Running head: attention efficiency and performance
The Effects of Text Exaggeration and Orthographic Similarity on
Correct Identification of Drug Names
KRAIG L. SCHELL, KYLE KELLEY & CORY HUNSAKER
Angelo State University
San Angelo, Texas U.S.A.
This study tested the effect of text exaggerations (capital letters, color enhancements) of specific
segments of drug names and orthographic similarity differences on correct name recognition. One
hundred and three participants saw 80 pairs of drug names over six trial blocks (480 pairs total)
where they saw an exaggerated prime word, a black mask for 450 milliseconds, and a target word
which was either the same word or a potentially confusable pair word. Participants decided
whether the prime and target words were the same. Results indicated that name identification was
less accurate at higher levels of orthographic similarity as expected. However, the text
exaggerations did not reliably improve performance on the task, contrary to expectations. The
practical and theoretical importance of these results is explored, especially regarding current
federal mandates concerning exaggerated labeling on drug products.
This research was supported by a grant from the Institute for Safe Medication Practices,
Huntington Valley, PA. The authors thank Liz Bankhead, Paul Turner and our undergraduate
research team for assistance in data collection.
Correspondence concerning this article should be addressed to Kraig Schell, Department
of Psychology, Angelo State University, Box 10907 ASU Station, San Angelo, TX, 76909.
Electronic mail can be sent to
The Effects of Text Exaggeration and Orthographic Similarity on
Correct Identification of Drug Names
Medication errors and patient safety have been salient issues in health care since
the Institute of Medicine released their disturbing report on the frequency with which
these errors were being made (Kohn, Corrigan & Donaldson, 1999). Since that time,
many recommendations and mandates have been put forward in hospitals and pharmacies
in an attempt to reduce the incidence of these errors. One of these mandates instated by
the Food and Drug Administration required that labeling for a short list of drugs be
changed. Specifically, the requirement addressed pairs of drug names that were similar
either in orthography or phonology under the assumption that they were vulnerable to
confusion. To remedy this, the parts of the names that made them unique were to be
exaggerated using capital letters. For example, "hydralazine" and "hydroxyzine" are one
of the pairs on the FDA list, and it is clear that they are very similar to one another. So,
current labeling for these drugs would change the appearance of the name to be
"hydrALAzine" and "hydrOXYzine" to emphasize the unique aspects of each name.
Though the idea sounds good in principle, we know of no controlled research that
supports the effectiveness of the mandate. Not only could such a line of research evaluate
the utility of exaggerated labeling, it could help to identify other interventions to further
prevent medication mishaps where the name of the drug is a contributing factor. It is with
this purpose in mind that the current study was designed. Orthographic similarity and text
exaggeration will be examined to determine whether they affect the ability to detect drug
names correctly in a rapid prime-target word matching task.
Background Literature: Reading, Attention and Human Factors
Of course, the intended effect of text exaggeration is to draw the attention of the
pharmacist to the part of the drug name that is unique in hopes of reducing name
confusability. However, reading words involves more than just attention, and the
assumed effect also implies that the rest of the word (the non-exaggerated portion) will
command little attention during performance. Fortunately, the reading literature provides
some well-documented premises that bear directly on this investigation, summarized
below in two main points.
