Neuropsychologia 62 (2014) 220232

Contents lists available at ScienceDirect

Neuropsychologia
journal homepage: www.elsevier.com/locate/neuropsychologia

Evidence for general right-, mixed-, and left-sidedness in self-reported

handedness, footedness, eyedness, and earedness, and a primacy
of footedness in a large-sample latent variable analysis
Ulrich S. Tran n, Stefan Stieger, Martin Voracek
Department of Basic Psychological Research and Research Methods, School of Psychology, University of Vienna, Liebiggasse 5, A-1010 Vienna, Austria

art ic l e i nf o

a b s t r a c t

Article history:
Received 27 November 2013
Received in revised form
22 July 2014
Accepted 25 July 2014
Available online 2 August 2014

Lateral preferences are important for the study of cerebral lateralization and may be indicative of
neurobehavioral disorders, neurodevelopmental instability, and decits in lateralization. Previous
studies showed that self-reported preferences are also concordantly interrelated, suggesting a common
genetic or biological origin, sidedness. However, with regard to the assessment and classication of
lateral preferences, there is a dearth of psychometric studies, but a need for psychometrically validated
instruments that can be reliably used in applied research. Based on three independent large samples
(total N4 15,100), this study investigated the psychometric properties of widely-used lateral preference
scales of handedness, footedness, eyedness, and earedness. Preferences were consistently and replicably
categorical, consisting of right, mixed, and left preferences each, underlining that primarily qualitative,
rather than quantitative, differences differentiate lateral preferences. Right-, mixed-, and left-sidedness
underlay the individual preferences, but sidedness alone could not fully explain the observed interrelations. Footedness was the single most important indicator of sidedness. Our data were further
consistent with predictions of right shift theory and corroborated a pull-to-concordance in handfoot
preferences. We recommend the use of psychometrically validated scales and of a trichotomous
classication of lateral preferences in future research, but conclude that handedness may be a biased
indicator of underlying sidedness. Footedness needs to be examined more closely with regard to cerebral
lateralization, neurodevelopmental disorders, and neurodevelopmental instability.
& 2014 Published by Elsevier Ltd.

Keywords:
Laterality
Sidedness
Latent variable analysis
Psychometric analysis
Taxometric analysis

1. Introduction
The preference for one side of the body with regard to limbs
(hand and foot), eyes, and ears is of long-standing interest for the
study of cerebral lateralization, but also in its own regard.
Handedness, probably the most investigated of these so-called
functional lateralizations or asymmetries, is indicative of language
lateralization in the brain (Szaarski et al. 2002; Szaarski,
Holland, Schmithorst, & Byars, 2006) and is of ubiquitous importance in every-day life. With regard to its distribution, it is wellknown that right-hand preference is dominant in the population,
with prevalences around 90% (Coren, 1993; Peters, Reimers, &
Manning, 2006).
Footedness refers to the dominant or preferred foot when
performing manipulative or mobilizing actions in a bilateral
context (Chapman, Chapman, & Allen, 1987; Gabbard & Iteya,
1996; Sadeghi, Allard, Prince, & Labelle, 2000), like kicking a ball

Corresponding author. Tel.: 43 1 4277 47119; fax: 43 1 4277 47192.

E-mail address: ulrich.tran@univie.ac.at (U.S. Tran).

http://dx.doi.org/10.1016/j.neuropsychologia.2014.07.027
0028-3932/& 2014 Published by Elsevier Ltd.

(skilled movement) or standing on one foot (unskilled/stabilizing/

balancing movement). Footedness is of importance in sports (e.g.,
Carey et al. 2009), but has also been reported to be an indicator of
language and other cerebral lateralizations, even superior to
handedness (Chapman et al., 1987; Gabbard & Iteya, 1996; Elias
& Bryden, 1998; Elias, Bryden, & Bulman-Fleming, 1998;
Searleman, 1980; Strauss, 1986), probably because of less social
pressure regarding side preference (Chapman et al., 1987). Rightfoot preference is considerably lower than right-hand preference
in the population, averaging around 80% (Porac & Coren, 1981), but
remarkably independent of foot skills (Carey et al., 2009).
Eyedness (ocular or sighting dominance) refers to the preference of one eye for monocular activities, such as looking through a
telescope, and must be distinguished from sensory dominance in
binocular activities and acuity dominance (visual acuity differences between the eyes), with both of which it is uncorrelated
with (Porac & Coren, 1976). Only about two thirds of the population is right-eyed (Bourassa, McManus, & Bryden, 1996; Porac &
Coren, 1976). Earedness refers to the preference of one ear in
monaural activities, such as placing an ear against a closed door to
listen in to a conversation, and appears to be the least investigated

