On Writing Legibly: Processing Fluency Systematically Biases Evaluations of Handwritten Material

Abstract

Evaluations of handwritten essays or exams are often suspected of being biased, such as by mood states or individual predilections. Although most of these influences are unsystematic, at least one bias is problematic because it systematically affects evaluations of handwritten materials. Three experiments revealed that essays in legible as compared to less legible handwriting were evaluated more positively. This robust finding was related to a basic judgmental mechanism that builds on the fluency with which handwriting can be processed. The present research further revealed that this evaluative bias is not inevitable but can be controlled for. Given the importance of evaluations based on handwritten work samples for individual success throughout school, college, university, and work life, it is important for individuals to be aware of this bias.

Full text

Handwriting and fluency - 1 -
Please cite as: Greifeneder, R., Alt, A., Bottenberg, K., Seele, T., Zelt, S., & Wagener, D. (2010).
On writing legibly: Processing fluency systematically biases evaluations of handwritten material.
Social Psychological and Personality Science, 1, 230-237. doi: 10.1177/1948550610368434
Address correspondence to: rainer.greifeneder@unibas.ch
Running head: HANDWRITING AND FLUENCY
On writing legibly:
Processing fluency systematically biases evaluations of handwritten material

Handwriting and fluency - 2 -
Abstract
Evaluations of handwritten essays or exams are often suspected of being biased, such as by
mood states or individual predilections. Whereas most of these influences are unsystematic, at
least one bias is problematic, because it systematically affects evaluations of handwritten
materials. Three experiments revealed that essays in legible as compared to less legible
handwriting were evaluated more positively. This robust finding was related to a basic
judgmental mechanism that builds on the fluency with which handwriting can be processed. The
present research further revealed that this evaluative bias is not inevitable, but can be controlled
for. Given the importance of evaluations based on handwritten work samples for individual
success throughout school, college, university, and work life, it is important for individuals to be
aware of this bias.
Keywords: handwriting; legibility; evaluation; fluency; education

Handwriting and fluency - 3 -
On writing legibly:
Processing fluency systematically biases evaluations of handwritten material
Both those who grade exams and those whose exams are graded have long suspected
that evaluations are not always objective, but biased by a host of influences, including mood
states and individual predilections. Most of these biases are unsystematic because they vary
across time and graders, and may thus not be particularly worrisome on an aggregate level.
However, at least one bias is problematic because it systematically affects evaluations of
handwritten material. Specifically, handwriting legibility biases evaluation, with essays in legible
as compared to less legible handwriting being evaluated more positively (e.g., James, 1929).1
This bias—hereafter referred to as legibility bias—can be highly consequential, because
individual success throughout school, college, university, and work life depends only in part on
standardized multiple-choice or computer-based performance assessments, and often on
evaluations of handwritten materials such as essays, exams, or even handwritten résumés, such
as in France. Understanding the source of this bias, and ways to control it, therefore appears
critical. The present research seeks to fulfill these goals, thus moving the field from
acknowledging the legibility bias to understanding its cause and cure.
Asked about the source of this bias, laypersons’ explanation seems based on the
assumption that handwriting is indicative of personality, as handwriting apparently comprises the
wealth of characteristics needed to mirror trait differences (e.g., Ben-Shakhar, Bar-Hillel, Bilu,
Ben-Abba, & Flug, 1986). Yet, graphology ratings have virtually zero predictive validity (Neter &
Ben-Shakhar, 1989), and, even more importantly here, are not reliable across raters (e.g.,
Bayne & O'Neill, 1988). Because of this, graphology-like inferences cannot account for the
observed systematic effect of legibility on evaluations.

Handwriting and fluency - 4 -
Another explanation for why legible handwritten materials receive higher grades is the
consideration of penmanship. James (1929) reported that essays in legible versus less legible
handwriting were evaluated more positively. Presumably this was because evaluators awarded a
premium for legible handwriting, or penalized less legible handwriting, as good penmanship was
a virtue at that time. Similar reasoning may explain later replications, in which elementary school
teachers—who teach penmanship and spontaneously take its mastery into account—were
recruited as participants (e.g., Briggs, 1970; Markham, 1976), or in which essay topics such as
“Hopes and aspirations for the next decade“—which lack objective content criteria and may
therefore invite the consideration of penmanship—were used (e.g., Hughes, Keeling, & Tuck,
1983; Klein & Taub, 2005). Nowadays, however, it would appear that mastery of penmanship is
less important, at least beyond elementary school, and when objective content criteria are
available. If evaluations are still biased by differences in legibility, penmanship is unlikely to be
the culprit.
As a new explanation, we propose that the legibility bias results from a basic judgmental
mechanism that takes the fluency of information processing into account. Specifically, we
suggest that individuals spontaneously form inferences such as, “If it can be processed fluently,
it is probably good.“ These inferences are then used as information in evaluation. This new
explanation is much broader than prior theorizing, because it suggests that the legibility bias is a
pervasive phenomenon. What follows seeks to substantiate this fluency hypothesis.
Fluency refers to the felt ease or difficulty with which mental processes can be executed
(e.g., Greifeneder & Bless, in press). A large body of evidence suggests that individuals
spontaneously recruit feelings of fluency when forming judgments of various kinds (for
overviews, e.g., Alter & Oppenheimer, 2009; Schwarz & Clore, 2007). For instance, statements
are endorsed as more probably true when processing is fluent (e.g., Reber & Schwarz, 1999; for
a review, Dechêne, Stahl, Hansen, & Wänke, in press), liking is enhanced when stimuli are

