psychological consequences

Enhancing feedback and improving feedback: subjective perceptions, psychological consequences, behavioral outcomes Constantine Sedikides1, Michelle A. Luke2, Erica G. Hepper3

1Psychology Department, University of Southampton 2School of Business, Management and Economics, University of Sussex 3School of Psychology, University of Surrey

Correspondence concerning this article should

be addressed to Constantine Sedikides,

Psychology Department, Center for Research

on Self and Identity, University of

Southampton, Southampton SO17 1BJ,

England, UK. E-mail: cs2@soton.ac.uk

doi: 10.1111/jasp.12407

Abstract

Three experiments examined subjective perceptions, psychological consequences,

and behavioral outcomes of enhancing versus improving feedback. Across experi-

ments, feedback delivery and assessment were sequential (i.e., at each testing junc-

ture) or cumulative (i.e., at the end of the testing session). Although enhancing

feedback was seen as more satisfying than useful, and improving feedback was not

seen as more useful than satisfying, perceptions differed as a function of short-term

versus long-term feedback delivery and assessment. Overall, however, enhancing

feedback was more impactful psychologically and behaviorally. Enhancing feedback

engendered greater success consistency, overall satisfaction and usefulness, optimism,

state self-esteem, perceived ability, and test persistence intentions; improving feed-

back, on the other hand, engendered greater state improvement. The findings pro-

vide fodder for theory development and applications.

Feedback is a common occurrence in daily life. Employees,

students, actors, or athletes receive it frequently from their

managers, instructors, directors, or coaches, respectively. A

body of literature attests to its relevance. Feedback, for exam-

ple, may contribute to the formation of competence self-

views and intrinsic task values (Gniewosz, Eccles, & Noack,

2014; Harackiewicz, 1979). It may also influence subsequent

responses, including job performance (Brown, Hyatt, & Ben-

son, 2010; Whitaker & Levy, 2012) and educational attain-

ment (Hattie & Timperley, 2007; Kluger & DeNisi, 1996).

Such responses, however, may not be what the feedback

giver (e.g., manager, teacher) had in mind (Fisher, 1979;

Gabriel, Frantz, Levy, & Hilliard, 2014; Kluger & DeNisi,

1996) and may not necessarily be in the recipient’s (e.g.,

employer’s, student’s) best interest (Gregory & Levy, 2012;

Ilgen & Davis, 2000; Kulhavy, 1977). Therefore, understand-

ing how recipients perceive the feedback in the first place is

crucial, if well-meaning evaluators wish to shape effectively

recipient responding for organizational or educational bene-

fit, or if recipients wish to maximize feedback-derived advan-

tages (Atwater & Brett, 2005; Brett & Atwater, 2001; Hattie &

Timperley, 2007). Do recipients, for example, perceive feed-

back as satisfying or useful? Perceptions of satisfaction and

usefulness are arguably prerequisites for recipients to engage

with and benefit from feedback. Understanding the psycho-

logical consequences and behavioral outcomes of feedback is

equally important. How do recipients, for example, feel about

and respond to feedback that aims at satisfying them versus

improving them? We explore, in this article, comparative per-

ceptions of enhancing and improving feedback, as well as

some of its potential psychological consequences (i.e., opti-

mism, state self-esteem, state improvement, perceived ability)

and behavioral outcomes (i.e., persistence intentions).

Background and scope

The bulk of the literature has been concerned with the critical

(i.e., negative) versus enhancing (i.e., positive) dimension of

feedback. This literature, for example, has examined critical

and enhancing feedback in terms of recall, goal pursuit, or

performance (Fishbach, Eyal, & Finkelstein, 2010; Sedikides,

Green, Saunders, Skowronski, & Zengel, 2016), perceptions

of one’s competence or the evaluator (Aronson & Linder,

1965; Vallerand & Reid, 1984), and judgments of test validity

or credibility (Campbell & Sedikides, 1999; Wyer & Frey,

1983). A generalized statement based on this large literature

VC 2016 Wiley Periodicals, Inc Journal of Applied Social Psychology 2016, 46, pp. 687–700

Journal of Applied Social Psychology 2016, 46, pp. 687–700

is that, on balance, enhancing feedback is seen as more satis-

fying and useful than critical feedback (Brett & Atwater,

2001; Hepper & Sedikides, 2012; Hsee & Abelson, 1991; Sedi-

kides & Gregg, 2008; Sutton, Hornsey, & Douglas, 2012).

