity between these t'wo types of investigators usually ends. Whereas the situatic l,',

attitude scale constructor will test his assumptions by performing an item analysis, behavi

! )

the behavioral observer will usually just accept his assumptions and decide on some entries arbitrary way of combining his numbers to arrive at a behavioral criterion score."

a gene

{

For example, he may first decide that people who have contributed more than $50 procec should be given a score of 5, that those who have contributed more than $25 but combi l ;;

ic '

less than $50 should be given a score of 4, and so on. If he does this for each of would j

! !E his continuous variables, he can then simply sum his set of numbers and arrive at time t j ]-

his multiple-act criterion.

presen In contrast, the attitude scale constructor first submits his items to a standard in a [

't scaling piocedure. As noted in Chapter 3,if using a Likert scale, he eliminates would g

i <,

those items that fail to discriminate between subjects with favorable and un-t me,

- 11 ; !,

favorable attitudes toward the church, or items that do not correlate with this at-

&*at

. l titude. In a sense, then, the investigator recognizes that some of the opinion items have -

, ". l he selected do not serve as good indicants of the particular attitude he is measur.

attitut ing, Clearly, if the behavioral observer were to follow the same procedure, he

, ; di too might find that some of the behaviors he has observed do not covary with the attitui 8

C0"S" t

underlying dimension of religiosity. To put it a bit more bluntly, he might find that of a r some of the behaviors he chose to observe have little to do with the degree of an a grvt individual's religiosity.

and b Two important conclusions can be drawn on the basis of these considerations, The first should be obvicas: Not every behavior with respect to some object is peric with i i*

ii related to the attitude toward that object. An investigator usually chooses to ob-bein il' serve a given behavior because he assumes that it is relevant to the attitude under e

man!

j

i consideration. What we have tried to show is that an investigator's intuition can be

{

{l wrong. Tests of the relation between attitude and a given behavior, therefore, can C

"5' to a large extent be viewed as tests of the investigator's intuition. Given the as-attitu '

1!

sumption that attitude toward an object determines all responses to that object, an be vi investigator is clearly free to choose any response to the object in testing the atti-PT C'

,, 1, tude-behavior relation. The considerations above indicate that this assumption is t th ii definitely invalid.

I

8. Frequently, the behavioral observer will not even try to construct a multiple.act 9.

l i

criterion score but will metely treat each of his single act observations as a different beha criterion. This procedure is as inappropriate as treating each item on a Likert scale piric.

j' or each bipolar adjective pair on a semantic differential as a separate dependent criter variable.

on w i

.I l

l$

t l

i.

s a

u

u 4

(4 l..

1..

t i i

s.

~

${J

}

370 Chapter 8. Prediction of Behavior

!i b,

It would be possible to continue with these examples, but the point should b clear: the lower the correspondence between the intention's and the behavio h't 4

m levcis of specificity, the poorer the prediction will be.

• s' on ity, one expects a high relation between these varia

. f;-

bc mc ceptual framework suggests that a person's performance of some behavior at a thi given point in time is determined by his intention to perform the behavior at l

te point in time.

0, Stability of the intention. Clearly, a person's intention may change over time Inl follows that a measure of intention taken some time prior to observation of t

[s i ',

behavior may differ from the perun's intention at the time that his behavior is 11: (

observed. The longer the time interval between measurement of intention a ht j( !l observation of behavior, the greater the probability that G.c individual ma J

new information or that certain events will occur which will change his inten f

g g

llI Thus the longer the time interval, the lower the correlation between intent l-Aq.

behavior.