First, it is generally accepted that reading is largely an automatic process taking
the form of a skill and requiring little focused attention once well learned (see Just &
Carpenter, 1980 for example). The expected effect of the text exaggeration is to make the
reading process more effortful, but it is unclear as to whether a few capital letters are
salient enough to disrupt the automaticity of reading. If they are not, then the entire drug
name is likely to be read automatically and the text exaggeration may actually harm
performance rather than help it by altering the expected form of the word. Second, it is
largely accepted that there are two reading systems: orthography and phonology (Frost,
1998; Lukatela & Turvey, 1994a, 1994b; Stone & Van Orden, 1993). There is some
debate about which system is more dominant under which circumstances (Carreiras,
Perea, & Grainger, 1997; Katz & Frost, 2001; Pugh, Rexer, & Katz, 1994), but the mere
presence of these two reading systems is enough to make the effects of text exaggeration
more complicated. The capital letters alter the word shape – they do not alter the actual
orthography of the word (the letter sequences). The shape of the word being read is not
generally considered to be very important to reading (Frost, 1998; Perea & Rosa, 2002),
so it remains to be seen whether such exaggerations will really have much of an effect. In
other words, despite the graphical change, the same letter sequences exist. So, it is likely
that text exaggerations must overcome some elemental reading principles in order to have
the effect intended by the pharmacy industry, and we are concerned about their ability to
Human factors research is also relevant to this issue in a number of ways. First, if
text exaggerations are a salient intervention for name recognition, then they should have
this effect in the initial stages of processing. Research shows that we are drawn toward
unique stimuli in our visual field, sometimes called "singletons" (Duncan & Humphreys,
1989; Yantis & Hillstrom, 1994; but see Egeth & Yantis, 1997 for clarification), so the
capital letters amongst "normal" letters must be unique enough to capture attention if
they are to work as intended. However, it is uncertain if such text exaggerations can
attract attention so powerfully. Second, there are several other types of exaggerations that
should be as capable of attracting attention if not more so. Research indicates that color,
for example, can capture attention under certain visual search paradigms (Treisman &
Gelade, 1980; Yantis, 1998). Furthermore, though there are no universal meanings for
colors, red is typically interpreted as "stop" or "wait", at least in many Western cultures
(Gage, 1995), which is the message that the text exaggeration intervention is intended to
send. So, red letters may be just as useful as capitals, if not more. Third, there is the
problem of sensitivity versus response bias. There is no theoretical basis to predict how
text exaggerations should affect these two measures. On the one hand, it could be argued
that sensitivity should increase and response bias should become more conservative if the
text exaggerations provide more salient target information as they are meant to do. On the
other hand, one could also say that each exaggeration is essentially a signal, which
dramatically increases event probability and should have significant negative effects on
response bias and sensitivity (Davies & Parasuraman, 1982; See, Warm, Dember, &
Howe, 1997; Wickens & Hollands, 2000). In short, there are reasons in the relevant
literature to question whether text exaggerations are the preventative measures that they
are meant to be.
Calculating Orthographic Similarity: The Trigram Formula
A second issue which is relevant to this investigation is orthographic similarity.
We define this as a mathematical ratio of trigrams (3-letter sequences) in common
between two pair words, expressed in Equation 1:
where A = the number of trigrams in Word 1, B = the number of trigrams in Word 2, and
C = the number of trigrams in common between the two words. The leading space in
each word is counted as a character. For example, Midol has the following trigrams:
[_mi, mid, ido, dol
] and Haldol has the following trigrams: [_ha, hal, ald, ldo, dol
bolded trigrams are in common, so using the formula above: 2(1) / (4 + 5) = .22. Trigram
scores range from 0 (no trigrams in common) to 1 (word is exactly the same). Also, note
that shorter words will typically result in higher trigram scores than longer words with the
same number of trigrams in common. The trigram formula is not the only approach to
orthographic similarity nor it is necessarily ideal, but it provides a good starting point for
this line of study and is well-used in the medication error literature (Lambert, 1997;
Lambert, Chang & Gupta, 2003; Lambert, Chang & Lin, 2001; Lambert, Lin, Chang &
Though it is feasible that word pairs with higher trigram scores might be more
easily confused than word pairs with lower trigram scores, this may not necessarily be the
case. One reason already mentioned is the potential confound concerning the length of
the words in the pairing, but a second (and perhaps more interesting) reason is that some
research which has argued that name confusability is one of the primary suspects in
medication errors is built on self-report data from medical professionals (Lambert, 1997).
Thus, the medication error events studied were subject to self-report bias. Furthermore,
although the research points out that, among the events studied, trigram scores tended to
be high, the only events selected for study were those where the reporting pharmacist
argued a priori
that he/she must have confused the correct drug with the one actually
dispensed. Thus, the fact that the drugs implicated in these events showed high trigram
scores is not surprising – it is likely an artifact of the sampling process.