U.S. Tran et al. / Neuropsychologia 62 (2014) 220232

of the lateral preferences that are of concern here (Porac & Coren,
1981). Similarly to eyedness, sensory dominance or differences in
acuity in binaural activities need to be conceptually distinguished
from earedness (Noonan & Axelrod, 1981). Earedness is only
weakly functional asymmetric, about 60% of the population being
right-eared (Porac & Coren, 1981), but is apparently a better
predictor of language lateralization than handedness, footedness,
or eyedness (Strauss, 1986).
A plethora of studies has provided evidence that, overall, lateral
preferences in handedness, footedness, eyedness, and earedness
are concordantly interrelated with associations between handedness and footedness being often strongest (Bourassa et al., 1996;
Dellatolas, Curt, Dargent-Par, & De Agostini, 1998; Dittmar, 2002;
Kang & Harris, 2000; McManus, Porac, Bryden, & Boucher, 1999;
Noonan & Axelrod, 1981; Porac, 1997; Reiss, 1999; Reiss & Reiss,
1999; Suar, Mandal, Misra, & Suman, 2007). Thus, lateral preferences may share a common genetic or biological origin, primary or
overall sidedness (Annett, 2002; Corballis & Morgan, 1978;
McManus, 1985; Previc, 1991). While family and genetic studies
suggest (strong) familial aggregations and (weak) genetic associations (Bourassa et al., 1996; Dellatolas et al., 1998; McManus et al.,
1999; Reiss, 1999; Reiss & Reiss, 1999; Warren, Stern, Duggirala,
Dyer, & Almasy, 2006), some twin studies have failed to conrm
the genetic determination of sidedness (Reiss, Tymnik, Kgler,
Kgler, & Reiss, 1999). To date, a complex multigenetic and multifactorial model of lateral preferences and their interrelations
appears most likely (McManus, Davison, & Armour, 2013; Reiss,
1999; Warren et al., 2006).
However, research into lateral preferences faces some important methodological problems. First, given the skewed distributions of lateral preferences, sample size and study power are an
issue. For example, the 10:90 handedness ratio lowers the power
of statistical tests by about 50% compared to a 50:50 ratio. Large
samples are thus required.
Second, classication of lateral preferences frequently adopts either
a dichotomy (right/left and right/non-right) or a trichotomy (right/
mixed/left), often based on arbitrary criteria and cutoffs on dimensional, continuous measures. This heterogeneity may prohibit direct
comparisons between studies (Beaton, 2008) and may also lead to
vastly different results even with the same data (Kelley, 2012). Most of
the above reviewed studies used dichotomies for classication (but see
Dittmar (2002), Gabbard and Iteya (1996) and Kang and Harris
(2000)). However, there is evidence from latent variable analyses that
at least for handedness a trichotomy may be more adequate (Dragovic
& Hammond, 2007; Dragovic, Milenkovic, & Hammond, 2008).
Third, with regard to assessment itself, the use of multi-item
inventories is recommended. The use of single items (such as writing
hand for handedness) may entail the underestimation of interrelations of lateral preferences (Bourassa et al., 1996; McManus et al., 1999;
Warren et al., 2006) and of associations with other variables, such as
sex (Papadatou-Pastou, Martin, Munafm, & Jones, 2008). Yet,
existing multi-item inventories, like the Edinburgh Handedness
Inventory (EHI; Oldeld, 1971) and the Lateral Preference Inventory
(LPI; Coren, 1993), differ with regard to item composition and response
format and rigorous psychometric analyses are scarce. Existing analyses suggest that some items of widely-used self-report inventories are inappropriate for the accurate assessment of handedness
(Dragovic, 2004; Dragovic & Hammond, 2007; Milenkovic & Dragovic,
2013; Veale, 2013) and that skilled and unskilled activities may
constitute separate factors in handedness (Healey, Liederman, &
Geschwind, 1986; Kang & Harris, 2000; Mikheev, Mohr, Afanasiev,
Landis, & Thut, 2002; Nicholls, Thomas, Loetscher, & Grimshaw,
2013; Steenhuis & Bryden, 1989) and footedness (Kalaycolu, Kara,
Atbaolu, and Nalac 2008; Kang & Harris, 2000; Mikheev et al.,
2002; Schneiders et al. 2010). However, the reported multidimensionality of handedness and footedness may have been spurious. Studies

221

relied on factor-analytic methods that were not suited for the highly
skewed item response distributions that are typically encountered in
lateral preference inventories. This may have resulted in an overextraction of factors (see Bernstein and Teng (1989)), caused by the
clustering of items with similar distributional properties in different
factors. With regard to response format, it is unclear whether three
categories, delineating right, left, and no preference as in the LPI, or
ve categories, differentiating within left and right between always
and usually as in revised versions of the EHI (Veale, 2013), are better
suited for the assessment of lateral preferences.
Overall, there is a dearth of psychometric and latent variable
analyses with regard to the assessment and classication of selfreported lateral preferences, even though such analyses are of
importance for the various and numerous elds of applied laterality
research.
Recent studies point out that mixed-handedness may be a risk
factor for neurodevelopmental and neurobehavioral disorders
(e.g., ADHD and language problems: Rodriguez et al. 2010; schizophrenia: Dragovic & Hammond, 2005; Sommer, Ramsey, Kahn,
Aleman, & Bouma, 2001; schizotypy: Somers, Sommer, Boks, &
Kahn, 2009), as mixed-handedness is considered an observable
manifestation of underlying neurodevelopmental instability
(Golembo-Smith et al. 2012; Rodriguez & Waldenstrm, 2008)
and decits in lateralization (Crow, 2013). Recently, Willems, Van
der Haegen, Fisher, and Francks (2014) have advocated specically
including left-handers in neuroscientic and neurogenetic studies,
instead of excluding them; this would further knowledge of brain functioning and allow a deeper insight into cerebral
lateralization and its genetic underpinnings as it is currently the
case. However, there is a need for psychometrically validated
instruments that can be reliably used in applied research
(Rodriguez et al., 2010).
The present study addressed the above issues. By design, our
study comprised three independent, large samples (total
N 4 15,100), following recent recommendations to counteract
potentially false-positive and thus irreproducible research ndings (Asendorpf et al. 2013). In genome-wide association studies, independent discovery and replication samples within the
same study are considered best practice, in order to guard
against false-positive ndings and to demonstrate the robustness of an effect, if the replication is successful (McCarthy et al.
2008). The present study included one sample (n 4 2400) in
which self-reported handedness was assessed with more items
than in the other two samples (total n 4 12,700). This sample
served for calibration purposes (calibration sample), whereas
the other two samples served for the purpose of crossvalidation (comparison sample 1 and 2). With regard to all
other lateral preferences, the three samples were full replication
samples of each other.
First, we investigated the dimensional structure, item properties, and optimal number of response categories of widely-used
self-report measures of lateral preferences in handedness, footedness, eyedness, and earedness with structural equation modeling
(SEM) and item response theory (IRT). Second, using two independent approaches, latent class analysis and taxometric analysis,
we determined whether lateral preferences were categorical or
dimensional, providing empirically derived cutoffs that may be
used in future research. Third, interrelations between lateral
preferences were examined, investigating evidence for underlying
overall sidedness, and also investigating the inuence of sex and
age on lateral preferences and sidedness. Fourth, the observed
pairwise associations of lateral preferences were utilized to probe
predictions of two specic single-locus genetic models, right shift
(RS) theory (Annett, 2000; Annett, 2002) and the dextral and
chance allele model (DC model; McManus, 1985) (for background
and details, see Section 2.3.6 below).

222

U.S. Tran et al. / Neuropsychologia 62 (2014) 220232

2. Material and methods

2.1. Participants
Three independent samples were used in this study, one calibration sample
(n2456) and two comparison samples, comparison sample 1 (n 7658) and
comparison sample 2 (n 5062). Descriptive characteristics of the samples are
given in Table 1. Overall, there was a broad age range and slightly more women
than men. A majority of the participants in the samples were Austrian or German.
Participants with nationalities other than Austrian or German were Italian (1.2%),
Turkish (.5%), and Romanian (.3%). Samples differed with regard to nationality
(2(4) 204.34, p o .001; more Germans in the calibration sample) and sex (2(2)
50.97, p o .001; more women in the comparison samples). Participants in comparison sample 1 were also on average 5.5 years younger than participants in the other
two samples (t(15,141) 23.51, p o .001, d .38).

