"PPPPPPPPI
, .

‘M

" " I. P ' Pl! -. o ‘7-
.3 .' ”1'31? $3.33!,- ~, ., . P!" 11";"1 h: 3'“. 3.
a 31.111. "Pfi‘.’

. ...P1PP.P§PPPPPPJ 1.2.

P!" o'P-P' ".931 ' '3
1.11 P3" f‘

3 1
1 P1P.

.3 Pl "113 P
P.1P'3'P $113
{1|

3 ‘11

V
.I
1.
I I1 3
O
o

. . .1 ..
' "‘1 PP‘ ': '9- "'P' P £3"

:3"

   

     
 
 

 
 

‘ I
m '-_
1- ." 0"-
W

 
 
  
  
      
  
    
   
     

 
  
       

H 0 ’
HI.

q...—

.-

Z.

Y

 

   
   
 
   
  
  
   
  
   
  
 
    
  
  
   
 

1'1
1}-

1

1'.
3.313?

1

J
l

 

 

'1.
3.

' " .115 ‘3 " -' ' 3
“31111111.?11'11311‘111 3 3 3" ° .3 P 3 13:3 3‘ z. 1 33': P1163333
PP'PP PI'PI "'1 PP "" ; .=Z:P ,:"Ias1.‘«‘aI.‘.:..-a2'.
3'331” P ‘ ..= . 12*.."11P -§;:!P;‘5.=:“.'1: :1.
PP. "PP PM 'P s ‘ P-

 

1P P 3P.31' {111131111 331133313
.' P '._ '.P'P
.111 P

 

”'11P!" 1111 1:1P3PP3. "3335‘

'3 '0 \P h" 110
. 3. 3 3'3'3 13331331111313‘1331'3333 ‘3ng3 51:31! 6&3“ 21133:}. 913 333333 $1 11. 3
. :1'. 3.. "'30" '1' ' 1'" 1‘ 51"1
3 P13“ '11:. '3. “1113911111 «PL-‘11.}. 3'3“ 1‘1'3‘11 1-111: 31111; 3:3; 41gb; 211' 32:13? 114'}: 9333 1;; ‘3 11 LP‘ 1:13:31 ""1" 3‘ 13.33,
1 13633323333333 P1313'331313'31P31'3‘3'3 '11 3"? :1 1%: '.}P3"33‘3§ 333.53., 93313. 3 3 3. P1 3‘3‘3'311 ‘3
1"1‘PP'P " PP'P'PPP‘P " '.'3-33333 "
.. a. '

' "'(u 331'11‘31 ~13 791

'PP'P’PP 11' 133' “.3111“ "'11“

1.

13133313 fl 1.1x ‘32:}? 331$?” 33333 31:3 31:1,. 1111139
I11 ‘“ " n, 21...?
.P'P "1"?" '3 '131.P"P.'A"31 33 1' ":P'U'PP-P‘qi'

L 9. ..,.

" 1 1. IP' ~ 1 I"' P'PXPP'PP'H'PP‘ 1131' {1'1 1‘1 33? ’1" "P'PP'PP?
. 3"- 3. 3.13.0353 33'. '3 . ”‘31... 3 33131;.{13 31333 33313333I133'3. .3; 3353:? 5319243311; 31:.“ ‘13:\1"3’3:3"‘3133331_'333P “31:13:53. 3J3: 'g33o.3:j
" w'b.--"':1.3.1."'-':'-'1’1:=.‘.1"":iPs'-\+IP‘."121»“1"113".*~21~.'.'.2‘.1.1'I:3'. 1.31'12.;11..o.33i.:3, 611". «3333.33.35,

" "' -' " - ‘PP'PJ ."."1P'1 .PP‘E'J' '1 P 113.13 . "'1'; '1“ 113201113611“ 333139.111 1' '3', "'3! ”11"3'3' 1P .. :éd'fivfl'b'fi' 3

' " ' ' "1 h ' 1! - .3 :_ 17P2.31'.PP“.11"'1:$.1‘.313”: 1.31.3.3133z3'3P‘sPPiI33P13'1'1'1‘31wP'PPPPP'."23P?«111....3113z313I333P33'2P.

. . P... i1%.ng33333393511333.1131]. 333:3 1333.. WP 313. 3 33 3.33%.3333333333.

0'1 ‘1' 1' 1"3'3 ”MP 33““ '1'"31'3""PP"" 13'313133‘3913!‘ "11' " .P 33.33% -1 .32313?"-313.13':'3."33.'.'33"31 3333133:}.1.333~1.~""'.':';'.3~111'

. 3h 1"'1 P. 1. '1' W '11 P... u.P“""'-P=. 1:11:21'."':!:‘PI.£'-"

113.154 33.3. 313111.. 3P3P3333311'1P33‘3.P33‘P.1.':1:~P;..l1|1P-'.1”"P. 331.1'1 1.11341 133'5‘ 3313333133313 .3333‘33y3‘3t3.333331§3

P ._.3 . .

. . _ .. 3 3.113313“? 13333.. $331121.
0 . 3 a 3.5:"; .313 391631313“th13'331'333" ‘3‘33hi‘fi 3’3 3 ‘1 L}. f:
31.....W.1 .1

"I .1 1 31'“; 3.:6 with-5 .1: . Z-
‘1 ”"2” 111.5. 11"”. 2.... PP'PP'WPu' ‘3'3‘3‘11'P1I1' ""1 11"'""'."'.‘11'J."' P'P'Pi'PfP-PP "' 3::‘\'
"1". . 33 1 33 ..3 3‘ .
1H1""'3£13‘111{';1 731313333 .1211 :" [1'13 '41“ 1%!"""""'\" PJ'PJL'L' "5111'. 11153;. "' 11'331 113133311
3‘ ‘P 1331

' " “111' P1333” "z' 1111-1311 33

 

11
.. «PP
1:11.11.

‘I
Q
04‘.

"H: . . 1'33: H.413.‘ o '3. 1 a '. '.""""
. » P PP PP“ --1
.3 .3. .3303.‘ . ' .3 . ‘3

.. . I 3
P '.11P1'P.'P"‘11'1

1'4. 641:1.

   
   
 

        
  

 

THESTS

QT“ “'33“? ff]

jivbtfb‘fi‘

 

-. 3),“?59
. ___n -.. .41. q M3 to 0 ‘
Eifié‘ijgfl M1“ up ‘w- .59.-
.. 3 Ugh?“ fiflnméig’ \
«‘ I'Mv- '
\ir‘rv.‘ _ 1”

 

This is to certify that the

dissertation entitled

THE EFFECTS OF DIRECTIONS TO CHANGE ANSWERS
ON THE ANSWER CHANGING BEHAVIOR ’
0F COLLEGE STUDENTS

presented by

Patricia Marie Wilson

has been accepted towards fulfillment .,
of the requirements for - -.v-:: .4 " ’

 

Ph . D. degree in Teacher Education

 

 

    

Major pro ssor

Date October 24 , 1983

 

MSU is an Affirmative Action/Equal Opportunity Institution 0-12771

 

llll llllllllllllllllllllllllllllllllll

3 1293 01739 8508

 

 

usmuuas remove this checkout from
your record. FINES will
be charged if book is

 

 

)V1531_J RETURNING MATERIALS:
Place in book drop to

 

returned after the date
07'2 4 Q.§ a stamped below.

sip? 7! 2335

 

 

 

 

 

 

 

Copyright by
PATRICIA MARIE WILSON
1983

THE EFFECTS OF DIRECTIONS TO CHANGE ANSWERS ON THE ANSWER
CHANGING BEHAVIOR OF COLLEGE STUDENTS

By

Patricia Marie Wilson

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Teacher Education

1983

ABSTRACT

THE EFFECTS OF DIRECTIONS TO CHANGE ANSWERS ON THE ANSWER
CHANGING BEHAVIOR OF COLLEGE STUDENTS

By

Patricia Marie Wilson

This study explored the effect of directions to change
answers on the answer changing behavior of college students on an
objective final examination for a religion course. Also explored
were the differences in the concomitant variables of sex of subjects,
Survey of Study Habits and Attitudes (SSHA) scores, and various
achievement and ability measures among treatment groups and effec-
tiveness of change groups. The relationship between the level of item
difficulty and the effectiveness of answer changing was also inves-

tigated.

Procedures

 

Subjects were 129 students who were enrolled in a liberal arts
college in the northeastern United States. Subjects were assigned at
random to one of three groups where they were given different sets of
directions regarding answer changing on the final examination: no
directions, directions to change with caution, and directions to
revise to improve scores. Data were analyzed using the analysis of

variance and chi-square procedures.

Patricia Marie Wilson

Results and Suggestions

Within the limitations of the methodology and p0pulation, the
major results of the study were:

1. Directions affected the degree of change effectiveness
on easy test items (p < .01).

2. The degree of change effectiveness was associated with
SSHA work methods scores on difficult items (p < .05).

3. Study attitudes and study orientation SSHA scores were
associated with the degree of change effectiveness on total items
(p < .05).

4. Grade point averages (p < .05) and final examination
scores (p < .01) were associated with the degree of change effective-
ness on difficult items.

5. SAT mathematics scores and SAT verbal scores were asso-
ciated with the degree of change effectiveness on difficult items
(p < .01).

Further investigation into the effects of directions to change
answers using different populations and content areas was suggested.
Also suggested was further investigation into the reasons students
have changed responses and the personality characteristics related to

nonchangers and countereffective changers.

To my family,
Dan, Kitty, Sarah, and Kent.

ii

ACKNOWLEDGMENTS

I wish to express my warmest appreciation to Dr. Lois A. Bader,
my chairperson and academic advisor. Her wisdom and support have con-
tributed immeasurably to the completion of this doctoral program. I
am also grateful to the members of my committee, Dr. James Costar,

Dr. Dick Featherstone, and Dr. Eugene Pernell, for their support and
flexibility.

I sincerely thank Dr. Lloyd Carr, who contributed his time toward
this project, and Dr. Robert 0055 and Chris Lamson, who provided excel—
lent dialogue and support throughout this study. Of great help in the
planning stages were Dr. Arvo Juola, Dr. Bill Mehrens, and Dr. Tyrus
Wessell. They contributed to my thinking and were accessible for
dialogue.

The support of my family through this project has been heart-
warming. A very special thanks to Mark and Sue Wilson and Byron and
Nancy King for entertaining nieces and nephew during summers of course-
work and writing. Most of all, my deepest thanks to my husband, Dan,
who supported and encouraged me to be a steward of the talents and

gifts I have been given.

iii

TABLE OF CONTENTS

Page

LIST OF TABLES ......................... vi
Chapter

I. THE PROBLEM ....................... l

Background of the Study ................ 1

Need for the Study .................. 3

Statement of Purpose ................. 5

Research Questions .................. 6

Generalizability ................... 6

Limitations ...................... 7

Definition of Terms .................. 7

Organization of Subsequent Chapters .......... 9

11. REVIEW OF RELATED LITERATURE .............. l0

Introduction ..................... lO

Beliefs About Answer Changing ............. lO

Subjects, Test Content, and Changes .......... l5

Test Directions .................... l9

Achievement and Aptitude ............... 23

Item and Test Characteristics ............. 26

Sex Differences .................... 28

Personality ...................... 28

Summary ........................ 32

III. METHODOLOGY ....................... 34

Introduction ..................... 34

Population ...................... 34

Procedures ...................... 35

Design ........................ 38

Instruments for Data Gathering ............ 38

Hypotheses ...................... 41

Data Analysis ..................... 45

Summary ........................ 45

iv

IV. PRESENTATION AND ANALYSIS OF DATA ............

V. SU

APPENDICES
A. LE

B. TE

Introduction .....................
Statistical Analysis .................
Directions and Number of Changes ..........
Directions and Effectiveness Of Changes .......
The SSHA and Effectiveness of Changes ........
Item Difficulty and Change Effectiveness ......
Sex Differences and Effective Changing .......
Achievement and Answer Changing ...........
Ability and Answer Changing .............
Summary ........................

MMARY, DISCUSSION, AND RECOMMENDATIONS ........

Summary ........................
Limitations ......................
Discussion ......................
Implications of the Study ...............
Recommendations for Future Research ..........

TTER TO STUDENTS WHO TOOK THE SSHA WITH THE
FINAL EXAMINATION ...................

ST DIRECTIONS TO SUBJECTS ...............

C. ITEM DIFFICULTY AND DISCRIMINATION DATA FOR THE

OBJECTIVE TEST ....................

D. GROUP EQUIVALENCY DATA .................

E. NUMBER OF CHANGES PER SUBJECT BY GROUP .........
F. TABLES 83 AND 84 ....................

REFERENCES

I41
T43

T44
146

ISO
T52
155
158
160

Table

10.
11.

12.
13.

14.
15.

LIST OF TABLES

Mean Number of Changes and Standard Deviations for
Each Group .......................
ANOVA Table for Number Of Changes ............

Mean Percentage Correct Scores and Standard Deviations

on Total Items .....................

ANOVA Table for Effectiveness of Changes on Total Items

Mean Percentage Correct Scores and Standard Deviations

on Easy Items .....................

ANOVA Table for Effectiveness of Changes on Easy Items . .

Mean Percentage Correct Scores and Standard Deviations

on Difficult Items ...................

ANOVA Table for Effectiveness of Changes on Difficult

Items .........................

Mean and Standard Deviation Scores for Delay Avoidance
Subtest for Degree of Change Effectiveness on Total

Items .........................

ANOVA Table for Delay Avoidance--Total ..........

Mean and Standard Deviation Scores for Delay Avoidance
Subtest for Degree of Change Effectiveness on Easy

Items .........................

ANOVA Table for Delay Avoidance--Easy ..........

Mean and Standard Deviation Scores for Delay Avoidance
Subtest for Degree of Change Effectiveness on

Difficult Items ....................
ANOVA Table for Delay Avoidance--Difficult ........

Mean and Standard Deviation Scores for Work Methods
Subtest for Degree of Change Effectiveness on

Total Items ......................

Page

47
48

49
49

5O
50

51

51

53
54

55
56

57
58

59

16.
17.

18.

19.

20.

21.

22.
23.

24.

25.

26.

27.

28.
29.

30.

31.

32.

33.

34.

ANOVA Table for Work Methods--Total ............

Mean and Standard Deviation Scores for Work Methods

Subtest for Degree of Change Effectiveness on Easy Items .

ANOVA Table for Work Methods-~Easy .............

Mean and Standard Deviation Scores for Work Methods
Subtest for Degree of Change Effectiveness on Difficult

Items ..........................

ANOVA Table for Work Methods--Difficult ..........

Mean and Standard Deviation Scores for Education Acceptance

Subtest for Degree of Change Effectiveness on Total Items .

ANOVA Table for Education Acceptance--Total ........

Mean and Standard Deviation Scores for Education Acceptance

Subtest for Degree of Change Effectiveness on Easy Items .

ANOVA Table for Education Acceptance--Easy .........

Mean and Standard Deviation Scores for Education Acceptance
Subtest for Degree of Change Effectiveness on Difficult

Items ..........................

ANOVA Table for Education Acceptance--Difficult ......

Mean and Standard Deviation Scores for Teacher Approval
Subtest for Degree of Change Effectiveness<n1Total Items

ANOVA Table for Teacher Approval--Total ..........

Mean and Standard Deviation Scores for Teacher Approval

Subtest for Degree of Change Effectiveness on Easy Items .

ANOVA Table for Teacher Approval--Easy ...........

Mean and Standard Deviation Scores for Teacher Approval
Subtest for Degree Of Change Effectiveness on

Difficult Items .....................
ANOVA Table for Teacher Approval--Difficult ........

Mean and Standard Deviation Scores for Study Habits Subtest
for Degree of Change Effectiveness on Total Items

ANOVA Table for Study Habits--Total ............

vii

Page

61
62

63

64

65
66

67

68

69

7O

71
72

73

74

75

76

77
78

35.

36.
37.

38.
39.

40.
41.

42.

43.

44.

45.

46.
47.

48.

49.

50.
51.
52.

53.

Mean and Standard Deviation Scores for Study Habits Subtest

for Degree of Change Effectiveness on Easy Items .....

ANOVA Table for Study Habits--Easy .............

Mean and Standard Deviation Scores for Study Habits Subtest
for Degree of Change Effectiveness on Difficult Items

ANOVA Table for Study Habits--Difficult ..........

Mean and Standard Deviation Scores for Study Attitudes
Subtest for Degree Of Change EffectiveneSSINTTotal Items

ANOVA Table for Study Attitudes--Total ...........

Mean and Standard Deviation Scores for Study Attitudes
Subtest for Degree of Change EffectivenesschEasy Items

ANOVA Table for Study Attitudes--Easy ...........

Mean and Standard Deviation Scores for Study Attitudes
Subtest for Degree Of Change Effectiveness on

Difficult Items .....................
ANOVA Table for Study Attitudes--Difficult .........

Mean and Standard Deviation Scores for Study Orientation
Subtest for Degree of Change Effectiveness on Total Items

ANOVA Table for Study Orientation--Total ..........

Mean and Standard Deviation Scores for Study Orientation
Subtest for Degree of Change EffectivenesscNTEasy Items

ANOVA Table for Study Orientation--Easy ..........

Mean and Standard Deviation Scores for Study Orientation
Subtest for Degree of Change Effectiveness on

Difficult Items .....................
ANOVA Table for Study 0rientation--Difficult ........

Degree of Change Effectiveness on Easy and Difficult Items .

Percentage of Correct Changes on Total Items for Males

and Females .......................

ANOVA Table for Percentage of Correct Changes on Total

Items for Males and Females ...............

viii

Page

79
80

81
82

83
84

85

86

87

88

89
9O

91

92

93
94
95

96

97

54.

55.

56.

57.

58.

59.
60.

61.
62.

63.
64.

65.
66.

67.
68.

69.
70.

71.
72.

Mean Percentage of Correct Changes on Easy Items for

Males and Females ....................

ANOVA Table for Percentage of Correct Changes on Easy

Items for Males and Females ...............

Mean Percentage of Correct Changes and Standard Deviations
on Difficult Items for Males and Females by Treatment

Group ..........................

ANOVA Table for Percentage of Correct Changes on

Difficult Items for Males and Females ..........

Means and Standard Deviations for Objective Test Scores
for the Degree of Change Effectiveness on Total Items

ANOVA Table for Objective Final--Total ...........

Means and Standard Deviations for Objective Test Scores

for the Degree of Change Effectiveness on Easy Items . . .

ANOVA Table for Objective Final Score--Easy ........

Means and Standard Deviations for Objective Test Scores

for the Degree of Change Effectiveness on DifficultItems.

ANOVA Table for Objective Final--Difficult .........

Mean Student Grade Point Averages and Standard Deviations
for Degree Of Change Effectiveness on Total Items

ANOVA Table for Grade Point Average--Total .........

Mean Grade Point Averages and Standard Deviations for

Degree of Change Effectiveness on Easy Items .......

ANOVA Table for Grade Point Average--Easy .........

Mean Grade Point Averages and Standard Deviations for
Degree of Change Effectiveness on Difficult Items

ANOVA Table for Grade Point Average--Difficult .......

Means and Standard Deviations for SAT Verbal Subtest

Scores for Degree of Change Effectiveness on Total Items .

ANOVA Table for Mean SAT Verbal--Total ...........

Mean SAT Verbal Scores and Standard Deviations for Degree

of Change Effectiveness on Easy Items ..........

ix

Page

98

98

99

99

101
102

103
104

105
106

107
108

109
110

111
112

73.
74.

75.
76.

77.
78.

79.
80.

81.
82.
83.

84.

ANOVA Table for SAT Verba1--Easy ..............

Mean SAT Verbal Scores and Standard Deviations for Degree
of Change Effectiveness on Difficult Items ........

ANOVA Table for SAT Verbal--Difficult ...........

SAT Mathematics Mean and Standard Deviation Scores for
Degree of Change Effectiveness on Total Items ......

ANOVA Table for SAT Mathematics--Tota1 ...........

SAT Mathematics Mean and Standard Deviation Scores for
Degree of Change Effectiveness on Easy Items .......

ANOVA Table for SAT Mathematics--Easy ...........

SAT Mathematics Mean and Standard Deviation Scores for
Degree of Change Effectiveness on Difficult Items

ANOVA Table for SAT Mathematics--Difficult .........
Summary of Research Findings for Null Hypotheses ......

Number of Students Who Gained Points, Lost Points, or
Were Not Affected by Changes, by Group ..........

Student Means and Standard Deviations for Positive and
Negative Changes, by Group ................

Page
116

117

123
124
129

159

159

CHAPTER I

THE PROBLEM

Background Of the Study,

 

Tests are used widely throughout schools to measure student
achievement, to revise curriculum, and to make educational decisions.
Millman and Pauk (1969) noted that student test-taking behavior may
affect scores and that some students lack sophistication in taking
tests. Some students do not make use of the characteristics of the
test or the testing situation to improve their scores. They are not
test-wise. Millman, Bishop, and Ebel (1965) stated that "it would
be desirable to seek ways to reduce differences in test-wiseness among
examinees in order to provide more valid estimates Of their actual
abilities and achievement levels" (p. 724).

According to the taxonomy of test-wiseness proposed by Millman
et a1. (1965), one aspect of test-wise behavior is going over the
questions on a test after one has finished going through the test
once and revising the answers that now appear to be incorrect. The
effects of such answer revision on objective test scores have been
investigated for more than 50 years (Mathews, 1929; McMorris,
Lichtstein, & Hoops, 1983). The results have consistently been that
students who go back and revise test items tend to raise their scores
more Often than they lower them (Mueller & Wasser, 1977). Students,
however, report that they believe that changing answers will lower

1

their scores (Foote & Belinky, 1972; Mathews, 1929; Sutton, 1982).
Contrary to their stated beliefs, approximately 61 percent tO

100 percent of the students revise answers anyway (Mercer, 1978).

But not all students revise answers (Staffer, Davis, 8 Brown, 1977),
and not all students profit from revision. Across the reported
studies, the ratio of students who profit from changes to students who
lose from changes ranges from 1.4:1 to 48:1 (Hill, 1937; McMorris et
a1., 1983). Additionally, Ballance (1977) found that only 16.5 per-
cent Of his subjects were able to predict their own answer changing
behavior.

Study-skills texts and measurement and evaluation texts give
mixed advice. Some study-skills texts state that students should stay
with their first response (Armstrong, 1975), and others recommend
checking answers and revising (Pauk, 1974). Some measurement texts
are silent on the matter (Gronlund, 1967; Stanley & Hopkins, 1972),
and others encourage teachers "to de-emphasize the conventional caution
'the first answer is invariably the best answer'" (Mehrens & Lehmann,
1978, p. 314), and "those who have studied it suggest that the best
strategy is to change one's response if after careful deliberation a
better answer or an answer more likely to be correct can be identified"
(Hills, 1981, p. 128).

In summary, going back over test responses is an aspect of test-
wise behavior. There is mixed advice from the literature in regard to
changing responses, and a plurality of students believe that changing

answers will tend to lower their scores.

Need for the Study_

 

The conflict between the research findings and student opinion
regarding answer changing is not easy to explain. Smith, White, and
Coop (1979) posited one theory to explain the discrepancy between
research findings and student-reported beliefs. These researchers
posited that there might be a relationship between cognitive develop-
ment and answer revision. Whether or not a student chooses to go back
and revise answers might be based on his ability to reverse his logi-
cal operations. Because children begin test taking before they are
capable of reversible thinking (Piaget, 1972), the response-set in
the child might be to stay with his first-choice answer. This response-
set might continue to operate.

Reinforcement from negative feedback the student gets when
he Observes a right answer changed to wrong might encourage the
response-set to stay with the first response. Juola (1968) noted
that of particular frustration to students are the correct answers
changed to incorrect answers. The student remembers that he has
lowered his score. He may not even attend to those answers that
raised his score. Thus, the tendency from childhood not to reverse
is reinforced by the occasional answer revised to a wrong response.