Handwriting and fluency - 5 -
processed fluently (e.g., Reber, Winkielman, & Schwarz, 1998), and events are judged as more
frequent and products as more positive when recollection is easy (e.g., Greifeneder & Bless,
2008; Wänke, Bohner, & Jurkowitsch, 1997). The diversity of these examples suggests that the
influence of fluency on judgment is a pervasive phenomenon (see also Alter & Oppenheimer,
2009).
But why is the influence of fluency on judgments pervasive? Supposedly, this is because
fluency is generally perceived to be “about“ whatever is the focus of attention (Clore et al., 2001;
Higgins, 1996), even if such attributions are not warranted. Once attributed, fluency is usually
interpreted as a signal of positivity, and disfluency as a signal of negativity (Schwarz, Song, &
Xu, 2008), because, over time, individuals have learned that positive states of the world are
associated with processing fluency, and negative states with disfluency (Unkelbach, 2006,
2007). Consistent with this evidence, we suggest that individuals spontaneously attribute
differences in fluency when processing handwritten material to the focus of attention, such as
the handwritten material or its author. Fluency is then used as information to draw inferences
about the positivity or negativity of these attribution targets. The present contribution seeks to
substantiate this new explanation by demonstrating a) that the legibility bias occurs even if
penmanship is not a concern, b) that it is related to processing fluency, and c) that it can be
controlled for by drawing participants’ attention to its source.
Experiment 1
Method
Participants and Design. Forty-four University of Mannheim students (22 female; mean
age 23.14 years, SD = 2.89) received 2 EUR and a chocolate bar to evaluate a good, medium,
and poor essay. As we wanted all participants to grade at least one legible essay and one less

Handwriting and fluency - 6 -
legible essay, participants were randomly divided into two experimental groups. Some
participants were presented the good, medium, and poor essay in, respectively, low, high, and
low legibility. Others were presented the same essays in the same order, but in high, low, and
high legibility. This resulted in a 3 (content quality: good, medium, poor) x 2 (group: low-high-low
vs. high-low-high legibility of essays) mixed-factorial design, with content quality as within factor
(see Table 1 for a concise overview of this design). As each of the three essays is presented in
both highly and less legible handwriting, the hypothesized effect of legibility should be apparent
when comparing the evaluations for each essay between groups. Importantly, because the
assignment of legibility to essays was reversed between experimental groups (low-high-low vs.
high-low-high), the direction of the hypothesized legibility effect should alternate from essay to
essay. Accordingly, across the three essays, the hypothesized effect of legibility should be
apparent in a significant interaction effect. Note that the present set-up of evaluating several
essays closely matches the situation graders usually encounter. Moreover, this set-up reduces
the likelihood that effects are due to handwriting features other than the systematically varied
legibility.
Essay construction. Initially, a series of typed essays of similar length was constructed,
with essays varying in the amount of correct information (content quality). Based on independent
pre-testing (N = 28, 12 female), three essays of good, medium, and poor content quality were
selected (6-point standard German grade scale; all pairwise |ts| > 2.10, ps < .04). A new sample
of students then copied these essays in their usual cursive handwriting, each on a separate,
blank sheet of paper. In second, independent pre-testing (N = 32, 18 female), participants
evaluated handwriting on a scale from 1, easy to read, to 6, difficult to read. For each essay a
highly and a less legible version was selected (all |ts| > 2.30, ps < .05), with the constraint that
all words in all essays be readable. In a final, independent pre-testing (N = 27, 16 female), which
followed the above described 3 x 2 mixed-factorial design, participants were first asked to read