Little research, however, has addressed another pivotal

feedback dimension, enhancing versus improving. For the

purposes of our research, enhancing feedback will refer to

consistently positive information linked to task performance,

whereas improving feedback will refers to an upward infor-

mation trajectory linked to task performance. How enhanc-

ing versus improving feedback is perceived, felt, and reacted

upon is not well understood. This is somewhat surprising,

given the growing presence of improvement motivation (e.g.,

the desire to improve) in the self-evaluation literature

(Breines & Chen, 2012; Collins, 1996; Green, Sedikides, Pin-

ter, & Van Tongeren, 2009; Heine & Raineri, 2009; Kurman,

2006; Pyszczynski, Greenberg, & Arndt, 2012; Sedikides,

2009). Do individuals perceive one type of feedback as more

satisfying or more useful than the other? Do the two feedback

types elicit different psychological and behavioral reactions?

Are perceptions, psychological consequences, and behavioral

outcomes influenced by repeated (i.e., multiple-occasion)

feedback delivery?

We explored, in three experiments, how subjective percep-

tions, psychological consequences, and behavioral outcomes

are impacted within a particular type of feedback and also

between types of feedback. We were concerned with task level

feedback (i.e., how well tasks are performed; Hattie & Tim-

perley, 2007) and externally-framed (rather than internally-

framed) feedback (MoEller, Pohlmann, Koeller, & Marsh,

2009). Further, we focused on feedback that was (a) based on

multiple testing occasions; (b) delivered to recipients sequen-

tially (i.e., at each testing juncture) or cumulatively (i.e., at

the end of the testing session); and (c) assessed (in terms of

perceptions, psychological consequences, and behavioral out-

comes) sequentially or cumulatively. Enhancing feedback was

consistently positive (e.g., percentile rankings in relation to

other test-takers could be 92, 90, 91, and 92 across four ses-

sions), whereas improving feedback tracked an upward per-

formance trajectory (e.g., percentile rankings in relation to

other test-takers could be 59, 68, 81, and 92 across four

sessions).

Theoretical and practical considerations

Our exploratory foray was informed by two contrasting theo-

retical perspectives. The self-enhancement perspective posits

that individuals strive mostly for information positivity, with

information improvement value playing a secondary hand

(Alicke & Sedikides, 2011; Brown & Dutton, 1995; Dunning,

2005; Hepper, Gramzow, & Sedikides, 2010; Sedikides &

Strube, 1997). This perspective predicts that enhancing (i.e.,

uniformly-positive) feedback will be perceived as more satis-

fying than improving (i.e., upward-trajectory) feedback, and

also as generally more satisfying than useful, because of its

hedonic tone. The perspective also anticipates that enhancing

feedback will exert stronger psychological and behavioral

impact than improving feedback. The self-improvement per-

spective, on the other hand, posits that individuals strive

mostly for improvement information, giving secondary

importance to information positivity (Gregg, Sedikides, &

Gebauer, 2011; Markman, Elizaga, Ratcliff, & McMullen,

2007; Prelec & Loewenstein, 1997; Sedikides & Hepper, 2009;

Taylor, Neter, & Wayment, 1995). This perspective predicts

that improving feedback will be perceived as more useful

than enhancing feedback, and also as generally more useful

than satisfying, because of its utilitarian value. Further, this

perspective anticipates that improving feedback will have

greater psychological and behavioral impact than enhancing

feedback. Although the two perspectives make general pre-

dictions about perceptions of feedback, they do not offer spe-

cific enough guidance about perceptions of feedback at

distinct junctures of delivery or assessment; this is a matter of

exploration.

Not only will the investigation of perceptions, psychologi-

cal consequences, and behavioral outcomes of enhancing and

improving feedback stretch the scope of the self-

enhancement and self-improvement perspectives, but it will

also address external validity issues. In ecological settings

(e.g., occupational environments, classrooms, artistic per-

formances, athletic events), feedback is often targeted toward

both enhancement and improvement, while being delivered

on multiple (as opposed to single) occasions. In addition, in

organizational settings, employees appear to desire, not just

self-enhancement feedback, but constructive or self-

improvement feedback, if one were to consult popular busi-

ness coaching and training books (e.g., Silberman & Hans-

burg, 2005). Self-improvement motivation has indeed been

investigated in such settings as organizations (Seifert, Yukl, &

McDonald, 2003), university enrolment (Clayton & Smith,

1987), the classroom (Harks, Rakoczy, Hattie, Besser, &

Klieme, 2014; Ryan, Gheen, & Midgley, 1998), volunteering

(Dickinson, 1999), correctional facilities (Neiss, Sedikides,

Shahinfar, & Kupersmidt, 2006), and enlistment in the army

(Pliske, Elig, & Johnson, 1986); however, perceptions of

improving feedback juxtaposed to perceptions of enhancing

feedback, as well as comparative psychological consequences

and behavioral outcomes, have not been addressed.

Perceptions of feedback satisfaction and usefulness ought

to be investigated for both theoretical and practical reasons.