A second theory regarding the discrepancy between student‘
reported beliefs and actual patterns of change was reported by
McMorris and Leonard (1976). These investigators pointed out that as
a test-taker goes through an examination his memory may be jogged,
and he is more able to reason the correct answer. This viewpoint,

they believed, is consistent with the theoretical work of Jenkins

(1974), which suggested that remembering is an active process of
elaboration in which stored information is synthesized to form a new
construct. That process takes time. The reevaluation of questions on
an examination could allow time to pass and thus lead to a more effec-
tive reconstruction Of the concept required to discern the correct
option. But because the reevaluation interrupts the original memory
trace, the test-taker may be discouraged from altering his initial
response. He does not know which memory to believe, and this conflict
may result in avoiding a revision or at least a feeling that his revi-
sion might be incorrect.

Recent research has centered on influencing students to change
answers and their subsequent answer changing patterns. There have
been mixed findings in the studies. The "influences" have varied from
reporting to students the findings of the research to rather extensive
treatment to alter beliefs (Foote & Belinky, 1972; McMorris et a1.,
1983; Sutton, 1982).

This study explored the effects of three different sets of
directions to change answers on students' answer-changing patterns
while considering the level of difficulty of the test, the sex dif-
ference of subjects, and the achievement and ability of the subjects.
Mueller and Wasser (1977) noted that we cannot predict with accuracy
which students will gain or lose from changing answers. Thus, giving
correct test-wiseness information is not entirely possible from exist-
ing research. While some studies have addressed the relationship
between sex differences, item difficulty, ability and achievement,

and personality and answer changing patterns, no study to date has

investigated the effect Of answer changing behavior on self-reported
study habits and attitudes. Can a student self-report Of study habits
and attitudes be used to indicate which students might profit or lose

from changing answers?

Statement Of Purpose

 

The purpose of this study was to explore the effects Of
directions to change responses on Objective tests and answer changing
patterns and to explore the answer changing patterns associated with
the Survey Of Study Habits and Attitudes (SSHA) (The Psychological
Corporation, 1965). Specifically, the investigator

l. Examined the effect of no specific directions, the effect
of directions to revise with caution, and the effect of directions to
revise because it will improve scores, on the answer changing patterns
of college students who were taking a comprehensive objective final
examination for a religion class in which they were all enrolled.

2. Examined the differences in the self-reported study habits
and attitudes scores among the direction groups and groups formed by
the effectiveness of changing answers.

3. Examined the different patterns of change between the
sex Of subjects among direction groups and the effectiveness of
change groups.

4. Examined the relationship between the level of item dif-
ficulty and effective answer changing.

5. Examined the differences in ability and achievement of

subjects among direction groups and effectiveness groups.

Research Questions

The major research questions explored were:

1. What is the effect of directions on the number of answers
changed?

2. What is the effect Of the directions on the percentage
of correct changes?

3. Are there differences in SSHA scores among the direction
groups (treatment) and the effectiveness groups (countereffective,
ineffective, and effective changers)?

4. Is there a difference between easy and difficult item
changes and the effectiveness of the change?

5. Are there any differences in the percentage of correct
changes among the direction groups and the groups formed by sex of
subjects?

6. Are there any differences in the achievement measures
among the direction groups and the effectiveness groups?

7. Are there any differences in the ability measures among

the direction groups and the effectiveness groups?

Generalizability

 

The population was one group of 129 students who were enrolled
in a required freshman-level religion class in a private four-year
liberal arts college north of Boston, Massachusetts, in the fall of
1982. The subjects were primarily of the Caucasian race and were
primarily from middle-class homes. The mean SAT score for the popula-
tion was 485 and the mean SSHA score was 101, which is approximately

the 30th percentile. There were 51 males and 78 females in the

population. Most of the subjects were in the first quarter Of their
freshman year. The course test used to collect data was constructed
by the course instructor. It may be concluded that the findings in
this study can be generalized to populations in other colleges with

similar enrollments.

Limitations

 

The limitations of this study were the level of difficulty of
the course test, the length of the course test, and the subsequent
willingness of the students to revise answers when they felt pressed
for time. Also, the willingness of the students to read the direc-
tions limited the findings. An additional limitation was the self-

reporting format of the SSHA.

Definition of Terms

 

The reader will better understand this study if certain terms
are clarified.

Answer changing or revision: Any observed blot out, erasure,

 

or crossed-out answer.
Answer changing behavior or pattern: The number, rate, or
frequency of changes and/or the profit or loss from the change.

Directions to change: The printed directions to subjects

 

regarding changing or revising answers.

Direction of change: Profit or loss from a change.

 

Net-gain changes: The gain from changes, which is calculated

 

by subtracting the right-tO-wrong changes from the wrong-to-right

changes.

Degree of change effectiVeness: A categorization of change
effectiveness based on a percentage of correct changes; 0-39 percent =
counter-effective, 40-59 percent = ineffective, 60-100 percent =
effective.

-R: Wrong-tO-right changes.

R- : Right-tO-wrong changes.

W- : Wrong-to-wrong changes.

Item difficulty: The percentage of the total group passing

 

the item.

Difficult item: Those items where 30 percent or more of the

 

students missed the item. The mean item difficulty was 29.
Easy item: Those items where 29 percent or fewer of the stu-
dents missed the item.

Total items: The total number of items changed as opposed to

 

easy items changed and difficult items changed.

Index of discrimination: Calculated by subtracting the per-

 

centage of the upper group marking the correct answer and the per-
centage of the lower group marking the correct answer.

Discriminating item: Items whose index of discrimination is

 

23 percent or above. The mean item discrimination was 23.
Test-wise: The ability to use the format or characteristics
of a test or testing situation to receive a higher score.

Effective changing: Changing 60-100 percent of answers from

 

wrong to right.

Ineffective changing: Changing 40-59 percent of answers from

 

wrong to right.

Countereffective changing: Changing 60-100 percent of answers
from right to wrong.

Penalty-for-guessing scoring_directions: Wrong answers are

 

subtracted from right answers; also called correction-for-guessing

scoring directions.

Organization of Subsequent Chapters
The content of Chapter I included background of the study,

need for the study, the research questions, the generalizability and
limitations of the study, the definition of terms, and organization of
subsequent chapters. In Chapter II the research and literature related
to the study are reviewed. A description of the design and methodology
used in the study is contained in Chapter III. In Chapter IV, the data
are presented and analyzed. In Chapter V, a summary of the study, dis-
cussion of results, and appropriate conclusions and recommendations for

future research are presented.

CHAPTER II
REVIEW OF RELATED LITERATURE

Introduction

The purpose of this study was to investigate the effects of
directions to change answers on the answer changing behavior of college
students. Also explored were the differences in the concomitant vari-
ables Of sex of subjects, Survey of Study Habits and Attitudes (SSHA)
scores, and various achievement and ability measures among treatment
groups and effectiveness of change groups. The relationship between
level of item difficulty and the effectiveness of answer changing
was also investigated.

The review of the literature is organized under the following
headings: (l) Beliefs About Answer Changing; (2) Subjects, Test
Content, and Changes; (3) Test Directions; (4) Achievement and Apti-
tude, (5) Item Characteristics, (6) Sex Differences, and (7) Per-

sonality.

Beliefs About Answer Changing,

 

Beliefs concerning the advisability of changing answers on
objective tests have been reported in empirical research, in textbooks
on measurement and evaluation, in journal articles, and in study-skills

textbooks. The empirical research is reviewed first.

10

11

In a survey, Mathews (1929) found that 86 percent of his
respondents felt that revising answers would lower their scores.
Eleven percent felt changing answers would raise scores, and 13 per-
cent felt that changes would cancel each other out. In an informal
show Of hands, Lynch and Smith (1975) found that 75 percent of the
300 subjects believed that changes would lower their scores. Jacobs
(1972) found 30 percent of his subjects believed changes would result
in a gain, 45 percent said that changes would produce a loss, and
25 percent said they did not know. Jacobs found no relationship
between the subjects' reported beliefs and mean net gains from their
answer changing.

Foote and Belinky (1972) assessed student beliefs concerning
answer changing before attempting to influence them to revise answers.
Ninety-three percent Of their subjects responded that they did check
over their answers after tests, and 99 percent said they had heard
first-impression answers were more apt to be correct. Thirty-two per-
cent believed revisions would improve scores, and 64 percent believed
revisions would lower scores or make no difference. After examining
their own answer changing on a returned examination, 72 percent of
the students responded that they were surprised at the results. How-
ever, only 48 percent Of the students said they would change answers
more Often.

Mueller and Shwedel (1975) found no one in their sample who
believed their answers would be improved with revisions. Stoffer,
Davis, and Brown (1977) found that 80 percent Of their subjects

believed changes would lower scores. Smith, White, and 600p (1979)

12

found only 7 percent of their sample who believed that answer revision
would increase scores. Eighty-six percent Of their sample, however,
did revise one or more answers. Similar beliefs and revision pat-
terns were noted by Foote and Belinky (1972), Jacobs (1972), and
Mueller and Shwedel (1975).

Ballance (1977) studied student expectations concerning answer
changing and student motivation to change answers. Findings were that
student expectations regarding the outcomes of their revisions were not
related to the actual outcomes. Only 16.5 percent of the students
were able to predict their own behavior. Also found was that 86 per-
cent Of the students would change answers because of reassessment of
the question. McMorris (1981), Sutton (1982), and McMorris, Lichtstein,
and Hoops (1983) also found no relationship between self-reported stu-
dent beliefs and answer changing practices. McMorris (1981) found a
mean of 3.5 on a seven—point scale that attempted to measure the
advisability of changing answers. Students reported they felt that
changing responses was "somewhat advisable.“ This finding was
replicated by McMorris et a1. (1983). McMorris (1981) also inves-
tigated the reasons for which students would change responses.

From 65 to 84 percent of the students reported they would change
responses if they needed to correct clerical errors, if they better
understood the question, and if they had rethought the question. The
relationship between these student reports was compared to student
gains. There was a tendency for those with a more positive atti-
tude toward changing to gain more, but the gains were not sig-

nificant. McMorris et a1. (1983) did find attitudes toward

13

changing answers on the course test (but not on other tests) signifi-
cantly related to the percentage of changes and gains. Their subjects
were instructed on the benefits of change. The strongest predictor
of change, however, was the number of answers changed.

Sutton (1982) collected data on student beliefs concerning
answer changing, established control and experimental groups, and
attempted to alter the subjects' beliefs concerning answer changing.
Results were that posttest measures were significantly higher for the
experimental group for beliefs concerning the value of answer chang-
ing, the frequency of answer changing, and the net gains from answer
changing. Sutton also found that students profited more from revi-
sions due to clerical errors than from changes of items that had
been rethought.

Sutton surveyed 11 different current measurement texts to
assess their comments concerning answer changing. He found that
Mehrens and Lehmann (1978) addressed the issue of answer changing in
a chapter on assembling and analyzing classroom tests. Mehrens and
Lehmann pointed out that the empirical evidence supports the view
that changing answers most often benefits the test-taker. Ebel
(1972) also stated that students should review answers and correct
any errors. Sutton did not locate other texts that included informa-
tion about answer changing.

The present investigator perused five current texts and
found one, in addition to Mehrens and Lehmann (1978) and Ebel (1972),
that advised students to gO back over and catch careless errors and

change responses after careful deliberation and also reported the

14

research evidence concerning answer changing. This information was
given in a section of the text devoted to developing test-taking
skills (Hill, 1981).

Millman, BishOp, and Ebel (1965) included answer-changing
behavior in their taxonomy of test-wiseness. These authors recom-
mended that students use time remaining after completion of the test
to reconsider answers. "Change answers if it seems desirable. Exam-
inees generally increase their scores if they do" (p. 714). Further,
Millman et a1. stated that it is desirable to reduce the differences
in test-wiseness among examinees in order to provide more solid esti-
mates of actual abilities and knowledge. "How can we change test
items and test directions, or other conditions of test-administration,
to minimize the harmful effects of differences in test-wiseness?"

(p. 724). However, in a 1979 review of literature on test-wiseness,
Sarnacki did not include a review of answer-changing research.

Finally, current study-skills texts were reviewed. Lynch
and Smith (1975) reviewed several texts that instruct students to
stay with their first answers. This investigator surveyed some later
editions of study-skills texts. Armstrong (1975) instructed students
not to change answers because one's first response is more reliable.
Maiorana (1980) followed the same general thought and advised students
to avoid impulse changes. Staton (1982) advised, "DO not change your
answer because of a mere doubt" (p. 69).

Shepherd (1982) recommended that students exercise care in
changing responses because "some students have a tendency to change

correct answers to incorrect answers" (p. 18). He further recommended

15

that students take note of their own behavior regarding changing
responses until they establish their own profit-or-loss patterns.
Gilbert (1982) stated that students have only three chances in ten
to change from the incorrect answer to the correct answer. Both
Pauk (1974) and Robinson (1970), however, recommended that students
change answers if they believe that another answer is correct.

In summary, the empirical research concerning answer changing
suggested that the majority of students report that they believe answer
changing will decrease their scores, but they revise their answers any-
way. Mixed results have been found when researchers have attempted to
influence answer changing behavior. Many current texts on measurement
and evaluation do not include sections on developing test-taking skills
and do not include information regarding answer-changing behavior.
Major reviewers of research on test-wiseness have not agreed on the
inclusion Of answer changing in their works. Finally, study-skills

texts give mixed advice concerning changing answers.

Subjects, Test Content, and Changes

Beginning with Lehman (1928) and Mathews (1929), the majority
Of studies concerning answer changing have been conducted with sub-
jects who were college undergraduates or graduates. Further, most
of the studies were conducted with college students who were enrolled
in education or psychology courses in which objective tests constructed
by the professors were used to collect data on answer changing patterns
(Archer&Pippert, 1962; Ballance, 1977; Bath, 1967; Berrien, 1939;
Clark, 1962; Foote & Belinky, 1972; Greene, 1981; Hill, 1937; Jacobs,

16

1972; Jarrett, 1947; Lamson, 1935; Lowe & Crawford, 1929; Lynch &
Smith, 1975; McMorris et a1., 1983; McMorris & Leonard, 1976;
Pascale, 1974; Penfield & Mercer, 1980; Range, Anderson, & Wesley,
1982; Reile & Briggs, 1952; Smith et a1., 1979; Sitton, Adams,
& Anderson, 1980; Sutton, 1982). Vidler and Hansen (1980) used
college psychology students as their subjects, but they used the
Watson-Glasser Critical Thinking Test as their data-gathering
instrument. Stoffer et a1. (1977) examined course examinations
Of college psychology students and course examinations of Air Force
personnel who were enrolled in a technical-vocational skills course.

A few researchers have studied answer changing behavior of
college students who were not enrolled in education or psychology
courses. Mallinson and Miller (1956) studied answer changing behavior
on the Objective examinations of students enrolled in liberal arts
courses. They collected data from 1,029 college students on tests
from courses in biology, economics, political science, and chemistry,
in addition to psychology and education. Copeland (1972) examined
the final examinations of chemistry students. Reiling and Taylor
(1972) examined economics tests, and Ballance and Gentzel (1980)
looked at junior-college students' business-principles examinations.
Davis (1975) examined tests for microbiology, pharmacology, physi-
ology, and pathology courses that were taught in a medical school.
Ballance (1975) examined the tests of 144 first-year professional
students from the allied-health-care field.

Two studies examined the General Education Development Tests

for answer changes (Smith, 1975; Smith 8 Moore, 1976). This

17

test contains subtests for English usage, social studies, natural
science, literature, and mathematics. These studies extended the
research beyond the college-age population to a noncollege adult
population.

The research was extended to the middle- and elementary-school
population in the late 19705. Crocker and Benson (1980) examined the
mathematics-computation subtest of the Metropolitan Achievement Test.
His subjects were in grade seven. Using subjects from grades seven
and eight, Cummings (1981) examined the Nelson Reading Skills Test
for answer changing. In New Zealand, Jackson (1978) examined the
Progressive Achievement Tests: Study Skills for response alterations
of 10-, 12-, and 14-year-old subjects. Four subtests of the Iowa
Test of Basic Skills for pupils in grades four, five, six, and seven
were examined by Mercer (1978). This study involved the first use of
a predominantly nonwhite population. Subjects in this study were
also grade-level and non-grade-level examinees who were administered
the test on grade level or below grade level, according to teacher
judgment.

Payne (1983) used science-course tests of subjects in grade
eight. Her sample included nonwhite subjects. Both Mercer (1978)
and Payne (1983) drew their samples from urban schools. Beck (1978)
extended the research to subjects in grade three who took the Metro-
politan Reading Test. In most of these studies, the format of the
objective tests was four-Option multiple-choice items. However,

Berrien (1939), Hill (1937), Jarrett (1948), Lamson (1935), Lehman

18

(1928), Lowe and Crawford (1929), Lynch and Smith (1975), Mathews
(1929), and Mueller and Shwedel (1975) examined changes on true-false-
items.

Whether standardized Or teacher-made examinations; true-false
or multiple-choice; college, adult, middle-school, or elementary-
school subjects, the investigators all found that when the changes
were tallied more answers had been changed from wrong-to-right than
from right-toewrong (Mueller & Wasser, 1977). Mercer (1978) reported
a range of 40 percent to 73 percent of answers changed from wrong to
right on studies conducted through 1978. This range has been repli-
cated in studies to the present (Ballance, 1975, 1977; Beck, 1978;
Best, 1979; Crocker 8 Benson, 1980; Cummings, 1981; Greene, 1981;
Jackson, 1978; McMorris, 1981; McMorris et a1., 1983; Mercer, 1978;
Payne, 1983; Penfield & Mercer, 1980; Range et a1., 1982; Smith et a1.,
1979; Stoffer et a1., 1977; Sitton et a1., 1980; Sutton, 1982; Vidler &
Hanson, 1980).

In a review of eight studies, Mueller and Wasser (1977)
reported a median gain-tO-loss ratio of 2.7:1 items. Mercer (1978),
summarizing the data from 23 studies, reported a range of .54 to 23.4
mean net gains for subjects. McMorris et a1. (1983) reported a student
gain-to-loss ratio of 48:1. The range of student gains appeared to
go from this high down to a student gain-to-loss ratio Of 1.4:1
reported by Hill (1937).

In summary, the majority of the studies concerning answer
changing have been conducted using subjects who are college students

enrolled in education or psychology classes. A few studies have

19

investigated answer changing on examinations from science courses and
liberal arts courses. Recently, the research has been extended to
younger students of middle school and elementary age. These studies
have used standardized tests as the data-gathering instruments. The
research has also been extended to the adult non-college-student pOpu-
lation. In all studies, results have indicated that more changes are

from wrong-to-right than from right-to-wrong.

Test Directions

 

In most of the studies on answer changing, the tests were
administered without specific directions regarding answer changing,
and the subjects were not aware that an experiment was being conducted.
The researchers collected the tests of the students and examined them
for erasures or crossed-out answers and then tallied the data into
categories of wrong-to-right changes (W-R), right-to-wrong changes
(R-W), and wrong-to-wrong changes (W-W). There were, however, several
studies in which special answer sheets or special directions regard-
ing changes were given.

Mathews (1929), who collected data from a true-false,
multiple-choice, objective examination, told students that changes
were more apt to be correct than incorrect. He also included penalty-
for-guessing scoring directions (wrong answers subtracted from right
answers) on the true-false section of the test. 0n the multiple-choice
section, 2.5 percent of the answers were changed, and 3.16 percent of
the answers on the true-false section were changed. Almost 60 percent

of the true—false items were changed to correct, whereas only 50

20

percent of the multiple-choice items were changed to correct. True-
false questions had 56 (9.8 percent) of the answers omitted, whereas
there were 5 (.9 percent) Omitted multiple-choice answers. Mathews
concluded that the higher true-false omission rate influenced the more
positive change rate for true—false items.

Hill (1937) and Ballance and Gentzel (1980) also included a
penalty-for-guessing scoring direction. However, they did not inform
their students Of a benefit for answer changers. Hill (1937) reported
a lower change rate and score improvement for changing than previous
researchers. There was a 1.43:1 ratio of correct to incorrect changes.
However, 2.3 percent of the answers to questions were omitted.
Ballance and Gentzel (1980) studied answer changing under three dif-
ferent sets of scoring directions. The NO Penalty group (N = 29)
was told that scoring would be based on the number of questions they
answered correctly. The Correction for Guessing group (N = 36) was
told that scoring would be based on the number correct minus one point
for each wrong response. The Severe Penalty group (N = 40) was told
that scoring would be based on the number right minus the number
wrong. There was a significant difference in the mean changes from
treatment to treatment. There was, however, no benefit for one group
over another in terms of profits from changes. The authors con-
cluded that correction-for-guessing directions appear to contribute to
"a testing atmosphere in which better decisions are made about answer

changing" (p. 654).

21

A number of researchers have included some form of directions
to change answers in their studies. Lehman (1928) studied the answer-
changing behavior of students by using a two-column answer sheet.
Lynch and Smith (1975) provided a similar sheet. Lowe and Crawford
(1929) developed a rotation experiment to assess the differences
between two groups Of subjects who took two parts of a test using two
different procedures; both groups used one procedure on the first
part Of the test and then switched procedures for the second section
of the test. One procedure involved reading through the test first
and waiting to mark items until the second reading. The other
procedure involved reading and marking choices and then reading and
revising. Significant gains were noted for those who read and marked
and then revised. In addition, 10 percent of the test items were
revised under the directions to read and mark and then revise.

Jacobs (1972) conducted an experiment to attempt to measure
changes from first impressions to final answer. He projected the
questions via slides for 30 seconds. Subjects were to record their
answers quickly and did not know they would have an Opportunity to
revise their work. Then subjects were given a mimeographed version
Of the projected questions and allowed to revise using a different-
colored pencil. The revision rate was 32 percent. The mean number
of changes was 14.7, and the ratio of points gained to points lost
was 2.8. The results of this study are limited by the unique data-
gathering technique of first-impression answers.

Foote and Belinky (1972) gave information and feedback con-

cerning answer changing to two classes of psychology students who had

22

already taken two course examinations on which their answer changing
patterns had been analyzed. Results showed that subjects' answer
changing patterns on the last two course examinations were not
significantly different from the patterns on the first two examina-
tions in regard to either the rate or effectiveness of changing.
On a questionnaire, only 48 percent of these students had indicated
that their future answer changing rate would increase.

Based on the no-change findings of Foote and Belinky,
Sutton (1982) attempted to influence student answer changing by
influencing and changing student beliefs. Subjects were 184 students
in a teacher-education program conducted by the Air Force. The
subjects were two intact groups that attended classes during the
first and then second five-week time period in a ten-week session.
Using small-group instruction, the experimental group was informed
about the benefits of answer changing and given individual feedback
concerning their own changes on tests. Pretreatment equality was
established between groups on beliefs concerning answer changing,
frequency of changing answers, and gains from answer changing.
Results at the end of the treatment showed significant differences
between the control and experimental groups for beliefs, rate of
changes, and effectiveness of changes. The mean net gain score for
the control group was 1.61, and for the experimental group, 2.3.

Using a sample that had been informed of the benefits of
answer changing, McMorris et a1. (1983) investigated the relationship
between the attitude toward and knowledge concerning answer changing

and the patterns Of answer changing. Results indicated that students

23

who were more positive toward answer changing and to the course and
its test were more likely to benefit from changes. The percentage
gain was a mean of 3.1, and the student gain-tO-loss ratio was 48:1.
Still, the most significant predictor of gain was the number of
changes made. This study was not designed to measure the effect of
influence to change answers on answer changing. The results were
limited by the design.

In summary, a few studies included directions to change
answers in their designs. Some of these studies also included a
penalty-for-guessing scoring instruction. Under penalty-for-guessing
scoring instructions, fewer answers were revised and more answers were
omitted. There were inconsistent findings in studies where there was
an attempt to influence student answer changing through instructions
to change answers or feedback concerning student answer changing.
Some of these studies found a higher rate of change, some found no

difference in the rate of change, and some found profits from changes.

Achievement and Aptitude

 

Several studies have investigated whether high-scoring
subjects or low-scoring subjects make more or fewer answer revisions
and whether those revisions tend to result in a profit or loss for
them. In addition, a few studies have investigated the relationship
between subject grade-point averages and answer changing, subject
course grade and answer changing, and subject ability and answer

changing.