't I. ' '

H.,

Very often, the behavior under consideration can occur only after some h"'

school sophomore may intend to go to college, he V8 l tc ;

Ii b

unc school, passed the college entrance exams, etc IO"

( 0i 1

C'"

Nl tervening steps, the lower the intention-behavior correlation will be He e d;

Pert M

y bi the problem is primarily one of the stability of the intention rather th r again, V"d' higher the probability that the completion of (o an of its h

ma3

[ [:i g

the L

step will result in new information which may produce a change in the ind do k

• *l j

intention."

h, intention is dependent on other people or even

.$i..

fom -

]'

lower the intention-behavior correlation is likely to be. If a person's int

$):

likel f

e behi based on the expectation that another person will behave in a certa

{#

s i

the expectation that some event will occur, and the expectation i i

g i a this information may welllead to a change in intention s not confirmed, e-hab,i observation of behavior, certain events can occ new

?h n

y take-individual's intentions. To predict the behavior from the initial

{ it a car-tion, it may be necessary to consider other variables in additio measure of-inten-P beha

. f This will be particularly true when (a) there is a long time interval betwe n to the intention.

5"i"'

l1 I

n the decis intervening behaviors, and the consideration of this set F.

ical (

,j, rm each of the

~

kindi better prediction of the ultimate behavior than the intention to p f e l; ns may lead to a the r <

havior per se.

er orm that be.

H I

down; Ib WPr a

4.,f 5

Q

=

f@s. lyW4,.

" ~~ #~5,Q4, _.-

., f["..*

j

\\

s

~^'

r y?,j%y34; < f'q :p,'r j '

-.,q g y

^~ j.

6

+

y, 4

n+

.; _ g

..c--

~

J_

_ ' t ' ;,

.g 3,

l.

{

I f

te l

l Conclusten 381 others were told and I

lowing delivery was Given a high degree of correspondence between a person's intention and his successful (enough actual behavior, one would expect the factors determining intentions also to be

. r the fourth hospital closely related to behavior. In Chapter 7 we discussed a theory for the prediction

itary formulas were of intentions, and we showed that behavioral intentions are predictable from the to breast-feed). The attitudinal and normative components of the theory. It follows that whenever a ast-feedmg behavior high intention-behavior relation is observed, the behavior in question should also Computing a mea-be predictable from attitude toward the behavior (An) and subjective norm (SN).

m a sigmficant co-Conversely, even when the two ecmponents accurately predict the intention, they itions supplied more w 11 not predict the behavior if the intention measured is itself inappropriate for than did those with prediction of the behavior in question. We have already seen that A, and SN can arly, however, milk predict behavior with a high degree of accuracy. McArdle (1972) did not obtain i posiuve mtentions, a direct measure of intention but instead measured these two components. The tilk for their babies.

multiple correlation of the two components with behavior was.77.

Additional evidence for these notions comes from several of the studies dis-j cussed earlier. We have discussed a number of studies that found a high cortcla-tion between intention and behavior, aad in the preceding chapter we showed l

that many of the same intentions could be predicted with high accuracy from the attitudinal and normative components. As would be expected, these studies also showed that the two components were highly predictive of overt behavior. For tive example, in their study using two Prisoner's Dilemma games, Fishbein and Ajzen (1970) found multiple correlations of.732 and.793 between the two compo-nents of the model and strategy choices in the two games. In fact, owing to the (10) high intention-behavior correlations, whatever factors were found to have signifi-(7) cant effects on intentions were also found to have the same effects on the corres-ponding behaviors.

ction of single act CONCLUSION

much human be-In this chapter we have discussed the prediction of overt behavior. We have seen

curately predicted f

rform the behavior that behavior can be measured at different levels of specificity and that it is im-l chavior to obtain, portant to distinguish between different types of behavioral criteria. Three major ins to be measured behavioral criteria were identified: single-act, repeated-observation, and multiple-i second, the mea-act criteria. We showed d it when properly constructed, repeated-observation c he performs the criteria are essentially beh vioral measures of attitudes toward behaviors, and or to performance multiple-act criteria are beh tvioral measures of attitudes toward objects.

intention and thus We have argued that th6 best predictor ' f a person's behavior is his intention o

nproveo by taking to perform the behavior, irreocctive of the nature of the behavioral criterion,

)

may intluence the Intentions and behaviors were both shown to vary in terms of behavior, target, situation, and time. Whereas repeated-observation criteria represent behavioral

. y^

should usually be i

vill be highly cor-measures across targets, situations, or time, multiple-act criteria represent mea-sures across different behaviors. An appropriate measure of intention corresponds in its level of specificity to the behavior that is to be predicted. Thus, to predict io) data.

such a single-act criterion as a person's attendance at the 7 A.M. Mass at St.