Handwriting and fluency - 7 -
the handwritten essays in their normal reading speed while processing latencies were recorded,
and then to revisit each essay and evaluate handwriting legibility. Results revealed that for each
of the three essays the highly legible version was not only perceived as more legible (all
|ts| > 2.80, ps < .01), but could also be read significantly faster (all |ts| > 2.08, ps < .05).2
Procedure. Upon arrival at the laboratory, participants were handed a questionnaire
including all instructions and materials. To foster a high level of accuracy motivation and thereby
decrease evaluation bias, participants were told that the experiment focused on interrater
reliability. Participants read a short paragraph (74 words) about a physics topic, “The emergence
of lightning,“ and were asked to evaluate the subsequent essays based on this standard. The
essays were supposedly written by students as part of a teaching assignment. Participants read
and evaluated the essays one by one, starting with an example before working on the three
target essays.
Evaluation. After each essay, participants were asked to evaluate the presumed author
regarding general academic competence, knowledge of other school subjects, diligence, time
spent studying, verbal expressiveness, and abilities in other domains.3 All six evaluations were
assessed on six-point Likert scales (1, high, to 6, low). Subsequently, participants assigned a
grade to the respective essay on a scale from 1, excellent, to 6, insufficient (standard German
grade scale).
Results and Discussion
Both evaluations of author abilities and assigned grades were individually rescaled such
that higher scores indicate more positive evaluations. For each essay, the six items targeting
author abilities were highly interrelated and averaged (all Cronbach’s  > .87). The indices for
author abilities and assigned grades were separately subjected to 3 x 2 mixed-factorial
ANOVAs.

Handwriting and fluency - 8 -
For author abilities, a main effect of content quality indicates that presumed authors were
evaluated more positively the higher the quality of the essays, reflecting that participants took
content quality into account, F(2, 84) = 45.34, p < .01, ² = .52. More importantly, across essays,
highly legible as compared to less legible handwriting led to more positive evaluations, as
apparent in the hypothesized interaction effect, F(2, 84) = 13.34, p < .01, ² = .24 (main effect
group, F < 1). Planned comparisons indicated that this legibility effect is strong for each of the
three essays, t(42) = 2.54, p < .02, Cohen’s d = 0.77; t(42) = 2.99, p < .01, Cohen’s d = 0.90;
t(42) = 2.62, p < .02, Cohen’s d = 0.79, thus ruling out alternative explanations that assume
differences for certain levels of content quality only. The cell means are depicted in Figure 1a;
for ease of presentation, means are not arranged by experimental group, but re-arranged by
handwriting legibility.
A parallel pattern of evidence was observed for assigned grades, with essays evaluated
more positively the better they were content-wise, F(2, 84) = 43.93, p < .01, ² = .51, but also
the more legible the handwriting was, as reflected in a significant interaction effect,
F(2, 84) = 6.10, p < .01, ² = .13 (main effect group, F < 1). Again, it is noteworthy that a
considerable percentage of the variance in grade ratings was due to systematic differences in
legibility. Note also that, on average, highly legible essays were evaluated 0.5 grade points more
positively than less legible essays, which should be considered serious given a 6-point grade
scale. Individual cell means are depicted in Figure 1b.
In sum, Experiment 1 revealed that handwriting legibility systematically biases
evaluations of both author abilities and content quality. This occurred even though penmanship
was unlikely to be a concern, because the sample consisted of students (instead of, e.g.,
elementary school teachers), and these students evaluated essays based on explicit content
criteria as provided in the standard paragraph. It was further shown that the effect occurred for

Handwriting and fluency - 9 -
good, medium, and poor essays, and was significant both between and within groups, thus
attesting to the legibility bias’ general nature.
Experiment 1 relied on a student sample to reduce the potential impact of penmanship
considerations. As a downside of this choice, the results may be perceived as less ecologically
valid. To address this concern, a follow-up study was conducted, which employed the same
design, procedure, and materials as Experiment 1, but recruited a convenience sample of
47 German secondary school teachers, who regularly evaluate handwritten exams. Replicating
the findings observed in Experiment 1, highly legible as compared to less legible handwriting
resulted in significantly more positive evaluations and assigned grades. This finding further
attests to the legibility bias’ general significance.
Experiment 2
Experiment 1 revealed that essays in legible as compared to less legible handwriting
were evaluated more positively. We suggest that this is because legible essays can be
processed more fluently. To empirically substantiate this hypothesis, we additionally measured
perceived legibility as a proxy for fluency. Furthermore, we wanted to rule out the alternative
explanation that legible essays are evaluated more positively because they are perceived as
more beautiful, reflecting a what-is-beautiful-is-good-heuristic (Dion, Berscheid, & Walster,
1972). Indeed, one could suppose that legible essays are evaluated more positively because
legibility and beauty often go together. Although a post-test suggests that this was not the case
in Experiment 1, it appeared desirable to address this concern experimentally. Accordingly, new
handwritings were selected for Experiment 2 such that legibility and beauty varied orthogonally.
We expected that only legibility, but not beauty would influence evaluations.