24

In a number of studies, high-scoring students made more gains
from changing answers than did lower-scoring students (Archer &
Pippert, 1962; Ballance, 1975; Bath, 1967; Berrien, 1939; Best, 1979;
Clark, 1962; Copeland, 1972; Crocker & Benson, 1980; Davis, 1978;
Jarrett, 1947; Lehman, 1928; Lynch & Smith, 1975; Mallinson & Miller,
1956; Mathews, 1929; McMorris, 1981; Payne, 1983; Reile & Briggs,
1952; Reiling & Taylor, 1972; Sitton et a1., 1980; Smith, 1975, 1977;
Stoffer et a1., 1977). It might be noted that in most of these
studies the total test score was used to investigate the net gains,
and this score included the net gains. Since answer changing tends
to raise the total test scores, there is a reciprocal causal effect
(Mueller & Wasser, 1977).

Pascale (1974), Penfield and Mercer (1980), and Smith et a1.
(1979) did not find gains for high-scoring subjects. Penfield and
Mercer (1980) found that high-scoring students changed from a wrong
alternative to another wrong alternative more often. Mallinson and
Miller (1956) noted a similar finding, although their high-scoring
subjects maintained top position for improvement. Smith et a1. (1979)
found that lower and average students profited more from change.

Archer and Pippert (1962), Lehman (1928), Lynch and Smith (1975),
and Mathews (1929) found that lower-scoring students had the highest
rate of changes. There was, however, an inverse relationship between
the number of changes and the gains. Johnson (1978) also noted a
higher rate of changes among lower-scoring students.

McMorris et a1. (1983), Mueller and Shwedel (1975), and

Reiling and Taylor (1972) reported a positive relationship between

25

the number of answers changed and the gains from changes. McMorris
et a1. noted that the most highly significant predictor of gain was
the number of answers that were changed (r_= .77)

Several authors reported that high-scoring students made
fewer changes than low-scoring students (Best, 1979; Johnston, 1978;
Lehman, 1928; Lynch & Smith, 1975; Mathews, 1929; Mueller & Shwedel,
1975; Reile & Briggs, 1952; Sitton et a1., 1980; Smith & Moore, 1976).
Bath (1967), Berrien (1939), and Mercer (1978) found no differences
between high- and low-scoring subjects. In all these studies, there
were inconsistencies across groups.

Jacobs (1972) found no relationship between verbal intelli-
gence, as measured by the Quick Word Test, Level II and net gains.
Stoffer et a1. (1977) found no relationship between net gains and the
verbal section of the Scholastic Aptitude Test for their college group
and no relationship between the Airman Qualifying Examination and net
gains for their technical-skills sample. McMorris and Leonard (1976)
found no relationship between student-reported grade-point averages
and net gains. This finding was replicated by Ballance (1975), and
McMorris et a1. (1983).

There have not been consistent findings in regard to the rela-
tionship of course grades and profits from changes. Reiling and
Taylor (1972) found no relationship between the final course grade
and answer changing, although the poorer students were slightly
better answer changers. Bath (1967) and Reile and Briggs (1952)
found the top and bottom quarters of subjects to profit more from

changes. Sitton et a1. (1980) found that there was a relationship

26

between A and C students and successful answer changing. Range et a1.
(1982) noted that B students made the greatest gains from answer
changing. ’

In summary, there have been inconsistent results concerning
the relationship between answer changing behavior and measures of
achievement and ability. Several studies have reported a relation-
ship between high scoring students and effective answer changing,
whereas others have reported a relationship between low scoring stu-
dents and high change rates. Studies to date have not reported a
significant relationship between answer changing behavior and grade
point averages. Additionally, the SAT scores have not been found to

be related to answer changing patterns.

Item and Test Characteristics

 

Studies concerning item difficulty have had inconsistent
findings. The majority of the studies found that difficult items
were changed more often than easier ones (Beck, 1978; Greene, 1981;
Jackson, 1978; Lynch 8 Smith, 1975; McMorris et a1., 1983; Vidler &
Hansen, 1980). Studies by Best (1979) and Jacobs (1972) found that
items of moderate difficulty and easy items were changed most often.
Ballance (1975) found that there were no differences in the rate of or
profit from changes of difficult items. Jacobs (1972) and Vidler
and Hansen (1980) found that difficult items were changed with less
success.

Reiling and Taylor (1972) and Smith et a1. (1979) investigated

the patterns of changes on higher cognitive questions and lower

27

cognitive questions. Findings were that higher cognitive questions
were changed more Often, but there was no relationship with net
gains or losses.

Two studies investigated the position of items in the test
and the frequency Of changes. The findings were conflicting.

Jackson (1978), studying the behavior of late-elementary-school
children, found that more items were changed at the end of the test
and at the beginning of sections within the subtest. Jackson's data-
gathering instrument was a standardized achievement test. Reile and
Briggs (1952), studying college students, found fewer changes were
made on items near the end of the test.

Crocker and Benson (1980) investigated test reliability and
item changing. They found the internal consistency Of the test was
virtually unaffected by examinees' response changes. They noted a
slight decrease in the standard error of measurement, but they felt
the decrease was too small to have practical significance. Cummings
(1981) replicated these findings. He noted that item-discrimination
power was reduced by changes, whereas Crocker and Benson (1980) found
that item-discrimination power was increased by changes.

In summary, there have been inconsistent findings regarding
item and test characteristics. Some studies have reported significant
gains and rate changes for easy items, and other studies have reported
significant gains and rate changes for difficult items. One study
reported more changes on beginning test items, and another reported
more changes on ending test items. In addition, there have been

inconsistent findings concerning item discrimination. Finally, two

28

studies that investigated test characteristics reported that the
internal consistency of the test was not changed due to answer revi—

sions.

Sex Differences

 

Authors who investigated sex differences reported inconsis-
tent findings. Bath (1967), Copeland (1972), and Stoffer et a1.
(1977) all found that females benefited more from changes than did
males. Copeland found that males made more changes, whereas Reile
and Briggs (1952) had found that females made more changes but the
males had more benefits from their fewer changes. Mueller and Shwedel
(1975) found that males made more changes, but they were right-to-
wrong changes. Mercer (1978) found that seventh- and eighth-grade
males made significantly more changes on the mathematics subtest
of the Iowa Test of Basic Skills. Pascale (1974) and Reiling and
Taylor (1972) found that males benefited from changes more Often
than did females. Beck (1978) and Sitton et a1. (1980) found no dif-
ferences in the number of changes or gains between males. Mueller
and Wasser (1977) noted that throughout the studies they reviewed,

the sex with the highest test mean had more profit from change.

Personality,

 

Ballance (1977) and Lynch and Smith (1975) suggested that the
relationship between personality and answer changing be explored.
From the mid-19705 to the present, a few studies have attempted to
explore that relationship (Greene, 1981; McMorris & Leonard, 1976;

Payne, 1983; Range et a1., 1982; Sitton et a1., 1980; Stoffer et a1.,

29

1977). For the most part, these studies have used varying personality
measures. The most studied variable has been anxiety.

McMorris and Leonard (1976), using students in four educational-
measurement and evaluation courses who were studied as four groups,
explored the relationship between item-response changes and four
cognitive styles: impulsivity-reflectivity, anxiety, preference for
complexity, and field independence. In this study, anxiety was meas-
ured by Spielburger, Gorsuch, and Lusheme's Self-Evaluation Question-
naire (1968). Results indicated that most of the relationships were
quite low and that the relationships were not consistent across the
four groups. Of the variables studied, impulsivity was the character—
istic most related to answer changing and the one the authors felt
might show replication in further studies. There did not appear to
be a relationship between anxiety and answer changing.

Stoffer et a1. (1977) found that 29 percent of their college
psychology students and 20 percent of their military technical-
vocational subjects made no answer changes on most of the course
tests during the semester. The Observation was true across several
tests, suggesting to the investigators a style of test taking as a
"changer" or a "nonchanger." They posited that those who perceived
that their test scores (reinforcements) were not under their control
would change fewer answers than those who perceived that their test
scores were under their control. Findings, however, were that there
was no correlation (r. = .00) between Rotter's personality measure

for internal-external locus of control and answer changing.

30

Sitton et al. (1980) studied undergraduate students enrolled
in adolescent psychology. They explored the relationship between
answer changing and depression, anxiety, assertiveness, introversion-
extroversion, and several demographic variables. The Taylor Manifest
Anxiety Scale was used to measure anxiety. Results were that assert-
ive students and top-scoring students changed fewer answers. Using a
multiple-regression procedure, the four best predictors of the number
Of changes were the LIE scale Of the Eysenck Personality Inventory,
assertiveness, course grade, and depression. These predictors
accounted for 22 percent of the variance in answer changing. The
three best predictors of wrong—to-right changes were the LIE scale,
marital status, and assertiveness. The LIE scale, according to the
inventory's authors, "may be used to identify subjects showing a
'desirability response-set'" (Eysenck & Eysenck, 1968). In discus-
sion, Sitton et a1. suggested that the connection between a desira-
bility response set and patterns Of answer changing might include a
motivation-level difference or a difference in ego investment in
academic achievement and thus represent more Of a desire tO respond
to a socially desirable pattern. Subsequent research by McMorris
et a1. (1983) found a significant relationship between student atti-
tude toward answer changing and the number of answers changed and
the number of answers changed wrong to right. The investigators
found no relationship between anxiety and answer changing.

Greene (1981), Payne (1983), and Range et a1. (1982) con-
tinued to look for a relationship between anxiety and answer changing.

Greene investigated the answer changing behavior of graduate subjects

31

enrolled in a statistics course and found that high-anxious students
made more changes than did low-anxious students. Both the high-
anxious and low-anxious students profited equally from their changes.
High-anxious students had a slight tendency to reduce their scores

by their changes. Greene used the Test Anxiety Scale to measure
anxiety.

Range et al. also found that high-anxious students changed
answers more often. The State-Trait Anxiety Scale (Spielberger)
was used to measure anxiety. In addition, these authors investigated
the relationship between depression, sex-role stereotyping, and
self-actualization and answer-changing patterns. In addition to the
results concerning anxiety, findings were that nondepressed students
made more positive changes and that the best predictors of these
positive changes were compulsivity, a positive view Of man, mascu-
linity (sex-role typing), lack of depression, and anxiety. These
accounted for 33 percent of the variance.

Studying younger students in grade eight, Payne (1983) repli-
cated the findings Of Green (1981) and of Range et a1. (1982) in
regard to anxiety and changes. Payne found that high-anxious sub-
jects changed more answers. The anxiety measure in the study was the
Survey of Feelings about Tests by Harnish, Hill, and Lyons (1980).
Payne also found that black subjects were more anxious than white
subjects, that their mean test scores were much lower, and that they
changed significantly more answers wrong to wrong and right to wrong
than did the white subjects. However, there was no correlation between

anxiety and answer changing for the group of blacks. The findings

32

of this study were consistent with Ebel's (1965) report that there is
a negative correlation between level of achievement and level of test
anxiety.

In summary, McMorris and Leonard (1976) and Sitton et a1.
(1980) were unable to find a strong relationship between anxiety and
answer changing patterns. Their findings were in conflict with those
of Greene (1981), Payne (1983), and Range et a1. (1982), who all
found that high-anxious students change more answers. McMorris and
Leonard found indications that impulsivity might have a relationship
with answer changing, and Sitton et al. found a relationship between
the LIE scale of the Eysenck Personality Inventory, which is thought
to measure desirability response set, and patterns of answer changing.
In addition, Range et al. found that nondepressed students made more

positive changes.

Summary

The related literature that was reviewed in this chapter was
concerned with answer changing on objective tests. More specifically,
studies were reviewed that focused on answer changing and beliefs
concerning answer changing, item and test characteristics and answer
changing, achievement and answer changing, ability and answer changing,
sex differences and answer changing, and personality and answer chang-
ing. On the basis of this information, those factors that appear to
be related to answer revision were selected for the present study.

The review of the literature suggested a need to explore the

effects of test directions to revise answers under experimental

33

conditions. A need to evaluate the differences between groups given
varying directions and groups formed by change effectiveness and
the variables of sex, item difficulty, achievement, and ability was
also suggested. Finally, a need to explore the differences in study
habits and attitudes scores and answer changing behavior under direc-
tions to change answers was also implied by the existing research.
Chapter III presents the population, methodology, and data-
gathering procedures, as well as the design, hypotheses, and

statistical treatment of the data.

CHAPTER III

METHODOLOGY

Introduction

 

This chapter identifies and describes the population and
elaborates on the methodology used in conducting the study. The data-
gathering procedures are outlined, the design of the study is pre-
sented, the data-gathering instruments are described, and the
hypotheses are presented. Finally, the statistical treatment of the

data is explained.

Population

 

The subjects in this study were college students who were
enrolled in a lOO-level religion class at a small liberal arts college
north of Boston, Massachusetts. The students who attend the college
are from middle- or upper-middle-class backgrounds. The majority Of
the students are from the northeastern United States. There were 159
students who originally enrolled in the religion course. Due to
attrition, 30 were eliminated from the study. Twelve students dropped
the course. Data were unavailable for 18 other subjects.

The religion course was selected because Of the number in the
class and the age Of the students enrolled. The course is a core
requirement for all students. The majority of the students were

entering freshmen. Thus, the data for this study were collected

34

35

on freshman students who were in their first three months of college

life.

Procedures

 

Permission for this study was Obtained from the academic dean
of the liberal arts college. The registrar granted permission to
Obtain the needed Scholastic Aptitude Test (SAT) scores and grade
point averages. The instructor of the course volunteered to allow
the investigator to obtain the data for the study.

The Survey of Study Habits and Attitudes (SSHA) by Brown and
Holtzman (Psychological Corporation, 1965) was administered by the
investigator to the subjects in a 35-minute block of class time the week
before the final examination. Permission from the college's retention
committee was obtained to tell the subjects that the survey was to be
used to study retention Of students at the college. Subjects were also
told that the results Of the survey would not influence their grades
and that their scores on the survey would be discussed with them if
they requested a conference.

Approximately 40 of the subjects were absent during the class
period when the survey was administered. These 40 took the survey at
the time of the final examination. A cover letter explaining the pur-
pose of the survey was given to these students. (See Appendix A).

The letter explained that the survey was being administered for reten-
tion purposes and gave directions for taking the test. These materials
were distributed at the door with the final examination. Most Of the

students received a personal word concerning the survey.

36

Subjects were assigned at random to one of three groups:
control, treatment 1, or treatment 2. Each of these groups received
a different set of printed directions on the Objective section of the
comprehensive final examination. The control-group subjects received
the following directions: "Please read the instructions carefully
and answer the questions asked. If you make an error, cross out (X)
the wrong answer and write another answer." The students in treat-
ment 1 were instructed, "If you feel that you have made an error,
cross out (X) the answer you believe is wrong and write another answer.

Change with caution answers you believe are incorrect." Students in

 

treatment 2 received these instructions: "If you find that you have
made an error, cross out (X) the answer you believe is wrong and mark
another answer. After you have worked through the test once, go back
and change the answers you now believe are incorrect. Research indi-
cates that revising answers will improve your score on this test."

In addition, treatment 2 students received the following direction

at the conclusion of the examination: "GO BACK AND REVISE YOUR
ANSWERS!!" Each set of instructions was headed with "INSTRUCTIONS,"

 

and an attempt was made in the printing to space and line so that the
instructions appeared to be of equal length. (See Appendix 8.)
Additionally, the objective section of the final was placed in the
middle of the test packet. A sentence-completion section preceded the
treatment portion, and an essay section followed.

The examinations were arranged in alphabetical order. Two
teacher assistants and the investigator distributed the examination

from outside the examining room. The investigator's presence was

37

explained by the cover letter of the survey. The teacher assistants
and the investigator attempted a word of explanation to each student.
Further, because the investigator had spoken in classes and taught
approximately 50 Of the subjects in two sections Of study skills,

the subjects did not question her presence.

The SSHA was collected with the final examination and then
separated from the finals, which were graded by the instructor of the
religion course. When the investigator examined the objective section
Of the final for changes, a tally was made of the answers that had
been changed from right—tO-wrong (R-W) and wrong-to-right (W-R). A
wrong-tO-wrong category was eliminated because 8 percent Of the changes
had been thoroughly blotted out. Items that were erased, blotted out,
and crossed out were tallied as changes. The objective section of the
final examination was then transferred from the printed-test answer
blanks to machine-scored answer sheets. The answer sheets were then
processed at the Testing Center of the Michigan State University
Computer Center. The examinations were scored; an item analysis was
computed; and the mean, standard deviation, variance, standard error
Of measurement, and reliability (Kuder-Richardson formula #20) were
computed.

The Scholastic Aptitude Test (SAT) scores were obtained for
all subjects, as were the end-Of—the-quarter or cumulative grade point
averages. The study was conducted with 129 subjects. There were 45
in the control group, 40 in the treatment 1 group, and 44 in the
treatment 2 group. Attrition in the pOpulation was due to 12 who

dropped the course or failed to take the final examination and 18 for

38

whom SAT or SSHA scores were not available. SAT scores were available
for all freshman students, but the SAT scores are not required by the
college for transfer students. Therefore, the majority of subjects

who were dropped from the study were transfer students.

Design

The study employed the randomized control—group posttest-only
design. The 159 subjects were assigned at random to one of three
groups: control, treatment 1, or treatment 2.

The test directions were the treatment or independent variable,
and the posttest or dependent variable was the answer changes. The
intention behind the use of the SSHA scores was to discover whether
any study habits or attitudes are associated with effective or inef-
fective answer changing. The associations between answer changing
effectiveness and sex, item difficulty, objective test scores, grade

point averages, and SAT scores were also explored.

Instruments for Data Gathering

 

Two instruments for gathering data were used. They were the
Survey of Study Habits and Attitudes (SSHA) by Brown and Holtzman
(Psychological Corporation, 1965) and a comprehensive final multiple—
choice and true-false test that was constructed by the professor of
the religion course.

The SSHA is a lOO-item survey of study habits and attitudes
for grades 12 through 14, which is a report of self-perceived study
habits and attitudes. The manual (1967) recommends its use as a

screening instrument, a diagnostic instrument, a teaching aid, and a

39

research tool (pp. 5-6). The administration time is approximately

35 minutes. Students record their answers to questions on a machine-
scored answer sheet that may be graded by hand or by machine. To

each statement, students respond to one of five Options: R--Rarely
(O%-15%), S--Sometimes (16%-35%), F--Frequent1y (36%—65%), G--Generally
(66%-85%), and A--Almost Always (86%-100%).

Four 25-item subtests are delineated. They are delay avoid-
ance, work methods, teacher approval, and education acceptance. Scores
on the first two subtests may be combined to yield a score for study
habits. The last two subtests combine to form study attitudes. The
total score is labeled "study orientation." Items under "delay avoid-
ance," for example, ask students how much they stick to work until it
is accomplished, how much their study is determined by mood, and how
able they are to attend to their work. "Work-methods" items query
about such habits and abilities as paper writing, test taking, and
reading. "Teacher approval" presents statements to the student such
as "Teachers enter teaching because they enjoy teaching" and "Teachers
are sarcastic toward poorer students" (Brown & Holtzman, 1965, pp. 3-4).
"Education acceptance" presents statements that deal with the reasons
students are in college and the behaviors that students may exhibit as
a result of their reasons for attending college.

Responses on the SSHA are based on student perception. The
manual suggests that the administrator should encourage the students
to express their true feelings about themselves (Brown & Holtzman,
1967). The self-rating technique on this survey has been listed as a

threat to the instrument's predictive validity by Wrenn (1959) and

4O

Shay (1972). Goldfried and D'Zurilla (1973), however, found a .55
correlation between the SSHA self-report and peer ratings. In estab-
lishing the concurrent validity of the instrument, Morris (1961)
compared teacher ratings of academic performance and SSHA scores.
Students who earned higher percentile ranks on the SSHA were rated as
superior students by their teachers, and those students who were rated
as poor students by their teachers had lower SSHA scores.

The manual also presents the evidence for the validity of the
survey in terms of low (.21) mean correlation coefficients between the
survey and aptitude tests and the somewhat higher mean correlation of
.36 between the SSHA and grades. The survey measures something differ-
ent from ability and appears to be somewhat more related to grades
than to ability (Brown & Holtzman, 1967). Wen and Lui (1976) also
found moderate, but significant, correlations (.34 to .49) between
course examination grades and the SSHA.

Test-retest reliability coefficients taken on the SSHA at
four-week intervals ranged from .93 to .95 for subtests and total
scores. These data were collected using 237 ninth graders who were
students in a Texas high school. More extensive reliability studies
were conducted with Form H, which is the form for grades 7 through 12
(Brown & Holtzman, 1967).

Evaluators of the instrument generally concur that the instru-
ment has been carefully devised and has satisfactory reliability and
adequate validity (Deese, 1959; Higgins, 1967; Shay, 1972; Wrenn, 1959).
However, the findings of this study are limited by the self-reporting

format and the validity and reliability of the instrument.

41

The final examination on which the changed answers were
Observed was an 80-item comprehensive objective examination.. It con-
tained 4O multiple-choice items and 40 true-false items. Item 12 in
the multiple-choice section was eliminated from the study because it
was constructed with seven foils. The other multiple—choice items
were constructed with four foils. This BO-item section was one of
three parts of the examination.

The examination was analyzed at Michigan State University.

The mean item difficulty was 29, and the mean item discrimination was
23. Of the 79 items on the examination, 45 items had a discrimination
index of 23 or higher. The raw scores on the examination ranged from
36 to 74; the mean was 55.84 and the standard deviation was 7.52. The

variance was 56.6. The Kuder-Richardson formula #20 reliability was .78.

Hypotheses

 

Ho 1: There is no effect due to directions on the number of
answers changed.

HO 2: ThereiSTHIeffectdue to directions on the percentage of
correct changes.

HO 2.1: There is no effect due to directions on the percentage
Of correct changes on total items.

HO 2.2: There is no effect due to directions on the percentage
of correct changes on easy items.

HO 2.3: There is no effect due to directions on the percentage
of correct changes on difficult items.

HO 3: There are no differences in SSHA scores among direction
groups and effectiveness groups.

Ho 3.1: There are no differences in SSHA delay avoidance scores
among direction groups and effectiveness groups on
total items.

Ho

Ho

Ho

Ho

HO

HO

HO

Ho

Ho

Ho

HO

Ho

Ho

.10:

.11:

.12:

.13:

.14:

42

There are no differences in SSHA delay avoidance scores
among direction groups and effectiveness groups on
easy items.

There are no differences in SSHA delay avoidance scores
among direction groups and effectiveness groups on dif-
ficult items.

There are no differences in SSHA work methods scores
among direction groups and effectiveness groups on
total items.

There are no differences in SSHA work methods scores
among direction groups and effectiveness groups on
easy items.

There are no differences in SSHA work methods scores
among direction groups and effectiveness groups on
difficult items.

There are no differences in SSHA education acceptance
scores among direction groups and effectiveness groups
on total items.

There are no differences in SSHA education acceptance
scores among direction groups and effectiveness groups
on easy items.

There are no differences in SSHA education acceptance
scores among direction groups and effectiveness groups
on difficult items.

There are no differences in SSHA teacher approval scores
among direction groups and effectiveness groups on total
items.

There are no differences in SSHA teacher approval scores
among direction groups and effectiveness groups on easy
items.

There are no differences in SSHA teacher approval scores
among direction groups and effectiveness groups on dif-
ficult items.

There are no differences in SSHA composite score study
habits (delay avoidance and work methods) among direc-
tion groups and effectiveness groups on total items.

There are no differences in SSHA composite score study
habits (delay avoidance and work methods) among direc-
tion groups and effectiveness groups on easy items.

HO

Ho

Ho

Ho

Ho

Ho

Ho

Ho

Ho

Ho

Ho

HO

.15:

.16:

.17:

.18:

.19:

.20:

.21:

43

There are no differences in SSHA composite score study
habits (delay avoidance and work methods) among direc-
tion groups and effectiveness groups on difficult
items.

There are no differences in SSHA composite score study
attitudes (education acceptance and teacher approval)
among direction groups and effectiveness groups on
total items.

There are no differences in SSHA composite score study
attitudes (education acceptance and teacher approval)
among direction groups and effectiveness groups on
easy items.

There are no differences in SSHA composite score study
attitudes (education acceptance and teacher approval)
among direction groups and effectiveness groups on
difficult items.

