<%BANNER%>

Competency Threat

Permanent Link: http://ufdc.ufl.edu/UFE0022212/00001

Material Information

Title: Competency Threat What Are the Effects on Working Memory?
Physical Description: 1 online resource (43 p.)
Language: english
Creator: Lai, Stephanie
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: Educational Psychology -- Dissertations, Academic -- UF
Genre: Educational Psychology thesis, M.A.E.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The purpose of our study is to challenge one's mental capabilities by categorizing learners into performance groups and investigating whether projecting certain expectations will affect their working memory performance. Results indicated that high working memory participants showed performance decrements when categorized in both our high and low performance/expectations condition. However, low working memory participants in our low performance/expectations condition showed a significant increase in performance after the manipulation. This result is in direct contrast to our hypothesis and is the direct opposite of the findings from previous studies. The added anxiety associated with threats and expectations illustrate differences between high- and low- working memory persons. Future research should assess why the changes in performance occurred for different types of learners.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Stephanie Lai.
Thesis: Thesis (M.A.E.)--University of Florida, 2008.
Local: Adviser: Therriault, David.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022212:00001

Permanent Link: http://ufdc.ufl.edu/UFE0022212/00001

Material Information

Title: Competency Threat What Are the Effects on Working Memory?
Physical Description: 1 online resource (43 p.)
Language: english
Creator: Lai, Stephanie
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: Educational Psychology -- Dissertations, Academic -- UF
Genre: Educational Psychology thesis, M.A.E.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The purpose of our study is to challenge one's mental capabilities by categorizing learners into performance groups and investigating whether projecting certain expectations will affect their working memory performance. Results indicated that high working memory participants showed performance decrements when categorized in both our high and low performance/expectations condition. However, low working memory participants in our low performance/expectations condition showed a significant increase in performance after the manipulation. This result is in direct contrast to our hypothesis and is the direct opposite of the findings from previous studies. The added anxiety associated with threats and expectations illustrate differences between high- and low- working memory persons. Future research should assess why the changes in performance occurred for different types of learners.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Stephanie Lai.
Thesis: Thesis (M.A.E.)--University of Florida, 2008.
Local: Adviser: Therriault, David.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022212:00001


This item has the following downloads:


Full Text





COMPETENCY THREAT:
WHAT ARE THE EFFECTS ON WORKING MEMORY?





















By

STEPHANIE LAI


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF ARTS IN EDUCATION

UNIVERSITY OF FLORIDA

2008

































2008 Stephanie Lai


































To all who inspired me to be a part of this field; and to my family and friends, for being the
support system that strengthens me as a person and motivates me to strive for success.









ACKNOWLEDGMENTS

I thank the members of my supervisory committee, Dr. David Therriault and Dr. Tracy

Linderholm, for their support and mentoring; the staff and members at the Department of

Educational Psychology for their assistance; the research assistants who volunteered their own

time to my project; the participants in my study for their cooperation; and everyone else who has

been involved with the process. I thank my parents for their love and encouragement, and their

unconditional support for my interests and dreams.










TABLE OF CONTENTS

page

A CK N O W LED G M EN T S ................................................................. ........... ............. .....

LIST OF FIGURES .................................. .. ..... ..... ................. .7

ABSTRAC T .........................................................................................

CHAPTER

1 IN TR OD U CTION .......................................................................... .. ... .... 10

2 M E T H O D S .......................................................................................................1 8

S a m p le ............................................................................1 8
Materials .......................................18
B e ll C u rv e S u rv ey ..................................................................................................... 1 8
W working M em ory T ests............................................................................................. 18
D difficulty Q u estionnaire .............................................................................. 19
D em graphic Q questionnaire ...................................................................................... 20
P ro c e d u re .............. .... ...............................................................2 0

3 RESULT S/DISCUSSION ................ ......... ......... ......... 23

S co rin g ........................................................ ............................... ................... 2 3
P perform ance ............. .............. ......................................... ...........................23
D ifficu lty Q u estion n aire ................................................................................................... 2 5
D em graphic Q u estionnaire ............................................................................................. 27
B ell C u rv e .....................................................2 7

4 GENERAL DISCUS SION ............ ...... ......... ...... ......... 28

5 L IM IT A TIO N S ....................................................34

6 FUTURE RESEARCH ............. ......... ........ ........35

APPENDIX

A BELL CURVE SURVEY ...................................................... 36

B DEMOGRAPHIC QUESTIONAIRE............................................... 37

C DIFFICULTY QUESTIONNAIRE ............... .... ......... ..... .........38

D SA M P L E SC R IP T ......................................................... ................................ 39









R E FE R E N C E S ....................................................... 40

B IO G R A PH IC A L SK E T C H ................................................................................ ............... ..... 43





















































6









LIST OF FIGURES

Figure page

3-1 P perform ance differences .......................................................................... ....................25









Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Arts in Education

COMPETENCY THREAT:
WHAT ARE THE EFFECTS ON WORKING MEMORY?

By

Stephanie Lai

August 2008

Chair: David Therriault
Major: Educational Psychology

We investigated the role of feedback and performance expectation as a potential mediator

of performance deficits in working memory tasks as a function of working memory capacity

span. After completing an RSPAN task, 173 participants were randomly assigned into three

different performance expectation conditions: 95th percentile condition, where we projected a

high expectation for participants to perform well on the subsequent task; 40th percentile

condition, where we projected a low expectation for participants to have more difficulty with the

subsequent task; and no feedback, where participants were not given any type of feedback and

moved on to the second task. After taking the second test, their scores were recorded to

determine their capacity span and changes in their performance. Under the 95th percentile

condition, we found that high working memory participants (HWM) scored significantly lower

on the second test, after projecting high expectations for their performance. Likewise, HWM

participants also showed a deficit in performance after projecting low expectations for their

performance. We speculate that HWM participants internalize feedback, both positive and

negative, in a counterproductive manner, so that their performance is hindered on future tasks.

Interestingly, low working memory participants (LWM) in the 40th percentile condition perform









significantly better on the second task. We think LWM participants use this negative feedback

productively as a motivator and draw reserved resources toward the second task. Our study

suggests that working memory capacity is not as stable a construct as previously believed, and

that different types of people interpret feedback differently to either be productive or

counterproductive on future tasks.









CHAPTER 1
INTRODUCTION

Within cognitive psychology, the idea of working memory is used to refer to the system

or systems involved in the temporary storage of information while simultaneously processing

information necessary for the performance of complex tasks such as learning, reading

comprehension, and reasoning (Baddeley, 2002). There have been many expansions of working

memory since Baddeley and Hitch's original model, but commonalities exist between models.

First, the varying models agree that working memory capacity is limited and these limitations are

due to different factors such as decay and interference (Engle, Tuholski, Laughlin, & Conway,

1999) and the ability to manage attentional resources (Redick & Engle, 2006). One source of

potential interference can be attributed to anxiety. A moderate amount of anxiety, or an optimal

level of arousal, is necessary to best complete complex cognitive tasks, but when the level of

anxiety exceeds the optimal threshold it results in a decline in performance. A second

commonality of the models is the inclusion of a management and control system. Working

memory capacity reflects the efficiency of executive functions, particularly the ability to

maintain task-relevant demands in the face of distracting task-irrelevant information (Engle et

al., 1999). The ability to monitor and control attention is a major source of individual

differences, differentiating high- and low- working memory capacity persons (Borella, Carretti,

& Mammarella, 2006). If it is assumed that working memory capacity is limited in attentional

resources, then an excess amount of anxiety will consume processing resources that would

normally be allocated to processing task-relevant information (Tobias, 1985).

One research approach to the study of attention has manipulated working memory

capacity by inducing test anxiety with either a stereotype threat or by projecting a strong

performance expectation that caused participants to "choke" (Beilock & Carr, 2001). Unlike past









research that has focused on manipulating social identity (stereotype threats), the purpose of our

study is to challenge one's conceptualization of their mental capabilities by categorizing

individuals into particular performance groups and investigating whether a characterization of

one's mental ability will affect their working memory capacity. Our present study investigates

the effect of anxiety associated with expectations and competency threat upon working memory

performance. Specifically, our study focuses on the anxiety that is created by a fictitious "label"

of competency (i.e., a percentile rank) and how that affects one's performance on subsequent

working memory tests. To help build our case for the investigation of anxiety and working

memory, we review research exploring the effects on working memory caused by anxiety, how

stereotype threat and expectations create performance pressure and fear of choking, and different

factors that reduce working memory capacity.

High levels of anxiety are said to cause interference with cognitive capacity if the

capacity required by a task exceeds the amount available (Tobias, 1985). Anxiety is usually

defined as an unpleasant, complex state of emotions that include cognitive, behavioral, somatic,

and emotional components (Sarason, 1984). Anxiety heightens physiological activity (e.g., heart

rate, sweating) and can also cause self-disparaging thoughts (Sarason, 1961). Specifically, high

levels of test anxiety can cause an adverse effect on task performance. Students who feel extreme

uneasiness and nervousness about an exam may be experiencing test anxiety. High anxiety

persons typically do more poorly on tests than do low anxiety persons, particularly under

stressful conditions. This performance difference is mainly due to the differences in the ability to

control and focus attention of high- and low- anxious persons during the task. The low anxious

person is more fully focused on task-relevant information while the high anxious person is

internally focused. The task can not be performed adequately when attention is divided between









internal (self) cues and external (task) cues (Wine, 1971). Cadinu, Maass, Rosabianca, and

Kiesner (2005) have demonstrated this detrimental phenomenon using stereotype threat. In their

study, participants completed the first half of a math test and then were introduced with a

stereotype that could elicit pressure and fear caused by the perceived stigmatism of their group.

This was expected to affect their performance on the second half of the test. Specifically, women

in the stereotype-threat condition were told "recent research has shown that there are clear

differences in the scores obtained by men and women in logical-mathematical tasks" (p. 574)

and the no-threat (control) condition were told "there are no differences between men and

women in logical mathematical tasks" (p. 574). Women in the stereotype-threat condition

reported a higher number of negative thoughts related to their inability to perform and their lack

of competence in math than their no-threat counterparts. Additionally, these participants also

showed a significant decrease in performance from the first half to the second half of the test

compared to the control group, who in fact, performed significantly better on the second half and

reported less negative math-related thoughts. Their results support previous work that not only is

there a reduction in working memory capacity, but an individual's disruptive thoughts, elicited

by a stereotype threat, can lead to detrimental performance.

Stereotype threat refers to the fear experienced by an individual that their performance on

a task will confirm an existing stereotype. An individual's performance is hindered when a

stigmatized social identity is made salient in the performance situation (Schmader & Johns,

2003). According to a Steele's (1997) model of stereotype threat, when one is in a situation

where a negative stereotype exists about one's group, the additional fear of being judged or of

self-fulfilling the stereotype will cause them to under-perform in a manner consistent with the

stereotype. An example that is commonly used is the stereotype that women have more difficulty









with math than men. If this stereotype becomes salient in a testing situation, women may

experience a pressure that degrades their performance. Interestingly, it only takes a subtle

reminder to make a well-known stereotype salient. Shih, Pittinsky, and Ambady (1999)

presented women with a questionnaire comprised of questions designed to make salient their

identity as women (e.g., Do you prefer single sex or coed college dormitories?) prior to taking a

math test. Without introducing a direct stereotype about their performance the researchers found

that participants who were primed to identify themselves as women achieved the lowest

performance than participants in the control condition, where they were not primed to identify

their gender. Similar to gender stereotypes, Schmader and Johns second Experiment (2003)

found that asking participants to identify their ethnicity on a demographic questionnaire, prior to

testing, affected performance. When informed the test was measuring general intelligence,

Latinos performed lower compared to Whites and a control group of Latinos (under the no-threat

condition). These results indicate that a stereotype can be made salient without an explicit

statement, and that priming one's ethnic identity can negatively affect performance. The

stereotype threat reroutes attentional focus from the task onto task-irrelevant worries and induces

anxiety caused by fear of confirming the negative stereotype (Steele, 1997). So, according to

Steele's model, increased anxiety caused by this stereotype will reduce cognitive resources, thus

leading to performance deficits. This idea is further supported by Schmader and Johns study,

where they found that women who completed a working memory test described as a measure of

mathematical ability showed a reduction in cognitive capacity and performed poorer than the

men.