Handwriting and fluency - 10 -
Methods
Participants and Design. Eighty-four University of Mannheim students (68 female; 4
unknown; mean age 21.54 years, SD = 4.21) received 2 EUR and a chocolate bar to evaluate
three essays of good, medium, and poor quality. Replicating the design of Experiment 1, some
participants were presented the good, medium, and poor essay in, respectively, low, high, and
low legibility. Other participants were presented the same essays in the same order, but in high,
low, and high legibility. In addition, handwriting beauty was varied between conditions: some
participants were presented with beautiful handwriting only, whereas other participants only
evaluated materials in less beautiful handwriting.4 This resulted in a 3 (content quality: good,
medium, poor) x 2 (group: low-high-low vs. high-low-high legibility of essays) x 2 (handwriting
beauty: high vs. low) mixed-factorial design, with content quality as within factor. Participants
were randomly assigned to the four conditions. Materials and procedures were similar to
Experiment 1, except for the following two changes.
Essay construction. A new sample of students was asked to copy the physics essays in
their usual cursive handwriting, each on a separate, blank sheet of paper. In independent pre-
testing, participants (N = 9, 6 female) evaluated each handwriting with respect to legibility
(1, easy to read, to 6, difficult to read) and beauty (1, beautiful, to 6, unsightly). Handwritings
were then picked to represent the four following styles: high legibility-beautiful, high legibility-less
beautiful, low legibility-beautiful, and low legibility-less beautiful. Highly legible versus less legible
handwritings were rated as more legible but not as more beautiful, F(1, 8) = 113.98, p < .01,
² = .93 (all other p > .17), and beautiful versus less-beautiful handwritings were rated as more
beautiful but not as more legible, F(1, 8) = 65.64, p < .01, ² = .89 (all other p > .12).

Handwriting and fluency - 11 -
Assessment of perceived legibility. After reading and evaluating all essays, participants in
Experiment 2 were asked to look a second time at every essay and to evaluate the respective
handwriting on a scale from 1, easy to read, to 6, difficult to read.
Results and Discussion
Evaluations of author abilities were individually rescaled so that higher values indicate
more positive evaluations, were averaged per essay (all Cronbach’s  > .83), and were
subjected to a 3 x 2 x 2 mixed-factorial ANOVA. Replicating Experiment 1, evaluations of author
abilities were strongly contingent on the essays’ content quality, with better essays being
evaluated more positively, F(2, 160) = 122.63, p < .01, ² = .61. Moreover, essays in highly
legible as compared to less legible handwriting were evaluated more positively, as reflected in a
significant legibility x content two-way interaction, F(2, 160) = 5.39, p < .01, ² = .06. Importantly,
handwriting beauty did not significantly influence evaluations of author abilities, F(1, 80) = 1.22,
p > .27 (all other Fs < 1). If anything, beautiful versus less beautiful handwritings were
associated with lower evaluations of author abilities (M = 3.93, SD = 0.60; M = 4.08, SD = 0.59),
thus refuting the alternative hypothesis that essays in legible handwriting are evaluated more
positively because what is beautiful is often also good. The 12 cell means are displayed in
Table 2 (for ease of presentation, means are not arranged by experimental group, but re-
arranged by legibility).
To more directly test the proposed fluency hypothesis, mediation analyses were
performed following Baron and Kenny (1986). Specifically, separately for each of the three
essays, but collapsing across beauty conditions, we analyzed whether the effect of the legibility
manipulation (independent variable) on evaluations of author abilities (dependent variable) is
mediated by perceived legibility, which supposedly constitutes a good proxy for fluency. Strong
mediation was observed for each of the three essays (Sobel’s Z = 2.23, p < .03; Z = 3.28,