There are no differences in SSHA study orientation scores
among direction groups and effectiveness groups on total
items.

There are no differences in SSHA study orientation scores
among direction groups and effectiveness groups on easy
items.

There are no differences in SSHA study orientation scores
among direction groups and effectiveness groups on dif-
ficult items.

There is no relationship between item difficulty and
effective answer changing.

There are no differences in percentage of correct
changes among direction groups and sex groups.

There are no differences in percentage of correct
changes among direction groups and sex groups on total
items.

There are no differences in percentage Of correct
changes among direction groups and sex groups on easy
items.

There are no differences in percentage of correct
changes among direction groups and sex groups on dif-
ficult items.

Ho

Ho

Ho

HO

Ho

HO

Ho

Ho

Ho

Ho

HO

Ho

HO

44

There are no differences in achievement measures
among direction groups and effectiveness groups.

There are no differences in objective test scores
among direction groups and effectiveness groups on
total items.

There are no differences in objective test scores
among direction groups and effectiveness groups on
easy items.

There are no differences in Objective test scores
among direction groups and effectiveness groups on
difficult items.

There are no differences in grade point averages
among direction groups and effectiveness groups on
total items.

There are no differences in grade point averages
among direction groups and effectiveness groups on
easy items.

There are no differences in grade point averages
among direction groups and effectiveness groups on
difficult items.

There are no differences in ability measures among
direction groups and effectiveness groups.

There are no differences in SAT verbal scores among
direction groups and effectiveness groups on total
items.

There are no differences in SAT verbal scores among
direction groups and effectiveness groups on easy
items.

There are no differences in SAT verbal scores among
direction groups and effectiveness groups on diffi-
cult items.

There are no differences in SAT mathematics scores
among direction groups and effectiveness groups on
total items.

There are no differences in SAT mathematics scores
among direction groups and effectiveness groups on
easy items.

45

HO 7.6: There are no differences in SAT mathematics scores
among direction groups and effectiveness groups on
difficult items.

Data Analysis

 

The information Obtained from answer changing, the SSHA
scores, the item analysis, grade point averages, and SAT scores was
entered into the computer at Michigan State University. The hypothe-
ses were tested with the analysis of variance and the chi-square
procedures, using the Statistical Package for the Social Sciences
(Nie, Hull, Jenkins, Steinbrenner, & Bent, 1978). The results Of the

data analysis are discussed in Chapter IV.

Summar
This chapter contained a discussion Of the methods and proce-
dures used in the study. The population was described, the data-
gathering procedures were outlined, and the design of the study was
presented. In addition, the hypotheses were presented and the sta-

tistical treatment of the data was explained.

CHAPTER IV

PRESENTATION AND ANALYSIS OF DATA

Introduction

 

The purpose Of this study was to investigate the effects of
three different sets Of directions to change answers on the answer
changing behavior of college students. In this chapter, the statis-
tical analysis of the data related to the major research questions is
presented. The writer attempted to answer the following questions:

1. What is the effect of directions on the number of answers
changed?

2. What is the effect of the directions on the percentage
Of correct changes?

3. Are there differences in SSHA scores among the direc-
tion groups (treatment) and the effectiveness groups (countereffec-
tive, ineffective, and effective changers)?

4. Is there a difference between easy and difficult item
changes and the effectiveness of the change?

5. Are there any differences in the percentage of correct
changes among the direction groups and the groups formed by sex of
subjects?

6. Are there any differences in the achievement measures

among the direction groups and the effectiveness groups?‘

46

47

7. Are there any differences in the ability measures among
the direction groups and the effectiveness groups?

In Chapter III, the null hypotheses were presented. The
hypotheses are restated in this chapter, and the findings are pre-
sented. The analysis of variance and the chi-square procedures were

used to test the hypotheses.

Statistical Analysis

 

Directions and Number of Changes

 

HO 1: There is no effect due to directions on the number
of answers changed.

The mean numbers of answer changes were calculated for each

group. These means are presented in Table 1.

Table 1.--Mean number of changes and standard deviations for each

 

 

group.
Group Size Mean Standard Deviation
Control 45 2.42 2.44
Treatment 1 40 2.20 1.77
Treatment 2 44 3.30 2.06

 

The analysis of variance failed to reject the hypothesis
(p > .05). There were no differences in the mean number of changes
due to the directions groups; F (2, 123 df) = 1.99, p > .05. (See
Table 2.) The directions given to the groups did not affect the

number of changes.

48

Table 2.--ANOVA table for number of changes.

 

 

 

Source df MS F
Treatment 2 9.189 1.99
Error 123 4.613
Total 128 587.395

p > .05

Directions and Effectiveness
of Changes

 

 

Ho 2.1: There is no effect due to directions on the per-
centage Of correct changes on total items.

The percentage Of correct changes was used as a measure of
answer changing effectiveness. This score was computed by dividing
the number of correct changes by the total number of changes and then
multiplying by 100. The range of percentage of correct changes, there-
fore, was 0-100 percent. This percentage-of-correct-changes score was
calculated for total items, for easy items, and for difficult items.
These divisions were made in the data because the test used as the
data-gathering instrument was judged as easy (lower standard devia-
tion than is suggested by testing authorities). The percentage correct
score was used because the net-gain scores used in other studies on
answer changing did not account for the total number of attempted
changes. Net-gain changes are calculated by subtracting the incorrect
changes from the correct changes. This procedure tends to equate
individuals who have the same net-gain score when, in fact, one may

have changed 18 items and the other, 4.

49

Easy items were those where 29 percent or fewer of the students
missed the question. Difficult items were those where 30 percent or
more Of the students missed the question. The phrases "effective
answer changing on total items" or "degree of change effectiveness on
total items" refer to all correct items changed by the subjects.

To test the null hypothesis, the mean percentage correct for

total items was calculated for each group. Table 3 lists these means.

Table 3.--Mean percentage correct scores and standard deviations on
total items.

 

 

Group Mean Percent Correct Standard Deviation
Control 40.24 41.57
Treatment 1 50.87 56.23
Treatment 2 56.63 36.75

 

The analysis of variance failed to reject the null hypothesis
(p > .05). There wasvunrelationship between the means and the direc-
tions. NO significant differences were found among these means;
F (2, 123 df) = 1.60, p > .05. (See Table 4.) The directions did

not affect the percentage of correct changes on total items.

Table 4.--ANOVA table for effectiveness Of changes on total items.

 

 

Source df MS F
Treatment 2 3219.16 1.60
Error 123 2007.56
Total 128 2059.72

 

p > .05.

50
Ho 2.2: There is no effect due to directions on the per-
centage of correct changes on easy items.
The mean percentage correct scores on easy items were calcu-
lated for students in each group. These means are presented in

Table 5

Table 5.--Mean percentage correct scores and standard deviations on
easy items.

 

 

Group Mean Percent Correct Standard Deviation
Control 37.34 43.57
Treatment 1 32.65 43.22
Treatment 2 61.74 44.72

 

The analysis of variance rejected the null hypothesis

(p < .05); F (2, 120 df) = 5.39, p < .05. (See Table 6.) The

directions did affect the percentage of correct changes on easy items.

It would appear that the strong directions affected the percentage of
correct changes. The mean percentage correct in treatment 2 was 29

percent higher than that in the treatment 1 group.

Table 6.-—ANOVA table for effectiveness Of changes on easy items.

 

 

Source df MS F
Treatment 2 10483.300 5.39*
Error 120 1943.354
Total 125 2059.254

 

*p < .05.

51
HO 2.3: There is no effect due to directions on the per-
centage Of correct changes on difficult items.
The mean percentage correct for difficult-item changes was

calculated for each group. These are listed in Table 7.

Table 7.--Mean percentage correct scores and standard deviations
on difficult items.

 

 

Group Mean Percent Correct Standard Deviation
Control 26.88 42.25
Treatment 1 43.39 44.27
Treatment 2 35.03 39.80

 

The analysis of variance failed to reject the null hypothesis
(p > .05); F (2, 123 df) = 1.60, p >.05. (See Table 8.) The direc-
tions did not affect the percentage of correct changes on difficult
items. The mean percentage correct scores Of students on difficult
items were not significantly different among groups. The directions
to change answers did not affect the percentage of correct changes on

difficult items.

Table 8.-—ANOVA table for effectiveness of changes on difficult items.

 

 

Source df MS F
Treatment 2 2891.18 1.60
Error 123 1800.58
Total 128 1788.37

 

p > .05.

52

In summary, the directions to change answers did not affect
the rate of answer changing or the percentage correct changes on
total items and difficult items. Directions to change answers did
statistically affect the percentage Of correct changes on easy
items.

The SSHA and Effectiveness
of Changes

 

 

Ho 3.1: There are no differences in SSHA delay avoidance
scores among direction groups and effectiveness
groups on total items.
The SSHA delay avoidance subtest means were calculated by
the directions groups and by the effectiveness groups. These means
are presented in Table 9.

The analysis of variance failed to reject the null hypothesis
(p > .05). There were no interactions between the directions groups
and the effectiveness groups on total items; F (4, 119 df) = .47,
p > .05. There was no main effect for directions; F (2, 119 df) =
.398, p > .05. There was no main effect for effectiveness on total
items; F (2, 119 df) = 1.57, p > .05. (See Table 10.) The directions

and the degree of change effectiveness were not associated with the

SSHA delay avoidance scores.

53

 

 

 

 

 

 

om.m w~.o~ cc mm.“ om.o~ mm mm.pp mp.om FF m¢.m Nn.wp PF N ucmEummgh
Nm.m mo.mp mm NN.oP mo.mp mp mo.op N¢.om n Nw.m om.n~ or F pcwsummgh
mN.m mm.o~ me mm.o o~.o~ up m~.n o¢.mN m mo.m mm.wp mm Fogucou
om cum: 2 am :mm: 2 mm cum: 2 am :mm: 2
Pmuohnl<o m>vpummmm m>wuumw$mcm m>wuumwwmgmpcaou asoco
pmuOH poachnu<n paychnu<o paychun<o

 

mo mmcmmv Low pmmpnsm A<ov mucmuwo>m

.memuw Page» co mmmcm>wuow44m mmcmgu
zmpmu Low mmcoom Aomv cowumm>mu cgmucmum use :moz--.m mpnmh

54

Table 10.--ANOVA table for delay avoidance-~total.

 

 

 

Source df MS F
Treatment group 2 33.79 .39
Degree of effectiveness 2 133.70 1.57
Treatment x Degree 4 40.60 .47
Error 119 84.91
Total 127 83.47
p > .05

HO 3.2: There are no differences in SSHA delay avoidance
scores among direction groups and effectiveness
groups on easy items.

The mean delay avoidance subtest scores were calculated by
directions groups and by effectiveness groups on easy items. These
means and standard deviations are presented in Table 11.

The analysis of variance failed to reject the null hypothesis
(p > .05). There were no interactions between the directions groups
and the effectiveness groups on easy items; F (4, 118 df) = .71,

p > .05. There was no main effect for directions; F (2, 118 df) =
.17, p > .05. There was no main effect for effectiveness; F (2, 118 df)
= .31, p > .05. (See Table 12.) The delay avoidance scores were not

associated with directions or with the degree of change effective-

ness.

55

 

 

 

 

 

 

m¢.m mp.o~ we oo.m oo.o~ mm m_.F oo.m_ FF No.pp mm.PN P? N pcmsummch
um.m em.mp mm mm.m Nu.wp mp mm.¢_ mu.m~ n ¢¢.m ow.m~ mp _ acmeummch
mm.w mm.o~ me vo.o pm.m_ up m¢.o_ mn.m~ m Nu.m ee.FN mm _ogpcoo
mm com: 2 am new: 2 mm com: 2 am new: 2
xmmmuu<o m>wuum$mm m>wuom$wo=H o>wpommmmgmucsou aaocw
Fmpoh ammwlu<o Amam-u<o Ammmu-<o

 

.mempw ammo co mmmco>wpumm$o mmcmzu
do mmgmmu Lo» pmmpasm A<ov mucmuwo>m zMan co$ mucoum Aamv cowumv>mv ugmucmum vcm cmozlu.pp mpnmp

56

Table 12.--ANOVA table for delay avoidance--easy.

 

 

 

Source df MS F
Treatment group 2 14.09 .17
Degree of effectiveness 2 27.22 .32
Treatment x Degree 4 60.89 .71
Error 118 85.51
Total 126 82.63

p > .05

Ho 3.3: There are no differences in SSHA delay avoidance
scores among direction groups and effectiveness
groups on difficult items.

The mean delay avoidance scores were calculated by directions
groups and by effectiveness groups on difficult items. These means
and standard deviations are presented in Table 13. Also presented are
the number of students in each degree-Of-change-effectiveness category.

The analysis of variance failed to reject the null hypothesis
(p > .05). There were no interactions between directions groups and
effectiveness groups; F (4, 120 df) = .37, p > .05. There was no
main effect for directions; F (2, 120 df) = .48, p > .05. There was
no main effect for effectiveness; F (2, 120 df) = 1.06, p > .05.

(See Table 14.) Delay avoidance scores were not associated_with the

directions or with the effectiveness of changing.

57

 

m¢.m m—.o~ #e m~.o— oo.- FF Fm.o~ FN.NN n m~.m mm.mp om N paprmmgh

 

 

 

 

 

o~.m um.m_ 04 mm.m mm.m_ m_ e~.m oe.- m o~.m m~.mp ON _ Semapmmzp
mm.m mm.o~ me om.m Pm._N pp mp.m oo.o~ m mm.m mN.m_ _m Peepeoo
om cum: 2 am new: 2 am cam: 2 am cam: z
upsuwccwo--<o m>wpumccm a>wpumcca=H m>wpomccmemp==ou azoeu
_moop opzuwccao--<o “_zoacewo--<o 32382443o--<o

 

.memuw pP=UW$mwu co mmmcm>_wum$$m wmcmzo
do mocmmc co; ummpnzm A<ov mocmuwo>m ampmu com mmcoom Aomv cowpmm>mc ucmucmpm use :mmzlu.m_ anmH

58

Table 14.-—ANOVA table for delay avoidance--difficult.

 

 

 

Source df MS F
Treatment group 2 41.35 .48
Degree of effectiveness 2 91.17 1.06
Treatment x Degree 4 31.92 .37
Error 120 85.64
Total 128 83.127

p > .05.

In summary, for this population, when delay avoidance subtest
scores were taken into account, there were no differences among
direction groups or effectiveness groups on total items, easy items,

or difficult items.

HO 3.4: There are no differences in SSHA work methods scores
among direction groups and effectiveness groups on
total items.

The mean and standard deviation work methods scores were cal-
culated by directions groups and by effectiveness groups on total
items. These are presented in Table 15.

The analysis of variance failed to reject the hypothesis
(p > .05). There was no interaction between the directions groups
and the effectiveness groups; F (4, 119 df) = .16, p > .05. There was
no main effect for directions; F (2, 119 df) = .50, p > .05. There
was no main effect for effectiveness; F (2, 119 df) = 2.45, p > .05.

(See Table 16.) Work methods scores were not associated with direc—

tions or with the degree Of change effectiveness on total items.

59

 

NN.m No.mm vv

mm.p_ mm.mm NN

mm.n oo.m~ FF

om.m w~.mw pp

N pcmspmmep

 

 

 

 

 

wo.m op.vm mm mo.- mu.m~ mp o~.m ~n.n~ m m¢.m om.m~ m_ P acmEummLh
eo.m Fu.mm mv mo.n oo.nm up mm.o ow.w~ m mm.op wo.¢~ mm Pogucou
om cum: 2 am cum: 2 am cum: 2 am cum: 2
Pmuopuuzz o>vpummmm m>wuummdwcH m>vuumw$mcmpczou azocw
pogop panchnnzz paychuuzz panchuuzz

 

4o mmcmmu Low

.msmpw quou :o mmm=m>wuummwm mucosa

pmmpnzm Azzv muozume xeoz cod mmcoom Aomv cowumw>mu ucmccmpm ucm :mmzuu.m~ epoch

60

Table 16.--ANOVA table for work methods—~total.

 

 

 

Source df MS F
Treatment group 2 44.05 .50
Degree of effectiveness 2 213.42 2.45
Treatment x Degree 4 14.39 .17
Error 119 86.91
Total 127 85.68

p > .05.

Ho 3.5: There are no differences in SSHA work methods
scores among direction groups and effectiveness
groups on easy items.

The mean and standard deviation work methods scores were cal-
culated by directions groups and by effectiveness groups on easy
items. These are presented in Table 17.

The analysis of variance failed to reject the null hypothesis
(p > .05). There was no interaction between the directions groups
and the effectiveness groups on easy items; F (4, 118 df) = 2.06,

p > .05. There was no main effect for directions; F (2, 118 df) =
.30, p > .05. There was no main effect for effectiveness on easy
items; F (2, 118 df) = 1.73, p > .05. Work methods scores were not
associated with the directions or with the degree of change effective-

ness on easy items.

61

 

 

 

 

 

 

Nm.m No.m~ cc mm.n pp.mm mm mm.w mm.¢~ m mm.PF m¢.mm mp N pcmEummLh
mm.m mm.¢~ mm om.m mo.mm pp em.w mn.mm e wN.w ww.- mm — pcmsummgh
¢o.m pm.mm mc mw.m mN.mN cp ¢~.m mn.¢~ e N¢.m no.0N mm pogpcoo
om cum: 2 am cum: 2 am cum: 2 am cum: 2
>m~wun£3 m>wpuo$wu m>wpomwwwcH m>wpumwwocmucsoo gzoco
—muoh ammMTTEz zmmmnszz xmmmulzz

 

we mmcmmu Lo;

.mswpw Ammo co mmwcm>wpummmm mmcmgu

ammuazm fizzy muocpws x203 Lo; mucoom Aomv cowpmw>mu ucmvcmum vcm :mmz--.mp mpomh

62

Table 18.--ANOVA table for work methods--easy.

 

 

 

Source df MS F
Treatment group 2 25.05 .31
Degree of effectiveness 2 145.20 1.77
Treatment x Degree 4 168.98 2.06
Error 118 81.95
Total 126 84.70

P > .05.

HO 3.6: There are no differences in SSHA work methods scores
among direction groups and effectiveness groups on
difficult items.

The means and standard deviations were calculated for work
methods subtest scores by directions groups and effectiveness groups
on difficult items. (See Table 19.)

The analysis Of variance failed to reject the hypothesis
(p > .05). There was no interaction between the directions and the
effectiveness; F (4, 120 df) = .86, p > .05. There was no main effect
for directions; F (2, 120 df) = .65, p > .05. There was a main effect
for effectiveness; F (2, 120 df) = 3.11, p < .05. (See Table 20.)
The work methods scores were associated with effective answer changing
on difficult items.

In summary, the SSHA work methods subtest was not associated
with direction groups or effectiveness of change on easy items or

total items. There was a relationship between the work methods

subtest and the effectiveness of change (N1 difficult items. Subjects

63

 

NN.m No.m~ we

ow.m mp.om —F mm.~ w~.m~ n mm.m om.F~ om N acmEummgh

 

 

 

 

 

Km.m NP.¢N oe em.op 84.4N my mo.m_ ow.mw m ~_.w mo.e~ om _ Semapmmeh

eo.m _~.m~ me .o.m mp.m~ _P oc.~ oo.- m mm.m mm.¢~ Fm Foeacoo

om com: 2 am emu: 2 am cow: 2 am cum: 2

“P3624429--zz m>mpomccm m>wouaccacH m>wuomccoew6==ou aaogm
Pooch “P3624420--23 opaowcewo--zz “pauwccwo--zz

 

mo mmgmou com

.msmpw upsowmwvc co mmmcm>wuuoddm mucosa
pmopnzm fizzy muosume xcoz so; mmcoom Aomv cowumw>wc ugmvcmpm use :mmz--.mp open»

64

who were higher scorers on the work methods subtest were the more

effective answer changers on difficult items.

Table 20.--ANOVA table for work methods--difficult.

 

 

Source df MS F
Treatment group 2 55.00 .65
Degree of effectiveness 2 250.01 3.11*
Treatment x Degree 4 71.93 .86
Error 120 83.56
Total 128 85.04

 

*p < .05.

Ho 3.7: There are no differences in SSHA education acceptance
scores among direction groups and effectiveness
groups on total items.

Mean and standard deviation scores for education acceptance
subtests were computed by directions groups and by effectiveness
groups. These scores are presented in Table 21.

The analysis of variance failed to reject the hypothesis
(p > .05). There was no interaction between the directions groups and
the effectiveness groups; F (4, 119 df) = .47, p > .05. There was no
main effect for directions; F (2, 119 df) = 1.92, p > .05. There was
a main effect for effectiveness; F (2, 119 df) = 4.54, p > .05. (See

Table 22.) The education acceptance subtests were associated with

effectiveness Of changing on total items.

65

 

ow.o~ Pw.wm ev

op.o mp.mm NN

mm.o mm.mm m—

N¢.mp Nn.¢m up N pcwapmmsh

 

 

 

 

 

up.m m¢.om mm n~.m Fm._m FF mm.“ Pn.Fm u oo.o~ mp.mm m p ecospmmch
eo.m Fm.om me mm.o n¢.mm FF mm.m oe.mm mp Pm.w co.w~ mm Focucou
om cum: 2 am cum: 2 am com: 2 am cum: 2
Pmpohlu<u m>wuumm$m o>wuumwwmcH m>wpuwmmmgoaczou aaocw
peach quoblu<m Papopnu<m Papohuu<m

 

.memuw Papop :o mmmcm>Ppum$$m mmcmsu 4o wmcmmv
com ammuazm A<uvmocmpamoomp:;pausuo Low mmcoom Aomv cowpmw>mv ugmvcmpm new :mmz-u._m mpamh

66

Table 22.--ANOVA table for education acceptance--total.

 

 

 

Source df MS F
Treatment group 2 153.67 1.92
Degree of effectiveness 2 363.55 4.54*
Treatment x Degree 4 38.28 .47
Error 119 79.96
Total 127 82.87

*p = .01.

Ho 3.8: There are no differences in SSHA education acceptance
scores among direction groups and effectiveness
groups on easy items.

The mean and standard deviation SSHA education acceptance
subtest scores were computed by directions groups and effectiveness
groups on easy items. These data are presented in Table 23.

The analysis of variance failed to reject the null hypothesis
(p > .05). There was no interaction between the directions groups and
the effectiveness groups on easy items; F (4, 118 df) = .64; p > .05.
There was no main effect for directions; F (2, 118 df) = .24, p > .05.
There was no main effect for effectiveness; F (2, 118 df) = .14,

p > .05. (See Table 24.) The mean education acceptance scores were

not associated with directions or with effectiveness of change on

easy items.

67

 

 

 

 

 

 

ow.oF Fm.wN cw om.m mF.mN mN mm.oF mm.mN FF om.NF mF.mN «F N pcmEummgF
Fm.n Fu.om mm mm.¢ em.mm m eo.¢ om.¢m e wo.m mm.wN v F ucmspmmgk
wo.w Fm.om m¢ om.n ¢o.Nm mF em.¢ mu.mN mN Nm.w mm.om RN Fogucoo
om cum: 2 am com: 2 mm com: 2 am com: 2
FmpoFlu<m m>FpumFFm m>FpomFchH m>Fuummegwu==ou gnoco
FMHOF memuu<m 4Amaul-(m Ammmuu<m

 

.memuF Ammo co mmocm>FuomFFm mmcmso do mmgmmu
com pmmaazm F<wvmocwpqmoumrz;umozum com mmcoum Fomv conmF>mc contempm use :mmz--.mm oFamF

68

Table 24.--ANOVA table for education acceptance--easy.

 

 

 

Source df MS F
Treatment group 2 96.17 1.14
Degree of effectiveness 2 49.30 .58
Treatment x Degree 4 55.57 .62
Error 118 83.92
Total 126 82.25

P > .05.

Ho 3.9: There are no differences in SSHA education acceptance
scores among direction groups and effectiveness
groups on difficult items.

The education acceptance mean and standard deviation scores
were computed by directions groups and by effectiveness groups on
difficult items. These data are presented in Table 25.