Our present study extends the work of stereotype threat by suggesting that the same

anxiety and worries can be induced by focusing participants' thoughts upon their academic









competency. Similar to the way one identifies themselves by gender or ethnic background, an

individual can also identify with a certain level of academic competency. Performance

information (feedback) aids students in making that identifying decision. Providing performance

information (i.e., low competency [40th percentile rank] or high competency [95th percentile])

might elicit the same worries of confirming an expectation that is elicited by a stereotype threat

and can overwhelm participants thereby negatively affecting their performance. Interestingly,

this type of threat is not dependent upon stereotypes related to either gender or race.

Past research has suggested that an expectation expressed by a stereotype (and the

personal value placed on a domain by an individual) may influence performance regardless of

stereotype salience (Smith & Johnson, 2006). In a stereotype threat situation, participants may

feel that because of their particular stereotype the experimenter has an expectation for them to do

poorly. It seems reasonable to suggest that whether stress or anxiety occurs depends on the

personal salience of a given situation for an individual. Personal salience, in turn, is a product of

those personality variables that shape perceptions of self and world (Sarason, 1984). When a

stereotype becomes salient during a testing situation, worry will divide the tester's attention,

between the task demands and internal concerns. Specifically, categorizing an individual's

competency level will produce self-focusing worries in trying to fulfill this expectation or prove

the expectation false. Situational conditions in which there are the greatest performance

differences occur when self-focusing tendencies are elicited (Wine, 1971).

Self-focusing tendencies are also elicited when high performance is expected and

individuals have a fear of choking (Smith & Johnson, 2006). Choking refers to a substantial

decreased in performing when burdened with performance pressure (Beilock & Carr, 2001).

According to their figure of participants' performance, Beilock and Carr (2005) operationally









classified choking as a performance decrement that suffers an estimated seven percent in

accuracy. Performance pressure has been defined as an anxious desire to perform at a high level

in a given situation (Beilock & Carr, 2001) and is thought to vary as a function of the personally

felt importance of a situation (Baumeister, Hamilton, & Tice, 1985). Performance pressure can

only elicit choking if the performer is aware of the need for good performance (Baumeister,

1984). Choking, like anxiety, creates mental distractions that compete for and reduce working

memory capacity that would otherwise be allocated to task execution (Beilock, Kulp, Holt, &

Carr, 2004).

While negative stereotypes can threaten intellectual performance, positive stereotypes

may also be harmful, negatively affecting performance by creating concern about failure to meet

high expectations held for a group (with which an individual identifies). For every group with a

negative stigmatism, members from the counter group are stereotyped as being comparably more

proficient (Smith & Johnson, 2006). For example, take the previously described stereotype that

women perform more poorly in math than men. While there is certainly a negative stereotype

about women, the same stereotype projects certain performance expectations for men. In a

similar stereotype threat condition, a study by Cheryan and Bodenhausen (2000) found that when

Asian American women were reminded of the stereotype that Asians are good at math and that

others may expect them to do well because of this stereotype, their performance suffered. This

suggests that an increased awareness of the personal relevance of a positive stereotype can also

hinder one's performance. Further, Jones and Berglas (1978) contended that performance

decrements were most likely to occur when a person is publicly expected to succeed, but

privately expects not to. When individuals are aware that others hold high expectations for their

performance, they may experience increases in performance pressure (Beilock & Carr, 2001). A









study by Baumeister et al., (1985) found that performers' own expectations of success improved

their performance, such that it motivated them, whereas an audience's expectations for success

lowered their performance. An audience expecting success will be especially disappointed by

failure, thus creating more pressure on the performer to meet the expectation. This research is

relevant to the current study because projecting an expectation on an individual based on their

performance compared to others will likely produce the same fear of falling short on subsequent

testing. Similar to one's level of competency, positive stereotypes (when they form the basis for

salient public expectations) can place significant performance pressure on members of that

group, thereby negatively affecting their performance (Cheryan & Bodenhausen, 2000).

Research on threat and working memory capacity has mainly focused on using social

identity, including sex, race, and age to manipulate academic performance. The present study

focuses on challenging one's mental identity as opposed to their social identity. Instead of a

direct stereotype manipulation, this study provides academic feedback (categorizing participants'

performance) and investigates whether suggesting one's competency will affect their

performance. Participants will be administered a working memory test, then given feedback on

how well they did (the manipulation) by assigning them randomly into one of three

groups/conditions: 95th percentile, 40th percentile, or no-feedback condition. The 95th percentile

condition is expected to create a high level of expectation for the second task and the 40th

percentile condition is expected to create a low level of expectation for the second task. After,

participants will complete a second working memory test. Participants in the no-feedback

condition will continue on to the second test without any explicit feedback or expectations. We

hypothesize (1) that low working memory participants assigned to the 40th percentile condition

will feel threatened in their capabilities and, similar to stereotype threat conditions, participants









will experience a fear of confirming the negative stigmatism. Consequently, we expect that their

performance will be hindered; (2) that low working memory participants assigned to the 95th

percentile condition will feel no added pressure by the expectation and their performance will be

stable across both tasks; (3) that high working memory participants assigned to the 95th

percentile condition will feel a weight of expectation and be overwhelmed with an anxious desire

to perform as one expects them to, in turn, allocating resources to this internal worry instead of

the task and hindering their performance (Beilock & Carr, 2005); (4) that high working memory

participants assigned to the 40th percentile condition will use the negative information about their

performance as motivation and likely increase in performance. The assumption is that high

working memory participants are likely to be academically high achievers (Engle, Kane, &

Tuholski, 1999) so they may have strategies to succeed in the face of pressure. Participants in the

no-feedback condition, both high- and low- working memory capacity, are expected (5) to have

stable performance across both tasks. The expectation of doing well and the fear of doing poorly,

as characterized by their conditions, will place a burden on the participants causing a reduction in

working memory capacity. Furthermore, we hypothesize (6) that participants in both of these

conditions will report perceiving the second test to be more difficult than the first because of the

added pressure.









CHAPTER 2
METHODS

Sample

The participants (N = 173) in this study were undergraduate educational psychology

students from the University of Florida who participated for credit toward a course requirement.

Materials

Bell Curve Survey

Upon signing the informed consent, participants were administered a Bell Curve Survey

(Appendix A) asking them to indicate where they believe their academic competency fell, on a

bell curve, compared to their peers. The survey served to help answer questions about individual

performance expectations and set the stage for the manipulation by highlighting comparisons

between individuals.

Working Memory Tests

To measure working memory capacity, a dual-processing test called the reading-span task

(RPSAN) was used (Conway et al., 2005). The RSPAN task was originally developed from

Daneman and Carpenter's (1980) work assessing working memory during reading. In the current

study, participants evaluated sentences while memorizing letters for later recall. Participants

were asked to read a sentence, presented on the computer screen, and decide whether or not that

sentence makes sense (in the real world) by clicking either the "yes" or "no" button on the screen

[e.g., Susan packed her stapler to protect her skin from burning in the sun.]. A letter is presented

after each sentence decision (yes or no) and at the end of a series of sentence/letter combination

trials (i.e., a set) participants were asked to recall the letters in the order presented. A set ranged

from two to six sentences. Participants are asked to recall, as best they could, the letters in serial

order. For example, if that particular set contained five sentences, than the participant would









have five letters that he/she would have to recall at the end. Having to decide whether or not the

sentences made sense is meant to further engage participants into a dual task: understanding the

sentences to make a judgment and recalling the letters at the ends of the sentences. Working

memory capacity is indexed as the number of letters that participants recall correctly from each

sentence/letter set.

The RSPAN task includes 18 sets of sentence/letter trials that contained two to six

sentence/letter combinations. The sets are presented in random order so participants do not know

how many sentences and letters they would be required to evaluate and recall at the beginning of

each set. The test is administered on a computer. Each set begins with the presentation of a

sentence and participants evaluate whether the sentence makes sense or not. After recording their

decision (clicking "yes" or "no"), participants are presented a to-be-remembered letter for 2

seconds. A blank screen lasting 1 second separates the presentation of each sentence and letter.

After presentation of all sentence/letter combinations in a set, participants are prompted to recall

all the letters in order in that set. Each set is separated by a prompt "next set", which is displayed

for 3 seconds. The computer records the letters recalled in the correct order, participants' correct

or incorrect responses to the sentences, and the time spent on each sentence.

We created the second RSPAN task to mirror the first. Between tasks, there were equal

number of sets and equal number of nonsense and sense sentences. The letters to be recalled at

the end of the sentence sets were also taken from the same sampling of letters. We compiled

sentences of equal length to the first task, that is, number of words per sentence, and also of

similar difficulty levels using the Flesch-Kincaid Grade Level readability scale.

Difficulty Questionnaire

Participants were administered a questionnaire following each test (twice). The Difficulty

Questionnaire (Appendix C) asked participants to rate the level of difficulty perceived on each









test on a 1-10 scale and they were asked to explain their response. The questionnaire was

evaluated for perceived difficulty on each test as a function of the condition they were assigned.

Demographic Questionnaire

Participants were administered a brief questionnaire regarding background information

upon completing the study. The Demographic Questionnaire (Appendix B) was administered at

the end of the study so there was no accidental priming of a stereotype threat that may affect

their performance. The questionnaire asked questions regarding age, gender, ethnicity, etc. The

questionnaire was evaluated for any demographic differences between conditions.

Procedure

Upon entering, a research assistant sat the participant at an individual computer in a quiet

laboratory room. Participants were administered an informed consent form to read and sign. The

research assistant then administered the Bell Curve Survey and asked participants to indicate

where the feel their academic competency falls, on a bell curve, compared to their peers. The bell

curve is used to introduce the idea of being compared to others that is highlighted when

participants are assigned to their conditions. Participants were advised that the experiment would

take about 1 hour and they would be completing two tests. The experimenter informed the

participant that the test is a reliable measure of working memory capacity and that their

performance is based on how accurately they evaluate sentences and the number of letters they

can recall in order correctly. The research assistant also described working memory capacity as

the ability to hold different pieces of information simultaneously while trying to process one

specific piece of information so that the participant was made fully aware of the task. The first

RSPAN test was then administered.

Following the first test, participants were administered the difficulty questionnaire. After

completing the difficulty questionnaire, the research assistant informed the participants of their









"performance". Participants were randomly assigned into one of the three conditions/groups.

Those in the 95th percentile condition were told that their test score was greater than 95% of the

scores of people taking the same test. Percentile ranks were specifically used in this study

because it highlights the participant's own performance, compared to others. Students are also

familiar with this type of reporting, as it is common in standardized tests with which they have

experience (i.e., the SAT or ACT). Additionally, the research assistant explained to them what

their percentile rankings mean and how their scores were compared to others. Explaining the

meaning of percentile rankings provided addition assurance that participants are well informed of

their performance and able to infer their own capabilities. Participants were told that based on

their percentile rank they were expected to do extremely well on the subsequent test. More

specifically, we used 95th percentile group because that is considered high achieving, but still

leaves room for improvement. In the 40th percentile condition the participants were told that their

score was 60% lower than others who took the same test. Participants were told that based on

their percentile rank they were expected to have difficulty on the subsequent test. The 40th

percentile was specifically used because it is considered below average, but we considered it a

realistic achievement group. In the no-feedback condition, participants completed the

questionnaire, but were not informed of any score or expectation, and continued onto the second

test. The no-feedback condition served as the control group and provides an important piece of

information: that the two different RSPAN tasks are of equal difficulty level.