2.2. Procedure and measures

Data were collected by a large number of data collectors in the course of a
larger project on individual difference variables. For the three samples, waves of
data collection were temporally separated by about half a year (comparison sample
1 vs. 2) and one year (comparison sample 2 vs. calibration sample), respectively,
and independent from each other with regard to the data collectors involved.
Participants were approached on a personal basis, using word-of-mouth and
personal contacts. Participants had to be uent in German, as this was the survey
language. Apart from insufcient language prociency, there were no further
exclusion criteria. Study participation was voluntary, anonymous, and participants
were not remunerated for participation. Participants completed the questionnaires
independently and unaided.
Participants in the calibration sample were presented with the 16 nonoverlapping handedness items of the 10-item EHI (Oldeld, 1971) and 12-item
LPI (Coren, 1993). Items asked for hand preference with regard to: writing, drawing,
throwing a ball to hit a target (throwing), holding a tennis or squash racket
(tennis; LPI), using a toothbrush (toothbrush), using a knife to cut something
without simultaneously using a fork (knife), using a hammer to drive a nail into
something (hammer; LPI), striking a match (match), using a pencil eraser
(eraser; LPI), dealing a card from a deck of cards (card; LPI), threading a needle
(needle; LPI), holding a y ap (y ap; LPI), using scissors (scissors; EHI),
holding a spoon (spoon; EHI), the upper hand when holding a broom (broom;
EHI), and opening the lid of a box (box; EHI). Response options were always right,
usually right, no preference, usually left, and always left (in this order), coded 2,
1, 0,  1, and  2, respectively. Cronbach of the 16-item scale was .96, .94 for the
10 EHI items and .96 for the 12 LPI items.
Participants in the comparison samples were presented with four of the above
16 items (writing, throwing, knife, and hammer) that had been selected
independently and a priori with respect to broad trait coverage (vs. redundancy),
item performance indicators, and balance of ne- vs. gross-motor skills. Response
options were left, either, and right (in this order), coded  1, 0, and 1, respectively.
Cronbach of the 4-item scale was .91 in comparison sample 1 and .92 in
comparison sample 2. For ease of presentation, options either and no preference are
uniformly referred to as no preference in the remainder of this text.
For the assessment of footedness, eyedness, and earedness, the three respective
self-report scales of Coren (1993) were used. Items asked for foot preference with
regard to: kicking a ball to hit a target (ball), picking up a pebble with the toes
(pebble), stepping on a beetle or a cigarette stump (beetle), stepping up onto a
chair (chair); eye preference with regard to: looking through a telescope (telescope), looking into a dark bottle to see how full it is (bottle), peeping through a
Table 1
Sample characteristics
Sample

n
Women, n (%)
Age, range (years)
Interquartile range
Mean (SD)
Nationality, n (%)c
Austria
Germany
Other
a
b
c

Calibration

Comparison 1

Comparison 2

2455
1240 (50.7%)a
1890b
2348b
35.64 (15.48)b

7658
4456 (58.2%)
1889
2235
30.24 (12.62)

5062
2749 (54.3%)
1892
2248
35.74 (16.05)

1397 (57.3%)
923 (37.8%)
119 (4.9%)

5339 (70.0%)
1773 (23.2%)
514 (6.7%)

3415 (67.7%)
1310 (26.0%)
316 (6.3%)

n 2448 due to missing data.

n 2423 due to missing data.
n2439, 7626, and 5041 due to missing data.

keyhole (keyhole), sighting down a gun or rie (gun); ear preference with regard
to: placing an ear against a closed door to listen in to a conversation going on
behind (door), placing an earphone (earphone), placing an ear against someone's
chest to hear his/her heartbeat (heartbeat), placing an ear against a small box to
hear a clock ticking within (clock). Items were presented with ve response
categories in the calibration sample and three response categories in the comparison samples. Cronbach was .81/.71/.73 for footedness in the calibration and the
two comparison samples, respectively, .94/.90/.90 for eyedness, and .87/.79/.81 for
earedness. Lateral preference scales were presented to participants in one block, in
the order of handedness, footedness, eyedness, and earedness.
2.3. Statistical analysis
2.3.1. Handedness: dimensionality and item selection
Dimensionality and item properties were investigated with SEM and IRT
methods with Mplus 6.11 (Muthn & Muthn, 2008). One-factor models were
tted to the data of the calibration sample, using robust weighted least squares
estimation with a mean- and variance-adjusted chi-square test statistic (WLSMV)
which is based on the items polychoric correlation matrix, conforming to the
items' ordered categorical format and also dealing adequately with the skewed
item response distributions (Beauducel & Herzberg, 2006).
In terms of IRT analysis, tting one-factor models with WLSMV to the data
corresponded to tting the graded response model (GRM; Samejima, 1969) to the
data, where a person's probability of endorsing a specic item response category is
modeled by a cumulative function involving the person's latent trait value and a set
of item parameters, corresponding to item discrimination and item thresholds, the
latter indicating the point on the latent trait continuum where the chance of
endorsing a given or a higher response category is 50% for a person with the same
latent trait value. Item discriminations in the GRM correspond to factor loadings in
SEM and may be estimated via SEM alongside threshold parameters. Item
parameters were estimated using Mplus DELTA parameterization and the variance
of the latent trait was set to unity.
Model t was assessed with the comparative t index (CFI), the Tucker-Lewis
index (TLI; CFI and TLI: good t: Z .95, acceptable t: Z.90), and the root mean
square error of approximation (RMSEA; good t: o .06, acceptable t: o.08), using
the benchmarks of Hu and Bentler (1999). In analyses with few degrees of freedom
(df), evaluation of model t was primarily based on CFI and TLI, because RMSEA,
penalizing for model complexity with the chi-square to df ratio, may then be
inated (Kenny, Kaniskan, & McCoach, 2014). Based on these analyses, items were
selected that loaded highest on the latent trait and were kept for further analysis.
2.3.2. Handedness: number of response categories
Item characteristic curves (ICCs) of the retained items in the calibration sample
were inspected with regard to whether response categories were sufciently
separated. Being based on a cumulative response function, threshold parameters
in the GRM are by denition in the same order as the response categories and
convey no information on category separation, i.e., whether each category has
highest probability of endorsement along a distinct section of the latent trait that
does not overlap with sections of the other categories. Indication of insufcient
separation may, however, be gained from inspection of the ICCs. Informed by these
analyses, the number of response categories was reduced, combining adjacent
response categories, and assessing the t of some alternative models in the
calibration sample. The best-tting alternative was kept as nal model and also
investigated in a 4-item short scale which was also used in the comparison
samples. Fit of one-factor models in this scale was then further assessed in the
comparison samples.
2.3.3. Handedness: latent classes and taxonity
Based on the nally obtained scale (calibration sample) and the 4-item short scale
(comparison sample 1 and 2), we determined whether the latent trait underlying
handedness was categorical or continuous with latent class analysis (LCA; e.g., Collins
and Lanza (2010)) and taxometric analysis (Meehl, 2004; Waller & Meehl, 1998),
independently in the calibration sample and the two comparison samples. In addition,
analyses with the 4-item short scale were also performed in the calibration sample to
enable direct comparisons with results in the two comparison samples. LCA employs a
latent variable model wherein associations between observed variables are explained by
a number of underlying discrete classes; given the latent classes, associations between
the observed variables disappear (conditional independence). Conditional independence
may be unrealistic in certain cases and can be relaxed, allowing also for residual
dependence of observed variables. LCA has been utilized in studies on handedness
before (Dragovic & Hammond, 2007; Merni, Di Michele, & Soffritti, 2013). We evaluated
the t of models with increasing numbers of latent classes, determining the smallest
number that explained the data best. Latent GOLD 4.5 was utilized for these analyses,
treating items as nominal, following previous results (Dragovic & Hammond, 2007;
Dragovic et al., 2008). Model t was assessed with (1) the Bayesian information criterion
(BIC), based on the log-likelihood value of the model; (2) percentages of classication
error; (3) the likelihood-ratio goodness of t statistic (L2). Indicator loadings, i.e.,
standardized linear regression coefcients for the indicatorcluster relationships, are
also reported.