More recent research has shown that, in controlled studies using pharmacists,
neighborhood density did significantly affect name recognition, particularly on names
presented more frequently during testing (Lambert, 2005). But, an earlier study showed
that name similarity facilitated
free recall of a list of drug names (Lambert, Chang & Lin,
2003). To further clarify this issue, this study will test two sets of word pairs – a low-
trigram and a high-trigram set – to determine under controlled conditions if name
confusability is more problematic as trigram score increases. In addition, stimulus
presentation frequency will be held constant at the word-pair level. Because we are using
pharmacy-naïve participants, we can eliminate the hindsight bias that may be
confounding pharmacist reports given that these participants will not be expected to have
preconceived notions of "which drug I probably confused the other drug for." Also, we
will be able to examine the effects of high orthographic similarity more completely
because pharmacy-naïve participants will not possess the rich knowledge structures for
these drug products that pharmacists do that may confound a pure perceptual analysis.
Because the goal of this study is to empirically test the logic behind exaggerated
text and whether that logic holds across levels of orthographic similarity, we propose the
H1: Measures of detection performance (d'
, raw error rate) will be significantly
higher in the capital-letter exaggeration group than in the control group.
H2: Measures of detection performance (d'
, raw error rate) will be significantly
higher in the color-enhancement exaggeration group than in the control group.
H3: Measures of detection performance (d'
, raw error rate) will be significantly
lower for the high-trigram group than for the low-trigram group in all text exaggeration
H4: Measures of detection performance (raw correct decision rate, c
) will be more
liberal in the high trigram group than in the low trigram group.
One hundred and three participants from a southwestern university volunteered to
participate in the research. Each participant was paid $6.50 USD for their time, which
ultimately totaled 3 hours. The sample was composed of 22 men and 81 women and had
a mean age of 22.2 years (SD = 5.6). All participants were enrolled full-time in college at
the time of testing and were recruited from a variety of different courses.
80 drug name pairings were selected as stimuli for this study. These name
pairings are listed in Appendix A sorted by their trigram scores and divided into "low"
and "high" groups. These groupings were constructed using several steps. First, the Food
and Drug Administration has mandated that certain drugs be labeled using exaggerated
labeling much like that being tested in this study, and these name pairs were included in
our name database. Second, a leading U.S. non-profit medication safety organization has
published a list of name pairs that have caused confusion in the past based on pharmacist
self-reports of errors. Third, a small team of researchers (including the lead author)
examined the 2003 version of the Physician's Desk Reference book of prescription drugs
for name pairs that appeared similar enough to each other to be potentially confusing.
The final database included 128 name pairs, although it should be noted (as is evident
from Appendix A) that some individual drug names were used in multiple pairs.
The 128 name pairs were filtered further using a trigram formula first used by
Lambert and his colleagues in the context of drug name confusability (Lambert, 1997;
Lambert, et al, 1999). The trigram formula parses each drug name into successive sets of
three letters and compares the trigram sets for the word pairs for identical members.
Generally speaking, the more identical trigrams there are between two names, their
greater their orthographic similarity and the higher their trigram score will be. To create a
complete trigram matrix, all drug names in the database were compared to all other drug
names, and then trigram scores were sorted from highest to lowest. This data also can be
found in Appendix A. To construct an independent variable based on orthographic
similarity, the highest and lowest 40 trigram score pairs were used, eliminating
approximately the middle third of the total trigram database.
The independent variable for text exaggeration was created by selecting a part of
every drug name and emphasizing it graphically using either capital letters (of the same
font and size) or color enhancement (the selected word segment printed in red instead of
black). Examples of these exaggerations using the two drug names discussed above
(hydralazine and hydroxyzine) are shown in Figure 1. Every drug name used in the study
could be emphasized using these techniques. Finally, the selected segment of the drug
name is the most distinct segment from the potentially confusable pair name; thus, the
"ala" in hydralazine would be emphasized to distinguish it from hydroxyzine. Though
this is not a foolproof method of selecting which parts of the name to emphasize, it is the
method recommended by the FDA and so we employed it in this study.