The second RSPAN test was then administered. Upon completing the second test all

participants were administered another difficulty questionnaire followed by the demographic

questionnaire.









Finally, participants were debriefed. Participants were informed that their percentile

rankings for the first test were fictitious. The manipulation (percentile groups) was not based on

their actual performance, but assigned at random.









CHAPTER 3
RESULTS/DISCUSSION

Scoring

We wanted to include the data of participants who were genuinely attempting to solve

both tasks, (i.e., making a plausibility decision regarding the sentence and remembering the

letters). We used an existing standard of 85% accuracy criterion for all participants (Conway et

al., 2005). Any participant who did not meet the 85% accuracy criterion was excluded from the

analysis. This resulted in the removal of 6 participants, less than 4% of the data. At the

conclusion of the task, five scores were reported: absolute score, total number correct, reading

errors, speed errors, and accuracy errors. Absolute score was the sum of all perfectly recalled sets

of letters. For example, if a participant correctly recalled 4 letters in a set size of 4, 5 letters in a

set size of 5, and 3 letters in a set size of 6, the participant's RSPAN absolute score would be 9.

Total number correct represented the total number of letters recalled in the correct position. The

types of errors reported were accuracy errors, speed errors, and reading errors. Accuracy errors

occurred when a participant did not correctly identify the sentence as sense or nonsense. Speed

errors occurred when the participant ran out of time in attempting to make a decision about the

sentence. Reading errors were the combined total amount of accuracy errors and speed errors.

Tercile ranges to classify participants as high-, middle-, or low working memory were

adopted from a previous experiment of 150 participants (using the same lab and participant pool

source).Based on this sample of participants, we classified absolute span scores 43-72, 27-42,

and 26 and below as high-, middle-, and low working memory participants, respectively.

Performance

To test whether performance changes occurred in the second RSPAN task, we conducted

a 2 x 3 x 2 analysis of variance (ANOVA) on the absolute span scores, with experimental









condition (performance groups 95th and 40th) and span type (high, middle, and low) as between-

subjects factors and time (test 1 and test 2) as a within-subjects factor. Span types corresponded

to participants' scores on their first RSPAN task. Condition and span type were treated as

independent variables. Participants' difference scores on test 2 (after the manipulation) were

treated as the dependent variable. Changes in span scores (test 1 to test 2) were analyzed using a

two-way ANOVA with two levels of condition (95th and 40th percentile) and three levels of span

type (high, middle, or low). The main effect of change (in scores from test 1 to test 2) yielded an

F ratio of F(2, 117) = 7.648, p < .001, such that overall change in span scores was significant.

When we looked at the differences in scores for participants in the control (no-feedback)

condition, there was no significant difference between their scores on the first test and their

scores on the second test. That is, one test was not inherently more difficult than the other, t =

.535; p < .595. This suggests that any changes in score in the 40th and 95th percentile conditions

can be attributed to the presence of the manipulation (feedback). Follow up t-tests were

conducted to examine differences within spans as a function of assigned condition. As

hypothesized, there was a significant decrease in performance from the first RSPAN score (M =

50.69, SD = 1.57) to second RSPAN score (M = 38.63, SD = 3.03) for high working memory

(HWM) participants placed in the 95th percentile condition, t = -4.242; p < .001. Consistent with

Beilock and Carr's (2005) findings and our hypothesis, there was no significant change in

performance for low working memory (LWM) participants from test 1 (M = 16.21, SD = 1.28) to

test 2 (M = 17.67, SD = 2.47). In contrast to our hypothesis, HWM participants assigned to the

40th percentile condition did significantly worse on their second test, t = -2.256; p < .038. After

receiving negative feedback about their performance on test 1, HWM participants' performance

on test 2 decreased by 7 points. Results for the LWM participants in the 40th percentile condition










did not support our hypothesis, however they comprised the most significant and surprising

finding. Low working memory participants in the 40th percentile condition did significantly

better on test 2 when given negative feedback on their performance for test 1. In other words,

when told their performance fell into the 40th percentile, LWM participants significantly

increased in performance from test 1 (M = 19.67, SD = 1.28) to test 2 (M = 26.46, SD = 2.47), t

= 3.497; p < .002. As hypothesized, there was no significant change in performance from the

first test to the second test for participants assigned to the no-feedback condition (control). We

made no specific hypotheses about mid-span participants, and an analysis of their performance

revealed no significant changes across the three conditions. Figure 3-1 illustrates the differences

in performance (by absolute span score) between HWM and LWM participants for the different

performance conditions: 95th percentile, 40th percentile, and control.


10
5

U HWM
-5 H LWM
-10
-15
95th 40th Control


Figure 3-1. Performance differences

Difficulty Questionnaire

There was a small but significant difference in ratings of difficulty from test 1 to test 2

for participants in the control condition, t = 2.888; p < .006, and experimental conditions 95th and

40th, t = 2.09; p < .041 and t = 2.174; p < .034, respectively. Participants in all conditions

perceived the second test to be more difficult than the first, but this difference was marginal, less

than one point on the ten-point Likert scale participants used to rate the difficulty of each test.









When we analyzed span capacity, we found that HWM participants in the control and 95th

percentile condition perceived the second test to be harder than that first, t = 2.54; p < .026 and t

= 3.5; p < .004, respectively. The other groups of participants did not find one test to be

significantly different in difficulty than the other.

We reviewed the differences in statements made by the participants when asked to justify

their difficulty ratings of the tests. Only four of the sixty questionnaires reviewed listed feedback

or expectation as an explanation for difficulty rating. All four of the questionnaires that listed

feedback or expectation as an explanation for difficulty rating were from participants assigned to

the 40th percentile condition: three HWM and one LWM participant. Interestingly, all three

HWM participants perceived the second test to be more difficult as a function of concentrating

on the negative feedback that was provided about their performance. The opposite was reported

by the LWM participant, who felt the second test to be easier than the first, but also reasoned it

to be because of concentration on the negative feedback. The three HWM participants reported,

"... Given my scores on the last test, I put more concentration into it", "It was difficult to

concentrate after receiving my score", and "After hearing what I got on the first experiment I got

nervous taking the second one." The LWM participant reported, "I found the test to be just a bit

easier than the last one because I felt I had to focus more because of the score I got on the

previous one." Another potential difference between high- and low- working memory individuals

could be attributed to how feedback, particularly negative feedback, is processed. It appears that

high working memory participants may be overwhelmed by the concentration on feedback and

use this information counterproductively, experiencing performance decrements on the second

task. In contrast, the low working memory participant concentrated on the feedback, similar to









HWM participants, but somehow used this information productively and exerted more effort to

perform better on the second task.

Demographic Questionnaire

There were no significant differences in scores from test 1 to test 2 for either females or

males. We were unable to analyze ethnicity because of the unequal distribution of ethnicity in

our sample (Asian: N = 6, Black: 28, Hispanic: 19, White: 110, Other: 4).

Bell Curve

When participants were asked to rate their own perceived level of competency at the

beginning of the experiment, the average across all participants was 80. Participants, on average,

rated themselves as being 30 points over the average. This is consistent with Dunning, Heath,

and Suls' (2004) review that on average, people will overrate themselves, saying their skills are

"above average," a statistical impossibility. In their empirical research review, Dunning, Heath,

and Suls found that self-assessments are, at best, modestly correlated with actual performance.









CHAPTER 4
GENERAL DISCUSSION

The purpose of our study was to introduce a challenge to participants' competency and to

test whether performance is affected by this challenge. Similar to the findings of stereotype

threat research, we expected to find that challenging participants' views of their own competency

would result in decreased performance. We proposed that, in addition to identifying with social

factors (i.e., race and gender), people identify with a level of competency. Therefore, challenging

one's believed competency would be similar to challenging a part of their identity. Our results

provide evidence that under certain conditions, performance on working memory tasks can

fluctuate substantially (See Figure 1). More importantly, the differences in type of feedback

given may either motivate or hinder people's performance as a function of their working memory

span.

We hypothesized that (1) high working memory participants in a high pressure

performance situation, like the 95th percentile condition, would perform lower on the second task

because they would choke under the pressure of maintaining performance in the 95th percentile.

In contrast, we hypothesized (2) that HWM participants in the 40th percentile condition would

excel after hearing negative feedback because of their ability to overcome pressure in real-life

performance pressure situations. Consistent with our hypothesis, HWM participants showed a

deficit in performance when given positive feedback and high expectations for their

performance. This is also consistent with Beilock and Carr's (2005) claim that under high

pressure performance situations, HWM participants show decrements in performance.

Surprisingly, HWM participants also did significantly worse after receiving negative feedback

and low expectations for future performance. We originally assumed that because HWM

participants are likely to be high-achievers in many real-life situations like learning and









comprehension (Engle, Kane, & Tuholski, 1999) they may possess skills or strategies to

overcome pressure. However, it may be the case that HWM participants view both conditions as

high pressure and their performance decrement can be explained through the reallocation of

attentional resources from task-relevant information to task-irrelevant worries. Recall that

Beilock and Carr (2005) classified their participants as choking when they suffered a seven

percent performance decrement. Based on the performance from our study, HWM participants in

both the 95th and 40th percentile condition experienced choking; their performance suffered 16%

and 9% in accuracy span scores, respectively.

Similar to other stereotype threat research, we hypothesized (3) that low working

memory participants in a competency threatening situation, similar to the 40th percentile

condition, may experience fear of confirming a negative stereotype, thus experiencing a

performance decrement after hearing the feedback and low expectation for future performance.

We also predicted (4) that LWM participants in the 95th percentile condition would not

experience any performance differences because they would not be threatened by the situation,

thus not experiencing the added pressure. Beilock and Carr (2005) concluded that LWM

participants do not suffer from the same disadvantage in high pressure situations as HWM

participants, as evidenced by their finding that performance remained relatively stable in their

high pressure condition. We expand on Beilock and Carr's results by demonstrating that under

certain performance situations, particularly when providing negative feedback and low

expectations for future performance, LWM participants may actually improve in performance,

scoring significantly higher after receiving feedback. This result is in direct contrast to our

hypothesis and is the direct opposite of the findings from previous stereotype threat studies

(Cadinu et al., 2005; Shih, Pittinsky, & Ambady, 1999). We speculate that negative feedback









might have acted as a motivator for our participants who had lower working memory scores. It is

important to note that while LWM participants' performance increased after being given

feedback, they still underperformed relative to HWM participants. Low working memory

participants may interpret feedback differently and use this information productively for future

performance. An alternative speculation would be that LWM participants may have expended

attentional reserves during the second test, utilizing extra, normally untapped, attentional

resources. Obviously, further empirical work will be necessary to substantiate these speculations.

It may be that certain types of interference (i.e., feedback information, expectations, feelings of

pressure, or threatening situations) tap into these attentional reserves, allowing LWM

participants to make use of normally untapped resources during subsequent performance

situations. Our results support our hypothesis about LWM participants in the 95th percentile

condition; they were unaffected by the positive feedback and high expectation for future

performance. Probably due to the absence of performance pressure to overcome competency

threat, LWM participants remained in a neutral state, performing similarly on test 1 and test 2.

We also predicted (5) that participants in the control condition would perform similarly

on both tests and (6) that all participants in the experimental conditions would report the second

test to be harder than the first. The results support our hypothesis about the control condition; all

participants performed similarly on test 1 and test 2 when given no feedback, suggesting that

there were no practice effects present, and that test 2 and test 1 were similar in difficulty level.

After analyzing the difficulty questionnaires, we found that all participants found the second test

to be more difficult than the first, irrespective of what condition they were assigned to. When we

examined ratings between high- and low- span participants in both experimental conditions, we









found that HWM participants in the 95th percentile condition perceived the second test to be

significantly more difficult than the first.