U.S. Tran et al. / Neuropsychologia 62 (2014) 220232

Taxometric analysis served to cross-validate the LCA solutions, providing an
alternative approach to assess the latent structure of handedness and relying on fewer
assumptions than LCA. Like LCA, taxometric analysis has been utilized in studies on
handedness before (Dragovic et al., 2008). Taxometric analysis divides observed
variables into input and output indicators, creates a series of ordered subsamples along
the input indicator, and graphs the results. MAMBAC analysis (mean-above-minusmean-below-a-cut; Meehl & Yonce, 1994), one of the most often used taxometric
methods (Haslam, Holland, & Kuppens, 2012), was used in this study, utilizing TaxProg
(Ruscio, 2012). MAMBAC sorts cases along the input indicator (x-axis) and plots at
evenly spaced cuts mean differences on the output indicator (y-axis). Mean differences
are calculated as means for cases above the cut minus means for cases below the cut.
Categorical data should yield a peaked curve, whereas dimensional data are expected to
yield a concave curve (Meehl & Yonce, 1994). TaxProg uses the observed variables in all
possible inputoutput pairings and 50 cuts as default. It also allows the generation of
categorical and dimensional comparison data (k 100 samples each, with size
n100,000 each, as default). With the comparison data, a quantitative index of curve
t (comparison curve t index [CCFI]; Ruscio & Kaczetow, 2009; Ruscio, Walters,
Marcus, & Kaczetow, 2010) may be obtained, ranging from 0 (dimensional) to 1
(categorical). For .4oCCFIo.6, results are inconclusive and should be interpreted with
caution. CCFI values have been found to be highly reliable in differentiating between
dimensional and taxonic simulated data even under unfavorable conditions (Ruscio &
Kaczetow, 2009; Ruscio et al., 2010) and use of CCFI values also reduced spurious
ndings in real data (Haslam et al., 2012). Evaluation of taxonity was therefore strictly
based on CCFI values.
Differences between obtained handedness classes/taxa were investigated (analysis
of variance; ANOVA) also with respect to laterality quotients (LQs) that are commonly
used with the EHI, computed for each scale and domain with the formula (Kelley, 2012)
LQ

kj 1 score of ith item

kj 1 absolute score of ith item 2  number of items scored '0'

 100;

where k refers to the number of items in the scale. LQ scores, ranging from  100
(exclusive left preference) to 100 (exclusive right preference), were further used
to derive cutoffs of obtained classes/taxa in all laterality domains with nonparametric receiver-operator-curve (ROC) analyses.

2.3.4. Footedness, eyedness, earedness

Similar analyses as for handedness were conducted with regard to scales of
footedness, eyedness, and earedness. Associations between lateral preferences in
all four investigated domains, and with sex and age were then investigated with
multinomial logistic regression analyses, controlling for sample and interactions of
sample with main effects to test for replicability of effects.
Cases with missing values in any items of the laterality scales were included in
all SEM and LCA analyses, but, due to methodological limitations, only cases with
complete data could be included in the taxometric analyses. More than 99.5% of
participants had none or at most one missing value per sample and scale.

2.3.5. Sidedness
The overall underlying structure of lateral preferences, sidedness, was nally
investigated with LCA. Associations of sidedness with sex and age, and associations
of the occurrence of inconsistent (i.e., not absolutely concordant preferences in any
pair of domains) and crossed (i.e., a right preference in one domain and a left
preference in another) lateral preferences in the four domains with sex, age, and
sidedness were investigated with (multinomial) logistic regression analyses, again
testing for replicability.

2.3.6. Probing predictions of single-locus genetic models

RS theory (Annett, 2002) assumes a continuum from left to right preferences and
asserts that lateral preferences are associated with cerebral dominance. A person's
position on the leftright continuum depends on a purely stochastic process, assuming
an underlying normal distribution, but is affected by a hypothesized RS-allele. Presence
of one allele (genotype RS  ) causes a right shift, presence of two alleles (RS ) a
further right shift, compared to individuals with RS  genotype. Annett (2000)
presented a statistical model to predict the bivariate distribution of lateral preferences
according to RS theory, using a dichotomous classication of right vs. left preferences. A
Supplementary table in Annett (2000) contains empirical estimates of the shift
introduced by the RS-allele and of the population proportions of RS  , RS  , and
RS genotypes, valid for Western populations. In the present study, we investigated
the t of predicted proportions of all pairs of lateral preferences in our data with chi
squared goodness-of-t tests.
The DC model (McManus, 1985) posits that left and right lateral preferences depend
on Dextral (D) and Chance (C) alleles. Genotype DD has a 0% probability for left
preferences, DC 25%, and CC 50%. In contrast to the RS model, the DC model does not
allow for mixed preferences (which may be combined with either left or right
preferences in the RS model), but excludes them from computations. A specic
prediction of the DC model is that for pairs of lateral preferences the proportions of
rightleft preferences and leftright preferences are the same (see Annett, 2000).
McManus et al. (1999) found that this equality held in groups of consistent-handers
(who use the same hand for writing and throwing) and inconsistent-handers separately.

223

We derived the respective proportions for all pairs of lateral preferences in our data and
investigated whether their condence intervals overlapped.