Handwriting and fluency - 12 -
p < .01; Z = 2.53, p < .02), suggesting that handwriting legibility biases evaluations because
fluency associated with legibility is used as information when forming judgments.
Experiment 3
Experiments 1 and 2 demonstrate that essays in legible handwriting are evaluated more
positively, and this was linked to spontaneous inferences based on processing fluency.
Experiment 3 was to test whether this basic judgmental mechanism can be controlled for if
participants know about the biasing influence of legibility. Such a possibility would provide an
important means to counter the legibility bias, and would further attest to the proposed fluency
hypothesis, as detailed below.
Methods
Participants and Design. One hundred and eight University of Mannheim students
(57 female; mean age 22.26 years, SD = 3.40) participated in return for 1.50 EUR and a
chocolate bar. Two participants had insufficient knowledge of the German language, six had
already participated in one of the previous studies. These eight participants were excluded from
further analyses.
Participants were randomly assigned to a 3 (content quality: good, medium, poor) x
2 (group: low-high-low vs. high-low-high legibility of essays) x 2 (control vs. information) mixed-
factorial design, with content quality as within factor. Half of the participants received the same
instructions as in the previous experiments (control); the other half (information) additionally
read: “Please note: Prior research revealed that the ease or difficulty with which handwritten
essays can be read strongly impacts their evaluation. Please try not to be influenced by how
easy or difficult it is to read the following essays.” The second sentence was subsequently
repeated above each essay. All other materials and procedures replicated those of

Handwriting and fluency - 13 -
Experiment 1, though the number of dependent variables per essay was reduced from six to four
for reasons of test efficiency. Because individuals generally cease to rely on fluency in judgment
when fluency is said to be undiagnostic (e.g., Schwarz et al., 1991), we hypothesized that the
influence of legibility on evaluations would be reduced or eliminated in the information condition.
Together, these predictions translate to an expected three-way interaction.
Results and Discussion
Evaluations of author abilities and grades were individually rescaled so that higher values
indicate more positive evaluations, were averaged per essay (all Cronbach’s  > .82), and were
subjected to a 3 x 2 x 2 mixed-factorial ANOVA. This analysis revealed a significant main effect
of content quality, F(2, 192) = 109.51, p < .01, ² = .53, a significant content quality by group
two-way interaction, F(2, 192) = 4.53, p < .02, ² = .05, a significant content quality by
information condition two-way interaction, F(2, 192) = 3.18, p < .05, ² = .03, and a significant
main effect of experimental group, F(1, 96) = 5.90, p < .02, ² = .06. Importantly, all of these
effects were qualified by the expected significant three-way interaction, F(2, 192) = 5.96, p < .01,
² = .06. The 12 cell means are displayed in Table 3.
To further investigate this pattern of results, the control condition and the information
condition were analyzed separately. Replicating prior findings, participants in the control
condition (no additional information) assigned more positive evaluations the higher the essays’
content quality, F(2, 100) = 39.65, p < .01, ² = .44, and the higher the essays’ legibility, as
reflected in a significant two-way interaction, F(2, 100) = 10.65, p < .01, ² = .18 (main effect
group, F < 1). Planned comparisons indicated that for every essay, the highly legible version led
to more positive evaluations than the less legible version, t(50) = 1.91, p < .07, Cohen’s

Handwriting and fluency - 14 -
d = 0.54; t(50) = 3.28, p < .01, Cohen’s d = 0.91; t(50) = 2.58, p < .02, Cohen’s d = 0.72, for the
three essays, respectively.
Participants in the information condition also assigned more positive evaluations, the
better the essays were content-wise, as reflected in a significant content main effect
F(2, 92) = 72.34, p < .01, ² = .61. Importantly, however, legible as compared to less legible
essays did not consistently result in more positive evaluations, as indicated by a non-significant
interaction term, F < 1. Thus, legibility did not systematically bias evaluations in the information
condition, presumably because participants, once informed about the deleterious impact of
handwriting legibility, were apt to correct for this bias. Unexpectedly, essays in the high-low-high
group led to more positive evaluations than essays in the low-high-low group (M = 4.28,
SD = 0.53; M = 3.91, SD = 0.53), as reflected in a significant main effect of experimental group,
F(1, 46) = 5.58, p < .03, ² = .11. Further planned comparisons proved reliable only for the
second essay, t(46) = 1.87, p < .07 (all other ps > .13), and for this second essay, lower legibility
led to more (and not less) positive evaluations. Potentially, this is because participants not only
corrected, but overcorrected for the supposed influence of fluency, resulting in the opposite of
the generally observed legibility bias (for evidence on overcorrection, e.g., Strack, Schwarz,
Bless, Kübler, & Wänke, 1993). Together, these results spark hope that the legibility bias may be
successfully countered by instructing individuals to do so.
While the primary objective of Experiment 3 was to identify a cure, the observed results
also attest to the validity of the fluency hypothesis. The logic is as follows: If directing
participants’ attention to fluency reduces the effect, one may conclude that fluency was
responsible for the observed legibility bias in the first place—for why else would the effect be
reduced when participants are aware of the presently undue influence of fluency? Experiment 3