Results from our study broaden the literature on individual differences between span

types. The capability to control attention, especially when there are competing demands, is a

determining factor in one's working memory capacity (WMC). After analyzing the data, we

ruled out the idea that fatigue negatively influenced WMC scores. There were no differences in

performance from test 1 to test 2 among any of the three different span types in our control

condition. Differences in performance were only elicited by the manipulation. One measurable

difference of WMC between high- and low- working memory persons is the ability to control

attention so that one can remain attentive to the task in the face of task-irrelevant distractions,

regardless of whether distractions are externally or internally generated (Barrett, Tugade, &

Engle, 2004). High working memory participants' ability to focus their attention (i.e., on

performance pressure) limits their awareness of other information, constraining the range that

they may attend to or process (Conway, Cowan, & Bunting, 2001). Additionally, Wiley (1998)

found that, in a creative problem-solving task, presenting misleading information for the purpose

of activating misleading solution attempts caused HWM participants to fixate their attention on

the misleading information. This reallocation of attentional resources may negatively impact

WMC and the ability to attend to all the information in the task. It may be the case that HWM

capacity can sometimes cause too much attentional focus and if the focus is on task-irrelevant

information, task performance will suffer. High working memory participants in both the 95th

and 40th percentile condition may have felt pressure to fulfill high expectations or pressure to

overcome failure, respectively, thus drawing their attention away from the working memory task.









Both external (i.e., feedback and expectation) and internal (i.e., thoughts and feelings) distracters

can draw attentional resources away from a task (Barrett et al., 2004).

An alternate explanation that could highlight the differences between high- and low-

working memory participants is the use of feedback. Based on a few explanations given by

participants, we found that both high- and low- working memory participants concentrated on the

negative feedback, but HWM participants felt the second test to be more difficult and

experienced a decrease in performance. The LWM participant who attributed their performance

on the second test to feedback felt the second test to be easier, so an argument could be made to

say that LWM participants use feedback information productively while HWM participants use

feedback counterproductively. We only collected data on one LWM participant, so this argument

is speculative and will need to be researched in the future.

Results from our study also add to the research on expectation and choking under

pressure. Literature suggests that when a certain level of performance is expected (either high or

low), the added pressure is believed to reduce one's working memory capacity causing a deficit

in performance by reallocating attentional resources to task-irrelevant concerns. The added

anxiety associated with threats and expectations illustrate differences in performance between

high- and low- working memory persons. In a high stakes testing situation, HWM participants

might "choke under pressure" or reallocate their attentional resources, but for LWM participants

the added pressure of performance may increase their performance. The performance differences

between high- and low- working memory capacity participants are especially highlighted in our

40th percentile condition, where participants received negative feedback and low expectations for

future performance. Grouping participants into performance groups and projecting expectations

in correlation with those groups is similar to the grouping people experience in real-life learning









situations. Consequently, the findings of our research are likely to apply to real-life testing

situations.

There has been considerable research examining stereotype threats in the education

system and ways to alleviate the corresponding performance pressure. Given our findings,

educators and policy makers may want to reexamine how students are placed into particular

scoring categories, such as percentile groups, or groups formed by ability; where performance is

highlighted in comparison to peers. Unknowingly, some students may interpret positive or

negative feedback productively or counterproductively. In accordance with our results, if it is

assumed that HWM participants are generally high achievers, high-achieving students may

interpret both positive and negative feedback as pressure, causing them to choke in both

situations. Consequently, teachers may want use caution in giving specific feedback to high-

achieving students because of the students' possible interpretations of the feedback. However,

for low-achieving students (if it assumed WMC correlates to school achievement) it may be

beneficial to highlight their performance in order to motivate them to perform past their own

expectations.

Working memory capacity is correlated with a variety of real-world cognitive tasks

(Engle et al., 1999) including reading comprehension, language comprehension, learning to spell,

and writing; a plethora of other skills needed to excel in an academic setting are affected. If

working memory is not as stable and consistent as originally believed, there may be a range of

motivators that affect performance on working memory span tasks. It is important to understand

how individual differences may affect performance, reactions to high-pressure situations, and

how feedback is interpreted, especially when discussing real-world cognitive tasks.









CHAPTER 5
LIMITATIONS

While the findings from our study help to move forward our understanding of working

memory, a few limitations and shortcomings do exist. For example, because some of our results

were unexpected, we do not have an accurate measure or assessment as to why the specific

changes in performance occurred for the different span types. Our difficulty questionnaire was

not as informative as we had hoped. Only four participants referred to the feedback and

expectations as a function of effort change toward the subsequent task. A more detailed self-

assessment could be used in future work to specifically prime responses related to feelings of

pressure, effort change, and motivation toward the second task. Without a detailed assessment,

our explanations for the surprising results should be interpreted with caution.









CHAPTER 6
FUTURE RESEARCH

Future research should assess, in more detail, the cognitive processes that underlie the

discrepancies in performance as a function of feedback and perceived performance pressure. A

self-report or questionnaire assessing motivation or feelings of pressure may provide additional

information as to why different types of achievers are affected differently by feedback and

projections of expectation. Likewise, a questionnaire may examine participants' reports of

potential effort changes between test 1 to test 2 and how they were affected by feedback and

expectations for future performance. Instead of asking open-ended questions, future researchers

should consider developing a more structured questionnaire, asking specific questions about

underlying processes that may have occurred (e.g., Did you put forth more effort/less effort?, Did

you feel more pressure given your feedback?).

Additionally, future research may want to include another cognitive task that is correlated

to working memory capacity and is more reflective of tasks in real-life learning situations. By

including a typical cognitive task, like a vocabulary test or reading comprehension assignment,

the findings from the study could more likely generalize to classroom settings. That is, findings

from the study can specifically relate to different types of achievers and how different types of

students are likely to interpret feedback and expectations.












APPENDIX A
BELL CURVE SURVEY


Subject#


Experiment:


Please indicate on this normal distribution bell curve where your competency lies compared to
your peers.


Paenlal of Cases
under Parlians of
the Nrmal CLur v 0.13%
Standard -
Devialions
N'irmaL Curve
Equval ents
PaEciLils
Equivalmlts


-3o -2c


I 10 20 30 40 50 60


14% 0.13%


lc 2c 3c 4<-


70 BO 940 99


I 5 30 20 40 50 60 70 o8 90


Date:


I I I I I I I I I I I I


95 9









APPENDIX B
DEMOGRAPHIC QUESTIONNAIRE

Date: Subject# Experiment:

What is your age?

What is your gender?

o Female
o Male

How would you classify yourself?

o American Indian or Native Alaskan
o Asian or Pacific Islander
o Black (not of Hispanic origin)
o Hispanic or Latino
o White
o Other:









APPENDIX C
DIFFICULTY QUESTIONNAIRE

Date: Subject# Experiment:

How difficult did you perceive this test to be?

1 2 3 4 5 6 7 8 9 10

Please give a brief explanation of why you gave the test this rating.









APPENDIX D
SAMPLE SCRIPT

"Here are the printed results of your performance on the reading span task (RSPAN). This is

your score and based on your performance you fall into the percentile group. This means that

your test score is greater/less than % of the scores of people taking the same test and

essentially, of measurable intelligence since WMC is related to fluid intelligence. So, based on

your performance we are expecting you to do very well/expecting you to have more difficulty

with this second test."









REFERENCES


Baddeley, A.D. (2002). Is working memory still working? [Electronic version].
European Psychologist, 7, 85-97.

Barrett, L.F., Tugade, M.M., & Engle, R.W. (2004). Individual differences in working
memory capacity and dual-process theories of the mind. Psychological Bulletin,
130(4), 553-573.

Baumeister, R.F. (1984). Choking under pressure: Self consciousness and paradoxical
effects of incentives of skillful performance. Journal of Personality and Social
Psychology, 46, 610-620.

Baumeister, R.F., Hamilton, J.C., & Tice, D.M. (1985). Public versus private
expectancy of success: Confidence booster or performance pressure? [Electronic
version]. Journal of Personality and Social Psychology, 48, 1447-1457.

Beilock, S.L., & Carr, T.H. (2001). On the fragility of skilled performance: What
governs choking under pressure? [Electronic version]. Journal of Educational
Psychology: General, 130, 701-725.

Beilock, S.L., & Carr, T.H. (2005). When high-powered people fail: Working memory
and "choking under pressure" in math [Electronic version]. American Psychological
Society, 16, 101-105.

Beilock, S.L., Kulp, C.A., Holt, L.E., & Carr, T.H. (2004). More on the fragility of
performance: Choking under pressure in mathematical problem solving [Electronic
version]. Journal ofExperimental Psychology: General, 133, 584-600.

Borella, E., Carretti, B., & Mammarella, I.C. (2006). Do working memory and
susceptibility to interference predict individual differences in fluid intelligence?
[Electronic version]. European Journal of Cognitive Psychology, 18, 51-69.

Cadinu, M., Maass, A., Rosabianca, A., & Kiesner, J. (2005). Why do women
underperform under stereotype threat?: Evidence for the role of negative thinking
[Electronic version]. American Psychological Society, 16, 572-578.

Cheryan, S.C., & Bodenhausen, G.V. (2000). When positive stereotypes threaten
intellectual performance: The psychological hazards of "model minority" status
[Electronic version]. American Psychological Society, 11, 399-402.

Conway, A.R., Cowan, N.C., & Bunting, M.F. (2001). The cocktail party
phenomenon revisited: The importance of working memory capacity. Psychonomic
Bulletin and Review, 8(2), 331-335.









Conway, A.R.A., Kane, M.J., Bunting, M.F., Hambrick, D.Z., Wilhelm, O., & Engle, R. (2005).
Working memory span tasks: A methodological review and users guide. Psychonomic Bulletin
andReview, 12(5), 769-786.

Daneman, M., & Carpenter, P.A. (1980). Individual-differences in working memory and
Reading [Electronic version]. Journal of Verbal Learning and Verbal Behavior, 19,
450-466.

Dunning, D., Heath, C., & Suls, J.M. (2004). Flawed self-assessment: Implications for
health, education, and the workplace. Psychological Science in the Public Interest,
5(3), 69-106.

Engle, R.W., Kane, M.J., & Tuholski, S.W. (1999). Individual differences in working
memory capacity and what they tell us about controlled attention, general fluid
intelligence, and function of the prefrontal cortex. In A. Miyake & P. Shah (Eds.),
Models of working memory: Mechanisms of active maintenance and executive control
(pp. 102-134). New York: Cambridge University Press.

Engle, R.W., Tuholski, S.W., Laughlin, J.E., & Conway, A.R.A. (1999). Working
memory, short-term memory, and general fluid intelligence: A latent-variable
approach [Electronic version]. Journal ofExperimental Psychology: General, 128,
309-331.

Jones, E.E., & Berglas, S. (1978). Control of attributions about the self through self-
handicapping strategies: The appeal of alcohol and the role of underachievement
[Electronic version]. Personality and Social Psychology Bulletin, 4, 200-206.

Redick, T.S., & Engle, R.W. (2006). Working memory capacity and attention network
test performance. Applied Cognitive Psychology, 20, 713-721.

Sarason, I.G. (1961). Text anxiety and the intellectual performance of college students
[Electronic version]. Journal of Educational Psychology, 52, 201-206.

Sarason, I.G. (1984). Stress, anxiety, and cognitive interference: Reactions to tests
[Electronic version]. Journal ofPersonality and Social Psychology, 46, 929-938.

Schmader, T., & Johns, M. (2003). Converging evidence that stereotype threat reduces
working memory capacity [Electronic version]. Journal ofPersonality and Social
Psychology, 85, 440-452.

Shih, M., Pittinksky, T.L., & Ambady, N. (1999). Stereotype susceptibility: Identity
salience and shifts in quantitative performance [Electronic version]. Psychological
Science, 10, 80-83.









Smith, J.L., & Johnson, C.S. (2006). A stereotype boost or choking under pressure?
Positive gender stereotypes and men who are low in domain identification [Electronic
version]. Basic andApplied Social Psychology, 28, 51-63.