Satisfaction reflects the affective and valence focus of the self-

enhancement motive, whereas usefulness reflects the con-

structive focus of the self-improvement motive. Moreover, in

organizational settings for example, it is arguably vital for

feedback (e.g., appraisals) to be perceived as useful in order

688 Enhancing and improving feedback

VC 2016 Wiley Periodicals, Inc Journal of Applied Social Psychology, 2016, 46, pp. 687–700

for staff to engage with both feedback and management in a

mutually beneficial manner. In addition, organizations, espe-

cially those competing for talent, are often under pressure to

devise ways to keep their staff satisfied.

Experiment 1: sequential feedback delivery and cumulative feedback assessment

In Experiment 1, we addressed, for the first time, subjective

perceptions of self-enhancing and self-improving feedback.

We note that in this and all subsequent experiments, we (a)

randomly assigned participants to between-subjects factors

of balanced designs, (b) tested participants in individual

cubicles, and (c) obtained no sex differences or counterbal-

ancing order effects.

Participants were under the impression that they were test-

ed in four key domains of human functioning: creativity, ver-

bal intelligence, social sensitivity, analytical ability. Numerical

feedback, either enhancing or improving, was delivered at

several (i.e., four) junctures, and feedback perceptions were

assessed cumulatively at the end of the testing session. The

starting point for enhancing and improving feedback was dif-

ferent (positive for enhancing, average for improving), but

the end-point was identical (i.e., positive). While providing a

preliminary test of the self-enhancement and self-improvement

perspective, the experiment simulated multiple-occasion feed-

back delivery to employees, students, actors, or athletes by a

supervisor, instructor, director, or coach, respectively. Would

such feedback be perceived as satisfying or useful at the end of

a business quarter, academic semester, rehearsal period, or ath-

letic event?

Method

Participants and design

Participants were 102 introductory psychology students at

University of North Carolina in Chapel Hill (71 female, 31

male), who volunteered for course credit. Information about

participant age is unavailable, due to a coding error. Never-

theless, the vast majority of participants were traditional stu-

dents, aged between 18 and 22 years. The design was a 2

(feedback type: enhancing, improving) 3 2 (feedback rating: satisfaction, usefulness) mixed factorial, with repeated mea-

sures on the latter factor.

Procedure and measures

Participants learned that they would be assessed on four piv-

otal domains of human functioning: creativity, verbal intelli-

gence, social sensitivity, analytic ability. The relevant tests had

ostensibly been standardized and administered to university

students since 1985 by the Educational Testing Service in

order to study the impact of the university environment on

social skills. Participants were then handed a booklet contain-

ing the tests, which were divided into four sections. They

received feedback (featuring an enhancing or improving tra-

jectory) after each section.

The first section, consisting of Raven’s Progressive Matrices

(RPM; 10 minutes), assessed creativity. Participants learned

that the RPM measures spatial perception and creativity, and

is a valid indicator of superior memory and innovative think-

ing. The RPM comprised eight questions. Participants deci-

phered a pattern in the displayed figures and selected, from

eight choices, the correct item to complete the pattern. Feed-

back followed.

The second section, consisting of the Verbal Fluency Test

(4 minutes) and the Analogies Test (5 minutes), assessed ver-

bal intelligence. Participants learned that better test scores

were associated with higher IQ and greater professional suc-

cess. For the Verbal Fluency Test, participants were given two

sets of four letters (L, C, E, N; F, O, S, P) and were asked to

generate as many 4-word sentences as possible using the

specified first letters for each word. For the Analogies Test,

participants were to complete 10 analogies. They received

three words, the first two of which were related. Their task

was to pick the word that related to the stimulus word in the

same way as the first two words. For example, the correct

answer for the analogy “Shoe: Foot:: Glove: (a. Arm, b.

Elbow, c. Hand)” would be Hand, because Hand is related to

Glove in the same way as Foot is related to Shoe. Feedback

followed.

The third section, consisting of the Perception of Relation-

ships Test (5 minutes) and the Perception of Deception Test (5

minutes), assessed social sensitivity. Participants learned that

individuals who performed well on these tasks were more

adept at solving interpersonal conflicts and had longer-

lasting relationships. We adapted the Perception of Relation-

ships Test from the Social-Cognitive Aptitude Test (Crocker,

Thompson, McGraw, & Ingerman, 1987). Participants read

paragraphs about two couples and indicated their impression

of each couple, whether the couple members were supportive

of each other, and the likelihood that each couple would still

be together in one year. In the Perception of Deception Test,

participants read two incidents (a man late for a date, a city

council member accused of neglecting to report campaign

contributions). Then participants indicated their impression

of each character, the quality of the relationship in the first

incident, the popularity of the city council member in the

second incident, and whether the main characters were lying.

Feedback followed.