3. Results
3.1. Handedness: dimensionality and item selection
A one-factor model and the graded response model yielded a good
t to the data (Table 2). Yet, writing and drawing were highly
redundant (r .99) and four other items (match, card, needle, broom,
and box) had standardized loadings o.90, while the rest displayed
loadings 4.90 (Table 3). Excluding these redundant and least tting
ve items increased the model t (Table 2). The remaining 10 items
were retained for further analysis.
3.2. Handedness: number of response categories
Item thresholds, distinguishing overall less frequently endorsed
response categories from exclusive right-handedness, all lay in the
negative latent trait spectrum (ranging from  1.58 to  .60 in the
DELTA parameterization) and appeared crowded (mean distance
between thresholds: .22). Inspection of ICCs (Fig. S1) revealed that
the response curves of the three middle categories (usually left, no
preference, and usually right) overlapped substantially, indicating that
these categories were insufciently separated from one another, and
that they were also insufciently separated from the extreme
categories in all retained items, with the exception of usually right
in all items save writing, and of no preference in y ap.
Three alternatives were tested with regard to the combination of
response categories: combining (a) always left and usually left, scoring
them both as  1left; (b) usually left, no preference, and usually right,
scoring them as 0no [exclusive] preference, considering usually left
and usually right preferences to indicate no exclusive preferences; and
(c) always left and usually left, scoring them as  1 left, and always
right and usually right, scoring them as 1 right. Alternative
(c) yielded the best model t, especially improving the RMSEA value
(Table 2), and further increasing the already high item factor loadings
(Table 3). However, according to ICCs (not shown), no preference was
only in y ap the most likely category in a distinct section of the
latent trait. Yet, with regard to the scoring of the LQ and the LPI that
incorporate in both cases a middle category, this solution still appeared
tenable and of heuristic value. All remaining analyses in the calibration
sample data were thus based on scoring three response categories,
combining always left and usually left, and always right and usually
right.
3.3. Handedness: latent class and taxometric analyses
3.3.1. Calibration sample
Latent class analysis initially suggested four handedness classes
in the calibration sample (Table 4). Class 1 (class size: .843)
pertained to exclusive right-handers, Class 2 (.035) to mixedhanders, Class 3 (.072) to left-handers, and Class 4 (inconsistent
right-handers; .051) to persons in-between Classes 1 and 2,
preferring less exclusively the right hand than members of Class
1. Clusters differed signicantly in LQ scores (right-handers:
M97.79, SD 6.34; inconsistent right-handers: M 54.22,
SD 15.35; mixed-handers: M 13.39, SD 28.08; left-handers:
M  81.67, SD 19.87; F(3, 2451) 19,278.92, po .001; p o.001
for all pairwise comparisons), with class explaining 95.9% of total
score variance.
MAMBAC analysis (using default settings) in the total sample
suggested that, overall, raw scores represented a categorical,
rather than a dimensional, trait, CCFI .69. In order to investigate
whether there were more than two taxa (as indicated by LCA)

224

U.S. Tran et al. / Neuropsychologia 62 (2014) 220232

across the spectrum of handedness scores, MAMBAC analyses

were applied to subsamples that contained two or more adjacent
latent classes. In order to be able to detect taxonity of small
classes, taxon base rates (derived from LCA) were supplied as input
and the number of cases to be set aside at each extreme along the
input indicator before making the rst and last cut was set to 10
(default 25).
In the 4-cluster solution, differences between right-handers
and inconsistent right-handers appeared to be not clearly taxonic,
CCFI .50. However, adding mixed-handers to the analysis sample
yielded evidence of taxonity, CCFI .63. Taxonity was also clearly
present in the subsample of left- and mixed-handers, CCFI .75.
In the 3-cluster solution, fusing inconsistent right-handers with
mixed-handers, MAMBAC analyses corroborated taxonity of all
three classes (Table 5). The 3-cluster solution was therefore kept as
the nal model. Class sizes and response probabilities are provided
in Table 5, along with LQ cutoffs. Indicator loadings ranged from
.73 (spoon) to .90 (hammer), with a mean of .79. Notably, loadings
of throwing (.83), tennis (.80), knife (.80), and hammer were
highest overall, all remaining loadings being o.80. Mean differences in LQ scores were large (right-handers: M 97.77, SD 6.33;
inconsistent right-handers/mixed-handers: M 42.54, SD 23.64;
left-handers: M  75.94, SD 26.44; F(2, 2452) 21,933.48,
p o.001; p o.001 for all pairwise comparisons), with class explaining 94.7% of total score variance. Among right-handers right-hand
preference was practically exclusive (see response probabilities in
Table 5). Inconsistent right-handers/mixed-handers had a higher
mean LQ score than mixed-handers in the 4-cluster solution and
had a relative right-hand preference across all indicators (see
response probabilities in Table 5). Left-handers had also a substantial probability of right-hand preference in writing and
showed otherwise less exclusive hand preference than righthanders.
Cross-validation with the 4-item short form in the calibration
sample corroborated a 3-cluster solution (Table 5; detailed LCA t
statistics omitted for brevity). However, based on four items,
persons previously classied as inconsistent right-handers/
mixed-handers were more often classied as right-handed and
the pattern of hand preference in the remaining class was more
indicative of mixed-handedness than mere inconsistent righthandedness, with probability of no preference in hammer now
even slightly higher than for right (Table 5). However, mixedhanders still had a relative or substantial right-hand preference in
most of the tasks. Indicator loadings were .76 (writing), .85
(throwing), .80 (knife), and .93 (hammer). Mean differences in
LQ scores were again large (right-handers: M 97.77, SD 10.04;
mixed-handers: M 5.47, SD 26.51; left-handers: M  75.23,

SD 31.73; F(2, 2452) 15,472.86, po .001; p o.001 for all pairwise comparisons), with class explaining 92.7% of total score
variance. Cutoffs in the 4-item solution were approximately
symmetric across the LQ continuum (Table 5). Moreover, whereas
in the 10-item solution right-handers endorsed no preference up to
two times and left-handers up to four times (91% of right- and 78%
of left-handers not endorsing no preference), these numbers were
reduced to one each in the 4-item solution, with 98% of right- and
99% of left-handers not endorsing no preference. In return, mixedhanders were characterized by endorsing with high probability
(87% of mixed-handers) at least once no preference. Apparently,
classication of handedness was thus more stringent in the 4-item
solution. It was therefore used for all ensuing analyses.
3.3.2. Comparison samples
One-factor models tted the data of the comparison samples
well (Table S1). 3-cluster solutions also proved the best tting
models in the two comparison samples (detailed LCA t statistics
omitted for brevity), clearly corroborated by MAMBAC analyses as
well (Table 5). Class sizes and obtained cutoffs were in good
accordance between samples and with results of the 4-item short
form in the calibration sample. Response probabilities agreed well
across the comparison samples and were also broadly comparable
Table 3
Handedness: standardized factor loadings (calibration sample).
Item

Writing
Drawing
Throwing
Tennis
Toothbrush
Knife
Hammer
Match
Eraser
Card
Needle
Fly ap
Scissors
Spoon
Broom
Box

Five response categories

Three response categories

16 items

10 items

10 items

4 items

.95
.96
.93
.94
.91
.92
.98
.88
.93
.80
.69
.93
.92
.91
.68
.78

.94

.96

.95

.94
.94
.91
.92
.98

.98
.98
.96
.96
.99

.99n

.93

.96

.93
.92
.91

.97
.97
.95

.97
.99n

Note. Items retained in item selection printed boldface. Items of the 4-item short
form are underlined.
n

Factor loadings constrained to equality.