Steele, C.M. (1997). A threat in the air: How stereotypes shape intellectual identity and
performance [Electronic version]. American Psychologist, 52, 613-629.

Tobias, S. (1985). Test anxiety: Interference, defective skills, and cognitive capacity
[Electronic version]. Educational Psychologist, 20, 135-142.

Wiley, J. (1998). Expertise as mental set: The effects of domain knowledge in creative
problem solving. Memory and Cognition, 26 (4), 716-730.

Wine, J. (1971). Test anxiety and direction of attention [Electronic version]. Psychology
Bulletin, 76, 92-104.









BIOGRAPHICAL SKETCH

I received a Bachelor of Science in psychology from the University of Florida in May

2006. I continued to the master's program in educational psychology at the University of Florida

in August 2006. Since then, I have worked on a number of successful research projects and was

given the opportunity to teach a course in fall 2007 and spring 2008. After completing my

master's degree I will continue to pursue my Ph.D. in educational psychology.





PAGE 1

1 COMPETENCY THREAT: WHAT ARE THE EFFECTS ON WORKING MEMORY? By STEPHANIE LAI A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORI DA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS IN EDUCATION UNIVERSITY OF FLORIDA 2008

PAGE 2

2 2008 Stephanie Lai

PAGE 3

3 To all who inspired me to be a part of this fi eld; and to my family and friends, for being the support system that strengthens me as a pers on and motivates me to strive for success.

PAGE 4

4 ACKNOWLEDGMENTS I thank the members of my supervisory co mmittee, Dr. David Therriault and Dr. Tracy Linderholm, for their support and mentoring; th e staff and members at the Department of Educational Psychology for their a ssistance; the research assistan ts who volunteered their own time to my project; the participan ts in my study for their cooperat ion; and everyone else who has been involved with the process. I thank my pare nts for their love and encouragement, and their unconditional support for my interests and dreams.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ............................................................................................................... 4LIST OF FIGURES .........................................................................................................................7ABSTRACT ...................................................................................................................... ...............8 CHAPTER 1 INTRODUCTION .................................................................................................................. 102 METHODS ....................................................................................................................... ......18Sample ........................................................................................................................ ............18Materials ..................................................................................................................... ............18Bell Curve Survey ........................................................................................................... 18Working Memory Tests ................................................................................................... 18Difficulty Questionnaire ..................................................................................................19Demographic Questionnaire ............................................................................................20Procedure ..................................................................................................................... ...........203 RESULTS/DISCUSSION ...................................................................................................... 23Scoring ....................................................................................................................... .............23Performance ................................................................................................................... .........23Difficulty Questionnaire .........................................................................................................25Demographic Questionnaire ...................................................................................................27Bell Curve ........................................................................................................................274 GENERAL DISCUSSION ..................................................................................................... 285 LIMITATIONS ................................................................................................................... ....346 FUTURE RESEARCH ...........................................................................................................35APPENDIX A BELL CURVE SURVEY .......................................................................................................36B DEMOGRAPHIC QUESTIONAIRE ..................................................................................... 37C DIFFICULTY QUEST IONNAIRE ........................................................................................ 38D SAMPLE SCRIPT ..................................................................................................................39

PAGE 6

6 REFERENCES .................................................................................................................... ..........40BIOGRAPHICAL SKETCH .........................................................................................................43

PAGE 7

7 LIST OF FIGURES Figure page 3-1 Performance differences ................................................................................................... .25

PAGE 8

8 Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Arts in Education COMPETENCY THREAT: WHAT ARE THE EFFECTS ON WORKING MEMORY? By Stephanie Lai August 2008 Chair: David Therriault Major: Educational Psychology We investigated the role of feedback and pe rformance expectation as a potential mediator of performance deficits in working memory tasks as a function of working memory capacity span. After completing an RSPAN task, 173 par ticipants were randomly assigned into three different performance expectation conditions: 95th percentile condition, where we projected a high expectation for participants to perform well on the s ubsequent task; 40th percentile condition, where we projected a low expectation for participants to have more difficulty with the subsequent task; and no feedback, where participants were not gi ven any type of feedback and moved on to the second task. After taking the second test, their scores were recorded to determine their capacity span and cha nges in their performance. Under the 95th percentile condition, we found that high working memory part icipants (HWM) scored significantly lower on the second test, after projecting high expect ations for their performance. Likewise, HWM participants also showed a defi cit in performance after project ing low expectations for their performance. We speculate th at HWM participants internal ize feedback, both positive and negative, in a counterproductive manner, so that their performance is hind ered on future tasks. Interestingly, low working memory participants (LWM) in the 40th percentile condition perform

PAGE 9

9 significantly better on the second ta sk. We think LWM participants use this negative feedback productively as a motivator and draw reserved resources towa rd the second task. Our study suggests that working memory capacity is not as stable a construct as previously believed, and that different types of people interpret feedback differently to either be productive or counterproductive on future tasks.

PAGE 10

10 CHAPTER 1 INTRODUCTION W ithin cognitive psychology, the idea of working memory is used to refer to the system or systems involved in the temporary storage of information while simultaneously processing information necessary for the performance of complex tasks such as learning, reading comprehension, and reasoning (Baddeley, 2002). Ther e have been many expansions of working memory since Baddeley and Hitchs original model, but commonalities exist between models. First, the varying models agree that working memory capacity is limited and these limitations are due to different factors such as decay and interference (Engle, Tuholsk i, Laughlin, & Conway, 1999) and the ability to manage attentional resources (Redick & Engle, 2006). One source of potential interference can be attr ibuted to anxiety. A moderate am ount of anxiety, or an optimal level of arousal, is necessary to best complete complex cognitiv e tasks, but when the level of anxiety exceeds the optimal th reshold it results in a decline in performance. A second commonality of the models is the inclusion of a management and control system. Working memory capacity reflects the efficiency of ex ecutive functions, particularly the ability to maintain task-relevant demands in the face of di stracting task-irrelevant information (Engle et al., 1999). The ability to monito r and control attention is a major source of individual differences, differentiating highand lowwork ing memory capacity persons (Borella, Carretti, & Mammarella, 2006). If it is assumed that worki ng memory capacity is limited in attentional resources, then an excess amount of anxiety will consume processing resources that would normally be allocated to processing task -relevant information (Tobias, 1985). One research approach to the study of attention has manipulated working memory capacity by inducing test anxiety with either a stereotype thre at or by projecting a strong performance expectation that caused participan ts to choke (Beilock & Carr, 2001). Unlike past

PAGE 11

11 research that has focused on manipulating social identity (stereotype threats), the purpose of our study is to challenge ones con ceptualization of their mental capabilities by categorizing individuals into particular pe rformance groups and investigati ng whether a characterization of ones mental ability will aff ect their working memory capacity. Our present study investigates the effect of anxiety associated with expectations and competen cy threat upon working memory performance. Specifically, our study focuses on the a nxiety that is created by a fictitious label of competency (i.e., a percent ile rank) and how that affects ones performance on subsequent working memory tests. To help build our case for the investigation of anxiety and working memory, we review research exploring the e ffects on working memory caused by anxiety, how stereotype threat and expectati ons create performance pressure and fear of choking, and different factors that reduce working memory capacity. High levels of anxiety are said to cause interference with cognitive capacity if the capacity required by a task exceeds the amount available (Tobias, 1985). Anxiety is usually defined as an unpleasant, complex state of emoti ons that include cogniti ve, behavioral, somatic, and emotional components (Sarason, 1984). Anxiet y heightens physiological activity (e.g., heart rate, sweating) and can also cause self-dispa raging thoughts (Sarason, 1961). Specifically, high levels of test anxiety can cause an adverse effect on task performance. Students who feel extreme uneasiness and nervousness about an exam may be experiencing test anxiety. High anxiety persons typically do more poorly on tests than do low anxiety persons particularly under stressful conditions. This performance difference is mainly due to the differences in the ability to control and focus attention of highand lowanxious persons during the task. The low anxious person is more fully focused on task-relevant information while the high anxious person is internally focused. The task can not be performe d adequately when attention is divided between

PAGE 12

12 internal (self) cues and external (task) cues (Wine, 1971). Cadinu, Maass, Rosabianca, and Kiesner (2005) have demonstrated this detrimental phenomenon using stereotype threat. In their study, participants completed the first half of a math test a nd then were introduced with a stereotype that could elicit pressure and fear caused by the perceived stigmatism of their group. This was expected to affect th eir performance on the second half of the test. Specifically, women in the stereotype-threat condition were told r ecent research has shown that there are clear differences in the scores obtained by men and women in logical-mathema tical tasks (p. 574) and the no-threat (control) condition were told there are no differences between men and women in logical mathematical tasks (p. 574). Women in the stereotype-threat condition reported a higher number of negative thoughts related to their inability to perform and their lack of competence in math than their no-threat count erparts. Additionally, these participants also showed a significant decrease in pe rformance from the first half to the second half of the test compared to the control group, who in fact, perf ormed significantly better on the second half and reported less negative math-related thoughts. Their re sults support previous work that not only is there a reduction in working memory capacity, bu t an individuals disruptive thoughts, elicited by a stereotype threat, can lead to detrimental performance. Stereotype threat refers to the fear experien ced by an individual that their performance on a task will confirm an existing stereotype. An individuals performance is hindered when a stigmatized social identity is made salient in the performance situ ation (Schmader & Johns, 2003). According to a Steeles (1997) model of stereotype threat, when on e is in a situation where a negative stereotype exists about ones gr oup, the additional fear of being judged or of self-fulfilling the stereotype w ill cause them to under-perform in a manner consistent with the stereotype. An example that is commonly used is the stereotype that women have more difficulty

PAGE 13

13 with math than men. If this stereotype beco mes salient in a testing situation, women may experience a pressure that degrades their perf ormance. Interestingly, it only takes a subtle reminder to make a well-known stereotype salient. Shih, Pittinsky, and Ambady (1999) presented women with a questionnaire comprised of questions designed to make salient their identity as women (e.g., Do you prefer single sex or coed college dormitori es?) prior to taking a math test. Without introducing a direct stereotype about their performance the researchers found that participants who were primed to identify themselves as women achieved the lowest performance than participants in the control condition, where they were not primed to identify their gender. Similar to gender stereotypes, Schmader and Johns second Experiment (2003) found that asking participants to identify their ethnicity on a dem ographic questionnaire, prior to testing, affected performance. When informed the test was measuring general intelligence, Latinos performed lower compared to Whites and a control group of Latinos (under the no-threat condition). These results indicate that a stereotype can be ma de salient without an explicit statement, and that priming ones ethnic identi ty can negatively affect performance. The stereotype threat reroutes attentional focus from the task onto task-irrelevant worries and induces anxiety caused by fear of confirming the negative stereotype (Steele, 1997 ). So, according to Steeles model, increased anxiety caused by this stereotype will reduce cognitive resources, thus leading to performance deficits. This idea is further supported by Schmader and Johns study, where they found that women who completed a work ing memory test described as a measure of mathematical ability showed a reduction in c ognitive capacity and performed poorer than the men. Our present study extends the work of stereo type threat by suggesting that the same anxiety and worries can be induced by focusing participants thoughts upon their academic