The fourth and final section, consisting of the Analytical

Ability Test (9 minutes), assessed logical reasoning. Partici-

pants learned that better performance was linked with success

in careers that involve critical thinking skills. The test asked

participants to determine in what grade each of eight

Sedikides et al. 689

VC 2016 Wiley Periodicals, Inc Journal of Applied Social Psychology, 2016, 46, pp. 687–700

children was and what costume they wore in the Thanksgiv-

ing pageant. Feedback followed.

The feedback, in the form of percentile rankings in rela-

tion to other university student test-takers, was either

enhancing or improving across the test sections. In the

enhancing condition, participants received feedback that

started at a high level and remained constant. The section

scores were: 92, 90, 91, 92. In the improving condition, par-

ticipants received feedback that started relatively low and

became progressively higher. The section scores were: 59, 68,

81, 92.

Finally, participants completed the satisfaction and useful-

ness scales in counterbalanced order (1 5 not at all, 9 5 very much). The satisfaction scale comprised three questions ask-

ing how pleased, satisfied, and content participants were with

the feedback (a 5 .95). The usefulness scale comprised three questions asking how useful, helpful, and constructive partic-

ipants considered the feedback (a 5 .95). Responses to the two scale indices were correlated, r(100) 5 .50, p< .001.

Results and discussion

Satisfaction and usefulness

Overall, participants in the enhancing condition (M 5 6.53, SD 5 1.78) rated the feedback higher (i.e., perceived it as more satisfying and useful) than those in the improving con-

dition (M 5 5.65, SD 5 1.78), feedback type main effect F(1, 100) 5 6.19, p 5 .015, g2partial 5 .06. Also, participants overall perceived the feedback as descriptively but not significantly

more satisfying (M 5 6.25, SD 5 1.96) than useful (M 5 5.92, SD 5 2.27), feedback rating main effect F(1, 100) 5 2.57, p 5 .112, g2partial 5 .03.

Crucially, the interaction was significant, F(1, 100) 5 4.38, p 5 .039, g2partial 5 .04. We proceeded to calculate four com- parison tests, using the Bonferroni correction (.05/

4 5 .0125). We examined the effects of feedback type sepa- rately on satisfaction and usefulness (i.e., each level of feed-

back rating). Participants in the enhancing condition

(M 5 6.91, SD 5 1.89) perceived feedback as more satisfying than those in the improving condition (M 5 5.59, SD 5 1.81), t(100) 5 3.58, p 5 .001, d 5 0.77; however, par- ticipants in the enhancing (M 5 6.14, SD 5 2.37) and improving (M 5 5.70, SD 5 2.16) conditions perceived feed- back as equivalently useful, t(100) 5 1.00, p 5 .321, d 5 0.19. We also examined the effects of feedback rating separately for

each feedback type condition (i.e., enhancing, improving).

Participants in the enhancing condition perceived the feed-

back as more satisfying than useful, t(50) 5 2.86, p 5 .006, d 5 0.40; however, participants in the improving condition perceived the feedback as equivalently satisfying and useful,

t(50) 5 20.32, p 5 .750, d 5 20.04.

Summary

Overall, participants regarded enhancing (compared to

improving) feedback as more satisfying. Furthermore, they

regarded enhancing feedback as more satisfying than use-

ful, whereas they regarded improving feedback as equiva-

lently useful and satisfying. Although these findings are

generally consistent with the self-enhancement perspec-

tive, it is possible that the design of Experiment 1 did not

allow for a fair test of the self-improvement perspective. In

particular, the delivery and assessment of the feedback

may have afforded limited opportunities for improvement,

thus reducing the feedback’s utilitarian value. Experiment

2 addressed this potential limitation.

Experiment 2: sequential feedback delivery and sequential feedback assessment

In Experiment 2, we asked a more focused question: Do par-

ticipants perceive the two feedback types (i.e., enhancing and

improving) differently when feedback is both delivered and

assessed at each performance juncture? Participants were

under the impression that they were tested in the same four

key domains as in the previous experiment. We delivered

feedback, either enhancing or improving, at several junctures

and assessed feedback perceptions separately at each juncture

(Ariely, 1998; Ilies, Nahrgang, & Morgeson, 2007; Tonidan-

del, Qui~nones, & Adams, 2002). This experiment simulated

situations such as the appraisal of multiple-occasion

(enhancing or improving) feedback administered to employ-

ees, students, actors, or athletes over the course of a business

quarter, academic term, rehearsal period, or athletic event.

Will recipients perceive such feedback as satisfying or useful

on each occasion? In addition, this experiment examined a

potential psychological consequence of feedback, optimism

about performance on future aptitude tests. Will enhancing

or improving feedback elicit higher optimism at the end of

the testing session (i.e., cumulatively)? This was an open-

ended question, as the relevant literature is equivocal (Sedi-

kides, 2012; Sedikides & Hepper, 2009; Taylor & Brown,

1988).