Handwriting and fluency - 15 -
thus further attests to the suggested fluency hypothesis by way of moderation evidence (for
details on this reasoning, see Schwarz et al., 1991; Spencer, Zanna, & Fong, 2005).
General Discussion
Three experiments revealed that handwriting legibility systematically biases evaluations
of author abilities and grades assigned to the respective essays. This effect was observed in
contexts where penmanship is unlikely to be considered an important part of performance, for
different levels of content quality (good, medium, and poor), and for different populations
(students and secondary school teachers), thus ruling out a series of potential alternative
explanations. Furthermore, the effect was not related to differences in handwriting beauty.
Rather, essays in more legible handwriting were evaluated more positively because of the
fluency associated with their processing. This conclusion dovetails with and extends findings in
the domain of social and cognitive psychology, holding that the fluency associated with
information processing influences judgments of various kinds and constitutes a frequent source
of information in conditions of daily life (e.g., Greifeneder, Bless, & Pham, 2010; Schwarz et al.,
2008).
The present research set out to investigate the process underlying the legibility bias.
While the observed evidence sustains the suggested fluency hypothesis, it should be
acknowledged that the mediation in Experiment 2 was based on perceived rather than measured
fluency, and thus a subjective measure. Although an objective measure would have been
desirable, this appears secondary in light of the fact that the fluency hypothesis was supported
with all three primary methodological approaches known to the field: a) by refuting a series of
alternative hypotheses, as in Experiments 1 and 2, b) by testing mediation in Experiment 2, and
c) by testing moderation in Experiment 3.

Handwriting and fluency - 16 -
It should also be noted that alternating highly legible and less legible handwriting may
have helped fluency experiences to be salient (e.g., Hansen & Wänke, 2008). Yet, this set-up is
likely to be of high ecological validity, because essays or exams are rarely ordered by legibility,
so that varying legibility levels is the very situation that graders usually encounter.
By relating the effect of legibility on evaluations to a fluency-based judgmental
mechanism, the present findings suggest that the legibility bias is a pervasive phenomenon. This
conclusion extends the scope of prior research on handwriting, and points to a potentially
harmful source of error, because evaluations of handwritten material may be consequential at all
stages of life. From this perspective, the final piece of evidence furnished by the present
research is of particular relevance: Participants did not show the legibility bias when alerted to its
existence, suggesting that the bias is not inevitable, but can be controlled for.
This final piece of evidence also resolves a seeming discrepancy between the present
findings and the notion of “wisdom in feelings” (e.g., Schwarz, 2002), which suggests that
reliance on feelings is a generally sensible judgment mechanism—provided correct attribution
(Schwarz, 2004). When appropriate attribution targets are unknown or unnoticed, however,
feelings may lead astray, because attribution is then primarily guided by temporal contiguity
(e.g., Clore et al., 2001). Presumably this is what causes the legibility bias, as fluency from
legibility appears to be influential simply because it is perceived when forming evaluations. The
present contribution thus sheds light on the Achilles’ heel of reliance on fluency and emphasizes
that for fluency feelings to be “wise,” appropriate conceptions of causation are important. Such
conceptions are available in certain domains and situations, for instance, with respect to undue
influences of media coverage on perceived name frequency (Oppenheimer, 2004), but appear to
be missing for legibility. For the sake of fair performance assessments based on handwritten
material, it therefore appears critical that individuals know about legibility’s potential for bias.

Handwriting and fluency - 17 -
References
Alter, A. L., & Oppenheimer, D. M. (2009). Uniting the tribes of fluency to form a metacognitive
nation. Personality and Social Psychology Review, 13(3), 219-235.
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social
psychological research: Conceptual, strategic, and statistical considerations. Journal of
Personality and Social Psychology, 51(6), 1173-1182.
Bayne, R., & O'Neill, F. (1988). Handwriting and personality: A test of some expert graphologists'
judgements. Guidance & assessment review, 4(4), 1-3.
Ben-Shakhar, G., Bar-Hillel, M., Bilu, Y., Ben-Abba, E., & Flug, A. (1986). Can graphology
predict occupational success? Two empirical studies and some methodological ruminations.
Journal of Applied Psychology, 71(4), 645-653.
Briggs, D. (1970). The influence of handwriting on assessment. Educational Research, 13(1),
50-55.
Clore, G. L., Wyer, R. S., Dienes, B. P. A., Gasper, K., Gohm, C., & Isbell, L. (2001). Affective
feelings as feedback: Some cognitive consequences. In L. L. Martin & G. L. Clore (Eds.),
Theories of mood and cognition: A user's guidebook (pp. 27-62). Mahwah, NJ: Lawrence
Erlbaum Associates.
Dechêne, A., Stahl, C., Hansen, J., & Wänke, M. (in press). The truth about the truth: A meta-
analytic review of the truth-effect. Personality and Social Psychology Review.
Dion, K., Berscheid, E., & Walster, E. (1972). What is beautiful is good. Journal of Personality
and Social Psychology, 24(3), 285-290.
Greifeneder, R., & Bless, H. (2008). Depression and reliance on ease-of-retrieval experiences.
European Journal of Social Psychology, 38(2), 213-230.