PAGE 14

14 competency. Similar to the way one identifie s themselves by gender or ethnic background, an individual can also identify with a certai n level of academic competency. Performance information (feedback) aids students in making that identifying decisi on. Providing performance information (i.e., low competency [40th percentile rank] or high competency [95th percentile]) might elicit the same worries of confirming an exp ectation that is elicited by a stereotype threat and can overwhelm participants thereby negativel y affecting their performance. Interestingly, this type of threat is not dependent upon stereo types related to either gender or race. Past research has suggested that an exp ectation expressed by a stereotype (and the personal value placed on a domain by an individual) may influence performance regardless of stereotype salience (Smith & J ohnson, 2006). In a stereotype thre at situation, participants may feel that because of their particular stereotype the experimenter has an expectation for them to do poorly. It seems reasonable to suggest that whet her stress or anxiety occurs depends on the personal salience of a given situa tion for an individual. Personal sa lience, in turn, is a product of those personality variables that shape percep tions of self and world (Sarason, 1984). When a stereotype becomes salient during a testing situ ation, worry will divide the testers attention, between the task demands and internal concer ns. Specifically, categorizing an individuals competency level will produce self-focusing worries in trying to fulfill this expectation or prove the expectation false. Situational conditions in which there are the greatest performance differences occur when self-focusing te ndencies are elicited (Wine, 1971). Self-focusing tendencies are also elicited when high performance is expected and individuals have a fear of choking (Smith & Johnson, 2006). Choking refers to a substantial decreased in performing when burdened with performance pressure (Beilock & Carr, 2001). According to their figure of pa rticipants performan ce, Beilock and Carr (2005) operationally

PAGE 15

15 classified choking as a performance decrement that suffers an estimated seven percent in accuracy. Performance pressure has been defined as an anxious de sire to perform at a high level in a given situation (Beilock & Carr, 2001) and is thought to vary as a function of the personally felt importance of a situation (Baumeister, Ha milton, & Tice, 1985). Performance pressure can only elicit choking if the performer is aware of the need for good performance (Baumeister, 1984). Choking, like anxiety, create s mental distractions that compete for and reduce working memory capacity that would otherwise be alloca ted to task execution (B eilock, Kulp, Holt, & Carr, 2004). While negative stereotypes can threaten intellectual perf ormance, positive stereotypes may also be harmful, negatively affecting perfor mance by creating concern about failure to meet high expectations held for a group (with which an individual identifies). For every group with a negative stigmatism, members from the counter group are stereotyped as being comparably more proficient (Smith & Johnson, 2006). For example, ta ke the previously described stereotype that women perform more poorly in math than men. While there is certainly a negative stereotype about women, the same stereotype projects certain performance expectations for men. In a similar stereotype threat c ondition, a study by Cheryan and Bodenhausen (2000) found that when Asian American women were reminded of the stereo type that Asians are good at math and that others may expect them to do well because of th is stereotype, their perf ormance suffered. This suggests that an increased awareness of the person al relevance of a positive stereotype can also hinder ones performance. Further, Jones a nd Berglas (1978) contended that performance decrements were most likely to occur when a person is publicly expected to succeed, but privately expects not to. When individuals are aware that others hold high expectations for their performance, they may experien ce increases in performance pre ssure (Beilock & Carr, 2001). A

PAGE 16

16 study by Baumeister et al., (1985) found that performers own exp ectations of success improved their performance, such that it motivated them, whereas an audiences expectations for success lowered their performance. An audience expec ting success will be especially disappointed by failure, thus creating more pressu re on the performer to meet th e expectation. This research is relevant to the current study b ecause projecting an expectation on an individual based on their performance compared to others will likely produce the same fear of falling short on subsequent testing. Similar to ones level of competency, pos itive stereotypes (when they form the basis for salient public expectations) can place significant performance pressure on members of that group, thereby negatively affecting thei r performance (Cheryan & Bodenhause n, 2000). Research on threat and working memory cap acity has mainly focused on using social identity, including sex, race, and age to manipulate academic performance. The present study focuses on challenging ones mental identity as opposed to their social identity. Instead of a direct stereotype manipulation, this study provide s academic feedback (categorizing participants performance) and investigates whether sugge sting ones competency will affect their performance. Participants will be administered a working memory test, then given feedback on how well they did (the manipulation) by a ssigning them randomly into one of three groups/conditions: 95th percentile, 40th percentile, or no-f eedback condition. The 95th percentile condition is expected to create a high level of expectation for the second task and the 40th percentile condition is exp ected to create a low level of expe ctation for the second task. After, participants will complete a second working me mory test. Participants in the no-feedback condition will continue on to the second test without any explicit feedback or expectations. We hypothesize (1) that low working memory participants assigned to the 40th percentile condition will feel threatened in their capabilities and, sim ilar to stereotype threat conditions, participants

PAGE 17

17 will experience a fear of confirming the negative s tigmatism. Consequently, we expect that their performance will be hindered; (2) that low working memory participants assigned to the 95th percentile condition will feel no added pressure by the expectation and their performance will be stable across both tasks; (3) th at high working memory participants assigned to the 95th percentile condition will feel a weight of expectation and be overw helmed with an anxious desire to perform as one expects them to, in turn, alloca ting resources to this internal worry instead of the task and hindering their perf ormance (Beilock & Carr, 2005); (4) that high working memory participants assigned to the 40th percentile condition will use the negative information about their performance as motivation and likely increase in performance. The assumption is that high working memory participants are likely to be academically high achie vers (Engle, Kane, & Tuholski, 1999) so they may have strategies to su cceed in the face of pressu re. Participants in the no-feedback condition, both highand lowworki ng memory capacity, are expected (5) to have stable performance across both tasks. The expect ation of doing well and the fear of doing poorly, as characterized by their conditions, will place a burden on the part icipants causing a reduction in working memory capacity. Furthermore, we hypothesi ze (6) that part icipants in both of these conditions will report perceiving the second test to be more difficult than the first because of the added pressure.

PAGE 18

18 CHAPTER 2 METHODS Sample The participants (N = 173) in this st udy were undergraduate educational psychology students from the University of Florida who particip ated for credit toward a course requirement. Materials Bell Curve Survey Upon signing the informed consent, participan ts were adm inistered a Bell Curve Survey (Appendix A) asking them to indicate where they believe their academic competency fell, on a bell curve, compared to their peers. The survey served to help answer questions about individual performance expectations and set the stage for the manipulation by highlighting comparisons between individuals. Working Memory Tests To m easure working memory capacity, a dual-pr ocessing test called the reading-span task (RPSAN) was used (Conway et al., 2005). The R SPAN task was originally developed from Daneman and Carpenter's (1980) work assessing working memory during reading. In the current study, participants evaluated sentences while memo rizing letters for later recall. Participants were asked to read a sentence, presented on the co mputer screen, and decide whether or not that sentence makes sense (in the real world) by clicking either the yes or no button on the screen [e.g., Susan packed her stapler to protect her skin from burning in the sun.]. A letter is presented after each sentence decision (yes or no) and at the end of a series of sentence/letter combination trials (i.e., a set) participants were asked to re call the letters in the order presented. A set ranged from two to six sentences. Particip ants are asked to recall, as best they could, the letters in serial order. For example, if that particular set cont ained five sentences, than the participant would

PAGE 19

19 have five letters that he/she w ould have to recall at the end. Havi ng to decide whether or not the sentences made sense is meant to further engage participants into a dual task: understanding the sentences to make a judgment and recalling the le tters at the ends of the sentences. Working memory capacity is indexed as the number of lett ers that participants r ecall correctly from each sentence/letter set. The RSPAN task includes 18 sets of sentence /letter trials that contained two to six sentence/letter combinations. The sets are presen ted in random order so participants do not know how many sentences and letters they would be requi red to evaluate and r ecall at the beginning of each set. The test is administered on a computer Each set begins with the presentation of a sentence and participants evaluate whether the se ntence makes sense or not. After recording their decision (clicking yes or no), participants are presented a to-be-remembered letter for 2 seconds. A blank screen lasting 1 second separate s the presentation of each sentence and letter. After presentation of all sentence/l etter combinations in a set, part icipants are prompted to recall all the letters in order in that set. Each set is separated by a prompt next set, which is displayed for 3 seconds. The computer record s the letters recalled in the correct order, participants correct or incorrect responses to the sentences, and the time spent on each sentence. We created the second RSPAN task to mirror the first. Between task s, there were equal number of sets and equal number of nonsense and sense sentences. The letters to be recalled at the end of the sentence sets were also taken from the same sampling of letters. We compiled sentences of equal length to the first task, that is, number of words per sentence, and also of similar difficulty levels using the Flesch-K incaid Grade Level readability scale. Difficulty Questionnaire Participants were adm inistered a questionnair e following each test (twice). The Difficulty Questionnaire (Appendix C) asked participants to rate the level of difficulty perceived on each

PAGE 20

20 test on a 1-10 scale and they were asked to explain their response. The questionnaire was evaluated for perceived difficulty on each test as a function of the c ondition they were assigned. Demographic Questionnaire Participants were adm inister ed a brief questionnaire rega rding background information upon completing the study. The Demographic Ques tionnaire (Appendix B) was administered at the end of the study so there was no accidental pr iming of a stereotype threat that may affect their performance. The questionnai re asked questions regarding age, gender, ethnicity, etc. The questionnaire was evaluated for any dem ographic differences between conditions. Procedure Upon entering, a research assistant sat the partic ipant at an individual computer in a quiet laboratory room Participants were administered an informed cons ent form to read and sign. The research assistant then administered the Bell Curve Survey and asked participants to indicate where the feel their academic comp etency falls, on a bell curve, comp ared to their peers. The bell curve is used to introduce the idea of being co mpared to others that is highlighted when participants are assigned to their conditions. Participants were advised that the experiment would take about 1 hour and they would be completing two tests. The experimenter informed the participant that the test is a reliable meas ure of working memory capacity and that their performance is based on how accurately they evaluate sentences and the number of letters they can recall in order correctly. The research assist ant also described working memory capacity as the ability to hold different pieces of inform ation simultaneously while trying to process one specific piece of information so that the particip ant was made fully aware of the task. The first RSPAN test was then administered. Following the first test, participants were ad ministered the difficulty questionnaire. After completing the difficulty questionnaire, the research assistant informed the participants of their

PAGE 21

21 performance. Participants were randomly assi gned into one of the three conditions/groups. Those in the 95th percentile condition were told that their test score was greater than 95% of the scores of people taking the same test. Percent ile ranks were specifically used in this study because it highlights the particip ants own performance, compared to others. Students are also familiar with this type of reporting, as it is comm on in standardized tests with which they have experience (i.e., the SAT or ACT). Additionally, th e research assistant explained to them what their percentile rankings mean and how their sc ores were compared to others. Explaining the meaning of percentile ranki ngs provided addition assurance that participants are well informed of their performance and able to infer their own capabilities. Partic ipants were told that based on their percentile rank they were expected to do extremely well on the subsequent test. More specifically, we used 95th percentile group because that is considered high achieving, but still leaves room for improvement. In the 40th percentile condition the partic ipants were told that their score was 60% lower than others who took the same test. Participants were told that based on their percentile rank they were expected to have difficulty on the subsequent test. The 40th percentile was specifically used because it is cons idered below average, but we considered it a realistic achievement group. In the no-feedb ack condition, participants completed the questionnaire, but were not informed of any scor e or expectation, and con tinued onto the second test. The no-feedback condition se rved as the control group and pr ovides an important piece of information: that the two different RSP AN tasks are of equal difficulty level. The second RSPAN test was then administ ered. Upon completing the second test all participants were administered another difficulty questionnai re followed by the demographic questionnaire.

PAGE 22

22 Finally, participants were debriefed. Particip ants were informed that their percentile rankings for the first test were fictitious. The manipulation (per centile groups) was not based on their actual performance, but assigned at random.