Method

Participants and design

Sixty University of Southampton undergraduates (35 female,

6 male, 19 undeclared; MAGE 5 19.27, SDAGE 5 3.21) partici- pated in exchange for course credit. We excluded (on an a

priori basis) 10 additional participants due to incomplete

responses (n 5 3), errors during data collation (n 5 6), or suspicion (n 5 1). The design was a 2 (feedback type: enhancing, improving) 3 2 (feedback rating: satisfaction,

690 Enhancing and improving feedback

VC 2016 Wiley Periodicals, Inc Journal of Applied Social Psychology, 2016, 46, pp. 687–700

usefulness) 3 4 (time: 1, 2, 3, 4) mixed factorial, with repeat- ed measures on the last two factors.

Procedure and measures

Under a pretext similar to that of Experiment 1, participants

completed four testing sections via computer and received

feedback (enhancing or improving) following each one. Dis-

tinctly from Experiment 1, they also indicated their percep-

tions of feedback following each section.

The first section, consisting of the Uses Test (6 minutes),

assessed creativity. Participants generated as many uses as

possible for a candle, a brick, and a spoon (Sedikides, Camp-

bell, Reeder, & Elliot, 1998). The second section, consisting

of the Verbal Fluency Test (4 minutes) and the Analogies Test

(5 minutes), assessed verbal intelligence and was the same as

in Experiment 1. The third section, consisting of the Percep-

tion of Relationships Test (5 minutes) and the Perception of

Deception Test (5 minutes), assessed social sensitivity and was

virtually identical to that of Experiment 1. The fourth and

final section, consisting of an Analytical Capacity Test (10

minutes), assessed logical thinking by asking participants to

decipher the full names and habitual situations of several per-

sons who had recently moved house.

After each section, participants received computer-

administered feedback, which represented a percentile rank-

ing in relation to other university student test-takers. In the

enhancing condition, the feedback started and ended at a

high level (92, 90, 91, 92). In the improving condition, the

feedback started low and increased steadily (59, 68, 81, 92).

Four times (i.e., once after each feedback administration),

participants completed the satisfaction (as> .88) and then usefulness (as> .86) scales used in Experiment 1. Responses to the two scales at each administration time were correlated,

rs(58)> .44, ps< .001.

At the end of the testing session, participants completed a

3-item optimism measure. The items assessed optimism

about performance on future aptitude tests (10 5 low, not at all, 100 5 high, very much). They were: “Using the percentile scores below, how do you expect to perform on aptitude tests

in the future?,” “How confident are you about your ability to

successfully perform on aptitude tests in the future?,” and

“How optimistic are you about your ability to excel at apti-

tude tests in the future?” (a 5 .78). Finally, given the positive relation between optimism and

mood (Cheung et al., 2013; Segerstrom, Taylor, Kemeny, &

Fahey, 1998), we included a mood measure in order to rule

out the possibility that participants in the improving condi-

tion were in a negative mood due to their low performance

(e.g., 59th percentile) on a valued dimension and therefore

less optimistic. Specifically, all participants indicated how

sad, blue, content, happy, pleased, and unhappy (Martin,

Abend, Sedikides, & Green, 1997) they were currently feeling

(1 5 not at all, 5 5 extremely; a 5 .86). Participants in the improving condition (M 5 3.79, SD 5 0.75) did not differ significantly from those in the enhancing condition

(M 5 4.06, SD 5 0.61), F(1, 58) 5 2.42, p 5 .125, g2partial 5 0.04. Thus, the reported results cannot be attributed to

between-condition mood differences and the mood variable

is not discussed further.

Results and discussion

Satisfaction and usefulness over time

In replication of Experiment 1, overall participants in the

enhancing condition (M 5 6.37, SD 5 1.07) perceived the feedback as more satisfying and useful compared to those in

the improving condition (M 5 5.74, SD 5 1.07), feedback type main effect F(1, 58) 5 5.15, p 5 .027, g2partial 5 0.08. Also, consistent with Experiment 1’s directional pattern, par-

ticipants overall perceived the feedback as more satisfying

(M 5 6.81, SD 5 0.96) than useful (M 5 5.30, SD 5 1.50), feedback rating main effect F(1, 58) 5 76.80, p< .001, g2partial 5 0.59. Neither the time main effect, F(2, 116) 5 0.38, p 5 .685, g2partial 5 0.007, nor the feedback type 3 feedback rating interaction, F(1, 58) 5 1.15, p 5 .289, g2partial 5 0.02, were significant. However, the feedback type 3 time interaction, F(2, 116) 5 22.50, p< .001, g2partial 5 0.28, as well as the feedback rating 3time interaction, F(3, 154) 5 8.64, p< .001, g2partial 5 0.13, were significant.