Handwriting and fluency - 18 -
Greifeneder, R., & Bless, H. (in press). The fate of activated information in impression formation:
Fluency of concept activation moderates the emergence of assimilation versus contrast.
British Journal of Social Psychology.
Greifeneder, R., Bless, H., & Pham, M. T. (2010). Moderators of the reliance on affective and
cognitive feelings in judgment: A review. Unpublished manuscript, University of Mannheim.
Hansen, J., & Wänke, M. (2008). It's the difference that counts: Expectancy/experience
discrepancy moderates the use of ease of retrieval in attitude judgments. Social Cognition,
26, 447-468.
Higgins, E. T. (1996). Knowledge activation: Accessibility, applicability, and salience. In E. T.
Higgins & A. W. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp.
133-168). New York, NY: Guilford Press.
Hughes, D. C., Keeling, B., & Tuck, B. F. (1983). Effects of achievement expectations and
handwriting quality on scoring essays. Journal of Educational Measurement, 20(1), 65-70.
James, H. W. (1929). The effect of handwriting upon grading. The English Journal, 16(3), 180-
185.
Klein, J., & Taub, D. (2005). The effect of variations in handwriting and print on evaluation of
student essays. Assessing writing, 10, 134-148.
Landy, D., & Sigall, H. (1974). Beauty is talent: Task evaluation as a function of the performer's
physical attractiveness. Journal of Personality and Social Psychology, 29(3), 299-304.
Markham, L. R. (1976). Influences of handwriting quality on teacher evalution of written work.
American Educational Research Journal, 13(4), 277-283.
Neter, E., & Ben-Shakhar, G. (1989). The predictive validity of graphological inferences: A meta-
analytic approach. Personality and Individual Differences, 10(7), 737-745.
Oppenheimer, D. M. (2004). Spontaneous discounting of availability in frequency judgment
tasks. Psychological Science, 15(2), 100-105.

Handwriting and fluency - 19 -
Reber, R., & Schwarz, N. (1999). Effects of perceptual fluency on judgments of truth.
Consciousness and Cognition: An International Journal, 8(3), 338-342.
Reber, R., Winkielman, P., & Schwarz, N. (1998). Effects of perceptual fluency on affective
judgments. Psychological Science, 9(1), 45-48.
Schwarz, N. (2002). Situated cognition and the wisdom in feelings: Cognitive tuning. In L. F.
Barrett & P. Salovey (Eds.), The wisdom in feeling: Psychological processes in emotional
intelligence (pp. 144-166). New York, NY, US: Guilford Press.
Schwarz, N. (2004). Metacognitive experiences in consumer judgment and decision making.
Journal of Consumer Psychology, 14(4), 332-348.
Schwarz, N., Bless, H., Strack, F., Klumpp, G., Rittenauer-Schatka, H., & Simons, A. (1991).
Ease of retrieval as information: Another look at the availability heuristic. Journal of
Personality and Social Psychology, 61(2), 195-202.
Schwarz, N., & Clore, G. L. (2007). Feelings and phenomenal experiences. In E. T. Higgins & A.
W. Kruglanski (Eds.), Social psychology: Handbook of basic principles (2nd ed., pp. 385-
407). New York, NY: Guilford Press.
Schwarz, N., Song, H., & Xu, J. (2008). When thinking is difficult: Metacognitive experiences as
information. In M. Wänke (Ed.), Social psychology of consumer behavior (pp. 201-223). New
York: Psychology Press.
Spencer, S. J., Zanna, M. P., & Fong, G. T. (2005). Establishing a causal chain: Why
experiments are often more effective than mediational analyses in examining psychological
processes. Journal of Personality and Social Psychology, 89(6), 845-851.
Strack, F., Schwarz, N., Bless, H., Kübler, A., & Wänke, M. (1993). Awareness of the influence
as a determinant of assimilation versus contrast. European Journal of Social Psychology,
23(1), 53-62.

Handwriting and fluency - 20 -
Unkelbach, C. (2006). The learned interpretation of cognitive fluency. Psychological Science,
17(4), 339-345.
Unkelbach, C. (2007). Reversing the truth effect: Learning the interpretation of processing
fluency in judgments of truth. Journal of Experimental Psychology: Learning, Memory and
Cognition, 33, 219-230.
Wänke, M., Bohner, G., & Jurkowitsch, A. (1997). There are many reasons to drive a BMW:
Does imagined ease of argument generation influence attitudes? Journal of Consumer
Research, 24(2), 170-177.