PAGE 23

23 CHAPTER 3 RESULTS/DISCUSSION Scoring We wanted to include the da ta of participants who were genuinely attempting to solve both tasks, (i.e., making a plausi bility decision regarding the sentence and remembering the letters). We used an existing standard of 85% accu racy criterion for all participants (Conway et al., 2005). Any participant who did not meet the 85% accuracy criterion was excluded from the analysis. This resulted in the removal of 6 pa rticipants, less than 4% of the data. At the conclusion of the task, five scores were report ed: absolute score, total number correct, reading errors, speed errors, and accuracy errors. Absolute score was the sum of all perfectly recalled sets of letters. For example, if a part icipant correctly recalled 4 letters in a set size of 4, 5 letters in a set size of 5, and 3 letters in a set size of 6, th e participants RSPAN absolute score would be 9. Total number correct represented the total number of letters recalled in the correct position. The types of errors reported were accuracy errors, sp eed errors, and reading errors. Accuracy errors occurred when a participant did not correctly identify the sentence as sense or nonsense. Speed errors occurred when the participant ran out of time in attempting to make a decision about the sentence. Reading errors were the combined to tal amount of accuracy errors and speed errors. Tercile ranges to classify participants as high-, middle-, or low working memory were adopted from a previous experiment of 150 partic ipants (using the same lab and participant pool source).Based on this sample of participants, we classified absolute span scores 43-72, 27-42, and 26 and below as high-, middle-, and low wo rking memory participants, respectively. Performance To test whether performance changes occurr ed in the second RSPAN task, we conducted a 2 x 3 x 2 analysis of variance (ANOVA) on th e absolute span scores, with experimental

PAGE 24

24 condition (performance groups 95th and 40th) and span type (high, middl e, and low) as betweensubjects factors and time (test 1 and test 2) as a within-subjects factor. Span types corresponded to participants scores on their first RSPAN task. Condition and span type were treated as independent variables. Participan ts difference scores on test 2 (after the manipulation) were treated as the dependent variable. Changes in span scores (test 1 to test 2) were analyzed using a two-way ANOVA with two levels of condition (95th and 40th percentile) and thre e levels of span type (high, middle, or low). The main effect of cha nge (in scores from test 1 to test 2) yielded an F ratio of F (2, 117) = 7.648, p < .001, such that overall change in span scores was significant. When we looked at the differences in scores for participants in th e control (no-feedback) condition, there was no significant difference between their scores on the first test and their scores on the second test. That is, one test was not inherently more di fficult than the other, t = .535; p < .595. This suggests that an y changes in score in the 40th and 95th percentile conditions can be attributed to the presence of the ma nipulation (feedback). Follow up t-tests were conducted to examine differences within sp ans as a function of assigned condition. As hypothesized, there was a significant decrease in performance from the first RSPAN score (M = 50.69, SD = 1.57) to second RSPAN score (M = 38.63, SD = 3.03) for high working memory (HWM) participants placed in the 95th percentile condition, t = -4.242; p < .001. Consistent with Beilock and Carrs (2005) findings and our hypot hesis, there was no significant change in performance for low working memory (LWM) participants from test 1 (M = 16.21, SD = 1.28) to test 2 (M = 17.67, SD = 2.47). In contrast to our hypothesis, HWM participants assigned to the 40th percentile condition di d significantly worse on their second test, t = -2.256; p < .038. After receiving negative feedback about their performance on test 1, HWM participants performance on test 2 decreased by 7 points. Results for the LWM participants in the 40th percentile condition

PAGE 25

25 did not support our hypothesis, however they comprised the most significant and surprising finding. Low working memory participants in the 40th percentile condition did significantly better on test 2 when given negative feedback on their performance for test 1. In other words, when told their performance fell into the 40th percentile, LWM participants significantly increased in performance from test 1 (M = 19.67, SD = 1.28) to test 2 (M = 26.46, SD = 2.47), t = 3.497; p < .002. As hypothesized, there was no significan t change in performance from the first test to the second test for participants as signed to the no-feedback condition (control). We made no specific hypotheses about mid-span partic ipants, and an analysis of their performance revealed no significant changes ac ross the three conditions. Figure 3-1 illustrates the differences in performance (by absolute sp an score) between HWM and LWM participants for the different performance conditions: 95th percentile, 40th percentile, and control. -15 -10 -5 0 5 10 95th40thControl HWM LWM Figure 3-1. Performance differences Difficulty Questionnaire There was a small but significant difference in ratings of difficulty fr om test 1 to test 2 for participants in the control condition, t = 2.888; p < .006, and experimental conditions 95th and 40th, t = 2.09; p < .041 and t = 2.174; p < .034, respectively. Particip ants in all conditions perceived the second test to be more difficult than the first, but this difference was marginal, less than one point on the ten-point Likert scale particip ants used to rate the difficulty of each test.

PAGE 26

26 When we analyzed span capacity, we found th at HWM participants in the control and 95th percentile condition perceived the second test to be harder than that first, t = 2.54; p < .026 and t = 3.5; p < .004, respectively. The other groups of participants did not find one test to be significantly different in di fficulty than the other. We reviewed the differences in statements ma de by the participants when asked to justify their difficulty ratings of the test s. Only four of the sixty questi onnaires reviewed listed feedback or expectation as an explanation for difficulty ra ting. All four of the quest ionnaires that listed feedback or expectation as an e xplanation for difficulty rating were from participants assigned to the 40th percentile condition: three HWM and one LWM participant. Interestingly, all three HWM participants perceived the second test to be more difficult as a f unction of concentrating on the negative feedback that was provided about their performance. The opposite was reported by the LWM participant, who felt the second test to be easier than the first, but also reasoned it to be because of concentration on the negative feedback. The three HWM participants reported, Given my scores on the last test, I put more concentration into it, It was difficult to concentrate after receiving my score, and After hearing what I got on the first experiment I got nervous taking the second one. The LWM participan t reported, I found the test to be just a bit easier than the last one because I felt I had to focus more because of the score I got on the previous one. Another potential difference between highand lo wworking memory individuals could be attributed to how feedb ack, particularly negative feedback, is processed. It appears that high working memory participants may be overw helmed by the concentration on feedback and use this information counterproductively, expe riencing performance decrements on the second task. In contrast, the low working memory partic ipant concentrated on the feedback, similar to

PAGE 27

27 HWM participants, but somehow used this inform ation productively and exerted more effort to perform better on the second task. Demographic Questionnaire There were no significant differe nces in scores from test 1 to test 2 for either females or males. We were unable to analy ze ethnicity because of the unequa l distribution of ethnicity in our sample (Asian: N = 6, Black: 28, Hi spanic: 19, White: 110, Other: 4). Bell Curve When participants were asked to rate thei r own perceived level of competency at the beginning of the experiment, the av erage across all participants wa s 80. Participants, on average, rated themselves as being 30 poi nts over the average. This is consistent with Dunning, Heath, and Suls (2004) review that on average, people will overrate themselves, saying their skills are above average, a statistical impossibility. In their empirical research review, Dunning, Heath, and Suls found that self-assessments are, at best, modestly correlated with actual performance.

PAGE 28

28 CHAPTER 4 GENERAL DISCUSSION The purpose of our study was to introduce a chal lenge to participants competency and to test whether perform ance is aff ected by this challenge. Similar to the findings of stereotype threat research, we expected to find that challenging participants views of their own competency would result in decreased performance. We propos ed that, in addition to identifying with social factors (i.e., race and gender), people identify w ith a level of competency. Therefore, challenging ones believed competency would be similar to challenging a part of thei r identity. Our results provide evidence that under ce rtain conditions, performance on working memory tasks can fluctuate substantially (See Figure 1). More impo rtantly, the differences in type of feedback given may either motivate or hinder peoples performance as a function of their working memory span. We hypothesized that (1) high working memo ry participants in a high pressure performance situation, like the 95th percentile condition, would perf orm lower on the second task because they would choke under the pressure of maintaining performance in the 95th percentile. In contrast, we hypothesized (2) th at HWM participants in the 40th percentile condition would excel after hearing negative feedback because of their ability to overcome pressure in real-life performance pressure situations. Consistent w ith our hypothesis, HWM pa rticipants showed a deficit in performance when given positive feedback and high expectations for their performance. This is also consistent with Beilock and Carrs (2005) claim that under high pressure performance situations, HWM particip ants show decrements in performance. Surprisingly, HWM participants also did signifi cantly worse after recei ving negative feedback and low expectations for future performance. We originally assumed that because HWM participants are likely to be high-achievers in many real-lif e situations like learning and

PAGE 29

29 comprehension (Engle, Kane, & Tuholski, 1999) they may possess skills or strategies to overcome pressure. However, it may be the case that HWM participants view both conditions as high pressure and their performance decrement can be explained thr ough the reallocation of attentional resources from task-relevant information to task-irrelevant worries. Recall that Beilock and Carr (2005) classified their participants as choki ng when they suffered a seven percent performance decrement. Based on the pe rformance from our study, HWM participants in both the 95th and 40th percentile condition experienced choki ng; their performance suffered 16% and 9% in accuracy span scor es, respectively. Similar to other stereotype threat resear ch, we hypothesized (3) that low working memory participants in a competency threatening situation, similar to the 40th percentile condition, may experience fear of confirming a negative stereotype, thus experiencing a performance decrement after hear ing the feedback and low expect ation for future performance. We also predicted (4) that LWM participants in the 95th percentile condition would not experience any performance differences because th ey would not be threatened by the situation, thus not experiencing the added pressure. Beilock and Carr (2005) concluded that LWM participants do not suffer from the same disa dvantage in high pressure situations as HWM participants, as evidenced by their finding that pe rformance remained relatively stable in their high pressure condition. We expand on Beilock a nd Carrs results by demonstrating that under certain performance situations particularly when providing negative feedback and low expectations for future performance, LWM par ticipants may actually improve in performance, scoring significantly higher afte r receiving feedback. This result is in direct contrast to our hypothesis and is the direct opposite of the findings from previous stereotype threat studies (Cadinu et al., 2005; Shih, Pittinsky, & Ambady, 1999). We speculate that negative feedback

PAGE 30

30 might have acted as a motivator for our participan ts who had lower working memory scores. It is important to note that while LWM participan ts performance increased after being given feedback, they still underperformed relative to HWM participants. Low working memory participants may interpre t feedback differently and use this information productively for future performance. An alternative sp eculation would be that LWM pa rticipants may have expended attentional reserves during the second test, utilizing ex tra, normally untapped, attentional resources. Obviously, further empirical work will be necessary to substantiate these speculations. It may be that certain types of interference (i.e., feedback information, expectations, feelings of pressure, or threatening situations) tap in to these attentional reserves, allowing LWM participants to make use of normally unta pped resources during subsequent performance situations. Our results su pport our hypothesis about LWM participants in the 95th percentile condition; they were unaffected by the positiv e feedback and high expectation for future performance. Probably due to the absence of pe rformance pressure to overcome competency threat, LWM participants remained in a neutral st ate, performing similarly on test 1 and test 2. We also predicted (5) that participants in the control condition would perform similarly on both tests and (6) that all part icipants in the experimental conditions would report the second test to be harder than the first. The results support our hypothesis about the control condition; all participants performed similarl y on test 1 and test 2 when given no feedback, suggesting that there were no practice effects present, and that te st 2 and test 1 were similar in difficulty level. After analyzing the difficulty questionnaires, we found that all participants found the second test to be more difficult than the firs t, irrespective of what condition they were assigned to. When we examined ratings between highand lowspan pa rticipants in both experimental conditions, we

PAGE 31

31 found that HWM participants in the 95th percentile condition percei ved the second test to be significantly more difficult than the first. Results from our study broaden the literatur e on individual differences between span types. The capability to contro l attention, especially when th ere are competing demands, is a determining factor in ones working memory capacity (WMC). After analyzing the data, we ruled out the idea that fatigue negatively influe nced WMC scores. There were no differences in performance from test 1 to test 2 among any of the three different span types in our control condition. Differences in performance were onl y elicited by the manipulation. One measurable difference of WMC between highand lowworking memory persons is the ability to control attention so that one can remain attentive to the task in the face of task -irrelevant distractions, regardless of whether distractions are externally or internally generated (Barrett, Tugade, & Engle, 2004). High working memory participants ability to focus thei r attention (i.e., on performance pressure) limits their awareness of other information, constraining the range that they may attend to or process (Conway, Cowan, & Bunting, 2001). Additionally, Wiley (1998) found that, in a creative problem-solving task, pr esenting misleading information for the purpose of activating misleading solution attempts caused HWM participants to fixate their attention on the misleading information. This reallocation of attentional resources may negatively impact WMC and the ability to attend to all the information in the task. It may be the case that HWM capacity can sometimes cause too much attentiona l focus and if the focus is on task-irrelevant information, task performance will suffer. High working memory participants in both the 95th and 40th percentile condition may have felt pressure to fulfill high expectations or pressure to overcome failure, respectively, thus drawing their attention away from the working memory task.