Mary's Cathedral on the coming Sunday morning, the measure of intention has to N

1

(

d.,

5

G b

i I

i..I.

j ',

382 Chapter 8. Predletion of Behavior H-refer to exactly the same behavior. That is, the person's intention to attend the

_ Unlike n 7 A.M. Mass at St. Mary's Cathedral this coming Sunday has to be measured.

that attit Similarly, the repeated-observation criterion " number of worship services at St.

behavior oHi Mary's Cathedral attended in the course of one year" requires a measure of in-analysis d1 tention such es "How many worship services at St. Mary's Cathedral do you intend related t to attend during the coming year?" To predict a multiple-act criterion, it is usually these sai t

necessary to obtain an even more general measure of intention. A multiple-act

'lI criterion based, for example, on observation of several religious behaviors at St.

Mary's Cathedral (e.g., number of worship services attended, amount of money i

contributed, singing in the church choir, and teaching Sunday school) could be

. r predicted from the following measure of intention: "I intend /do not intend to act supportive toward St. Mary's Cathedral,"

(1

.7 Within our conceptual framework, intentions are viewed as the immediate antecedents of corresponding overt behaviors. The apperent simplicity of this na-tion is somewhat deceptive, however. Since it is often impossible or impractical oO

,I to measure a person's intention immediately prior to his performance of the be-havior, the measure of intention obtained may not be representative of the per-son's intention at the time of the behavioral observation. Intervening events that L !.,

may lead to changes in intentions will therefore also have to be taken into con-sideration. For example,if a person intends to buy a car three months hence, any l

y j;.

change in his financial position, the price of the car, or the asaiLbility of gasoline Hi-may influence his intention and must therefore be taken into account if accurate

[

behavioral prediction is to be achieved. Barring such changes in intentions, an appropriate measure of intention will usually allow accurate prediction of behavior.

,[

Understanding a person's behavior, however, requires more than just knowl-lj cdge of his intention. It is not very illuminating to discover that people usually do i

i what they intend to do. If behavioral prediction is the primary objective, the simplest and probably most efficient way to accomplish this is to obtain an appro-I

[ ;*,

priate measure of the person's intention. If understanding his behavior is the primary objective, the factors determining his intention must be specified. Chapter i

[

7 was devoted to a discussion of these factors We presented a theoretical model y[

which specifies two major determinants of intentions: attitudes toward the be-havior and subjective norms. These two components must be measured at the j[

. same level of specificity as the intention. Given high correspondence between in-i

~,,

tention and behavior, one can also view the attitudinetand normative components.

v. ; _

l

' [i Ui as the determinants of the behavior. In fact, when intention and behavior are h ht :

. highly relate'd, everything we have.said about the factors influencing intentions i[ j,,";

can also be applied to an underdariding of the'dEcfminants of b'ehavior. Thus,.

~

4 n,

it should not be surprising to find that attitude toward the behavior is often re-l ':

Icted to performance of the behavior.