Crucially, the significant effects were qualified by the

three-way interaction, F(3, 154) 5 4.56, p 5 .006, g2partial 5 0.07 (Figure 1). We conducted two 2 (feedback type) 3 4 (time) Analyses of Variance (ANOVAs), followed by pairwise

comparisons with Bonferroni correction, for each level of

feedback rating—that is, separately for satisfaction (.05/

4 5 .0125) and usefulness (.05/4 5 .0125). First, we examined satisfaction. A 2 (feedback type) 3 4

(time) mixed ANOVA revealed a significant interaction, F(2,

129) 5 32.86, p< .001, g2partial 5 0.36. The linear trend for time differed by feedback type, F(1, 58) 5 54.97, p< .001, g2partial 5 0.49. Although the linear trends were significant for the enhancing condition, F(1, 29) 5 10.19, p 5 .003, g2par- tial 5 0.26, and improving condition, F(1, 29) 5 54.33, p< .001, g2partial 5 0.65, the effect of the trend was greater for the improving condition. Thus, participants perceived the

enhancing feedback as less satisfying over time, but perceived

improving feedback as more satisfying over time (Figure 1).

Pairwise comparisons of feedback type showed that partici-

pants in the enhancing condition were more satisfied than

those in the improving condition at time 1, t(58) 5 8.52, p< .001, d 5 2.20, and at time 2, t(58) 5 4.29, p< .001, d 5 1.10, but not at time 3 or 4, ts(58)< |1.50|, ps> .141, ds< |0.40|.

Sedikides et al. 691

VC 2016 Wiley Periodicals, Inc Journal of Applied Social Psychology, 2016, 46, pp. 687–700

We proceeded with examining usefulness. The feedback

type 3 time interaction was again significant, F(2, 125) 5 7.71, p 5 .001, g2partial 5 0.12, with the linear trend differing by feedback type, F(1, 58) 5 12.64, p 5 .001, g2partial 5 0.18. The linear trend was significant in the enhancing condi-

tion, F(1, 29) 5 11.61, p 5 .002, g2partial 5 0.29, but not in the improving condition, F(1, 29) 5 2.72, p 5 .110, g2partial 5 0.09. Given that the means decreased over time, we conclude

that participants perceived enhancing feedback as less use-

ful over time (Figure 1). Pairwise comparisons revealed

that participants in the enhancing condition found feed-

back marginally more useful than those in the improving

condition at time 1, t(58) 5 3.60, p 5 .001, d 5 0.92, but not at time 2, t(58) 5 1.60, p 5 .115, d 5 0.45, nor time 3 or 4, ts(58)< |.661|, ps> .510, ds< |0.21|. Together, as

illustrated in Figure 1, these patterns demonstrate that

feedback was perceived as more satisfying over time in the

improving condition but not in the enhancing condition,

and was perceived as less useful over time in the enhancing

condition but not in the improving condition.

Optimism

Participants in the enhancing condition expressed more opti-

mism (M 5 73.11, SD 5 11.84) compared to those in the improving condition (M 5 67.11, SD 5 8.79), F(1, 58) 5 4.97, p 5 .030, g2partial 5 0.08.

Summary

Consistent with the findings of Experiment 1 and the self-

enhancement perspective, participants regarded enhancing

feedback as more satisfying and useful compared to improv-

ing feedback. However, several effects, which emerged due to

sequential feedback assessment, added texture to this conclu-

sion. First, participants in the enhancing feedback condition

rated the feedback as less satisfying and useful over time. Sec-

ond, participants in the improving feedback condition rated

the feedback as more satisfying but not more useful over

time. Third, participants in the enhancing condition began

by rating the feedback as more satisfying and useful than

those in the improving condition, but by time 3 and 4 this

was no longer the case. In all, participants regarded enhanc-

ing (compared to improving) feedback as more satisfying

and useful, but they did so in the short-term rather than

long-term. Finally, participants reported higher levels of opti-

mism following enhancing than improving feedback.

Experiment 3: subjective perceptions, psychological consequences and behavioral outcomes as a function of sequential feedback delivery and feedback assessment

Experiments 1–2 delivered enhancing or improving feedback

on several domains (i.e., creativity, verbal intelligence, social

sensitivity, analytical ability), although these domains were

said to exemplify “human functioning.” Nevertheless, in aca-

demic and employment settings, repeated feedback often per-

tains to a single ability domain. Moreover, arguably the

improvement value of feedback is highest when that feedback

targets a specific domain instead of spreading over multiple

domains. Therefore, in Experiment 3 we tested the replicabil-

ity of Experiment 2 findings while delivering feedback, at sev-

eral (i.e., five) junctures, about participants’ performance in

one domain: cognitive flexibility. How do recipients perceive

single-domain feedback when it is delivered and assessed

sequentially?

Experiment 3 additionally aimed to extend our prior work

in two ways. To begin, it expanded the measures of psycho-

logical outcomes to include not only optimism about future

performance, but also overall satisfaction and usefulness,

state self-esteem and state improvement, as well as perceived

ability. Also, it included a behavioral outcome, test persis-

tence intentions. Do enhancing and improving feedback

Figure 1 Satisfaction and usefulness as a function of feedback type and

time in Experiment 2. Error bars indicate standard errors of the mean.