Handwriting and fluency - 21 -
Footnotes
1 Another systematic bias is the “beauty-is-talent” effect (Landy & Sigall, 1974), which holds
that essays are evaluated more positively when authored by presumably attractive as
compared to non-attractive students.
2 Due to space limitations, only general results are reported for the pre-tests, Replication 1, and
the mediation analysis in Experiment 2. Full results are available from the authors.
3 Asking for “diligence” may have been unfortunate, because individuals potentially had a naïve
theory that less readable handwriting directly reflects less diligence. There is reason to
believe, however, that this alternative process did not produce the observed results. First, the
same significance levels are obtained if author ability is computed without diligence. Second,
a similar pattern of results is observed for grades. Third, evaluations in Experiment 3 are not
influenced by differences in legibility once individuals are made aware of fluency, suggesting
that fluency is the mediating link.
4 Varying legibility within participants, but beauty between participants may be perceived as an
unfair test of alternative hypotheses. Note, however, a) that legibility is also varied between
participants, since each essay is presented in highly legible handwriting to some participants,
and less legible handwriting to other participants, and b) that the between-participants simple
contrasts in Experiment 1 proved significant for every essay.

Handwriting and fluency - 22 -
Acknowledgements
Authors 2 to 5 have contributed equally and are listed in alphabetical order. This research
was supported by a post-doctoral grant from the Deutsche Forschungsgemeinschaft and a post-
doctoral grant from the Landesstiftung Baden-Württemberg, both awarded to the first author.

Handwriting and fluency - 23 -
Tables
Table 1
Design of Experiment 1
Essay content quality
Good
(Essay 1)
Medium
(Essay 2)
Poor
(Essay 3)
Legibility of essays in group 1 Low High Low
Legibility of essays in group 2 High Low High

Handwriting and fluency - 24 -
Table 2
Mean Author Evaluations (with Standard Deviations) in Experiment 2 as a Function of Content
Quality, Legibility of Essays, and Handwriting Beauty
Essay content quality
Good
(Essay 1)
Medium
(Essay 2)
Poor
(Essay 3)
Beautiful handwriting
Highly legible essays 5.08 (0.81) 3.93 (0.92) 3.17 (0.79)
Less legible essays 4.76 (0.62) 3.44 (0.79) 3.22 (0.85)
Less beautiful handwriting
Highly legible essays 5.21 (0.52) 4.11 (1.06) 3.39 (0.98)
Less legible essays 4.78 (0.80) 3.90 (0.97) 3.08 (0.87)
Notes. Author evaluations were assessed on 9-point Likert-scaled items. Higher values indicate
more positive evaluation. Means are not arranged by experimental group, but re-arranged by
handwriting legibility.

Handwriting and fluency - 25 -
Table 3
Mean Evaluations (with Standard Deviations) in Experiment 3 as a Function of Content Quality,
Legibility of Essays, and Information Condition
Essay content quality
Good
(Essay 1)
Medium
(Essay 2)
Poor
(Essay 3)
Control (no additional information)
Highly legible essays 4.76 (0.63) 4.62 (0.44) 3.70 (0.81)
Less legible essays 4.41 (0.69) 4.01 (0.86) 3.11 (0.84)
Additional information
Highly legible essays 4.98 (0.75) 4.08 (0.84) 3.35 (0.80)
Less legible essays 4.69 (0.63) 4.50 (0.72) 2.98 (0.90)
Notes. Evaluations were assessed on 9-point Likert-scaled items. Higher values indicate more
positive evaluation. Means are not arranged by experimental group, but re-arranged by
handwriting legibility.

Handwriting and fluency - 26 -
Figure Legend
Figure 1. Mean evaluations (with standard errors) of author abilities (1a) and assigned
grades (1b) for each essay in Experiment 1. Higher ratings indicate more positive evaluation.
Means are not arranged by experimental group, but re-arranged by handwriting legibility.
Evaluations of legible essays are displayed as white bars, evaluations of less legible essays as
black bars.

Handwriting and fluency - 27 -
Figure 1
Figure 1a Figure 1b
2,5
3,0
3,5
4,0
4,5
5,0
5,5
Essay 1 (good) Essay 2 (medium) Essay 3 (poor)
Mea n eva luation of author a bilities
2,5
3,0
3,5
4,0
4,5
5,0
5,5
Essay 1 (good) Essay 2 (medium) Essay 3 (poor)
Mea n assigne d grade