T,'

in contrast, traditional measures of attitude toward an object can influence a given behavior only indirectly, and thus low and inconsistent relationr. between l'

these attitudes and single-act or repeated-observaticin criteria are to be expected.

t I

r

$ 4 nf q A s g ' A

+

i y

Toward a Cumuladve Body of Knowledge St9 ly relatively trivial ating socially rele-The conceptualization proposed by Hovland and his associates has a decep-4 noted in Chapter tive elegance. Since the yielding mediator is never directly assessed, it is always possible to account for any obtained result by making suitable post hoc assump-which research is i

tions about this mediator. The crucial test of this theory rests on the investigator's It is just as easy ability to demonstrate the consistency and validity of his assumptions concernin it is :o develop the yielding parameter. These problems are seen most clearly in McGuire's v of the literature (1968) formal two. factor model:

ritfut than labora-

d his poutioa on p(g) = pgg)p(y),

cognize that this Probability of opinion change is viewed as a function of probability of reception Id study, which I and probability of yielding. If opinion change and reception are measured, it i meet the basic becomes possible to solve this equation and obtain a value for yielding. This procedure, however, adds little to our understanding of persuasion unless the d neither for an estimated value is consistent with psychological theory.

or the testing of Our discussion above concerning inconsistent results with respect to different dependent variables provides one example in which estimates of the yielding ach has dealt for parameter would be incompatible with psychological theory. Specifically, such relevance. Most nndings are inconsistent with the assumption that yielding is invariant with re tTects of source, to the kind of dependent variable under consideration. A similar problem exists

e produced by a

' ailed to produce in relation to the etYects of various independent variable manipulations For example, to account for the inconsistent effects of high-and low-fear appeals one

er

d by the would have to argue that fear arousal sometimes increases yielding and at other set mdependent times reduces yiciding. This argument is inconsistent with the assumption that given manipulation, such as fear arousal, should have consistent effects on ic neglect of the yiciding.

3ach shares with s, book we have These considerations apply whenever a theory is, tested by. estimating one o

.5 more.of its ruajor variables or parameters. Although such an spprosch isprfectl atGtudes, inten-legitimate, it will contribute to'a cum' lative body of k'nowledge only if it is 3een that these u

accompanied by a psychological theory that allows the investigate'r to derive 3 instances of a

.g. a testable hypotheses about the theory's parameters. The psychological assu

~

Je that they are arbitrarily from one study to another or changed to e approach his attitudes, inten-set of experimental findings. We have discussed some of these problems with ge content and information integration (see Chapters 5 and 6). respect to the appl apparent when-(Terent kinds of

-h

. r.2 TOWARD*CUMCLATIVt BODYOF 150$EEMY luenced by the a

We have adc pted in this book an approach based on our conceptual framework mplies that the which we hope will facilitate the accumulation of kriowledge in' the attitude area s, or behaviors This approach makes a clear distinction between beliefs, attitudes intentions an often obteined behaviors it indicates how these variables can be measured; and it specifies th oach does not relations among them. This set of concepts and their specified interrelations vide a conceptual framework which can be used to analyze various phenomena 4

g3

.r.

A l

~

w*,

1 p.

l, 520 Chapter 12. Conclusion and lines of research in a systematic and consistent fashion. We have scea that J

i this approach leads to the formulation of hypotheses that are consistent with a

[

wide array of empirical findings. At the same time, it provides explanations for the apparently inconsistent findings in different areas of investigation.

i, Most important, our approach permits the investigator to test his hypotheses Ng' i {.

about theoretical processes directly by obtaining measures of his explanatory con-W l:

N structs. Most studies are designed to test a hypothesis about the effects of some 4

manipulation on a dependent variable. According to our approach, the influence 4

F beliefs and the direct or indirect relations of these beliefs to the dependent f'

of the manipulation on the dependent variable is mediated by its effects on certain a

5

-(

I' i variable. Hypotheses about the effects of the manipulation on a given set of beliefs and about the exact nature of the relation between these beliefs and the dependent I; 1 variable can be directly tested. This is done by first measuring the beliefs in f, j question and examining whether the manipulation had the hypothesized effects.

f The links intervening between these beliefs and the dependent variable can also p

be assessed, allowing a test of the hypothesized relations.