Table 2
Fit of one-factor models (calibration sample).

Handedness: ve response categories

16 items
10 items
Handedness: combining response categories (10-item measure)
(a) always left and usually left
(b) usually left, no preference, and usually right
(c) always left and usually left, and always right and usually right
Short measures, combining response categories as in (c)
Handedness (4-item short form)
Footedness
Eyedness
Earedness
excluding headphone
Note. Constraining factor loadings of items

throwing and hammer/

2 (df)

CFI

TLI

RMSEA [90% CI]

852.15 (104)
281.41 (35)

.994
.998

.993
.997

.054 [.051,.058]
.054 [.048,.059]

246.27 (35)
234.13 (35)
162.46 (35)

.998
.998
.999

.997
.997
.999

.050 [.044,.056]
.048 [.042,.054]
.039 [.033,.045]

22.66 (3)a
4.73 (2)
19.49 (2)
73.31 (2)
10.35 (1)b

1.000
.999
1.000
.994
.999

.999
.998
.999
.982
.997

.052 [.033,.072]
.024 [.000,.052]
.060 [.037,.085]
.121 [.098,.145]
.062 [.032,.098]

heartbeat and clock to equality to attain convergence.

U.S. Tran et al. / Neuropsychologia 62 (2014) 220232

225

Table 4
Handedness: t of latent class models (calibration sample).
Model

LL

Npar

BIC(LL)

L2

df

Classication error, %

1-cluster
2-cluster
3-cluster
4-cluster
5-cluster

 9655.53
 5300.52
 4866.44
 4766.13
 4713.53

20
41
62
83
104

19,467.17
10,921.07
10,216.82
10,180.12
10,238.83

11,365.47
2655.45
1787.31
1586.67
1481.46

2434
2413
2394
2371
2350

o .001
o .001
1.000
1.000
1.000

.00
.17
1.64
1.91
2.40

Note. LL log-likelihood; Npar number of parameters; BIC(LL) Bayes information criterion based on LL; L2 likelihood ratio test statistic. Retained solution printed
boldface.

to those of the calibration sample (Table 5). Indicator loadings in

the comparison samples compared to the calibration sample
(writing: .77/.74, throwing: .77/.81, knife: .77/.79, and hammer:
.85/.87). However, mixed-handers in the comparison samples had
higher response probabilities for no preference than in the calibration sample. Using originally three instead of ve categories
apparently entailed a more frequent endorsement of no preference
among mixed-handers. Accordingly, 100% of mixed-handers in the
comparison samples endorsed at least once no preference, whereas
right- and left-handers at most once (overall 96% of right- and 90%
of left-handers not endorsing the middle category).

were also clearly corroborated by MAMBAC analyses (Table 6). Class

sizes, response probabilities, and cutoffs were all in good accordance
between samples; mixed preference classes were, however, consistently larger in comparison sample 1 than in the other two samples.
One hundred percent of respondents of mixed-preference classes
endorsed no preference at least once per scale, whereas respondents
of right-preference classes at most once and respondents of leftpreference classes at most twice (overall 74%, 83%, and 77% of rightand 53%, 78%, and 80% of left-preference respondents not endorsing
no preference in footedness, eyedness, and earedness, respectively).
3.6. Associations between lateral preferences

3.4. Footedness, eyedness, earedness: calibration sample

One-factor models, combining always left and usually left as well as
always right and usually right, tted the data well (Table 2), again
better than when utilizing all ve response categories (not shown).
However, factor loadings of some items were relatively low (chair in
footedness, Table 6; headphone in earedness, .73). Excluding chair
did not raise model t (not shown); hence, this item was kept in the
footedness scale. Excluding headphone did raise model t (Table 2); it
was therefore excluded from the earedness scale.
LCA provided evidence of 3-cluster solutions each for footedness, eyedness, and earedness (detailed LCA t statistics omitted
for brevity), consisting each of right-, left-, and mixed-preference
classes (Table 6). Notably, size of right-preference classes was
considerably smaller than in handedness, and mixed-preference
classes were large in footedness and earedness. Preference classes
in all domains were characterized by a consistent endorsement of
preferred side (no preference in the mixed-preference classes) with
highest probability across all respective indicators (see response
probabilities in Table 6). Exceptions to this occurred in ball in
footedness and gun in eyedness: mixed-footed respondents had a
strong right-foot preference in ball and mixed-eyed respondents
a right-eye preference in gun. Note, however, that class membership in footedness and eyedness depended less on items ball and
gun than on the remaining items in the respective scales (see
indicator loadings in Table 6). Chair had also a low indicator
loading in footedness; however, excluding this item did not
substantially alter the cluster solution.
Otherwise, mixed-preference classes were again clearly distinct
from right- and left-preference classes across all three domains
regarding the use of the no preference option: 100% of respondents
of mixed-preference classes endorsed no preference at least once
per scale, whereas respondents of right-preference classes at most
once and respondents of left-preference classes at most twice
(73%, 84%, and 79% of right- and 51%, 76%, and 82% of leftpreference respondents not endorsing no preference in footedness,
eyedness, and earedness, respectively).
3.5. Footedness, eyedness, earedness: comparison samples
One-factor models also tted the data in the comparison samples
well (Table S1). Again, 3-cluster solutions could be replicated and