PAGE 32

32 Both external (i.e., feedback and expectation) and internal (i.e., thoughts and feelings) distracters can draw attentional resources away from a task (Barrett et al., 2004). An alternate explanation that could highlig ht the differences between highand lowworking memory participants is the use of feedback. Based on a few explanations given by participants, we found that both highand lowworking memory participan ts concentrated on the negative feedback, but HWM participants felt the second test to be more difficult and experienced a decrease in performance. The LW M participant who attrib uted their performance on the second test to feedback felt the second test to be easier, so an argument could be made to say that LWM participants use feedback info rmation productively while HWM participants use feedback counterproductively. We only collected da ta on one LWM participant, so this argument is speculative and will need to be researched in the future. Results from our study also add to the research on expectation and choking under pressure. Literature suggests that when a certain level of performance is expected (either high or low), the added pressure is believed to reduce ones working memory capacity causing a deficit in performance by reallocating attentional reso urces to task-irrelevant concerns. The added anxiety associated with threats and expectations illustrate diffe rences in performance between highand lowworking memory persons. In a hi gh stakes testing situation, HWM participants might choke under pressure or re allocate their attentional resour ces, but for LWM participants the added pressure of performance may increase th eir performance. The performance differences between highand lowworking memory capacity pa rticipants are especially highlighted in our 40th percentile condition, where participants receiv ed negative feedback and low expectations for future performance. Grouping participants into performance groups and projecting expectations in correlation with those groups is similar to the grouping people experience in real-life learning

PAGE 33

33 situations. Consequently, the findings of our rese arch are likely to appl y to real-life testing situations. There has been considerable research exam ining stereotype thr eats in the education system and ways to alleviate the correspondi ng performance pressure. Given our findings, educators and policy makers may want to reexam ine how students are placed into particular scoring categories, such as perc entile groups, or groups formed by ability; where performance is highlighted in comparison to peers. Unknowi ngly, some students may interpret positive or negative feedback productively or counterproductively. In accordance with our results, if it is assumed that HWM participants are genera lly high achievers, highachieving students may interpret both positive and negative feedback as pressure, causing them to choke in both situations. Consequently, teachers may want use caution in giving specific feedback to highachieving students because of the students possibl e interpretations of th e feedback. However, for low-achieving students (if it assumed WMC correlates to school achievement) it may be beneficial to highlight their performance in or der to motivate them to perform past their own expectations. Working memory capacity is correlated with a variety of real-world cognitive tasks (Engle et al., 1999) including reading comprehens ion, language comprehension, learning to spell, and writing; a plethora of other skills needed to excel in an academic setting are affected. If working memory is not as stable and consistent as originally believed, there may be a range of motivators that affect performance on working memory span tasks. It is important to understand how individual differences may affect performan ce, reactions to high-pressure situations, and how feedback is interpreted, especially when discussing real-world cognitive tasks.

PAGE 34

34 CHAPTER 5 LIMITATIONS While the findings from our study help to move forward our understanding of working memory, a few limitations and shortcomings do exist. For example, because some of our results were unexpected, we do not have an accurate m easure or assessment as to why the specific changes in performance occurred for the different span types. Our difficulty questionnaire was not as informative as we had hoped. Only four participants referred to the feedback and expectations as a function of effort change toward the subsequent task. A more detailed selfassessment could be used in future work to spec ifically prime responses related to feelings of pressure, effort change, and motivation toward the second task. Without a detailed assessment, our explanations for the surprising resu lts should be inte rpreted with caution.

PAGE 35

35 CHAPTER 6 FUTURE RESEARCH Future research should assess, in more deta il, the cognitive processes that underlie the discrepancies in performance as a function of fe edback and perceived performance pressure. A self-report or questionnaire assess ing motivation or feelings of pressure may provide additional information as to why different types of achie vers are affected differently by feedback and projections of expectation. Li kewise, a questionnaire may exam ine participants reports of potential effort changes between test 1 to test 2 and how they were affected by feedback and expectations for future performance. Instead of asking open-ended questi ons, future researchers should consider developing a mo re structured questionnaire, asking specific questions about underlying processes that may have occurred (e.g. Did you put forth more effort/less effort?, Did you feel more pressure given your feedback?). Additionally, future research may want to incl ude another cognitive task that is correlated to working memory capacity and is more reflectiv e of tasks in real-life learning situations. By including a typical cognitive tas k, like a vocabulary test or re ading comprehension assignment, the findings from the study could more likely generalize to classroom settings. That is, findings from the study can specifically relate to differe nt types of achievers a nd how different types of students are likely to interpre t feedback and expectations.

PAGE 36

36 APPENDIX A BELL CURVE SURVEY Date:_______________Subject#_________________Experim ent:________________ Please indicate on this normal distribution bell curve where your competency lies compared to your peers.

PAGE 37

37 APPENDIX B DEMOGRAPHIC QUESTIONAIRE Date:_______________Subject#_________________Experim ent:________________ What is your age? _____ What is your gender? o Female o Male How would you classify yourself? o American Indian or Native Alaskan o Asian or Pacific Islander o Black (not of Hispanic origin) o Hispanic or Latino o White o Other: __________________

PAGE 38

38 APPENDIX C DIFFICULTY QUESTIONNAIRE Date:_______________Subject#_________________Experim ent:________________ How difficult did you perceive this test to be? 1 2 3 4 5 6 7 8 9 10 Please give a brief explanation of why you gave the test this rating.

PAGE 39

39 APPENDIX D SAMPLE SCRIPT Here are the printed results of your perform ance on the reading span task (RSPAN). This is your score and based on your perfor mance you fall into the perc entile group. This means that your test score is greater/less than % of the scores of people taking the same test and essentially, of measurable intel ligence since WMC is related to fl uid intelligence. So, based on your performance we are expecting you to do very well/expecting you to have more difficulty with this second test.

PAGE 40

40 REFERENCES Baddeley, A.D. (2002). Is working mem ory still working? [Electronic version]. European Psychologist, 7, 85-97. Barrett, L.F., Tugade, M.M., & Engle, R.W. (2 004). Individual differences in working memory capacity and dual-process theories of the mind. Psychological Bulletin, 130(4), 553-573. Baumeister, R.F. (1984). Choking under pressu re: Self consciousness and paradoxical effects of incentives of skillful performance. Journal of Personality and Social Psychology, 46, 610-620. Baumeister, R.F., Hamilton, J.C., & Tice, D.M. (1985). Public versus private expectancy of success: Confidence booster or performance pressure? [Electronic version]. Journal of Personality and Social Psychology, 48, 1447-1457. Beilock, S.L., & Carr, T.H. (2001). On the fragility of skilled performance: What governs choking under pre ssure? [Electronic version]. Journal of Educational Psychology: General, 130, 701-725. Beilock, S.L., & Carr, T.H. (2005). When high-powered people fail: Working memory and choking under pressure in math [Electronic version]. American Psychological Society, 16, 101-105. Beilock, S.L., Kulp, C.A., Holt, L.E., & Carr, T.H. (2004). More on the fragility of performance: Choking under pressure in mathematical problem solving [Electronic version]. Journal of Experimental Psychology: General, 133, 584-600. Borella, E., Carretti, B., & Mammarella, I.C. (2006). Do working memory and susceptibility to interference predict individual differences in fluid intelligence? [Electronic version]. European Journal of Cognitive Psychology, 18, 51-69. Cadinu, M., Maass, A., Rosabianca, A., & Kiesner, J. (2005). Why do women underperform under stereotype threat?: Evidence for the role of negative thinking [Electronic version]. American Psychological Society, 16, 572-578. Cheryan, S.C., & Bodenhausen, G.V. (2000). When positive stereotypes threaten intellectual performance: The psychol ogical hazards of model minority status [Electronic version]. American Psychological Society, 11, 399-402. Conway, A.R., Cowan, N.C., & Bunti ng, M.F. (2001). The cocktail party phenomenon revisited: The importance of working memory capacity. Psychonomic Bulletin and Review, 8(2), 331-335.

PAGE 41

41 Conway, A.R.A., Kane, M.J., Bunting, M.F., Hamb rick, D.Z., Wilhelm, O., & Engle, R. (2005). Working memory span tasks: A me thodological review and users guide. Psychonomic Bulletin and Review, 12 (5), 769-786. Daneman, M., & Carpenter, P.A. (1980). Indivi dual-differences in working memory and Reading [Electronic version]. Journal of Verbal Learning and Verbal Behavior, 19, 450-466. Dunning, D., Heath, C., & Suls, J.M. (2004). Flawed self-assessment: Implications for health, education, and the workplace. Psychological Science in the Public Interest, 5(3), 69-106. Engle, R.W., Kane, M.J., & Tuholski, S.W. (1999). Individual diffe rences in working memory capacity and what they tell us about controlled atte ntion, general fluid intelligence, and function of the prefrontal cortex. In A. Miyake & P. Shah (Eds.), Models of working memory: Mechanisms of active maintenance and executive control (pp. 102). New York: Cambridge University Press. Engle, R.W., Tuholski, S.W., Laughlin, J. E., & Conway, A.R.A. (1999). Working memory, short-term memory, and gene ral fluid intelligence: A latent-variable approach [Electronic version]. Journal of Experimental Psychology: General, 128, 309-331. Jones, E.E., & Berglas, S. (1978). Control of attributions about the self through selfhandicapping strategies: The appeal of alcohol and the role of underachievement [Electronic version]. Personality and Social Psychology Bulletin, 4, 200-206. Redick, T.S., & Engle, R.W. (2006). Worki ng memory capacity and attention network test performance. Applied Cognitive Psychology, 20 713-721. Sarason, I.G. (1961). Text anxiety and the in tellectual performance of college students [Electronic version]. Journal of Educational Psychology, 52, 201-206. Sarason, I.G. (1984). Stress, anxiety, and c ognitive interference: R eactions to tests [Electronic version]. Journal of Personality and Social Psychology, 46, 929-938. Schmader, T., & Johns, M. (2003). Converging ev idence that stereotype threat reduces working memory capacity [Electronic version]. Journal of Personality and Social Psychology, 85, 440-452. Shih, M., Pittinksky, T.L., & Ambady, N. (1999). Stereotype susceptibility: Identity salience and shifts in quantit ative performance [Electronic version]. Psychological Science, 10, 80-83.

PAGE 42

42 Smith, J.L., & Johnson, C.S. (2006). A ster eotype boost or choking under pressure? Positive gender stereotypes and men who are low in domain identification [Electronic version]. Basic and Applied Social Psychology, 28, 51-63. Steele, C.M. (1997). A threat in the air: How stereotypes shap e intellectual identity and performance [Electronic version]. American Psychologist, 52, 613-629. Tobias, S. (1985). Test anxiety: Interference, defective skills, and cognitive capacity [Electronic version]. Educational Psychologist, 20, 135-142. Wiley, J. (1998). Expertise as mental set: The effects of domain knowledge in creative problem solving. Memory and Cognition, 26 (4), 716-730. Wine, J. (1971). Test anxiety and direc tion of attention [Electronic version]. Psychology Bulletin, 76, 92-104.

PAGE 43

43 BIOGRAPHICAL SKETCH I received a Bachelor of Science in psyc hology from the University of Florida in May 2006. I continued to the masters program in edu cational psychology at the University of Florida in August 2006. Since then, I have worked on a number of successful research projects and was given the opportunity to teach a course in fall 2007 and spring 2008. After completing my masters degree I will continue to pursu e my Ph.D. in educational psychology.