Stated somewhat differently, adoption of our conceptual framework forces the investigator to make explicit assumptions about the processes intervening be-1 h'< hi framework specifies the intervening processes appropriate for a given dependent tween his manipulation and the dependent variable. Moreover, our conceptual t

lt N

variable. On the basis of this approach, an investigator studying the effects of com-Abelso k

I municator credibility on behavioral change might make the following assump.

and P.

i l'

y tions: (1) Communicator credibility will influence acceptance of the statements Chica; contained in the message. (2) Differential acceptance of source beliefs will pro-Abelse

{

3 jy, j

duce different amounts of change in the corresponding proximal beliefs. (3) tions. I p

Differential cha,nges in proxirnal. beliefs will influggps. amount.of change in beliefs Bchavi

.g f.;

about performance of the behavior. (4) These$ anges'will produce correspond-i h

ing changes in attitude toward the ; thavior (5) As a result, communicator 1. Abelso T

attitud!

credibility will affect intentions to perform the behavior. (6) The corresponding I

behasfor will also be affected. Each variable in this sequence can be directly Adams i

,[

measured, and the hypothesized effects of communicator credibility can be tested.

Adams l,,

When communicator credibility fails to influence behavioral change, it is possible learmn lj,

to discover where in this chain of effects the investigator made inappropriate Adams assumptions. For example, the assumption that changes in beliefs about the verbal l; ;

_[

~

con' sequences of performing the behavior will lead to change in attitude toward Adams that behaviM.rn;y not hav,,e,.been supported in tgcular study.

a

.w

. g pg W4.hsq-seeit that mmfy/c4h6cumradir,ggr.urtnconclusive-findingrin--

7-374 p'

.p em. -

the attitude area appear to be attributable to such inappropriate assumptions about 4 l:

+

4 the relaticas-between different kinds of variables. One clear example of such a A.Jzen, y

{ ;;

~

' fallacious asstimption 'is the hypoth8iis tha't att]tude ioward an object is related to freedo:

4 L

specific behaviors with respect to that object. It ismur hope that by distinguishing 197I' between beliefs, attitudes, intentions, and behaviors and by specifying the deter.

Ajzen, i

minants of these variables, our conceptual frr.mework will contribute to the effects J

development of a cumulative body of knowledge in the attitude area.

280.(I n

N Il 4

N ll l!

o m

e MW Mmm.n!@g%c LEA,R REG.ULATORY,;CO,MM,ISSIONMV%+%+d,Qg@g$m$

MMQ

1. NUC g4W y R dW 4.

a g%

~

g$h!p@%Mk$EM0RANDUMM ame@Qec)$s]igangNQy e

MM ohn f1 gg gh hM e%DWM A m p o. soc 1atasyW g g g hg g

g$$gd$

nas

% m}M@ y A g W % {j Q d W W'e A

d 4

k%mWC'thjef $ $ $ g Q @Ww % Bg g m *w 4

W 29n+m. WMQ W R.am,h%

syrge o y

Q fY N6f ?

Q Mn$WQOMM i MMP Q WY9QW w&W "eMJ u11 a n )g iEm11 t;i.ngi y

$ @&,%Any% Q M e % @W M E R $ $ $ $:

MS'9 T

4&

JBWM wW.m.

d J docume nts&f31 e dT,onst.h.e&s,o sen g recW W M 4 % %1M W W~ e W"te x h i b.n g g a n d at e d e ncE,p,a r tt o fit

,1 el. o f f i o.r:d l s h ~e a~r i ~n g e d nN they S HO M M pro dh..,Wp%w@

ync -ceedi cia

-~

v p: gF $m?ith e? c o u nt Mp r e's s w$ b e Qn w it& mis -- ~ide~ red exempttfro~pth_e9royisionss fsth~etunitedwE d

b<7JJ 4 cons o

WW 6%S t a"te W C 6pi r i s h't%c t ku n is s sii tM w a s% H g j n E 1;1 ynfi l ehu n dir h

1th l

yrigwp gyn qm@ym my:kh x

wm

n w

p t h'e w d o c u.... c.,<.. a,, ~..: w,,mentsM entttoNT U.

y+..

w.