Signal probability for all groups was held constant at 50%; thus, half of the name
pairs presented to participants were mismatches and half were matches. This was done to
maximize the number of times that each name pair was observed by the participants, so
that later analyses can approach the data from the standpoint of individual word pairs to
determine the shared characteristics of the most confusable drug names.
. Two Windows-based computers were used for testing, each running
Windows 2000 at 1 Ghz with a 17-inch tube monitor (dot pitch .22) at 1024 x 768
resolution. Superlab Pro was used to display the drug name stimuli and RB-834 8-button
response pads were used to record participants' decision data. All visual stimuli were
Figure 1. Examples of text exaggerations applied to word pairs in this study.
presented as JPEG files 3" from the screen bottom and beginning 2.2" from the left edge
of the screen. Each stimulus filled a rectangular space 1" high; the width of this space
varied with the length of the word displayed. The approximate distance from eye to
screen was 32" using a prototypical 66" tall human participant as a reference. Stimuli
were presented in black type using Arial font (except for words displayed using the color
enhancement exaggeration, in which case part of the word was displayed in red) against a
white screen background.
Participants were randomly assigned to an experimental group based on which
text exaggeration (control, capital letters, color enhancement) and orthographic similarity
(high, low) to which they would be exposed. Stimuli for each experimental condition
were saved in a separate experiment file but all files followed the same general procedure
as described in this section.
Figure 2 shows the screen progression for a given trial in the experiment. The first
image in any trial was a series of black plus signs against a white background. This image
was displayed for three seconds, after which a prime word appeared containing any
necessary text exaggerations. The prime word was displayed for 450 ms, and then
immediately replaced with a black bar which covered the entire area where the prime
word was located. The black bar was intended to clear the participant's iconic memory
registers and encourage short-term memory usage. The final screen in each trial was the
target word, which was always presented in standard type with no exaggerations and in
the same physical position as the prime word. Participants were instructed to, as quickly
as possible, decide whether the target and the prime words were identical. If they were,
participants were to press the green button on the response pad; if they were not, the red
button on the pad was the correct response. Once a response was made, the plus sign
image returned and a new trial began immediately. In an effort to simulate the feedback
isolation of a pharmaceutical environment no feedback concerning correct or incorrect
decisions was provided at any point during the task.
Participants were exposed first to four practice trials using common English
words to orient them to the exaggeration type they would see. After these four trials,
remaining questions were answered and participants were allowed to relax with eyes
closed for one minute. Next, they began the main experiment, which was composed of
480 total word pairs (80 unique word pairs) over six trial blocks. Thus, each unique word
Figure 2. Graphic depiction of the screen sequence involved in one experimental trial.
pair was seen six times by each participant. However, within trial blocks, the sequencing
of the word pairs was randomized to control for practice and order effects. For each trial,
response data was recorded in addition to response time data. Also, information
concerning which word pair was displayed at which time was also recorded, allowing for
future analyses to examine which particular word pairs were most confusable in the
study. Finally, between each trial block, participants were asked to relax for three minutes
with eyes closed before beginning the next block. Experimenters were on hand to track
this relaxation time and initiate the next block of stimuli, but during the experimental
blocks, participants were left alone in the primary testing room. Experimenters could be
contacted in the event of unique circumstances with two-way radios – this occurred in the
case of three participants who were ultimately removed from the database.
When the final trial block was finished, participants were asked to complete a
post-test (the content of which is not relevant to the research questions in this study) and
were released after being informed that their payment for participation would be arriving
Mean differences on all primary dependent variables for categorical variables not
of specific interest to our hypotheses (gender, age) were conducted to rule out any
unintended differential effects. Men and women did not differ significantly on any
dependent measure, and there were no significant correlations between participant age
and the dependent variables. Thus, all analyses beyond this point will collapse across
these categorical factors.