692 Enhancing and improving feedback

VC 2016 Wiley Periodicals, Inc Journal of Applied Social Psychology, 2016, 46, pp. 687–700

affect differentially psychological consequences and behavior-

al outcomes?

Method

Participants and design

Participants (n 5 50; 32 females, 18 males; MAGE 5 20.64, SDAGE 5 2.39) were recruited from several academic depart- ments at the University of Southampton in return for course

credit or £5 payment. We excluded on an a priori basis 11 additional participants for suspicion. The design was a 2

(feedback type: enhancing, improving) 3 2 (feedback rating: satisfaction, usefulness)3 5 (time: 1, 2, 3, 4, 5) mixed factori- al design with repeated measures on the last two factors.

Procedure and measures

Participants were led to believe that they were involved in the

establishment of normative UK data on an index of cognitive

flexibility, integrative orientation (IO), which predicted per-

formance on IQ and GRE tests as well as successful manage-

ment of relational conflict. They responded to all measures

on computer.

Participants began by completing a 3-item pre-test mea-

sure of perceived IO ability. Each item required them to move

a sliding scale between two opposing anchors (e.g., 0 5 I have extremely low IO ability . . . 9 5 I have extremely high IO ability; a 5 .87).

Subsequently, participants took the ostensible IO test,

which consisted of five rounds of nine Remote Associates

Test (Mednick & Mednick, 1967) items, and lasted 10–25

minutes. Participants in the enhancing condition responded

to test items that were relatively easy in every round (as per

normative data: Bowden & Jung-Beeman, 2003; McFarlin &

Blascovich, 1984). Participants in the improving condition

responded to test items that were difficult in round 1 and

became increasingly easy, with those in round 5 being identi-

cal to those in round 5 of the enhancing condition. We

recorded the number of correct responses as a manipulation

check index of test performance.

After each round, participants received feedback in the

form of percentile scores. In the enhancing condition, feed-

back started at a relatively high level and remained there

(92, 90, 93, 91, 92). In the improving condition, feedback

started at a relatively low level and became progressively posi-

tive (54, 65, 77, 84, 92). Following each round, participants

rated the feedback on satisfaction (pleased, satisfied; a> .85) and usefulness (useful, helpful; a> .78) by moving a sliding scale between two anchors (0 5 not at all, 100 5 extremely). These ratings constituted the satisfaction and usefulness over

time measure. Responses to the two scales were weakly or

moderately correlated at each time-point, rs(48) ranging

from .22, p 5 .122 (time 5), to .48, p< .001 (time 3). Finally, at the conclusion of the testing session, participants complet-

ed, in randomized order, psychological consequences mea-

sures (i.e., overall satisfaction and usefulness, optimism,

state self-esteem and state improvement, perceived ability)

and a behavioral outcomes measure (i.e., test persistence

intentions).

Overall satisfaction and usefulness

These scales were identical to the ones used in Experiment 1

(as 5 .90). Responses to the two scales were uncorrelated, r(48) 5 .21, p 5 .149.

Optimism

This scale was similar to the one used in Experiment 2. We

reworded the three items to reflect optimism about perfor-

mance on future integrative orientation tests (0 5 low, not at all, 100 5 high, very much; a 5 .89).

State self-esteem and state improvement

One item assessed state self-esteem: “Right now, I am feeling

good about myself” (0 5 strongly disagree, 9 5 strongly agree). Six items assessed how much participants believed they had

improved during the session (0 5 not at all, 9 5 extremely; a 5 .80). Examples are: “To what extent did your ability to solve IO questions improve during the course of the test?,”

“How much progress do you feel you made over the

session?,” “To what extent was your ability to solve integra-

tion orientation questions stuck in a rut during the test?”

(reverse-scored).

Perceived ability

The same three items as the relevant pre-test measure

assessed perceived IO ability (a 5 .87).

Test persistence intentions

One item assessed test persistence intentions by asking how

willing participants would be to complete a similar test in the

future (0 5 not at all, 9 5 extremely).

Results and discussion

Test performance

We began by examining the effectiveness of the manipula-

tion. Were participants in the enhancing condition consis-

tently successful at the IO test, and did participants in the

improving condition improve over time? To address these

questions, we conducted a 2 (feedback type) 3 5 (time) mixed ANOVA on number of correct responses in the test.