The analysis of variance failed to reject the "U11 hypothesis
(p > .05). There was no interaction between directions and effective—
ness on difficult items; F (4, 120 df) = .22, p > .05. There was no
main effect for directions; F (2, 120 df) = .88, p > .05. There was
no main effect for effectiveness; F (2, 120 df) = 2.69, p > .05.

(See Table 26.)

In summary, the education acceptance subtest scores were

associated with the effectiveness of change on total items. Educa-

tion acceptance scores were not associated with effectiveness Of

change on easy items or difficult items.

69

 

ow.oF om.mN we

mo.m mo.Nm FF

Fm.oF Ne.wN mF

mm.FF mF.mN FF N pamEpmeF

 

 

 

 

 

mo.w om.om ow we.“ em.Pm A om.~ oo.mm m Ne.m mm.mm m _ Semapmmzh
eo.m om.om me ou.m em.em em oo.m oo.~m om om.e Pm.m~ Fm _oepcou
am new: 2 gm cum: 2 cm cam: 2 am cum: 2
“Fauwecwo--<w m>wpumecu m>wuomceo=H m>wbomccoemp==ou asoew
Pepe» p_=uwccwo--<m “_zuwcevo--<m Spauwecwo--<m

 

.meouF quuFFFFu :o mmmcm>FpumFFm mmcmzo we mmcmmu

Low ammunzm F<mvmucmpamoomrszpbusum Low mmcoum Fomv :onmF>wu ugmucmum use cmmzuu.mN mFamh

70

Table 26.--ANOVA table for education acceptance--difficu1t.

 

 

 

Source df MS F
Treatment group 2 72.86 .88
Degree of effectiveness 2 222.33 2.69*
Treatment x Degree 4 18.30 .22
Error 120 82.36
Total 128 82.29

*p = .01.

HO 3.10: There are no differences in SSHA teacher approval
scores among direction groups and effectiveness
groups on total items.

The means and standard deviations for teacher approval scores
were calculated by directions groups and by effectiveness groups on total
items. (See Table 27.) In all three treatment groups, the highest mean
teacher approval scores were earned by the ineffective changers.

The analysis of variance failed to reject the null hypothesis
(p > .05). There was no interaction between directions groups and
the effectiveness groups on total items; F (4, 119 df) = 1.08, p >
.05. There were no main effects for directions; F (2, 119 df) =
.35, p > .05. There was a main effect for effectiveness; F (2, 119 df)
= 3.54, p < .05. (See Table 28.) The teacher approval means were
associated with the degree of effectiveness of changes on total items.

The subjects with higher mean teacher approval scores were the inef-

fective changers.

71

 

 

 

 

 

 

mm.m ww.mN we mm.m NF.NN NN wm.m NF.mN FF oe.mF oo.eN FF N pewsummLF
mm.w om.NN mm Fm.w Fo.wN mF No.m Ne.mm N mm.m wN.NN mF F pcmsummgF
mm.N ew.mN me eN.N wo.wN NF om.m ow.Fm m om.F Nm.eN mN Focucou
om cum: 2 am new: 2 am cum: 2 am new: 2
quoFuu<F m>FpumFFm w>FpomFchF m>FpumFFmmeczoo qzoco
FoucF quoFTT<F FmpoF--<F Fmaohuu<F

 

.mEmuF Fmpop co mmmcm>Fuommmm mmcmgu mo mmcmmu
Low ammuazm FwFV Fe>ocgam cmzummp com mmcoum Fomv coFumF>oe ecmvcwpm ecu cmmz--.NN anmF

72

Table 28.--ANOVA table for teacher approval--tota1.

 

 

 

Source df MS F
Treatment group 2 24.69 .35
Degree of effectiveness 2 244.01 3.54*
Treatment x Degree 4 75.04 1.08
Error 119 68.916
Total 127 70.95

*p = .032.

HO 3.11: There are no differences in SSHA teacher approval
scores among direction groups and effectiveness
groups on easy items.

The mean teacher approval subtest scores and standard devia-
tions were computed by directions groups and by effectiveness groups
on easy items. These data are presented in Table 29.

The analysis Of variance failed to reject the null hypothesis
(p > .05). There were no interactions between directions and effec-
tiveness; F (4, 118 df) = .64, p > .05. There was no main effect for
directions; F (2, 118 df) = .24, p > .05. There was no main effect
for effectiveness; F (2, 118 df) = .14, p > .05. The teacher approval

scores were not associated with directions or with change effective-

ness on easy items.

73

 

 

 

 

 

 

mm.m mm.e~ we mN.m 8N.FN 8N mm.m oo.mN m Fw.FF oe.o~ mF N “caspaoeF
NF.m oo.mN mm MF.m mF.FN FF me.o mN.em e mm.F om.FN mm F FemsmecF
mm.F em.oN mm mo.w mm.FN eF em.m oo.mN e Fm.F mm.mN FN Foeocou
am cum: 2 am com: 2 am com: 2 am cam: z
meunu<F w>FpomFFm m>FuumFmocF m>FpumFchmucaoo asocw
FMFOF swam--<F xmmm--<F mem--<F

 

.mEmFF ammo co mmwcm>FuuwFFm mmcmgo Fo

mmgmmu com pmmanzm FeFv Fm>ocqam Logomwp L04 mmcoum Famv :onmF>mu execcmpm vcm cmw2-1.mN anmF

74

Table 30.--ANOVA table for teacher approval--easy.

 

 

 

Source df MS F
Treatment group 2 18.01 .24
Degree of effectiveness 2 10.40 .14
Treatment x Degree 4 46.73 .64
Error 118 72.42
Total 126 69.75

p > .05.

Ho 3.12: There are no differences in SSHA teacher approval
scores among direction groups and effectiveness
groups on difficult items.

The means and standard deviations for teacher approval scores
were calculated by directions groups and by effectiveness groups.
These data are presented in Table 31.

The analysis of variance failed to reject the
(p > .05). There were no interactions between the directions groups
and the effectiveness groups on difficult items; F (4, 120 df) = 2.63,
p > .05. There was no main effect for directions; F (2, 120 df) =
.07, p > .05. There was no main effect for effectiveness; F (2, 120 df)
= 2.63, p > .05. (See Table 32.) The teacher approval scores were
not associated with the directions or the change effectiveness on
difficult items.

In summary, SSHA teacher approval subtest scores were asso-

ciated with the effectiveness of change on total items. The teacher

75

 

 

 

 

 

 

am.m NN.©N we Ne.F em.mN FF NF.oF eF.NN N NN.N Ne.mN 8N N chsmeEF

mN.N NF.FN oe mm.m oo.NN mF FN.© oe.mm m om.e om.oN ON F FemeFemeF

mm.F eN.eN me FN.F mo.mN FF mm.e oo.NN m mo.F em.mN Fm Foeaeou

om cum: 2 am new: 2 am com: 2 am cam: z

FFauFFFFQ--<F a>FFumFFN m>FFomFFa=F m>FFuaFFmLmF==ou asoee
FFaoFFFFa--<F FFauFFFFQ--<F

Fence szuFFFFouu<h

ITI

 

.mEmFF “FauFFFFu co mmmcm>FuomFFm mmcmgu mo

mmcmme coF meunam F<Fv Fm>ocaam Locommp com mwcoom Fomv conmF>me ugmucmgm new cmmzsl.Fm anme

76

approval subtest was not associated with the degree of change effective-

ness on easy or difficult items, and there were no effects for directions.

Table 32.--ANOVA table for teacher approva1--difficult.

 

 

 

Source df MS F
Treatment group 2 5.18 .07
Degree of effectiveness 2 186.08 2.63
Treatment x Degree 4 39.30 .56
Error 120 70.52
Total 128 70.41

p > .05.

Ho 3.13: There are no differences in SSHA composite score
study habits (delay avoidance and work methods)
among direction groups and effectiveness groups
on total items.

The means and standard deviations for the study habits scores
by directions groups and effectiveness groups on total items were cal-
culated. These data are presented in Table 33.

The analysis of variance failed to reject the hypothesis
(p > .05). There were no interactions between directions and effec-
tiveness; F (4, 119 df) = .10, p > .05. There was no main effect for
directions; F (2, 119 df) = .59, p > .05. There was no main effect
for effectiveness of change; F (2, 119 df) = 2.29, p > .05. The

SSHA means were not associated with the treatment or with the degree

Of change effectiveness on total items.

77

 

Fc.oF mN.mw we

wN.mF mm.ww NN wo.NF NF.mw FF mm.NN mo.Nw FF N ucmaumwgh

 

 

 

 

 

mo.mF Fm.Nw mm mw.FN mm.Nw mF mm.mF oo.mw m oo.mF FN.ow mF F “cwEummsh
mN.mF No.0e me 0N.NF wo.mw NF mF.oF ON.©m m mN.mF oo.Nw MN Fogpcou
om cum: 2 am cum: 2 am cum: 2 am com: 2
FmFoFluzm m>FpomFFm m>FpomFFocF m>FuumFFmquczou azocw
FNHOP FMFOFTTIm FMHOHIIIm FMHOHITIm

 

mo mmcmmu cow

.mspr FmFoF co mmmcm>wuumewm mmcmcu
pmmpazm Fzmv mFan; zusum Foe mmgoom Fomv :oFFmF>me ucmvcmum ecu :mm211.mm mFQMF

78

Table 34.--ANOVA table for study habits--tota1.

 

 

 

Source df MS F
Treatment group 2 169.18 .59
Degree of effectiveness 2 653.63 2.29
Treatment x Degree 4 30.08 .11
Error 119 285.35
Total 127 250.81

p > .05.

HO 3.14: There are no differences in SSHA composite score
study habits (delay avoidance and work methods)
among direction groups and effectiveness groups
on easy items.

Study habits means and standard deviations were computed by
directions groups and by effectiveness groups on easy items. These
means and standard deviations are given in Table 35.

The analysis of variance failed to reject the null hypothesis
(p > .05). There were no interactions between the directions groups
and the effectiveness groups; F (4, 118 df) = 1.21, p > .05. There
was no main effect for directions; F (2, 118 df) = .35, p > .05.
There was no main effect for effectiveness; F (2, 118 df) = .60,

p > .05. (See Table 35.) The directions and the degree of effective-

ness Of change were not associated with the study habits scores on

easy item changes.

79

 

 

 

 

 

 

so.oF mN.mw ww mN.wF FF.mw oN ou.m mm.mm m om.ON mF.om mF N pzmsummgF
me.wF we.me mm No.mF Nn.Fw FF Nm.mN oo.mm e mw.oF NF.Fw mN F ucmsummLF
mN.mF Ne.ww me eN.wF Ne.ww eF mo.m om.Fe w mm.oF Fm.me NN Fogacou
am new: 2 am cum: 2 am cum: 2 cm cum: 2
Ammmuuzm m>FFumFFu m>FuumFFm=F m>FpumFFmLmucsou mecca
FmNOF Nmmu--zm ameu--=m Nmmm--zm

 

mo mwemmc cow

.msmuF Ammo :o mmmcm>Fpomme mmcmnu
pmmpnzm Fzmv mFan; xezum Low mmcoom Fomv :onoF>me ugmucoum ecu comzsu.mm «FamF

80

Table 36.--ANOVA table for study habits--easy.

 

 

Source df MS F
Treatment group 2 98.07 .35
Degree of effectiveness 2 169.92 .60
Treatment x Degree 4 341.48 1.21
Error 118 280.54
Total 126 277.397

 

p > .05.

HO 3.15: There are no differences in SSHA composite score
study habits (delay avoidance and work methods)
among direction groups and effectiveness groups
on difficult items.

The means and standard deviations for the study habits com-
posite score by directions groups and by effectiveness groups on dif-
ficult items were computed. These means are presented in Table 37.

The analysis Of variance failed to reject the null hypothesis
(p > .05). There was no interaction between the directions groups and
the effectiveness groups; F (4, 120 df) = .50, p > .05. There was no
main effect for directions; F (2, 120 df) = .77, p > .05. There was
no main effect for effectiveness; F (2, 120 df) = 2.40, p > .05. (See
Table 38.) SSHA study habits scores were not associated with the
directions or with the degree of change effectiveness on difficult
items.

In summary, when the study habits scores were taken into

account, there were no differences among direction groups or effec-

tiveness groups on easy, difficult, or total items.

81

 

 

 

 

 

 

F0.0F ON.mw ww O0.0F OO.Nm FF F0.0F OO.FO F N0.0F ON.Ow ON N OOOOFOOOF
Ow.OF NN.Nw ow F0.0N OO.ww OF N0.0N ON.Fw m w0.0F ON.Nw ON F OOOOOOOOF
ON.OF N0.0w Ow wm.NF O0.00 FF wF.OF O0.00 m OF.OF F0.0w Fm FOOFOOO
Om OOOz 2 OO OOmz 2 OO OOOz 2 OO OOOz z
OFOOFFFFO--:O OOFOOOFFN O>FOOOFFOOF O>FOOOFFOOOOOOOO OOOOO
FOOOF OFOOFNFFO--:O OFOOFFFFO--:O “FOOFFFFO--IO

 

.msmuF FFOOFFFFO co mmmcm>FpumFFm mmcmgu
Fo wmcmmu Low ammunzm Fzmv OFFOO; meawm Low mOLOOm Fomv :oFFOF>mO ULOOOOFO use :Omz--.nm OFOOF

82

Table 38.-—ANOVA table for study habits-~difficult.

 

 

 

Source df MS F
Treatment group 2 217.16 .77
Degree of effectiveness 2 672.71 2.40
Treatment x Degree 4 141.42 .50
Error 120 279.84
Total 128 279.15

p > .05.

Ho 3.16: There are no differences in SSHA composite score
study attitudes (education acceptance and teacher
approval) among direction groups and effective-
ness groups on total items.

The mean and standard deviation scores were computed for the
study attitudes composite scores by directions groups and by effective-
ness groups on total items. These data are presented in Table 39.

The analysis of variance failed to reject the null hypothesis.
(p > .05). There were no interactions between the directions and
effectiveness groups; F (4, 119 df) = .62, p > .05. There was no main
effect for the directions; F (2, 119 df) = 1.30, p > .05. There was a
main effect for effectiveness; F (2, 119 df) = 3.87, p > .05. (See
Table 40.) There were significant differences among the study attitudes

means. The study attitudes means were associated with the degree Of

change effectiveness on total items.

83

 

 

 

 

 

 

mN.mF Fo.em we ON.mF wo.wm NN mw.m em.No FF om.ON Fm.ne FF N acoEFOmLF
mm.mF mN.Nm mm mo.mF mm.mm mF oo.eF mm.Fm N mw.FF mm.mm mF F pcmeummgF
mF.mF Fm.um me mo.NF FF.Nm NF mw.m ow.oo m om.oF om.Nm mN Focucou
om cum: 2 am new: 2 am cum: 2 am com: 2
FOFoFlu<m w>FFomFFu m>FpomFchF w>FpumFmocmpczou Ozocc
FOFOF FmpoFuu<m FOFoFuu<m FOFOFuu<m

 

.mEmpF quou :o mmmcm>FpomFFm mmzmgo mo
mmemoc cow ammunzm mev mmuapFupm avzpm Low mmcoum Famv conOF>mu ucmecmum ecu cmmz--.mm wFOOF

84

Table 40.--ANOVA table for study attitudes--total.

 

 

 

Source df MS F
Treatment group 2 341.61 1.30
Degree of effectiveness 2 1016.74 3.87*
Treatment x Degree 4 162.56 .62
Error 119 262.28
Total 127 269.01

*p < .05.

HO 3.17: There are no differences in SSHA composite score
study attitudes (education acceptance and teacher
approval) among direction groups and effective-
ness groups on each items.

The means and standard deviations were computed for the study
attitudes composite scores by directions groups and by effectiveness
groups on easy items. These scores are reported in Table 41.

The analysis of variance failed to reject the null hypothesis
(p > .05). There were no interactions between the directions groups
and the effectiveness groups; F (4, 119 df) = .60, p > .05. There
was no main effect by directions; F (2, 119 df) = .67, p > .05. There
was no main effect by effectiveness Of change on easy items; F
(2. 119 df) = .35, p > .05. (See Table 42.) The study attitudes scores

were not associated with directions or with change effectiveness

on easy items.

85

 

 

 

 

 

 

mN.NF Fo.em we mm.mF em.wm ON mN.mF om.em m No.mN mo.mm mF N acmEpmmgF
Nm.mF ou.~m mm Fw.mF Nm.mm FF om.oF mn.me w Ne.mF Fm.em mN F pcmspmmgF
oF.mF Fm.Fm me mm.eF oo.oo wF em.o mN.mm w Fm.mF ew.um FN Focucou
am cum: 2 am cum: 2 am com: 2 am new: 2
ammmlu<m o>Fuomem m>FpumFFocF m>FpomFFmFmocaou Ozocw
FOOOF Naem--<m NOON--<O NOON--<O

 

.msmuF Ammo co mmmcm>Fuuowwm mmcmgu Fo
mmgmme LOF ammunzm mev mmeapFFum zezpm LOF mmgoom Fomv :onOF>me ecmccmum ecu cmmz--.Fw anOF

86

Table 42.--ANOVA table for study attitudes--easy.

 

 

 

Source df MS F
Treatment group 2 185.03 .67
Degree of effectiveness 2 95.73 .35
Treatment x Degree 4 165.45 .60
Error 118 273.31
Total 126 265.12

p > .05.

HO 3.18: There are no differences in SSHA composite score
study attitudes (education acceptance and teacher
approval) among direction groups and effective-
ness groups on difficult items.

The study attitudes mean and standard deviation scores were
calculated by directions groups and by effectiveness groups on diffi-
cult items. (See Table 43.)

The analysis of variance failed to reject the null hypothesis

(p > .05). There were no interactions between directions groups and

effectiveness groups; F (4, 120 df) .15, p > .05. There was no main

effect by directions; F (2, 120 df) .62, p > .05. There was no main
effect by effectiveness of change on difficult items; F (2, 120 df) =
2.19, p > .05. (See Table 44.) Study attitudes scores were not
associated with directions or with the degree of change effectiveness
on difficult items.

In summary, the SSHA study attitudes composite scores were asso-

ciated with the degree of change effectiveness on total items. There

87

 

mN.mF Fo.wm we mF.eF oo.mm FF mm.mF mm.om n ww.mF MN.Nm 0N N pawsummLF

 

 

 

 

 

OO.OF NO.NO Ow OO.OF OO.OO OF NO.OF OO.NO O NO.OF OO.OO ON F FOOEOOOLF
OF.OF FO.FO Ow O0.0F OO.NO FF FN.F OO.wO N Fw.OF ON.OO Fm FOOOOOO
Om cum: 2 OO cum: 2 OO com: 2 OO Owe: z
OFOOFFFFO--OO O>FOOOOFO O>FOOOFFOOF O>FOOOFFOLOOOOOO OOOLO
FOOOF OFOOFFFFO--OO FFOOFFFFO--OO OFOOFFFFO--wm

 

.mempF FFOUFFFFO co mmmcm>Fpumme «mango
we mmcmme com ammunsm F<mv OOOOFFFFO anaum FOF mmcoum Fomv :onOF>me esmucmpm ecm cams--.mw mFOOF

88

was no association between the composite scores and the degree of

change effectiveness for easy items and difficult items.

Table 44.--ANOVA table for study attitudes--difficu1t.

 

 

 

Source df MS F
Treatment group 2 170.60 .63
Degree of effectiveness 2 595.30 2.19
Treatment x Degree 4 40.64 .15
Error 120 271.21
Total 128 266.96

p > .05.

Ho 3.19: There are no differences in SSHA study orientation
scores among direction groups and effectiveness
groups on total items.

The mean and standard deviation study orientation scores were
computed by directions groups and by effectiveness groups on total
items. These means are presented in Table 45. In all three groups,
the means for ineffective changers were the highest means.

The analysis Of variance failed to reject the null hypothesis
(p > .05). There were no interactions between the directions groups
and the effectiveness groups; F (4, 119 df) = .25, p > .05- There was
no main effect for directions; F (2, 119 df) = .68, p > .05. There was
a main effect for effectiveness of change; F (2, 119 df) = 3.34, p <

.05. (See Table 46.) 'Huastudy orientation scores were associated with

89

 

Nm.Nm mo.ooF we

wN.NN om.mm NN mm.mF NN.FFF FF

wo.mw wF.Om FF

N acoEpOmLF

 

 

 

 

 

mm.Fm em.mm mm Nm.Nm om.mm mF mN.Fm oo.oFF N FF.Nm mF.Nm OF F acmEpOmcF
mm.wN mm.moF me Fm.mN ON.OoF NF ew.mF oo.mNF m Nw.Nm Nm.mm MN Focucou
Om cum: 2 on cum: 2 am com: 2 am can: 2
FOFOFuuoo m>Fuoowwm m>FpomFchF w>FuomFFmgmpc=oo Ozogw
FOFOF FOFOFluom Fmpoellom FOFOFuuom

 

.mEmFF quou co mmmcw>FpomFFm «mango No
wmgmme Low ummpazm Fomv conmpcmFgo anspm com mmcoum Fomv coFFOF>mu csmccmum ecu cmmz-u.me mFaOF

90

the degree of change effectiveness on total items. The more effec-

tive changers had higher mean study orientation scores.

Table 46.--ANOVA table for study orientation--total.

 

 

 

Source df MS F
Treatment group 2 652.29 .68
Degree of effectiveness 2 3209.42 3.34*
Treatment x Degree 4 240.01 .25
Error 119 958.83
Total 127 960.18

*p = .039.

HO 3.20: There are no differences in SSHA study orientation
scores among direction groups and effectiveness
groups on easy items.

The means and standard deviations were computed for the study
orientation scores by directions groups and by effectiveness groups on
easy items. These scores are presented in Table 47.

The analysis of variance failed to reject the null hypothesis
(p > .05). There were no interactions between directions and effec-
tiveness Of change; F (4, 118 df) = 1.00, p > .05. There was no main
effect for directions; F (2, 118 df) = .14, p > .05. There was no
main effect for effectiveness of change; F (2, 118 df) = .15, p > .05.
(See Table 48.) The study orientation scores were not associated with

directions or with the degree of change effectiveness, nor was there an

interaction between these variables.

91

 

Nm.Nm mo.ooF we

Nm.mN mw.Nm mN Ne.wN oo.wm m mo.mw ow.moF mF N acmEummLF

 

 

 

 

 

Fm.om 0N.FoF mm em.Fm me.FoF FF om.mm mN.NNF w mm.mN mN.om mN F ucmEpOch
em.wN mm.moF me mm.NN Ne.woF eF FN.mF mN.mm w Fw.Fm oc.moF NN Focpcou
am and: 2 am new: 2 am cum: 2 am cum: 2
Nmmmluom m>FuumNNm m>FuumFchF m>Fpuowwmcmuczou Ozogc
FOOOF NOON--OO NOON--OO NOON--OO

 

.mEmuF ammo co mmmcm>FpumNFm mmcmco No

wmgmwc cow ammuaam Fomv coFumpcmFgo Nuzpm LON mwcoum Fomv conmF>wc ugmccmpm can :Omz--.Nw mFOOF

92

Table 48.--ANOVA table for study orientation--easy.

 

 

 

Source df MS F
Treatment group 2 138.23 .14
Degree of effectiveness 2 148.37 .15
Treatment x Degree 4 974.92 1.00
Error 118 970.88
Total 126 945.313

p > .05.

HO 3.21: There are no differences in SSHA study orientation
scores among direction groups and effectiveness
groups on difficult items.

The mean and standard deviation study orientation scores by
directions groups and by effectiveness groups on difficult items are
presented in Table 49.

The analysis of variance failed to reject the null hypothesis
(p > .05). There were no interactions between directions and effec-
tiveness groups; F (4, 120 df) = .21, p > .05. There was no main
effect for directions; F (2, 120 df) = .48, p > .05. There was no main
effect for effectiveness; F (2, 120 df) = 2.56, p > .05. (See
Table 50.)

Study orientation total scores were associated with the
degree of change effectiveness on total items. The study orientation

scores were not associated with the degree of change effectiveness

on easy or on difficult items or with the directions.