Table 7 presents results on handedness and Table 8 on footedness, eyedness, and earedness. For each of the models, sample was
included as a main effect to control for differences in preference
proportions across samples. Furthermore, stability and replicability of main effects was checked by testing interactions of sample
with each of the signicant main effects, combining mixed- and
left-handed in analyses on footedness, eyedness, and earedness in
order to avoid numerical problems in estimation. Interaction terms
were eliminated with a stepwise backwards procedure, keeping in
nal models only terms that were signicant at p o.05. With
regard to interpretation, we focus mostly on effects that could be
estimated with highest precision and reliability (po .001) and that
were replicable across samples (i.e., were not qualied by sample
interactions and absent in some of the samples). Overall, regression models explained sizable proportions of the lateral preference
distributions (pseudo R2 values in Tables 7 and 8).
Handedness (Table 7) showed largest associations with footedness and small associations with eyedness, smallest with earedness, with associations being mostly concordant. Earedness (left
preference) was reliably associated only with left-handedness.
Notably, men had overall a slightly higher left-hand preference
than women; however, this effect was only small and not stable
across samples.
Preferences for footedness, eyedness, and earedness (Table 8)
were again mostly concordant across the various domains of
laterality. Overall, handedness was a stronger predictor of footedness than for eyedness or earedness. Footedness and handedness
were equally predictive of eyedness. In earedness, footedness was
a stronger predictor than handedness. Mixed preferences in
footedness, eyedness, and earedness were more frequent in
younger participants and left-earedness was more frequent in
older participants. Moreover, men had more often mixed and left
preferences in footedness and earedness than women, whereas
women were more often left-eyed than men.
3.7. Sidedness
Classes of lateral preference served as indicators in this analysis. Table 9 displays ts of different cluster solutions, Table 10
results of the nal 3-cluster solution. The three samples were
collapsed as individual results differed only negligibly from each

226

U.S. Tran et al. / Neuropsychologia 62 (2014) 220232

other. A 3-cluster solution tted the data best, but residual

dependencies
of
eyedness
and
earedness
(bivariate
residual 45.87 in the initial 3-cluster model) and of handedness
and footedness (residual 11.37 after allowing for eyednessearedness dependence) needed to be allowed for (all bivariate
residuals r1.26 in the nal model), indicating that associations

between these domains could not be fully explained by overall

sidedness.
Around 60% of respondents were classied as right-siders, 30%
as mixed-siders, and slightly less than 10% as left-siders. According
to indicator loadings (handedness: .57, footedness: .94, eyedness:
.28, earedness: .31) and also suggested by response probabilities

Table 5
Handedness: class sizes and response probabilities in the latent class model, CCFI Values and LQ Cutoffs.
Comparison sample 1n

Calibration sample

Class size
CCFI
LQ cutoffs
Writing
Throwing
Tennis
Toothbrush
Knife
Hammer
Eraser
Fly ap
Scissors
Spoon

Comparison sample 2n

Right

Mixed

Left

Right

Mixed

Left

Right

Mixed

Left

.849
.900n
.66
.59n
472
441n
.00/.00/.99
.01/.00/.99n
.00/.00/.99
.01/.01/.99n
.00/.01/.99
.00/.01/.98
.01/.00/.99
.01/.01/.98n
.00/.00/.99
.00/.00/.99n
.00/.02/.98
.00/.02/.97
.00/.00/.99
.00/.02/.98

.071
.021n
.76
.76n
72 to  7
41 to  40n
.17/.02/.81
.33/.06/.60n
.13/.14/.73
.22/.37/.41n
.12/.17/.70
.15/.29/.57
.14/.14/.72
.36/.25/.38n
.09/.11/.80
.30/.37/.33n
.20/.41/.39
.05/.47/.47
.08/.20/.72
.17/.23/.59

.079
.078n

.895

.028

.077

.890

.026

.084

.67

.87

.88

.63

4 41

41 to  40

o  40

4 41

41 to  40

o  40

.01/.00/.99

.13/.21/.66

.80/.01/.19

.01/.00/.99

.24/.20/.56

.74/.02/.25

.01/.02/.98

.09/.55/.36

.83/.04/.13

.01/.01/.98

.14/.50/.36

.87/.03/.11

.01/.01/.97

.13/.62/.25

.83/.03/.13

.01/.01/.98

.12/.48/.41

.83/.04/.13

.00/.01/.99

.09/.59/.33

.91/.03/.06

.00/.01/.99

.09/.58/.33

.92/.02/.06

o 7
o  40n
.72/.00/.28
.72/.00/.28n
.89/.02/.09
.92/.01/.08n
.91/.05/.04
.90/.03/.07
.91/.01/.08
.90/.00/.10n
.97/.01/.03
.97/.00/.03n
.88/.04/.08
.85/.11/.03
.83/.02/.15
.83/.05/.11

n
Utilizing the 4-item short form. Response probabilities pertain to left/no/right preferences. CCFI values pertain to tests of right- vs. mixed-handers and mixed- vs. lefthanders, respectively. LQ scores were computed with 10 (calibrations sample) and four items (calibration and comparison samples) and a scoring scheme based on three
response categories (see text). Cutoffs were derived with nonparametric ROC analyses (area under the curve [AUCs] Z .957, p o.001), maximizing sensitivity and specicity,
rounding to the nearest integer. Responses with highest probability per class printed boldface.

Table 6
Footedness, eyedness, earedness: class sizes and response probabilities in the latent class model, CCFI Values, and LQ Cutoffs.
Scale

Footedness
Class size
CCFI
LQ cutoffs
Ball (.84/.60)
Pebble (.84/.66)
Beetle (.83/.66)
Chair (.67/.51)
Eyedness
Class size
CCFI
LQ cutoffs
Telescope (.94/.79)
Bottle (.95/.82)
Keyhole (.96/.84)
Gun (.90/.70)
Earedness
Class size
CCFI
LQ cutoffs
Door (.81/.67)
Heartbeat (.90n/.76)
Clock (.90n/.81)