Overall, participants in the enhancing condition (M 5 5.54,

Sedikides et al. 693

VC 2016 Wiley Periodicals, Inc Journal of Applied Social Psychology, 2016, 46, pp. 687–700

SD 5 1.76) performed better than those in the improving condition (M 5 3.96, SD 5 1.04), feedback type main effect F(1, 48) 5 14.99, p< .001, g2partial 5 0.24. Also, performance improved on average across rounds, time main effect F(4,

192) 5 28.24, p< .001, g2partial 5 0.37; linear trend F(1, 48) 5 102.02, p< .001 (Figure 2). Importantly, the feedback type 3 time interaction was significant, F(4, 192) 5 27.46, p< .001, g2partial 5 0.36. The linear trend differed significant- ly by feedback type, F(1, 48) 5 91.11, p< .001, g2partial 5 0.66. Performance did not increase over time in the enhancing con-

dition, F(1, 24) 5 0.15, p 5 .707, g2partial 5 0.01, but it did increase in the improving condition, F(1, 24) 5 205.88, p< .001, g2partial 5 0.90 (Figure 2). Pairwise comparisons with Bonferroni correction (.05/5 5 .01) confirmed that participants in the enhancing condition performed better

than those in the improving condition at time 1, F(1,

48) 5 67.69, p< .001, g2partial 5 0.59, and time 2, F(1, 48) 5 42.17, p< .001, g2partial 5 0.47, but not at time 3, 4, or 5, Fs< 1, ps> .346. In all, the manipulation was effective.

Satisfaction and usefulness over time

In replication of Experiment 2, overall participants in the

enhancing condition (M 5 72.49, SD 5 12.89) perceived the feedback as more satisfying and useful compared to those in

the improving condition (M 5 63.62, SD 5 12.71), feedback type main effect F(1, 48) 5 6.00, p 5 .018, g2partial 5 0.11. Also, consistent with Experiment 2, participants overall per-

ceived the feedback as more satisfying (M 5 71.60, SD 5 15.35) than useful (M 5 64.51, SD 5 16.32), feedback rating main effect F(1, 48) 5 9.31, p 5 .004, g2partial 5 0.16. Overall, evaluations of feedback increased over time, F(4,

192) 5 18.01, p< .001, g2partial 5 0.27 (Figure 3). The analysis also produced significant interactions between feedback type

and time, F(3, 141) 5 36.10, p< .001, g2partial 5 0.43, and

feedback rating and time, F(2, 116) 5 31.22, p< .001, g2partial 5 0.39.

Crucially, the significant effects were qualified by the three-

way interaction, F(2, 116) 5 19.03, p< .001, g2partial 5 0.28 (Figure 3). As in Experiment 2, we conducted two 2 (feed-

back type) 3 5 (time) mixed ANOVAs, followed by trend and pairwise analyses with Bonferroni correction (.05/

5 5 .01) for satisfaction and usefulness. First, we examined satisfaction. The feedback type 3 time

interaction was significant, F(3, 122) 5 52.23, p< .001, g2par- tial 5 0.52. The linear trend for time differed by feedback type, F(1, 48) 5 74.38, p< .001, g2partial 5 0.61: it was signifi- cant for the improving condition, F(1, 24) 5 88.75, p< .001, g2partial 5 0.79, but not for the enhancing condition, F(1, 24) 5 0.57, p 5 .458, g2partial 5 0.02. Thus, participants per- ceived improving (but not enhancing) feedback as more sat-

isfying over time (Figure 3). Pairwise comparisons of

feedback type showed that participants in the enhancing con-

dition were more satisfied than those in the improving con-

dition at time 1, F(1, 48) 5 61.46, p< .001, g2partial 5 0.56,

Figure 2 Task performance as a function of feedback type and time in

Experiment 3. Error bars indicate standard errors of the mean.

Figure 3 Satisfaction and usefulness as a function of feedback type and

time in Experiment 3. Error bars indicate standard errors of the mean.

694 Enhancing and improving feedback

VC 2016 Wiley Periodicals, Inc Journal of Applied Social Psychology, 2016, 46, pp. 687–700

time 2, F(1, 48) 5 10.60, p 5 .002, g2partial 5 0.18, and time 3, F(1, 48) 5 6.79, p 5 .012, g2partial 5 0.12, but not at time 4 or 5, Fs< 1, ps> .436, g2partial< 0.02.

We proceeded with examining usefulness. The feedback

type 3 time interaction was again significant, F(3, 143) 5 3.33, p 5 .012, g2partial 5 0.07, with the linear trend differing by feedback type, F(1, 48) 5 5.06, p 5 .029, g2partial 5 0.10. The linear trend was significant in the enhanc- ing condition, F(1, 24) 5 5.76, p 5 .024, g2partial 5 0.19, but not in the improving condition, F(1, 24) 5 0.69, p 5 .415, g2partial 5 0.03. Thus, participants perceived enhancing feed- back as less useful over time (Figure 3). Pairwise comparisons

showed that participants in the enhancing condition found

“Order a similar paper and get 20% discount on your first order with us Use the following coupon “FIRST20”