Table 1 displays means and standard deviations for raw error rate, raw correct
decision rate, raw reaction speeds, d'
, and c
, a measure of response bias considered to
possess the strongest psychometric properties (See, et al, 1997). These scores are broken
down further by exaggeration type and trigram group. Performance statistics for the high-
trigram, capital-exaggeration group were noticeably worse than for any other group.
was lower for all exaggeration types in the high-trigram groups. Finally, it is also
interesting that c
was higher in the capital-exaggeration group compared to the other
Addressing Hypotheses H1 and H2, Figure 3 shows means and 95% confidence
intervals for d'
and raw error rate across the three text exaggeration types. Separate lines
are plotted for trigram group assignments. No mean differences in performance across
text exaggeration types were observed for either performance measure, regardless of
Table 1 Descriptive Statistics for Performance Measures Within Text Exaggeration and Trigram Groups Measure Exaggeration
Trigram Group Low
97.31 95.20 96.90 95.50 98.28 94.02
Correct Decision Rate 94.29 91.00 95.42 90.36 93.23 85.61 Reaction Time 0.94 0.97 0.91 0.95 0.95 1.17 d-prime 3.82 3.16 3.86 3.20 4.02 2.80 Response Bias (c
) 0.19 0.18 0.17 0.21 0.33 0.25 Note: Reaction time in seconds per trial.
trigram group assignment (all p
's > .05). It should be noted that group differences in d'
within the high-trigram condition neared significance (p
= .063), but in the opposite
direction expected such that the capital-exaggeration group performed more poorly than
the other two groups. The maximum calculated power for these comparisons was
approximately .65. Therefore, Hypotheses H1 and H2, which suggested that task
performance would be different in specific directions between text exaggeration groups,
was not supported by these data.
Addressing Hypothesis H3, Figure 4 displays means and 95% confidence
intervals for d'
and raw error rate across trigram groups. With respect to d'
differences between trigram groups are notable and significant for all exaggeration
conditions (for control group, F(1, 35) = 9.65, p
< .01; for color group, F(1, 31) = 7.58, p
< .01; for capitals, F(1, 31) = 17.69, p
< .01). For raw error rate, the difference between
trigram groups is not as dramatic, but is still reliable across exaggeration conditions (F (1,
Figure 3. Means and 95% CIs for d'
and raw error rate for different orthographic similarities and within text exaggeration conditions.
Figure 4. Means and 95% CIs for d'
and raw error rate between text exaggeration conditions and within levels of orthographic similarity.
101) = 17.01, p
< .01). Within exaggeration groups, significant differences were observed
between trigram groups for the capital-letter exaggeration condition (F (1,31) = 16.34, p
< .01), but not for the other conditions. On the basis of these findings, we can report that
Hypothesis H3 received support from these data.
To address Hypothesis H4, raw correct decision rate and response bias (c
examined between trigram groups across the entire sample and within exaggeration
conditions. Participants showed greater liberality in terms of overall false alarm responses
in the high trigram group (F (1, 101) = 19.96, p
< .01), but no group differences in c
differences in overall false alarm responses were located in the two exaggerated-text
conditions (both p
's < .02). The maximum calculated power for comparisons involving
response bias was approximately .44. Thus, H4 was only partially supported by these
The purpose of this study was to determine the effectiveness of text exaggerations
on drug name identification and to assess whether the orthographic similarity of the
names moderated accurate identification. We used a prime-target name matching task
designed to measure the immediate quality of the memory trace and whether text
exaggeration improved that quality. The data showed that the text exaggerations,
specifically the capitalization of specific letter segments in the word, did result in higher
sensitivity (as measured by d'
) but only under specific circumstances. For the most part,
the text exaggerations did not improve accurate performance beyond a control condition
with no exaggerations. In addition, orthographic similarity as measured using trigrams
(matched pairs of three-letter segments between two words) was a moderator of accurate
name identification such that high-trigram names were identified less accurately than
Since 2001, the Food and Drug Administration has suggested that certain drug
names should be exaggerated in some way to differentiate them from a potentially
confusable similar name (i.e., glipizide and glyburide, Vioxx and Zyvox). Manufacturers
have in many cases assented to these changes but there is little evidence available to
indicate that these exaggerations have any value with respect to accurate name
identification. The present study proposed that if the exaggerations didn't create an initial
memory trace that improved name identification in a focused-attention task, then it
probably would not improve performance in a real-world setting like a pharmacy. The
results of this study pose serious questions regarding the benefits of cognitive
comparisons between exaggerated and unexaggerated versions of the same words. Of
course, this is an important question because, although the label on the stock bottle is
exaggerated, the drug name on the script form or on the computer screen is not.