93

 

 

 

 

 

 

NO.NO OO.OOF ww OO.wN OF.FFF FF FO.wN NO.NOF F ON.wO OO.OO ON N OOOEOOOEF
FN.FO OF.OOF Ow N0.00 OO.NOF OF wN.OO OO.wOF O Ow.ON OO.FO ON F OOOOOOOLF
O0.0N O0.00F Ow NF.FN OO.wFF FF OO.NF OO.NFF O ON.FO OO.wO Fm FOOOOOO
Om One: 2 OO One: 2 OO Owe: 2 OO Owe: z
FFOOFFFFO--OO O>FOOOFFN O>FOOOFFOOF OOFOOOFFOLOOOOOO OOOOO
FOOOF FFOOFFFFO--OO OFOOFFFFO--OO FFOOFFFFO--OO

 

.msepF “FOOFFFFe ee mmece>FpeeNFe emceee we

eegmee Lem umepeem Fomv :ereueeFee Nezpm Lem museum Fomv :eFueF>ee exeeceum eee ceez--.me mFeeF

94

Table 50.--ANOVA table for study orientation--difficult.

 

 

 

Source df MS F
Treatment group 2 467.53 .48
Degree of effectiveness 2 2472.83 2.56
Treatment x Degree 4 206.33 .21
Error 120 965.44
Total 128 953.56

p > .05.

In summary, the education acceptance and teacher approval
subtests and the composite score of those two subtests, study atti-
tudes, were associated with the degree of change effectiveness on
total items. The SSHA study orientation total score was also asso—
ciated with change effectiveness on total items. The SSHA scores
were not associated with change effectiveness on easy items. The
degree of change effectiveness on difficult items was associated with
the work methods subtest. It should be noted that the education
acceptance, teacher approval, and study attitudes SSHA subtest scores
and the total score, study orientation, were associated with the

degree of change effectiveness on difficult items at the .10 level.

95

Item Difficulty and
Change Effectiveness

 

 

Ho 4: There is no relationship between item difficulty and
effective answer changing.

A comparison was made between the number of students who were
effective, ineffective, and countereffective changers on easy items
and the number of students who were effective, ineffective, and
countereffective changers on difficult items. These data are presented

in Table 51.

Table 51.--Degree Of change effectiveness on easy and difficult items.

 

D'ff'
Easy Items 1 1CU]t Items Total

Countereffective Ineffective Effective

 

 

 

 

Countereffective 42 8 18 68
52.7%

Ineffective 7 2 2 11
8.5%

Effective 28 5 17 50
38.8%

Total 77 15 37 129
59.7% 11.6% 28.7% 100.0%

p < .05.

The chi-square test failed to reject the null hypothesis
(p < .05). There was no significant relationship between the effec-
tiveness of changers on easy items and on difficult items X2 (4 df) =
1.74, p > .05. Easy and difficult items were changed with the same

degree of change effectiveness.

96

Sex Differences and
Effective Changigg

 

 

HO 5.1: There are no differences in percentage of correct

changes among direction groups and sex groups.
Effective answer changing was defined as the percentage Of
correct changes. The mean percentage of correct changes for total
items for males and females in each direction group was calculated.
These are presented in Table 52. Treatment 1 group females had a much

higher mean percentage Of correct changes on total items than males.

Table 52.-~Percentage Of correct changes on total items for males
and females.

 

 

 

 

 

Male Mean Female Mean Total Mean
Group Percentage of Percentage of Percentage Of
Correct Changes Correct Changes Correct Changes
N Mean SD N Mean SD N .Mean SD
Control 17 40.39 44.95 28 40.15 40.56 45 40.24 41.57

Treatment 1

Treatment 2

16 31.25 34.35
18 53.57 36.28

24 63.95 64.38
26 58.76 37.64

40 50.87 56.23
44 56.63 36.75

 

The analysis Of variance failed to reject the null hypothesis

(p > .05).

There were no interactions between the directions groups

and the effectiveness groups on total items; F (2, 123 df) = 1.54,

p > .05.

1.60. p > .05.

There was no main effect for directions; F (1, 123 df) =

There was no main effect for effectiveness of change;

F (1, 123 df) = 2.17, p > .05.

(See Table 53.) There were no differ-

ences in the percentage of correct changes due to sex or directions.

97

Table 53.-—ANOVA table for percentage of correct changes on total
items for males and females.

 

 

 

Source df MS F
Treatment group 2 3219.16 1.60
Sex 1 4361.93 2.17
Treatment x Sex 2 3097.93 1.54
Error 123 2007.56
Total 128 2059.72

p > .05.

Ho 5.2: There are no differences in percentage of correct
changes among direction groups and sex groups on
easy items.

The mean percentage of correct changes on easy items was
computed for males and females in each Of the treatment groups. These
data are presented in Table 54. Treatment 1 males had a higher mean
percentage of correct changes on the easy items than did females.

The control group and treatment 2 group mean percentages of correct
changes for males and females on easy items appeared to be similar.

The analysis of variance failed to reject the hypothesis
(p > .05). There were nO interactions between the directions groups
and the sex groups; F (2, 123 df) = .64, p > .05. There was no main
effect for sex; F (1, 123 df) = .57, p > .05. There was a main effect
for directions; F (1, 123 df) = 5.39, p < .05. (See Table 55.) The
percentage of correct changes was affected by the directions. Easy

item changes were affected by the strong directions to go back and

revise answers.

98

Table 54.--Mean percentage of correct changes on easy items for
males and females.

 

 

Female Mean
Percentage of
Correct Changes

Total Mean
Percentage of
Correct Changes

 

 

N Mean 50

N Mean SD

 

Male Mean
G Percentage of
FOUP Correct Changes
N Mean SD
Control 17 36.60 44.18

Treatment 1

Treatment 2

13 39.93 32.52
18 62.03 43.83

28 37.79 44.00
24 19.23 47.07
26 61.53 46.19

45 37.34 43.57
37 32.65 43.22
44 61.74 44.72

 

Table 55.--ANOVA table for percentage of correct changes on easy items

for males and females.

 

 

 

Source df MS F
Treatment group 2 10483.30 5.39*
Sex 1 1116.076 .57
Treatment x Sex 2 1257.41 .64
Error 120 1943.35
Total 125 2059.25

*p < .05.

HO 5.3: There are no differences in percentage of correct

changes among direction groups and sex groups on

difficult items.

Mean percentage of correct changes on difficult items was

calculated for males and for females by directions groups. These

data are presented in Table 56.

99

Table 56.-~Mean percentage of correct changes and standard deviations
on difficult items for males and females by treatment

 

 

 

 

 

group.
Male Mean Female Mean Total Mean
G Percentage of Percentage of Percentage of
roup Correct Changes Correct Changes Correct Changes
N Mean SD M Mean SD N Mean SD
Control 17 20.58 39.76 28 30.71 43.96 45 26.88 42.25

Treatment 1 16 47.91 46.69 24 40.38 43.33 40 43.39 44.27
Treatment 2 18 36.11 40.82 26 34.29 39.881 44 35.03 39.80

 

The analysis Of variance failed to reject the hypothesis
(p > .05). There were no significant interactions between the direc-
tions groups and the sex groups; F (2, 123 df) = .46, p > .05. There
was no main effect for directions; F (1, 123 df) = 1.60, p > .05,
There was no main effect for sex; F (l, 123 df) = .004, p > .05. (See
Table 57.) The directions and sex of subjects did not affect and were

not associated with the percentage of correct changes on difficult items.

Table 57.--ANOVA table for percentage of correct changes on difficult
items for males and females.

 

 

Source df MS F
Treatment group 2 2891.13 1.60
Sex 1 7.80 .004
Treatment x Sex 2 828.09 .46
Error 123 1800.58
Total 128 1788.37

 

p > .05.

100

In summary, the degree of percentage of correct changes on
total items, easy items, or difficult items was not associated with
the sex of the subjects, nor was it affected by the directions to
change. Although there were some differences within treatment groups,
these differences were not significant. The mean percentages of cor-
rect changes for males and females did not differ significantly.

Achievement and
Answer Changing,

 

 

HO 6.1: There are no differences in the objective test scores
among direction groups and effectiveness groups on
total items.

The means and standard deviations for the final examination
by directions groups and by effectiveness groups on total items were
calculated. These means are presented in Ta 1e 58.

The analysis of variance failed to reject the "U11 hypothesis.
(p > .05). There were no interactions between the directions groups
and the effectiveness groups; F (4, 119 df) = .28, p > .05. There
was no main effect for directions; F (2, 119 df) = 1.07, p > .05.
There was no main effect for effectiveness; F (2, 119 df) = .84,

p > .05. (See Table 59.) Objective test scores were not associated

with directions or with the degree of change effectiveness on total

items.

101

 

 

 

 

 

 

FN.N Fm.¢m we mm.m mm.wm NN mN.m wF.mm FF mo.m om.wm FF N acmEumeF
ww.m NF.©m mm mm.m No.Nm mF mo.w wF.Nm N Nw.w FN.mm mF F acmEumme
em.m 00.0m me NN.© N¢.Nm NF w©.m ow.oo m wN.© 0N.mm MN Foxucou
om com: 2 am saw: 2 am cum: 2 cm cum: 2
FeueFllegeem m>FpeeFFN e>FueeFFecF e>FueemFeceuceeo eeeew
on: O8: :58 FBOF FSOF
FeueF --eceem FmeF ceez -Teceem umeF ceez -legeem umeF eeez

 

emcece Ne

.mEeuF Feueu :e mmeee>FueeNFe
eecmee ecu Fee meeeem amen e>Fpeewee New meeFueF>ee eceeeeum ece meeez--.mm eFeeF

102

Table 59.--ANOVA table for objective final--total.

 

 

 

Source df MS F
Treatment group 2 61.91 1.07
Degree of effectiveness 2 49.02 .85
Treatment x Degree 4 16.39 .28
Error 119 57.80
Total 127 56.089

p > .05.

Ho 6.2: There are no differences in objective test scores
among direction groups and effectiveness groups on

easy items.

The mean and standard deviation scores for the effectiveness

groups and the directions groups on easy items are presented in

Table 60.

The analysis Of variance failed to reject therunl hypothesis

(p > .05). There were no interactions between the directions groups

and the effectiveness groups; F (4, 118 df) = .52, p > .05. There

was no main effect for directions; F (2, 118 df) = .66, p > .05.

There was no main effect for effectiveness; F (2, 118 df)

.09,

p > .05. (See Table 61.) Objective test scores were not associated

with directions or with change effectiveness on easy items.

103

 

NN.N Fm.wm we

mm.o MN.wm oN oo.m oo.mm m mo.oF ON.om mF N acmsuemgh

 

 

 

 

 

 

eN.m Ne.ww mm Ne.FF me.Nm FF mm.N mN.Nm e om.N mm.em mN F pcesueegF
wm.e ee.em me Fm.e FN.Nm eF mm.N mN.wm w eN.m wN.em NN Feeuceu
om see: 2 am :ee: 2 am see: 2 am eeez z

xmemu-eceem e>FpeeFFm e>FFeeFFecF e>FpeeFFegeue=eu eeecw

umeF eeez Anew Nmem Nmem

FeueF -leceem pmeF eeez -ueeeem ameF :eez -neeeem FmeF :eez
.mEeFF ammo ee mmeee>FpeeFFe
emeege Fe eecmee on» Low meceem “mew e>Fueenee Lem meeFueF>ee eceeeeum eee meeez--.oe eFeeF

104

Table 61.--ANOVA table for objective final score--easy.

 

 

 

Source df MS F
Treatment group 2 39.85 .67
Degree of effectiveness 2 5.81 .1 0
Treatment x Degree 4 31.085 .52
Error 118 59.67
Total 126 57.57

p > .05.

Ho 6.3: There are no differences in objective test scores
among direction groups and effectiveness groups on
difficult items.

The mean and standard deviation objective test scores were
calculated for directions groups and for effectiveness groups on
difficult items. These means are presented in Table 62.

The analysis of variance failed to reject the null hypothesis
(p > .05). There were no interactions between the directions groups
and the effectiveness groups; F (4, 120 df) = .806, p > .05. There
was no main effect for directions; F (2, 120 df) = .97, p > .05.
There was a main effect for effectiveness of change; F (2, 120 df) =
4.78, p < .05. Objective test scores were associated with the degree of
effectiveness of changes on difficult items. (See Table 63.) There

was a difference among the mean Objective test scores for counter-

effective, ineffective, and effective changers.

105

 

 

 

 

 

 

FN.N Fw.wm we mN.m me.wm FF Fm.m oo.mm N oN.N 0N.mm 0N N pcmspemsh
mm.w Nw.mm ow mF.m mw.mm mF em.N ON.oo m me.N mm.Nm 0N F ucmspewLF
em.m om.om me em.m oo.mm FF FN.m mm.mm m mo.m mm.mm Fm Fosucou
am new: 2 am new: 2 am com: 2 am new: 2
FFeeFFNFOTTeceem e>FueeFFm e>FueeFFeeF e>FueeNNeLepeeeu eeecu
OOOF Owe: OFOOFFFFO FFOOFFFFO OFOOFFFFO
FeeeF uneceem FmeF eeez -Teceem umeF eeez uleeeem pmeF eeez

 

emeese Fe

.mEeFF pFeeFFFFe :e mmeee>FFeeFFe
eecmee esp New meceem peep e>Fueemee New meereF>ee eceeeepm use meeezln.No eFeeF

106

Table 63.-~ANOVA table for objective fina1--difficu1t.

 

 

 

Source df MS F
Treatment group 2 52.82 .97
Degree of effectiveness 2 258.86 4.78*
Treatment x Degree 4 43.63 .80
Error 120 54.12
Total 128 56.74

*p = .01.

Ho 6.4: There are no differences in grade point averages
among direction groups and effectiveness groups
on total items.
Grade point averages by directions groups and by effectiveness
groups were computed on total items. These are presented in Table 64.
The analysis of variance failed to reject the null hypothesis
(p > .05). There were no interactions between the directions and
the effectiveness of changes on total items; F (4, 119 df) = 1.62,
p > .05. There was no main effect for directions; F (2, 119 df) =
.77, p > .05. There was no main effect for effectiveness; F (2, 119 df)

= .19. p > .05. The grade point averages were not associated with

directions or with effectiveness of changes on total items.

107

 

 

 

 

 

 

No. mw.N ee om. mo.N NN mo. mN.N FF FN. ww.N FF N acmEpemLF
ow. mw.N mm oo.F em.N mF Nm. mm.N N am. om.N mF F pcmsuewLF
wN. wo.N me wN. wo.N NF mm. mN.m m cm. Nw.N MN Fogpcoo
om cum: 2 am com: 2 am cemz 2 am new: 2
FeeeF e>FFeeNFm e>FueeFFeeF e>FueeNFegeueeee eeecw
Tn<mw :emz FeFOF FeHOF FmpOF
FOFOF --wOO Owe: --<OO Owe: --<OO Owe:

 

Fe eecmee Lee Fomv mceFeeF>ee egeeceum

.meeeF Fepea ee mmece>FaeeFFe emeece
eee F<mwv memece>e ueFee eeegm peeeeum :eez-u.wm mFeeN

108

Table 65.--ANOVA table for grade point average--tota1.

 

 

 

Source df MS F
Treatment group 2 44.12 .77
Degree of effectiveness 2 11.15 .19
Treatment x Degree 4 94.09 1.65
Error 119 56.95
Total 127 57.16

p > .05.

HO 6.5: There are no differences in grade point averages
among direction groups and effectiveness groups on
easy items.

The mean grade point averages by direction groups and by

effectiveness groups on easy items were calculated. (See Table 66.)

The analysis of variance failed to reject the null hypothesis

(p > .05). There were no interactions between the directions and

the change effectiveness on easy items; F (4, 118 df) = 1.22, p > .05.
There was no main effect for directions; F (2, 118 df) = .99, p > .05.
There was no main effect for change effectiveness; F (2, 118 df) =

1.17, p > .05. (See Table 67.) Grade point averages were not associated

with directions or change effectiveness on easy items.

109

 

 

 

 

 

 

Ne. ww.N we om. Nm.N eN mm. mo.F m Ne. em.N mF N FeesueeFF
mm. Nw.N mm mm. me.N FF mm. we.N e mm. mN.N mN F peespeecF
wN. we.N me om. ew.N eF NN. NF.N e we. we.N NN Fegueeo
am new: 2 am eeez 2 am eeez 2 cm see: 2
NOON e>FpeeFFu e>FueeFFecF e>FueeFNeceuceeu eeecw
-uweo eeez NOON NOON NOON
FeueF --<eu :eez 11<ee eeez nuweo eeez

 

.mseuF Nmee ee mmeee>FueeFFe

emeece Fe eecmee New Fomv mceFueF>ee eceeeepm ece Fwewv memege>e peFee eeecm eeez--.ee eFeeF

110

Table 67.——ANOVA table for grade point average--easy.

 

 

 

Source df MS F
Treatment group 2 55.69 .99
Degree of effectiveness 2 65.79 1.17
Treatment x Degree 4 68.79 1.22
Error 118
Total 126

p > .05.

H0 6.6: There are no differences in grade point averages
among achievement groups and effectiveness groups
on difficult items.

The mean grade point averages were calculated by directions

groups and by effectiveness groups on difficult items. These are

presented in Table 68. The grade point averages were higher for

ineffective and effective changers than for countereffective changers.

The analysis of variance failed to reject the null hypothesis
(p > .05). There were no interactions between the directions groups
and the change effectiveness groups on difficult items; F (4, 120 df)
.90, p > .05. There was no main effect for directions; F (2, 120 df)
1.32, p > .05. There was a main effect for effectiveness of changes;
F (2, 120 df) = 4.26, p < .05. (See Table 69.) The grade point
averages were associated with the degree of change effectiveness on

difficult items.

111

 

 

 

 

 

 

 

No. we.N we em. NN.N FF mm. mm.N N Nm. mm.N mN N newspeeLF
mm. ew.N ow mm. me.N mF om. mo.m m NN. mF.N ON F newspeecF
wN. we.N mw No.F ew.N FF wN. mo.m m Ne. Fm.N Fm Feepeeu
om :ee: 2 am see: 2 am eeez 2 am see: 2
FFeeFFFFo e>FueeFFm e>FueeFNeeF e>FeeeNFecepeeeu eeeeu
-uwew eeez pFeeFFFFo uFeeFNFFo pFeeFFFFO
FepeF --<Ow eeez ulweu ewe: TT<Ow new:
.mEeFF FFOOFFNFe ce mmeee>FpeeFFe
emeege Fe eecmee Fem Fomv meereF>ee eceeeepm eee F<euv memece>e peFee eeecm eeez--.we eFeeF

112

Table 69.--ANOVA table for grade point average--difficult.

 

 

 

Source df MS F
Treatment group 2 72.05 1.32
Degree of effectiveness 2 231.85 4.26*
Treatment x Degree 4 49.00 .90
Error 120 54.37
Total 128 56.75

*p < .05.

In summary, student achievement, as defined by scores on the
Objective final examination and grade point averages, was associated
with the degree of change effectiveness on difficult items. However,
achievement measures were not associated with directions and the
degree of change effectiveness on total and easy items. Subjects who
had higher grade point averages and higher objective final examination

scores had a higher percentage of correct changes.

Ability and Answer Changing_

 

HO 7.1: There are no differences in SAT verbal scores among
direction groups and effectiveness groups on total
items.

The mean SAT verbal scores by directions groups and by effec-

tiveness groups on total items were computed. These means are pre-
sented in Table 70. The mean SAT scores increased with the degree

of change effectiveness.

113

 

Nm.om wo.wa we
Fm.ooF om.wa mm

Fw.woF oo.omw NN
wF.OFF wN.omw mF

wF.mw mw.mmw FF
mm.FoF Nm.me N

wm.om NN.wa FF
mw.moF Ne.www mF

N FeespeecF

F peespeecF

 

 

 

 

 

mN.NN mm.omw me NN.No mN.mFm NF Fo.oN OO.wom m FN.FN mF.mmw mN Fogucou
om :ewz 2 cm :em: 2 am com: 2 am new: 2
FepeF e>FeeeFFm e>FueeFFeeF e>FueeFFeceueeeu Oeecw
--FOOLO> Fem OOOz FOFOF FOFOF FOOOF
FeFOF uuFeaLm> F<m :emz uuFean> F<m sew: uuFean> F<m new:

 

.mEeFF Feuep ee mmeee>FpeeFFe

emeese Ne mecmee gem meceem Fmeeeem FeeLe> me New Fomv meeFueF>ee eceeeepm eee meeezll.oN eFeeF

114

The analysis of variance failed to reject the null hypothesis
(p > .05). There were no interactions between the directions groups
and the effectiveness groups; F (4, 119 df) = .35, p > .05. There
was no main effect for directions; F (2, 119 df) = 2.71, p > .05).
There was no main effect for effectiveness of changes; F (2, 119 df) =
1.05, p > .05. (See Table 71.) The SAT verbal scores were not asso-

ciated with the directions or with the degree of change effectiveness.

Table 71.--ANOVA table for mean SAT verbal-~total.

 

 

 

Source df MS F
Treatment group 2 22452.56 2.71
Degree of effectiveness 2 8713.95 1.05
Treatment x Degree 4 2936.89 .36
Error 119 8280.64
Total 127 8259.74

p > .05.

Ho 7.2: There are no differences in SAT verbal scores among
direction groups and effectiveness groups on easy
items.

The SAT verbal score means by directions groups and by effec-

tiveness groups on easy items were calculated. These are presented
in Table 72. Again, these scores gradually improved with the degree

of change effectiveness.

115

 

Nw.om wo.wa we

Nm.wm mo.Nww mN

em.NN oo.mow m

mw.moF mm.wa mF

N newspeecF

 

 

 

 

 

FF.woF mo.mme mm we.FmF om.oom FF oe.FN oo.ONm e ww.Nm me.mee mN F peeEpeeeF
mN.NN mm.omw mw Nm.mm Nw.eom wF Fo.Nm om.wa w oe.Nw mm.mww NN Feepceo
om com: 2 am new: 2 am see: 2 am new: 2
Feuee e>FueeFFm e>FueeFFecF e>FueereFeuceeu eeecw
-uFeece> me ewe: NOON NOON NOON
FeFeF --Feege> Fem eeez usFeece> me :eez -TFeece> F<m eeez

 

.msepF meme :e

mmeee>FpeeFFe emeese Fe eecmee Few Fomv mcereF>ee egeeceum ece mecoem Fence> me :eezuu.NN eFeeF

116

The analysis of variance failed to reject the null hypothesis
(p > .05). There were no interactions between the directions groups
and the effectiveness groups; F (4, 118 df) = 1.13, p > .05. There
was no main effect for directions; F (2, 118 df) = 2.40, p > .05.
There was no main effect for effectiveness of changes; F (2, 118 df)
.92, p > .05. (See Table 73.) The SAT verbal scores were not

associated with directions or the degree of change on easy items.

Table 73.--ANOVA table for SAT verba1--easy.

 

Source df MS F
Treatment group 2 19516.62 2.46
Degree of effectiveness 2 7358.90 .93
Treatment x Degree 4 8994.25 1.13
Error 118 7932.37
Total 126 8077.90

 

p > .05.

Ho 7.3: There are no differences in SAT verbal scores

among direction groups and effectiveness groups
on difficult items.

The mean SAT verbal scores by directions groups and by effec-
tiveness groups were computed. These means are presented in Table 74.
The mean SAT scores improved with the degree of change effectiveness.

The analysis of variance failed to reject the null hypothesis

(p > .05). There was no interaction between the directions and the

effectiveness of changes. There was no main effect for directions;

117

 

om.om o.wa ww
oo.moF o.mmw ow

oF.Nm m.omw FF
ow.ooF o.mmm mF

N peespeegF

oo.mm w.me N om.mw m.Fww 0N

F peeeFeecF

oo.om o.owm m om.mw m.mww ON

 

 

 

 

 

ON.NN m.omw me ow.wN N.NNm FF ow.woF o.owm m om.wm w.wNw Fm Fosucou
om com: 2 cm cum: 2 cm cow: 2 cm cum: 2
FeueF e>FpeeFFm e>FFeeFFeeF e>FueeFNeceueeee geese
TuFengm> F<m new: stuFFFFo szuFFFFo quuFFFFo
FOOOF --FOOLO> Fem Owe: --FOOLO> Fem OOOz --FOOLO> Fem OOOz

 

.msmeF OFOOFFFFO OO

mmeee>FueeFFe emeece we eeceee Few Fomv meeFueF>ee eceeeepm eee meeeem Feegm> F<m :eez-u.eN eFeeF

118

F (4, 120 df) = .59, p > .05. There was a main effect for effectiveness;
F (2, 120 df) = 7.20, p < .01. The SAT verbal scores were associated
With the degree Ofchange effectiveness on difficult items. (See Table
75.) Students who were more effective changers on difficult items

had higher SAT verbal scores. The increase of the mean SAT verbal

scores with the degree of change effectiveness was sufficient enough

to be significant.