,w:n w

p+,

,i c$11$ fx,.m.

n..x<.

m. w eg

,a., y.. z.

z w.yo m-roc e s s i n'gs. f a un. - n,i nR i t h, s... w w>,,,.Q M,tQ qW u,g,

/

c...,a.

.,r y

s

~s om 7

for

'e.xemptlcTa:ssificati,;o W W W ~, f?p %n& n o

6 Ms@w@ MVQP

.d %ge Ma, %,M_ggg$$qq n.

4 n:n m

jp g

y.<

'- ('x y y]O

+s

- d:q, y.

p.& y 9,3_

Q

_g; m:

n

.j jg[g;ja..

g- ' I 3

3

' }WJgy' ij V o W<

$q

.j

+

yg;.

m

.ya,+<g"pmy%xc gg@g,mi

m 8 3 y sy w, 'e,

w' ;

g Neh W

g

g y

w" ',,, %

g,4 o

4 h,$

s J

99 1

9p; ggg ypp;y gaq t

-u,

=,

m w%..n%m o

s m,.mm.

v.we. ~

wh -

f Y,,

s

?e y% 'l% % ] m gM[$'4

~

f',*'

..,. m :; Q^U 1%

i E

V'~'

&: i..... '

,tq.i a

i tW ; j N, W' ['W. u.

1 W'y c.

p '%s 7

a Y 'L W p' %p L

L mq n

h M~

<av

.K~ n an, pgjs, ;G x, 4,3 w

~

8

&q:'

9..

s > m% % y)l y;-g;;;;< g:Q,ihn., X"; < m:h.n.

'g:i m, u_t,

J: '

c

n wa a

s, m.w, a w rl}.

)u n;f, q

t

.'.y t

~ if Q-a 4

,e

..a

,~; n y

A,

,,+

p.

[

]

'i 5

I

+,$$

Q Qm., u..('.T f% K r

.y~

a y

~

i, 1,

MN'Y i >

3 x

, 1 3

c

,m.. :,..

m 1

1 i

a w

s y-

,.1 ?,

5 t 4;W x i

.,1 c g 4p e

6 2s w

Qk

+#

"'/

s R..

.,e 4

',4

' g%

s m.

y g.;

,3,.

'. I $, c i : -

'y

,,2 1,

y, t

. c.

f j' m

y 4,.

4

,,5 g

{

(

-}$' ' '

hl N.

[,

5 W' q ' u r n,

,p i

? (f g

'S' d

> Ul {: ~

j.

pi.

'a jh N '

u\\'

n

' [ Q:; f,, -

.; 3 4

(v Y

h 4

l lW

n

'R,

T

, fy t

e j

c i

~-

, i J _ y(:

q-OM w

%9 c +' '

wNu 4

m.,

s r

-; N?.

9 ;

w '

e y

i

'S, s

fQ.%.f: _

,i

, j h h;_

l,:

l Q,,m n p,i e .z L 1 '1 3 gpg .y4{,

• vt 4

.'u >wy g 6 g;p. m wo J ) L.3 l s k y] yl ]g iW %m i '1 1 ,.hi,'c Q.< l .i 'k L$[ %_; e j,d,mQp. y 4i! s ,/ ,J ( ^ j ,3, p j 4C ( r r y. 'e f.]pp/p,a %,.s s'

-M TM Wi

',7., 8 yI + i [q;y % py vvh,, 4.:1'. ' '\\, ,M,, :h, [e t.[l pr r i,.. 4.,. < g. ' ' [f

• t ll4:; 4f

' f ,<l Q O' 1 ,4 s < r s gM : +,' ^! r i a i f ' ' fhlq q vs ~ ~ ww, p-g?

p Off(;, 3, o.

r s, , ~ ,2 a '\\ 4 'f 5 I b 'w % m;mMy E 'i { p ,, ^ 4 3

w' 1

1 ~, 4: > 'r v a ~ ~ - - - $9 5 O. ~his . ~ M. a-

• }}