There are some limitations to the current study that should be noted. First, these
participants were not pharmacy-trained individuals. Many had probably never seen the
preponderance of these drug names before and perhaps didn't even know how to
pronounce some of them correctly. Thought this is a potential limitation it also controlled
for potential biases. In other words, trained pharmacy personnel could have introduced
two biases into the current investigation: increased dependence on phonological rather
than graphicological processing of the names (Frost, 1998; Pugh, et al, 1994) and
knowledge of which drug names were likely to appear on the screen due to their training.
Although the present simulation provides important insight into how novices respond to
the exaggerations, the response of trained professionals remains unclear. Given the
findings of this study and the safety sensitive nature of the pharmaceutical industry it
would seem prudent to examine exaggeration dynamics in a pharmaceutical setting.
A second limitation has to do with the structure of the task. In a real pharmacy,
names could be compared several times from the time a script is initiated to the point
when it is completed. In our task, the prime words were shown for 450ms and could not
be reviewed further. Therefore, it is possible that text exaggerations have their benefit
later in the prescription process, not right at the beginning or after only one name
comparison. On the other hand, the data do indicate that performance stabilized on the
task fairly quickly (probably after 120-150 pairs of names), so further exposure to the
names may not have made much difference. Indeed research that shows pharmacists to be
less accurate in checking scripts that they filled personally compared to scripts that co-
workers filled (eg., Grasha, Reilley, Schell, Tranum & Filburn, 2000). In the same way,
once the individual has decided that the names match, further checks may be biased by
their initial conclusions. Future research should be done to test this possibility.
A third limitation is that drug name pairs of moderate trigram similarity were
removed to create the two trigram groups. Thus, it remains unclear whether these drug
name pairs would be identified at different sensitivities than the high- and low-trigram
pairs. Nevertheless, constructing the experiment such that orthographic similarity was an
independent grouping variable did provide the benefit of simple and clear tests of the
orthographic similarity hypothesis that confirmed much of the work that Lambert and his
colleagues did which suggested that trigram similarities were influential on reported
medication errors (Lambert, 1997, 2005; Lambert, et al, 2001; Lambert, et al, 1999).
Treating similarity as a continuous measure would have complicated the interpretation of
the effects of the factor on name identification, and these effects were important given the
reliance of Lambert's work on pharmacist self-reports of errors and the likely sample bias
that results from such self-reports.
Finally, ceiling effects in the dependent measures are always a concern with this
sort of study. Examination of the means in Table 1 show that control group accuracy rates
are quite high, even after 480 presentations of the drug names. However, we also believe
that a gain of even one percentage point of accuracy is practically important when it
comes to medical mistakes, given that may extrapolate into millions of actual
The results of this study inform our understanding of whether text exaggerations
are actually helpful in preventing name-confusion medication errors. Given the findings
of the present study it appears that the beneficial effects of name exaggerations may be
questionable. More importantly, this study suggests that exaggerations can be detrimental
to name identification, especially at high levels of orthographic similarity. More research
is needed to make a final determination as to whether the exaggerations are actually
preventative of medication errors where drug name confusion may have played a
Carreiras, M., Perea, M. & Grainger, J. (1997). Effects of orthographic neighborhood in
recognition: Cross-task comparisons. Journal of Experimental
Psychology:Learning, Memory & Cognition, 23
Davies, D.R. & Parasuraman, R. (1982). The psychology of vigilance
Duncan, J. & Humphreys, G.W. (1989). Visual search and stimulus similarity.
Psychological Review, 96
Egeth, H.E. & Yantis, S. (1997). Visual attention: Control, representation, and time
Annual Review of Psychology, 48
Frost, R. (1998). Toward a strong phonological theory of visual word recognition: True
issues and false trails. Psychological Bulletin, 123
Gage, J. (1995). Colour and culture. In T. Lamb & J. Bourriau (Eds.), Colour, art and
. (pp. 175-193). New York: Cambridge University Press.