Table 75.--ANOVA table for SAT verba1--difficult.

 

 

 

Source df MS F
Treatment group 2 18168.42 2.44
Degree of effectiveness 2 53439.40 7.20*
Treatment x Degree 4 4436.44 .60
Error 120 7422.46
Total 128 8207.81

*p = .001.

Ho 7.4: There are no differences in SAT mathematics

scores among direction groups and effectiveness
groups on total items.

The mean SAT mathematics scores by directions groups and by
effectiveness groups on total items were calculated. (See Table 76.)
The ineffective changers in each group had a higher mean SAT mathe-
matics score than the effective changers.

The analysis of variance failed to reject the null hypothesis

(p > .05). There were no interactions between the directions groups

119

 

om.moF NN.wmw we
wo.Nm mN.omw mm

oe.woF oo.omw NN
No.NoF om.Fom mF

mm.wFF em.www FF
oF.ooF oo.oFm N

wo.mFF em.wNe FF
em.om wN.me mF

N FeesueeFF

F peeeFeecF

 

 

 

 

 

FN.Fm em.wmw me mm.oFF ON.Nme NF oe.me oo.emm m mo.eN wo.omw mN Fegpeeu
om see: 2 am see: 2 am eeez 2 am see: 2
FeueF e>FpeeFFN e>FpeeFNeeF e>FueeFFecepeeeu eeece
1-:F<z me eeez FeueF FeueF FeueF
FeeeF -aszz Fem eeez -Tszz Fem eeez TTIsz Fem eeez

 

.mEeFF Fepea co mmeee>FueeNFe

emceze Fe eecmee New meeeem Fomv eereF>ee eceeeeum ece sees er<zv meFueEesueE F<m-u.eN eFeeF

120

and the effectiveness groups; F (4, 119 df) = .36, p > .05. There
was no main effect for directions; F (2, 119 df) = .27, p > .05.
There was no main effect for effectiveness; F (2, 119 df) = .80,

p > .05. The SAT mathematics scores were not associated with directions

or the degree of change effectiveness on total items. (See Table 77.)

Table 77.--ANOVA table for SAT mathematics--tota1.

 

 

 

Source df MS F
Treatment group 2 28.15 .27
Degree of effectiveness 2 82.65 .80
Treatment x Degree 4 37.68 .37
Error 119 102.36
Total 127 98.57

p > .05.

HO 7.5: There are no differences in SAT mathematics scores
among direction groups and effectiveness groups on
easy items.

The means for SAT mathematics scores by directions groups

and by effectiveness groups on easy items were calculated. These are
presented in Table 78. The mean scores for the strong-directions
group (treatment 2) gradually increased, but the mean scores for the
other two groups were not consistent.

The analysis of variance failed to reject the null hypothesis

(p > .05). There were no interactions between directions groups and

effectiveness groups on easy items; F (4, 118 df) = 1.64, p > .05.

121

 

mm.moF NN.www ww
Nm.mm Nm.mwe mm

oo.NoF mm.NNw mN
ON.ow oo.oom FF

om.mFF oo.omw m
Nw.mm om.Nmm w

oo.oFF oo.wom mF
Nm.FNF ow.mow MN

N peesueeFF

F peesueegF

 

 

 

 

 

FN.Fm oe.wmw me Nw.eN wF.Nme wF oe.om oo.mow w mo.NNF om.Nom NN Fecuceu
om :em: 2 am new: 2 am new: 2 am cum: 2
FepeF e>FpeeFFm e>FueeFFeeF e>FueeFFeeeueeeu eeeco
-uszz F<m new: NOON NOON Amen
FMHOF uuzF<z F<m com: unzF<z F<m :ewz uazF<z N<m :ewz

 

.msepF Nmee :e mmeee>FFeeFFe

emeece Ne eegmee Lew meceem Fomv eeFueF>ee eceeeepm ece ewes Fwazv meFueEeepeE F<m11.NN eFeeF

122

There was no main effect for directions; F (2, 118 df) = .10, p > .05.
There was no main effect for change effectiveness; F (2, 118 df) =
.14, p > .05. (See Table 79.) The SAT mathematics scores were not

associated with directions or with change effectiveness on easy items.

Table 79.--ANOVA table for SAT mathematics--easy.

 

 

 

Source df MS F
Treatment group 2 10.53 .13
Degree Of effectiveness 2 13.95 .14
Treatment x Degree 4 162.29 1.64
Error 118 98.70
Total 126 98.01

p > .05.

Ho 7.6: There are no differences in SAT mathematics scores
among direction groups and effectiveness groups on
difficult items.

The means and standard deviations for SAT mathematics sub-
tests were calculated by directions groups and by effectiveness groups
on difficult items. These are presented in Table 80. The mean SAT
mathematics scores gradually increased with the level of change effec-
tiveness. Again, the ineffective changers scored the highest mean.

The analysis of variance failed to reject the null hypothesis
(p > .05). There were no interactions between the directions groups

and the effectiveness-Of-change groups; F (4, 120 df) = .64, p >>.05.
There was no main effect for directions; F (2, 120 df) = .42, p > .05.

123

 

 

 

 

 

 

O0.00F ON.wOw ww OO.wFF Ow.ONO FF Ow.OOF O0.000 F OF FOF Ow.OOw ON N OOOEOOOOF
OO.OO O0.00w Ow OO.wO O0.000 OF O0.00F OO.ONO O ON ON O0.0ww ON F OOOEFOOLF
ON.FO OO.wOw Ow ON.OFF ON.NFO FF NN.NOF OO.OOO O OO.OO ON.NNw Fm FOOOOOO
Om Owe: 2 OO OOOz 2 OO Owe: 2 OO Owe: z
FFOOFFFFO O>FFOOFFN O>FOOOFFOOF O>FOOOFFOLOOOOOO OOOLO
--=F<z Fem Owe: FFOOFFFFO OFOOFFFFO OFOOFFFFO
FOFOF --IF<z Fem OOOz --:F<z Fem Owe: --:F<z Fem OOOz

 

.mEeFF uFeeFFNFe :e mmece>FeeeFFm

emcese Ne eecmee Lew meceem Fomv :eFFeF>ee eceeeeum eee sees Fszzv meFueEegaeE F<mnu.ow eFeeF

124

There was a main effect for degree of change effectiveness; F
(2, 120 df) = 6.91, p < .01. The SAT mathematics scores were
associated with the degree of change effectiveness. (See Table 81.)
The students who were more effective changers had higher SAT mathe-

matics scores.

Table 81.--ANOVA table for SAT mathematics--difficult.

 

 

 

Source df MS F
Treatment group 2 38.77 .42
Degree of effectiveness 2 633.95 6.91*
Treatment x Degree 4 58.71 .64
Error 120 91.628
Total 128 97.81

*p = .001.

In summary, SAT verbal and mathematics subtest scores were
associated with the effectiveness Of changes on difficult items.
Effective answer changers on difficult items had higher mean mathe-
matics and verbal test scores than countereffective changers, but
ineffective changers had the highest mean SAT scores. The same rela-
tionships were not found for change effectiveness on total items or

easy items.

Sim

The major findings reported within the limitations Of the

study in this chapter were:

125

1. The directions to change answers did not affect the
number of answers changed.

2. The directions to change answers affected the percentage
of correct changes on easy items. However, that effect was not main-
tained when difficult items or easy items were considered.

3. The SSHA work methods scores were associated with the
effectiveness of changes on difficult items. The SSHA subtests, educa-
tion acceptance and teacher approval, were associated with answer chang-
ing effectiveness on total items. The SSHA composite score, study
attitudes, was also associated with the answer changing effectiveness
on total items, as was the SSHA total score, study orientation. The
SSHA scores were not associated with effective answer changing on easy
items. The SSHA subtest delay avoidance and effective answer changing
were not associated with answer changing effectiveness on total, easy,
or difficult items.

4. Item difficulty was not related to effective answer
changing.

5. Neither males nor females were more effective answer
changers or were more affected by directions.

6. Student achievement, as measured by the objective final
scores and the subjects' grade point averages, was associated with
the answer changing effectiveness on difficult items. Student achieve-
ment was not associated with answer changing effectiveness on easy or
total items or by directions to change answers.

7. Student ability, as measured by the SAT verbal and mathemat-

ics subtest scores, was associated with the degree of change effectiveness

126

on difficult items. Student ability was not associated with the
degree of change effectiveness on easy items or on total items.
In Chapter V, a summary of the study, a discussion of the

data, and recommendations for future study are presented.

CHAPTER V

SUMMARY, DISCUSSION, AND RECOMMENDATIONS

Summary

The purpose Of this study was to investigate the effects of
three different sets of directions to change answers on the answer
changing behavior of college students who were taking an objective
final examination in a required religion course in a liberal arts
college in the northeastern United States in fall 1982. The study
also investigated the differences in the Survey of Study Habits and
Attitudes (SSHA) scores among different direction groups and among
different change effectiveness groups. Also considered were differences
in student achievement and ability among direction groups and change
effectiveness groups. The relationship between item difficulty and
answer changing was also investigated.

Major findings were that the directions to change answers
affected answer changing behavior on easy items. The SSHA work
methods score was associated with the effective changing of difficult
items. The study attitudes subtests and study orientation total SSHA
scores were associated with effective changing on total items.

Grade point averages, Objective test scores, and SAT verbal
and mathematics subtest scores were associated with effective answer

changing on difficult items. NO ability measures and no SSHA subtest

127

128

scores were associated with effective answer changing on easy items or
on total items. Neither males nor females were more effective answer
changers, nor were there relationships between the level of item dif-

ficulty and effective answer changing. (See Table 82.)

Limitations

 

Discussion of the data cannot be undertaken without consider-
ing the many limitations of the study. When the investigator chose to
collect these data on this population, the decision was made with
some awareness of the potential difficulties: possible faking of
responses on the SSHA, a lengthy final examination that might inhibit
the perceived opportunity for answer revision, the unknown level Of
difficulty of the Objective final exam, and the possibility that stu-
dents would not read the directions concerning changing responses.

Students did find the test lengthy. Many stayed past the
hour of the examination to complete the test. The difficulty level
Of the final examination as measured by the standard deviation (7.52)
did reveal the test to be somewhat easy. In addition, there was some
concern that some items did not discriminate between the high—scoring
and low-scoring subjects. (See Appendix C.) Subjects blotted out

approximately 8 percent of their original responses before marking

the final answers, suggesting that some subjects might not have read
the directions. Although some researchers have assigned such revi-
sions at random and have used a wrong-tO-wrong category, this
investigator chose to assign the blotted wrong answers to one of

two categories: right-to-wrong or wrong-to-right.

129

Table 82.--Summary of research findings for null hypotheses.

 

Null Hypotheses

Findings

 

Ho 1

Ho 2.

Ho 2.

HO 2.

HO 3.

Ho 3.

Ho 3.

HO 3.

HO 3.

Ho 3.

HO 3.

There is no effect due to directions on

the number of answers changed.

There is no effect due to directions on

the percentage of correct changes on
total items.

There is no effect due to directions on

the percentage of correct changes on
easy items.

There is no effect due to directions on

the percentage of correct changes on
difficult items.

There are no differences in SSHA delay
avoidance scores among direction
groups and effectiveness groups on
total items.

There are no differences in SSHA delay
avoidance scores among direction
groups and effectiveness groups on
easy items.

There are no differences in SSHA delay
avoidance scores among direction
groups and effectiveness groups on
difficult items.

There are no differences in SSHA work
methods scores among direction groups
and effectiveness groups on total
items.

There are no differences in SSHA work
methods scores among direction groups
and effectiveness groups on easy
items.

There are no differences in SSHA work
methods scores due to effectiveness
Of answer changing on difficult
items.

There are no differences in SSHA
education acceptance scores due to
effectiveness of answer changing
on total items.

Failure to reject

Failure to reject

Rejected, p = .006

Failure to reject

Failure to reject

Failure to reject

Failure to reject

Failure to reject

Failure to reject

Rejected, p = .048

Rejected, p = .01

130

Table 82.--Continued.

 

Null Hypotheses

Findings

 

Ho 3.8

Ho

HO

HO

Ho

Ho

Ho

Ho

Ho

There are no differences in SSHA edu-
cation acceptance scores among
direction groups and effectiveness
groups on easy items.

There are no differences in SSHA
education acceptance scores among
direction groups and effectiveness
groups on difficult items.

There are no differences in SSHA
teacher approval scores due to
effectiveness Of answer changing
on total items.

There are no differences in SSHA teacher
approval scores among direction groups
and effectiveness groups on easy

items.

There are no differences in SSHA teacher
approval scores among direction groups
and effectiveness groups on difficult
items.

There are no differences in SSHA com-
posite score study habits (delay avoid-
ance and work methods) among direction
groups and effectiveness groups on
total items.

There are no differences in SSHA com-
posite score study habits (delay avoid-
ance and work methods) among direction
groups and effectiveness groups on

easy items.

There are no differences in SSHA com-
posite score study habits (delay avoid-
ance and work methods) among direction
groups and effectiveness groups on
difficult items.

There are no differences in SSHA study
attitudes scores due to effective-
ness of answer changing on total
items.

Failure to reject

Failure to reject

Rejected, p = .032

Failure to reject

Failure to reject

Failure to reject

Failure to reject

Failure to reject

Rejected, p = .023

131

Table 82.--Continued.

 

Null Hypotheses

Findings

 

HO 3.17

Ho

Ho

HO

Ho

Ho

Ho

HO

Ho

Ho

HO

.18

.19

.20

.21

There are no differences in SSHA com-
posite score study attitudes (education
acceptance and teacher approval) among
direction groups and effectiveness
groups on easy items.

There are no differences in SSHA com-
posite score study attitudes (education
acceptance and teacher approval) among
direction groups and effectiveness
groups on difficult items.

There are no differences in SSHA

study orientation scores due to effec-
tiveness Of answer changing on total
items.

There are no differences in SSHA study
orientation scores among direction
groups and effectiveness groups on
easy items.

There are no differences in SSHA study
orientation scores among direction
groups and effectiveness groups on
difficult items.

There is no relationship between item
difficulty and effective answer changing.

There are no differences in percentage of
correct changes among direction groups
and sex groups on total items.

There are no differences in percentage of
correct changes among direction groups
and sex groups on easy items.

There are no differences in percentage
of correct changes among direction
groups and sex groups on difficult
items.

There are no differences in the objec-
tive test scores among direction
groups and effectiveness groups on
total items.

There are no differences in the objec-
tive test scores among direction groups
and effectiveness groups on easy

items.

Failure to

Failure to

reject

reject

Rejected, p = .039

Failure to

Failure to

Failure to

Failure to

Failure to

Failure to

Failure to

Failure to

reject

reject

reject

reject

reject

reject

reject

reject

132

Table 82.-~Continued.

 

Null Hypothesis

Findings

 

HO 6.3

Ho 6.4

Ho 6.5

Ho 6.6

Ho 7.1

Ho 7.2

HO 7.3

HO 7.4

Ho 7.5

HO 7.6

There are no differences in objec-
tive test scores due to effective-
ness of answer changing on difficult
items.

There are no differences in the grade
point averages among direction groups
and effectiveness groups on total
items.

There are no differences in the grade
point averages among direction groups
and effectiveness groups on easy
items.

There are no differences in grade
point averages due to effectiveness
of answer changing on difficult
items.

There are no differences in the SAT
verbal scores among directions
groups and effectiveness groups

on total items.

There are no differences in the SAT
verbal scores among directions
groups and effectiveness groups on
easy items.

There are no differences in the SAT
verbal scores due to effectiveness
Of answer changing on difficult
items.

There are no differences in the SAT
mathematics scores among direction
groups and effectiveness groups on
total items.

There are no differences in the SAT
mathematics scores among direction
groups and effectiveness groups

on easy items.

There are no differences in SAT
mathematics scores due to effec-
tiveness Of answer changing on
difficult items.

Rejected, p = .01

Failure to reject

Failure to reject

Rejected, p = .016

Failure to reject

Failure to reject

Rejected, p = .001

Failure to reject

Failure to reject

Rejected, p = .001

 

133

The SSHA was administered the last class period before the
final examination. It was thought that few students would miss this
class. Contrary to expectations, approximately 40 subjects out Of
147 who were still enrolled in the class did not take the SSHA
during class. This meant that the SSHA had to be administered at the
time of the final examination. This had some potential problems.
First, it made the total test time 35 minutes longer for some subjects.
This might have inhibited their changing patterns. It also might have
encouraged subjects to rush through the SSHA, giving less accurate or
meaningful self-reports. Several subjects were dropped from the study
due to lack of completion of the SSHA. They accounted for a portion Of
the 18 subjects who were eliminated because of insufficient data. It
should be noted that more students were lost to the study because Of
missing SAT scores than missing SSHA scores. Finally, taking the SSHA
on the day Of the final meant that the investigator was present at
the time of the final examination. This might have alerted the sub-
jects to the potential existence of an experiment. Although a control
group was used in the design, this still could have affected the pat-
terns Of changing.

The last limitation of this study has to do with the missing
SAT scores on transfer students and the subsequent exclusion of that
group from the study. The effects of this on the study are really
not known. The data concerning the equivalency Of the groups showed
that all groups were equal. (See Appendix D.) However, treatment 2
did have SAT verbal scores of 457 as compared with 490 for the control
and 492 for treatment 1. When the effect on the achievement and ability

134

measures of answer changing effectiveness for higher achieving students
is considered, the lower SAT scores for treatment 2 might have influ-

enced the results of the study.

Discussion

 

Analysis of the data indicated that the directions to change
answers given to the three groups did not affect the number of
changes made. There was, however, a higher mean number of changes
in the strong-directions-to-change group (treatment 2) than in the
caution group (treatment 1) or the nO-specific-directions group
(control). These differences were not significant. (See Tables
1 and 2, p. 47.)

The subjects in this study changed an average Of 2.5 answers.
Mercer (1978), in a review of 23 studies, noted a range of 1.1 to
17.0 mean changes per student and a median of 4.4 changes per student.
In studies where directions to change have been incorporated into the
design of the study, higher mean changes have been noted. Jacobs
(1972) reported 14.7 mean changes per student, Foote and Belinky
(1972) reported 8.8 and 4.7 mean changes, Smith and Moore (1976)
reported 13.2 and 17.0 mean changes, and Sutton (1982) reported 5.2
mean changes. Subjects in the present study changed fewer responses
than would have been expected. The range of changes in the present
study was 0-12 changes.

Examination of the raw data indicated that 93 percent of
the subjects in the present study changed responses. Mercer (1978)

noted a range of 61-100 percent of subjects changing some responses

135

in her review. Mueller and Wasser (1977) noted a median of 89 per-
cent of the students changing responses in the seven studies they
reviewed. Findings concerning percentage Of students changing answers
appear to be in keeping with previous research.

Of the 20 subjects who did not revise any answers in the
present study, 10 were in the no—specific-directions group (22 percent
of the control group), 6 were in the caution group (15 percent of
treatment 1), and 4 were in the strong-direction group (9 percent of
treatment 2). Because data were unavailable to ascertain these sub-
jects' behavior on previous testing, the findings of Stoffer, Davis,
and Brown (1977)--that some students appear to have a nO-change
response-set--cannot be confirmed.

Jarrett (1947) noted that some students change a larger pro-
portion Of answers than others. In this study, there were five sub—
jects in the control group who accounted for 35 percent of the answer
changing. Three subjects changed 27 percent of the total responses
in the treatment 1 group, whereas three subjects changed 16 percent
of the responses in the treatment 2 group. In the no-specific-change
group the mode for change was 1, in the caution group the mode was 1,
and in the strong-direction group the mode was 2. (See Appendix E.)

Both the no-specific-directions group and the caution group ,
had more subjects who changed no answers and subjects who accounted
for a greater percentage of changes. The mode was higher for the
strong-direction group. Although the mean changes were not signifi- /’
cantly different among the groups, an examination of the data indi-

cated that there were some different patterns within the groups.

136

One explanation for the possible lower rate of changes in
the present study is the difficulty level of the objective final on
which the answer changing was Observed. The examination had a mean
of 55.84 and a standard deviation of 7.52. The examination might not
have been difficult enough to encourage much revision. Another pos-
sible explanation was that the test was lengthy enough that students
did not perceive that they had the time to revise their answers.

Because the test was thought to be too easy to provoke a
higher rate Of answer changing, changes were examined on easy items
and difficult items as well as total items. The thought was that
patterns Of answer changing might differ on easy and difficult items.

Analysis Of the data concerning directions to change answers
and the degree of change effectiveness indicated that strong direc-
tions to change answers significantly affected answer changing on
easy items. Students who were told to go back and revise answers
profited from revision of easy items. Analysis Of data also
indicated that there were no benefits for the strong-direction group
on difficult or total items. One possible explanation for this find-
ing could be that when the students were instructed to go back and
revise answers, they found and corrected their clerical errors on
easy items. And they found and corrected errors from misreading
questions and perhaps gained points from information from other items.
The answers that were more difficult were not revised as effectively
because subjects simply did not know the correct answers. Students
in the control and treatment 1 groups might have specifically gone

back and checked selectively those answers that were difficult for

137

them. Perhaps this caused them to miss the opportunity to revise
their clerical errors and errors caused by misreading the easier
questions or to pick up points on the easier items by using knowledge
they gained on subsequent questions. Sutton (1982) found more benefit
for his adult subjects from such clerical types of revisions.

Table 83 in Appendix F presents data concerning the number Of
students who gained, lost, or were not affected by changes. Table 84
presents data concerning mean positive and negative changes by group.

Subjects in all three treatment groups who had higher work
methods SSHA scores were the superior answer changers on difficult
items. A possible explanation was that those with more self-reported
study skill knew more and could therefore revise answers on difficult
items more easily. There were no other significant effects on SSHA
scores by effective changing of difficult items. Apparently, student
self-reported time-management strategies and attitudes toward teachers
and education were not associated with changing difficult answers effec-
tively. It should be noted that there were no interactions among the
work methods score, the degree of change effectiveness, and the direc-
tions to change answers.

Answer changing effectiveness on total items was asso-
ciated with education acceptance, teacher approval, and the
composite study attitudes SSHA scores. Students with better atti-
tudes toward schooling did a more effective job of changing answers
than those who did not express positive attitudes toward schooling.
One explanation for this finding is that those with more positive

attitudes possibly found it easier to spend more time attempting to

138

go back and proofread their examinations. The assumption here is that
the subjects believed that answer revision would increase their scores.
The finding concerning attitude in the present study supported the find-
ings of McMorris, Lichtstein, and Hoops (1983), who reported a rela—
tionship between a positive attitude toward the course and course

tests and effective answer changing. A similar finding was reported

by Sitton, Adams, and Anderson (1980).

The total score for the SSHA, study orientation, was associated
with the degree Of change effectiveness on total items. Countereffec-
tive changers did have the lowest study orientation scores. (See
Table 45, p. 89.) The highest scores on study orientation, however,
were held by those subjects who were ineffective changers. Examina-
tion of the data revealed that the number of students in this inef-
fective group was not great enough to generate meaningful implications.
This phenomenon, however, suggested a need for further study. Possibly
there was a group of students who had studied well, and when they
attempted to change answers they read too much into the questions and
changed right answers to wrong ones. This group could also have reached
the "ceiling" effect on the examination, and their changes could, there-
fore, have been from wrong-tO-wrong. Even though the ineffective
changers had the highest mean study orientation scores, the study orien-
tation scores were significantly higher for the effective changers than
for countereffective changers. 'There was no interaction with the treat-
ment. (See Table 44, p. 88.)