Grasha, A.F., Reilley, S., Schell, K.L., Tranum, D. & Filburn, J. (2000). A cognitive
systems perspective on human performance in the pharmacy: Implications for
accuracy, effectiveness, and job satisfaction. (Technical Report 062100-R).
Cincinnati, OH: Cognitive-Systems Performance Laboratory, University of
Just, M.A. & Carpenter, P.A. (1980). A theory of reading: From eye fixations to
Psychological Review, 87
Katz, L. & Frost, S.J. (2001). Phonology constrains the internal orthographic
Reading & Writing: An Interdisciplinary Journal, 14
Kohn, L.T., Corrigan, J.M. & Donaldson, M.S. (Eds.) (1999). To err is human - building
a safer health care system
. Washington, DC: National Academy Press.
Lambert, B.L. (2005). Designing safe drug names. Drug Safety, 28
Lambert, B.L., Chang, K.Y. & Gupta, P. (2003 ). Effects of frequency and similarity
neighborhoods on pharmacists' visual perception of drug names. Social Science &
Lambert, B.L., Chang, K.Y. & Lin, S.J. (2001 ). Effect of orthographic and phonological
similarity on false recognition of drug names. Social Science & Medicine, 52
Lambert, B.L., Chang, K.Y. & Lin, S.J. (2003 ). Immediate free recall of drug names:
effects of similarity and availability. American Journal of Health-System
Lambert, B.L., Lin, S.J., Chang, K.Y. & Gandhi, S.K. (1999). Similarity as a risk factor
in drug-name confusion errors: the look-alike (orthographic) and sound-alike
Medical Care, 37
Lukatela, G. & Turvey, M.T. (1994a). Visual lexical access is initially phonological: 1.
Evidence from associative priming by words, homophones and
Journal of Experimental Psychology: General, 123
Lukatela, G. & Turvey, M.T. (1994b). Visual lexical access is initially phonological: 2.
Evidence from phonological priming by homophones and pseudohomophones.
Journal of Experimental Psychology: General, 123
Perea, M. & Rosa, E. (2002). Does "whole-word shape" play a role in visual word
Perception & Psychophysics, 64
Pugh, K.R., Rexer, K. & Katz, L. (1994). Evidence of flexible coding in visual word
Journal of Experimental Psychology: Human Perception &
See, J.E., Warm, J.S., Dember, W.N. & Howe, S.R. (1997). Vigilance and signal
detection theory: An empirical evaluation of five measures of response bias.
Human Factors, 39
Stone, G.O. & Van Orden, G.C. (1993). Strategic control of processing in word
Journal of Experimental Psychology: Human Perception &
Treisman, A.M. & Gelade, G. (1980). A feature-integration theory of attention. Cognitive
Wickens, C.D. & Hollands, J.G. (2000). Engineering psychology and human
. (3rd Ed.). Upper Saddle River, NJ: Prentice Hall.
Yantis, S. (1998). Control of visual attention. In H. Pashler (Ed.), Attention
. (pp. 223-
256). Hove, UK: Psychology Press.
Yantis, S. & Hillstrom, A.P. (1994). Stimulus-driven attentional capture: Evidence from
equiluminant visual objects. Journal of Experimental Psychology: Human
Perception & Performance, 20
Drug Name Pairs and Trigram Scores Used In This Study
Trigram x 100
DRAFT: March 5, 2007 Do not cite or quote without permission of the author. THE DANGERS OF SUMMARY JUDGMENT: GENDER AND FEDERAL CIVIL LITIGATION Elizabeth M. Schneider* The interconnections of procedure and gender have been a subject of much national attention, as many federal and state Gender Bias Task Force Reports have documented ways in which gender bias impacts on procedure.1 These issues have also been the focus of considerable scholarship.2 In this Article, I turn to one of the most important procedural devices in federal civil procedure – summary
ACC/AHA/ESC Practice Guidelines ACC/AHA/ESC Guidelines for the Management of Patients With Supraventricular Arrhythmias*—Executive Summary A Report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines and the European Society of Cardiology Committee for Practice Guidelines (Writing Committee to Develop Guidelines