It should be noted that the SSHA scores were not associated

with effective answer changing on easy items. This finding was of

139

particular interest because strong directions to change answers did
significantly affect answer changing on easy items. High scores on
self-reported study habits and attitudes were not associated with effective
changes on easy items. This finding suggested a need for further
research into the reasons that students revised answers. Were the
revisions made by the strong-directions group a result of revision Of
clerical errors? If this is true, it appears that directions to
revise answers tended to help some students to be test-wise--to make
use of the format Of the test and the testing situation to improve
their scores.

Neither males nor females were more effective answer changers.
This finding replicated the finding of Mueller and Wasser (1977), who
reported that sex differences were usually present in studies where
one sex or the other had a higher mean test score. In this study, there
were no differences between the mean test scores of males or females.
Data analysis also indicated that there was no difference in effective
answer changing on difficult or on easy items. This finding replicated
the finding of Ballance (1975).

Data analysis indicated that Objective test scores, grade point
averages, and SAT scores were associated with the degree of effective
answer changing on difficult items. Achievement and ability measures
were not associated with effectiveness of changes on easy or total
items. The finding that higher-achieving students were more effective
answer changers may indicate that higher-ability or higher-achieving
students are more likely to be more effective answer changers on the

more difficult test questions, but their effectiveness is not superior

140

on the easier item changes. Jacobs (1972) and Vidler and Hansen
(1980) did not find a relationship between effective changing on
difficult items and high achievement. Others who have investigated
achievement and ability measures and effectiveness of answer revision
have not found relationships between these variables when total item
changes have been studied (McMorris 8 Leonard, 1976; Pascale, 1974;
Reiling 8 Taylor, 1972; Stoffer et a1., 1977). This study replicated

these findings.

Implications of the Study,

 

Within the limitations of population and methodology of this
study, implications are suggested for educators. Educators should
encourage students to go back. u: check their answers on objective
tests. Answers that have been incorrectly marked, misread, or that
now appear incorrect should be revised. Such revisions will probably
improve scores.

A college instructor of a study skills course might include
the following information concerning answer changing: "Go back and
revise your answers; you will improve your scores. However, the
improvement will probably be from the changes that you have made on
easy items. You will not be likely to make as many points from revi-
sion of difficult items if you do not have good study skills and if you
do not usually score well on classroom tests and standardized tests,
such as the SAT. If you go through the test a second time, particu-
larly be aware Of the items on which you feel you have made a clerical

error, you think you have misread, and items where you now know the

141

answer because a subsequent item gave you a clue about the answer.
These will most likely be the easier test items. Carefully consider
changing items on which you are still not sure between two answers.
These are probably the more difficult items, and your chances of chang-
ing them from a wrong to a right answer are more related to your level

of achievement and study ability."

Recommendations for Future Research

 

The findings of the present study justify further investiga-
tion into the effects of directions to change answers on answer
changing behavior of students.

1. Replication of the design of the present study using a
course test that is not as lengthy and that has a higher difficulty
and discrimination level is suggested. Further, replication with a
younger population is also needed. Such research might attempt to
test for the developmental stage at which children are capable Of
reversing answers (Piaget, 1972). Replication of the finding for
effective revision of easy items is needed.

2. Investigation into the reasons students revise answers
is also suggested. Interviewing students immediately after an examina-
tion for explanations of their revisions might give insight into the
present finding concerning the effectiveness of students on easy
items in the strong-directions groups.

3. Investigation into the reasons that students in the
ineffective answer changing category achieved high scores on different

achievement and ability measures but lost points changing is also

142

suggested. The small number of ineffective changers in this study
inhibits the report of meaningful implications.

4. Investigation into the students‘ interpretation of the
words "change with caution" is needed. In this study, the behavior
of subjects in the caution group was more similar to subjects in the
nO-specific-directions group than it was similar to the strong-
directions group. Perhaps "change with caution" in a testing situa-
tion has a different connotation to students than it does to the
instructors and study-skills-textbook writers.

5. Investigation into personality factors associated with
nonchangers, countereffective changers, and changers who change larger

proportions Of answers is also needed.

APPENDICES

143

APPENDIX A

LETTER TO STUDENTS WHO TOOK THE SSHA
WITH THE FINAL EXAMINATION

144

145

Dear Student,

Because you were not present in class on Monday to complete
the Survey of Study Habits and Attitudes, you need to do that now.
The retention committee of Gordon College is interested in helping

you do well at Gordon.

Instructions:

 

1. Use the special pencil provided.

2. Fill in the information on the answer sheet. On the answer
sheet, print your name, your age, your sex (M or F), and
the date, which is 11/10/82. The Form is Form 9, Now
circle your year in school.

3. Now read the directions in the answer booklet on the first
page and begin working.

Turn in the Survey of Study Habits and Attitudes booklet,

answer sheet, and pencil with your final examination.

Thank you,

Pat Wilson

PW:jr

APPENDIX 8

TEST DIRECTIONS TO SUBJECTS

146

147

CONTROL

Name:

I.D.:

Bible 101 UNDERSTANDING THE OLD TESTAMENT

Final Examination: Part B,,November 1982

This is Part B of your final examination. It consists of 40 multiple-
choice questions and 40 true or false statements, worth one point each.
This section will account for 20% of your final course grade.

INSTRUCTIONS

 

Please read the instructions carefully, and answer the questions asked.
If you find that you have made an error, cross out (X) the wrong answer
and write another answer.

148

TREATMENT 1

 

Name:

I.D.:

Bib1e 101 UNDERSTANDING THE OLD TESTAMENT

Final Examination: Part 8, November 1982

This is Part B of your final examination. It consists of 40 multiple-
choice questions and 40 true or false statements, worth one point each.
This section will account for 20% Of your final course grade.

INSTRUCTIONS

 

If you feel that you have made an error, cross out (X) the answer you
believe is wrong and write another answer. Change with caution answers
you believe are incorrect.

 

149

TREATMENT 2

Name:

I.D.:

Bib1e 101 UNDERSTANDING THE OLD TESTAMENT

Final Examination: Part 8, November 1982

This is Part B of your final examination. It consists of 40 multiple-
choice questions and 40 true or false statements, worth one point each.

This section will account for 20% of your final course grade.

INSTRUCTIONS

If you find that you have made an error, cross out (X) the answer you
believe is wrong and mark another answer. After you have worked

through the test once, go back and change the answers you now believe
to be incorrect. Research indicates that revising answers will improve

your score on this test.

GO BACK AND REVISE YOUR ANSWERSII

APPENDIX C

ITEM DIFFICULTY AND DISCRIMINATION DATA
FOR THE OBJECTIVE TEST

150

151

Discrimination Data

 

 

 

 

 

 

 

Item Difficulty, Discrimination Item Difficulty_ Discrimination
1 12 ll 41 60 28
2 75 23 42 9 8
3 61 25 43 23 37
4 71 32 44 22 -12
5 23 28 45 59 6
6 42 49 46 18 31
7 15 20 47 32 54
8 20 34 48 10 8
9 29 51 49 48 28

10 27 38 50 10 O
11 7 17 51 21 15
12 omitted 52 7 23
13 21 22 53 12 20
14 32 49 54 48 —48
15 13 3 55 30 38
16 18 23 56 23 26
17 10 23 57 10 17
18 O O 58 95 - 6

19 5 14 59 6O - 3

20 68 43 60 12 3

21 20 51 61 24 O

22 23 48 62 4 6

23 12 8 63 72 - 9

24 72 54 64 19 4O
25 36 4O 65 l8 17
26 32 6 66 11 11
27 32 23 67 62 20
28 ll 20 68 48 40
29 24 23 69 8 14
30 66 63 7O 72 31
31 65 62 71 12 23
32 14 28 72 15 17
33 52 26 73 28 54
34 13 31 74 2 6
35 15 34 75 18 34
36 35 4O 76 6O 25
37 30 51 77 18 34
38 33 46 78 17 - 6
39 10 17 79 35 12
4O 12 3 80 12 22

 

APPENDIX D

GROUP EQUIVALENCY DATA

152

153

Group Equivalency for SAT--Verbal

 

 

 

Source df MS
Between groups 2 17207.41
Within groups 125 8116.58
Total 127
p = .12

Group Equivalency for SAT-—Mathematics

 

 

 

Source df MS
Between groups 2 10.92
Within groups 125 99.97
Total 127
p = .89

Group Equivalency for Study Orientation

 

 

Source df MS
Between groups 2 215.68
Within groups 126 965.27
Total 128

 

p = .80

154

Group Equivalency for Objective Final Scores

 

 

 

Source df MS F
Between groups 2 38.21 .67
Within groups 126 57.03
Total 128
p = .51

Group Equivalency for Grade Point Averages

 

 

Source df MS F
Between groups 2 39.71 .69
Within groups 126 57.02
Total 128

 

p = .50

APPENDIX E

NUMBER OF CHANGES PER SUBJECT BY GROUP

155

156

Control Group (N = 45) Treatment 1 (N = 40)
. Number Of . Number of

§H§l§9§- Changes §E§l§££~ Changes
011 9 021 2
012 3 022 1
013 1 023 4
014 2 024 3
015 0 025 3
016 1 026 12
017 6 027 4
018 0 028 l
019 5 029 6
110 1 210 0
111 O 211 4
112 O 212 1
113 4 213 O
114 0 214 1
115 l 215 O
116 l 216 3
117 3 217 0
118 0 218 5
119 O 219 3
120 5 220 5
121 0 221 2
122 3 222 8
123 7 223 1
124 1 224 3
125 2 225 O
126 0 226 1
127 2 227 3
128 5 228 3
129 1 229 l
130 1 230 3
131 3 231 2
132 1 232 2
133 4 233 1
134 1 234 2
135 9 235 l
136 2 236 l
137 4 237 0
138 4 238 2
139 1 239 2
140 2 240 2
141 7
142 l
143 O
144 2
145 2

157

Treatment 2 (N = 44)

. Number of
§991§££— Changes

031
O32
O33
O34
035
036
037
038
039
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
334
335
336
337
338
339
340
341
342
343
344

mmNmewmmOO—‘h-‘(flwwmNCDOOOdeNNNNdNNNhNNOWNm—‘(flm

APPENDIX F

TABLES 83 AND 84

158

159

Table 83.--Number of students who gained points, lost points, or were
not affected by changes (neutral), by group.

 

 

 

Group N Lost Gained Neutral
Control 45 7 16 22
Treatment 1 40 10 12 18
Treatment 2 44 13 19 12

Total 129 30 47 52

 

Student gain-to-loss ratio = 1.56:1.

Table 84.--Student means and standard deviations (SD) for positive (+)
and negative (-) changes, by group.

 

 

Group N Mean + Changes SD Mean - Changes SD
Control 45 1.31 1.66 1.55 1.41
Treatment 1 40 1.30 1.44 1.15 , 1.44

Treatment 2 44 1.90 1.44 1.18 1.22

 

REFERENCES

160

REFERENCES

Archer, N. S., 8 Pippert, R. (1962). Don't change the answer!
Clearing House, 31, 39-41.

 

Armstrong, W. H. (1975). Study tips, Woodbury, NY: Barron's Education
Series, Inc.

 

Ballance, C. T. (1975). An investigation of the relations of the altera-
tion of responses to objective test questions to selected student and
test item characteristics. Dissertation Abstracts International, gg,
5220A (University Microfilms No. 76-4592).

 

Ballance, C. T. (1977). Students' expectations and their answer-
changing behavior. Psychological Reports, 41, 163-166.

 

Ballance, C. T., 8 Gentzel, W. E. (1980). Effects of several scoring
techniques on answer-changing by students. Psychological Reports,
46, 653-654.

Bath, J. A. (1967). Answer-changing behavior on Objective examinations.
Journal of Educational Research, 61, 105-107.

 

Beck, M. D. (1978). The effect of item response changes on scores on
an elementary reading achievement test. Journal of Educational
Research, 21, 153-156.

 

Berriens, F. K. (1939). Are first impressions best on objective tests?
School and Society, §Q,_319-320.

 

Best, J. B. (1979). Item difficulty and answer changing. Teaching of
Psychology, 6, 228-230.

 

 

Brown, W. F., 8 Holtzman, W. H. (1965). Survey of stggy habits and
attitudes. New York: Psychological Corporation.

Brown, W. F., 8 Holtzman, W. H. (1967). Survey of study habits and
attitudes: SSHA manual. New York: Psychological Corporation.

 

Clark, C. A. (1962). Should students change answers on Objective tests?
Chicago Schools Journal, 43, 382-385.

 

Copeland, D. A. (1972). Should chemistry students change answers on
multiple-choice tests? Journal of Chemical Education, 49, 258.

161

162

Crocker, L., 8 Benson, J. (1980). Does answer-changing affect test
quality? Measurement and Evaluation in Guidance, lg, 233-239.

Cromack, T. R. (1981, April). Some theoretical implications of chang-
jgg answers on multiple-choice tests. Paper presented at the meéting
of the New EnglandTEducational Research Organization, Lenox, MA.

 

Cummings, O. W. (1981). Impact Of response changes on objective test
characteristics and outcomes for junior high students. Measurement
and Evaluation in Guidance, 14, 32-37.

 

Davis, R. E. (1975). Changing examination answers: An educational
myth? Journal of Medical Education, §Q, 685-687.

Deese, J. (1959). In 0. K. Buros (Ed.), The fifth mental measurement
yearbook reviews: Reading7study skills (p. 272). Highland, NJ:
Gryphon Press.

 

 

Ebel, R. L. (1965). Measuring educational achievement. Englewood
Cliffs, NJ: Prentice-Hall.

 

Ebel, R. L. (1972). Essentials Of educational measurement (2nd ed.).
Englewood Cliffs, NJ: Prentice-Hall.

 

Eysenck, H. J., 8 Eysenck, S. B. G. (1968). Eysenck personality inven-
tory. San Diego: Educational and Industrial Testing Service.

 

Foote, R., 8 Belinky, C. (1972). It pays to switch? Consequences of
changing answers on multiple-choice examinations. Psychological
Reports, 41, 667-673.

 

Gilbert, H. W. (1982). Pathways: A guide to reading and study skills.
Boston: Houghton-Mifflin.

 

Goldfried, G., 8 D'Zurilla, T. (1973). Prediction of academic compe-
tence by means Of the Survey of Study Habits and Attitudes. Journal
of Educational Psychology, g4, 116-122.

 

Greene, K. (1981). Item-response changes on multiple-choice tests as
a function of test anxiety. Journal of Experimental Education, 42,
225—228.

Gronlund, N. E. (1967). Measurement and evaluation in teaching, New
York: Macmillan.

Harnish, D. L., Hill, K. T., 8 Fyans, L. J. (1980, April). Development

 

of a shorter, more reliable, and more valid measure of test motiva-
tion. Paper presented at the annual meeting of the National Council
on Measurement in Education, Boston, MA.

163

Higgins, M. J. (1967). Test reviews. Journal of Counseling Psychology,
14, 392-393.

Hill, G. E. (1937). The effects Of changed response in true-false
tests. Journal of Educational Psychology, 28, 308-310.

Hills, J. R. (1981). Measurement and evaluation in the classroom.
Columbus, OH: Charles E. Merrill.

 

Honig, F. (1973). Taking tests and scoring high. New York: Arco.

 

Jackson, P. F. (1978). Answer changing on objective tests. Journal of
Educational Research, 11, 313-315.

 

 

Jacobs, S. S. (1971, February). An experimental analysis of answer
changing behavior on Objective tests. Paper presented at the annual
meeting of the American Educational Research Association, New York.

 

 

Jarrett, R. E. (1948). The extra-chance nature Of changes on students'
response to objective test-items. Journal of General Psychology, §§,
243-250.

 

Jenkins, J. J. (1974). Remember that old theory Of memory? Well,
forget it! American Psychologist, 29, 785-795.

 

Johnston, J. J. (1975). Sticking with first responses on multiple-
choice exams: For better or for worse? Teaching of Psychology, 2,
178-179.

 

Johnston, J. J. (1978). Answer-changing behavior and grades. Teaching
of Psychology, 5, 44-45.

 

Juola, A. E. (1968). Examination skills and techniques. Lincoln, NE:
Cliff's Notes, Inc.

 

Kellighan, T., Madaus, G. F., 8 Airasian, P. W. (1982). The effects of

 

standardized testing. Boston: Kliewer-Nyhoff.

Lawson, E. E. (1935). What happens when the second judgement is
recorded in a true-false test? Journal Of Educational Psychology,
26, 223-227.

Lehman, H. C. (1928). Does it pay to change initial decisions in true-
false tests? School and Society, 48, 456-458.

Lindeman, R. H., 8 Merenda, P. F. (1979). Educational measurement.
Glenview, IL: Scott Foresman.

Lowe, M. L., 8 Crawford, C. C. (1929). First impression versus second
thought in true-false tests. Journal Of Educational Psychology, 29,
192-195.

164

Lyman, H. B. (1978). Test scores and what they mean. Englewood Cliffs,
NJ: Prentice-Hall.

 

Lynch, 0. )., 8 Smith, 8. C. (1975). Item response changes: Effects on
test scores. Measurement and Evaluation in Guidance, 7, 220-224.

 

Maiorana, V. P. (1980). How to study and learn in college. Englewood
Cliffs, NJ: Prentice-Hall.

 

Mallinson, G. G., 8 Miller, D. J. (1956). The effects of second guess-
ing in achievement scores of college tests. Thirteenth yearbook of
the National Council on Measurement in Education, 24:26.

 

 

Matthews, C. 0. (1929). Erroneous first impressions on objective tests.
Journal of Educational Psychology, g9, 280-286.

 

McMorris, R. F. (1981, April). Answer changing,and attitudes. Paper
presented at the meeting of the American Educational Research Asso-
ciation, Los Angeles.

McMorris, R. F., 8 Leonard, G. (1976, April). Item response changes
and cognitive style. Paper presented at the meeting of the National
Council on Measurement in Education, San Francisco.

 

 

McMorris, R. F., Lichtstein, S. F., 8 Hoops, A. J. (1983, April).
Answer changing after instructions on answer changing. Paper pre-
sented at the conference of the’American EdficationaTTResearch
Association, Montreal, Canada.

Mehrens, W. A., 8 Lehmann, I. J. (1978). Measurement and evaluation in
education and psychology (2nd ed). New York: Holt, Rinehart 8
Winston.

 

 

Mercer, M. (1977). An update Of 'Implications of changing answers on
objective test items.‘ (ERIC Document Reproduction Service NO. ED 163
051).

 

Mercer, M. (1978). Answer changing and students' test scores (Doctoral
dissertation, Rutgers University, 1978). Dissertation Abstracts
International, 49, 0214A.

 

Millman, J., Bishop, C. H., 8 Ebel, R. (1965). An analysis of test-
wiseness. Educational and Psychological Measurement, 25, 707-726.

Millman, J., 8 Pauk, W. (1969). How to take tests. New York: McGraw-
Hill.

Morris, F. L. (1961). The validity of the Brown-Holtzman Survey of
Study Habits and Attitudes 1960 experimental revision for grades
7-12. Unpublished masterTs thesis, Wayne State University, Detroit,
MI.

165

Mueller, 0. J., 8 Shwedel, A. (1975). Some correlates of net gain
resultant from answer changing on Objective achievement test items.
Journal of Educational Measurements, 1e, 251-254.

Mueller, 0. J. ,8 Wasser, V. (1977). Implication of changing answers
on objective test items. Journal of Educational Measurement, 14,
9-13.

 

Nie, N. H., Hull, C. H., Jenkins, J. G., Steinbrenner, K., 8 Bent,
. H. (1975). Statistical package for the social sciences (2nd ed.).
New York: McGraw-Hill.

 

Noll, V. H., Scannell, D. P., 8 Craig, R. C. (1979). Introduction to
educational measurements. Boston: Houghton-Mifflin.

 

 

Norman, M. H., 8 Norman, E. S. K. (1976). How to read and study for
success in college, New York: Holt, Rinehart 8 Winston.

 

 

Pascale, P. J. (1974). Changing initial answers on multiple-choice
achievement tests. Measurement and Evaluation in Guidance, g,
236-238.

 

Pauk, W. (1974). How to study in college (2nd ed.). Boston: Houghton-
Mifflin.

 

Payne, 8. D. (1983, April). Test anxiety, answer changing behavior,
and race. Paper presented at the American Educational Research
Association, Montreal, Canada.

 

Penfield, D., 8 Mercer, M. (1980). Answer changing and statistics.
Educational Research Quarterly, 5, 50- 57.

 

Piaget, J. (1972). The child's concept of physical causality. Potowa,
NJ: Littlefield, Adams, 8 Co.

Range, L. M., Anderson, H. N., 8 Wesley, A. L. (1982). Personality
correlates Of multiple-choice answer-changing patterns. Beychological

 

Reports, 51, 523-527.

Reile, P. J., 8 Briggs, L. J. (1952). Should students change their
initial answers on objective-type tests? More evidence regarding an
old problem. Journal Of Educational Psychology, 43, 110-115.

Reiling, E., 8 Taylor, R. (1972). A new approach to the problem of
changing initial responses to multiple choice questions. Journal of
Educational Measurement, 9, 67-70.

 

Robinson, F. P. (1970). Effective study_(4th ed.). New York: Harper
8 Row.

166

Sarnacki, R. E. (1979). Examination of test-wiseness in the cognitive
test domain. Review of Educational Research, 2, 252-279.

Shay, C. B. (1972). In 0. K. Burow (Ed.), The seventh mental measure-
ment yearbook (pp. 782-783). Highland, NJ: Gryphon Press.

 

Shepherd, J. F. (1982). The Houghton-Mifflin study skills handbook.
Boston: Houghton-Mifflin.

 

Sitton, L. R., Adams, I. G., 8 Anderson, H. N. (1980). Personality
correlates of students' patterns of changing answers on multiple-
chOice tests. Eeychological Reports, 42, 655-660.

 

Smith, A., 8 Moore, J. C. (1976). The effects of changing answers on
scores of non-test-sophisticated examinees. Measurement and Evalua-
tion in Guidance, 8, 252-254.

 

 

Smith, J. D. (1977). The relationship of content area and ability with
multiple-choice test response alterations. Dissertation Abstracts
International, 31, 5066A. (University Microfilms No. 77-3154).

 

 

Smith, M., White, K. P., 8 Coop, L. H. (1979). The effect of item type
on the consequences of changing answers on multiple-choice tests.
Journal of Educational Measurements, 1g, 206-208.

 

Stanley, J. C., 8 Hopkins, K. D. (1972). Educational and psychological

 

measurements and evaluation. Englewood Cliffs, NJ: Prentice-Hall.

 

Staton, T. F. (1982). How to study (7th ed.). Nashville, TN: How to
Study.

Stoffer, G. R., Davis, K. E., 8 Brown, J. B., Jr. (1977). The conse-
quences of changing initial answers on objective tests. A stable
effect and a stable misconception. Journal of Educational Research,
70, 272-277.

 

Sutton, J. P., Jr. (1982). Influencing answer-changing behavior:
Changing beliefs about the consequences of answer changing through
consultation (Doctoral dissertation, Auburn University, 1982).
Dissertation Abstracts International, 4e, 5029A.

Thorndike, E. L. (1951). Reliability. In E. F. Linquist (Ed.),
Educational measurement. Washington, 0.0.: American Council on
EducatiOn.

 

Vaas, C. E., 8 Nungester, R. J. (1982, March). An investigation of the

 

answer changing behavior of adult learners. Paper presented at the
meeting of theTAmerican Educational Research Association, New York.

Vidler, D., 8 Hansen, R. (1980). Answer changing on multiple choice
tests. Journal of Experimental Education, 42, 18-20.

167

Wen, S., 8 Lui, A. (1976). The validity of each of the four subtests
of the Survey of Study Habits and Attitudes for each of two samples
of college students and under each Of two treatment conditions involv-
ing use of released class time. Educational and Psychological Meas-
urement, 44, 565-568.

 

Wilson, P. (1980). Answer changing behavior on objective tests: What
is our responsibility? Epistle, p, 27-31.

Wrenn, C. G. (1959). In 0. K. Buros (Ed.), The fifth mental measure-
ment yearbook reviews: Reading-study skills (pp. 272-273). Highland,
NJ: Gryphon Press.

 

 

"mmANN