Calibration sample

Comparison sample 1

Comparison sample 2

Right

Mixed

Left

Right

Mixed

Left

Right

Mixed

Left

.632
.76
441
.01/.02/.98
.01/.10/.89
.01/.08/.91
.08/.08/.84

.261
.87
41to  16
.05/.19/.76
.02/.72/.27
.00/.72/.28
.10/.51/.39

.106

.566
.78
441
.01/.02/.97
.03/.05/.92
.03/.11/.86
.11/.09/.80

.347
.85
41 to  16
.04/.27/.69
.02/.70/.27
.01/.79/.21
.09/.55/.37

.086

.595
.77
4 41
.01/.02/.97
.03/.05/.92
.02/.11/.87
.11/.07/.82

.314
.76
41 to  16
.05/.26/.69
.04/.67/.30
.01/.80/.19
.09/.54/.37

.091

.630
.79
441
.01/.05/.94
.01/.06/.93
.01/.04/.95
.03/.03/.94

.141
.81
41 to  16
.05/.72/.23
.04/.83/.13
.04/.80/.16
.11/.33/.56

.229

.567
.72
441
.02/.06/.93
.02/.06/.92
.02/.06/.92
.03/.02/.95

.211
.89
41 to  16
.05/.72/.23
.03/.84/.13
.04/.77/.19
.09/.40/.50

.220

.572
.76
4 41
.01/.05/.94
.02/.06/.92
.02/.04/.94
.03/.02/.95

.201
.84
41 to  16
.08/.71/.22
.04/.84/.12
.04/.77/.19
.11/.38/.51

.227

.517
.79
426
.04/.10/.86
.03/.09/.89
.03/.03/.94

.286
.85
26 to  26
.06/.73/.20
.04/.81/.15
.08/.74/.18

.197

.471
.82
426
.08/.10/.83
.05/.09/.86
.03/.05/.92

.315
.85
26 to  26
.07/.75/.18
.04/.82/.14
.06/.80/.14

.214

.498
.85
4 26
.06/.08/.83
.04/.08/.88
.03/.04/.93

.286
.79
26 to  26
.08/.73/.18
.05/.82/.13
.07/.79/.14

.216

o  16
.68/.09/.23
.71/.18/.11
.55/.30/.15
.69/.15/.16

o  16
.85/.09/.05
.86/.08/.06
.88/.06/.06
.80/.06/.14

o  26
.74/.12/.14
.83/.09/.09
.92/.02/.06

o  16
.65/.13/.23
.73/.19/.08
.53/.31/.16
.71/.15/.15

o  16
.82/.08/.10
.86/.07/.07
.85/.07/.08
.76/.06/.18

o  26
.70/.10/.20
.80/.10/.10
.89/.05/.06

o  16
.70/.08/.22
.78/.13/.09
.57/.28/.16
.72/.13/.15

o  16
.83/.05/.11
.83/.10/.07
.87/.06/.07
.77/.05/.19

o  26
.74/.09/.17
.83/.08/.09
.88/.06/.06

Note. Numbers in parentheses pertain to factor loadings (SEM) and indicator loadings (LCA) in the calibration sample. Response probabilities pertain to left/no/right
preferences. CCFI values pertain to tests of right- vs. no preference and no vs. left-preference, respectively. LQ scores were computed with four (three in earedness) items per
scale and a scoring scheme based on three response categories (see text). Cutoffs were derived with nonparametric ROC analyses (area under the curve [AUCs] Z .944,
p o.001), maximizing sensitivity and specicity, rounding to the nearest integer. Responses with highest probability per class printed boldface.
n

Factor loadings constrained to equality.

U.S. Tran et al. / Neuropsychologia 62 (2014) 220232

(Table 10), footedness, rather than handedness, discriminated

most strongly between sidedness classes. Right-handedness was
strongly favored among both right- and mixed-siders, but also was
prominent among left-siders. Earedness overall was more concordant with sidedness than eyedness; mixed-siders were more
likely right- than mixed-eyed. Thus, whereas right- and left-siders
were consistently characterized across all four domains with
highest probability by right and left preferences, respectively,
mixed-siders were with the highest probability mixed-footed
and mixed-eared, but right-handed and right-eyed.
Sidedness was dependent on sex (multinomial logistic regression analysis, controlling for sample and possible interactions of
sample with main effects with a stepwise backwards procedure as
above; overall 2(8) 262.27, p o.001, Nagelkerke R2 .021), with
higher proportions of mixed- (OR1.28 [1.19, 1.38], p o.001) and
left-siders (OR1.34 [1.19, 1.52], po .001) among men than
women. Moreover, mixed-siders were also younger than both
right- (OR.98 [.98, .99], p o.001) and left-siders (OR .99 [.98,
.99], po .001).
Probability means (Table 10) show that among right-handers a
substantial proportion was not right-sided, 30% being mixed- and
3% left-sided. Proportions of non-left-siders among left-handers
(20% mixed- and 12% right-sided) and non-mixed-siders among
mixed-handers (19% right- and 8% left-sided) were also strikingly
similar, around 30% each. Within the three handedness classes
consistency with sidedness was thus similar, despite vastly different handedness rates. A majority (4 50%) of left-eyed and lefteared respondents was right-sided, and 24% and 20% of right-eyed
and right-eared respondents, respectively, were mixed-sided.
Overall, inconsistencies in any pair of lateral preferences affected
69.2% of respondents, most frequent in handear preferences (47.8%),
and least frequent in handfoot preferences (32.9%). Inconsistencies
were more frequent (logistic regression analysis, but otherwise as
above; overall 2(6)3,264.29, po.001, Nagelkerke R2 .273) in
younger respondents (OR.99 [.99, .99], po.001), and, compared
to right-siders, in mixed-siders (OR32.04 [26.29, 39.05], po.001)
and left-siders (OR2.18 [1.90, 2.49], po.001). Crossed preferences
affected in total 32.9% of respondents, most frequent in handeye
preferences (19.1%), and least frequent in handfoot preferences
(4.6%). Excluding mixed-siders, crossed preferences (logistic regression analysis as above; overall 2(5)228.65, po.001, Nagelkerke
R2 .029) were slightly more frequent in men than women (OR 1.09
[1.00, 1.18], p.048), and more frequent in left- than right-siders
(OR2.43 [2.15, 2.75], po.001).

227

3.8. Single-locus genetic models

Table 11 lists the observed and predicted proportions, as
derived from RS theory (Annett, 2000), of the combinations of
left and non-left preferences (combining mixed and right preferences) for all pairs of lateral preferences, collapsing all three
samples. The results were noticeably consistent with the predictions of RS theory (predicted proportions deviated from observed
proportions mostly by o1%) and also supported Annett's notion of
a pull to concordance (i.e., combinations of lateral preferences
were more often concordant than discordant, as predicted by the
model) with regard to handfoot preferences. Due to the large
sample size, some goodness-of-t tests were nominally signicant.
However, deviations of predicted from observed proportions were
mostly small. Applying the 1/3-rule (i.e., adding one-thirds of the
predicted proportions of the discordant combinations to the
respective proportions of the two concordant combinations each;
Annett, 2000), did not sufciently improve the model t with
regard to handfoot preferences (Table 11). Apparently, the 1/3rule overcorrected the discordant combinations, thus yielding
predicted values that were too low. However, applying an exploratory 1/4-rule sufciently improved the model t. The observed
good t of the predictions of RS theory with the data depended on
combining mixed preferences with right preferences. Combining
mixed and left preferences, data t was substantially lowered
(details omitted for brevity).
In contrast, predictions of the DC model (McManus, 1985) were
not supported by the data. Excluding subjects with mixed preferences, as required by the DC model, the predicted equality of
proportions of rightleft and leftright preferences was disconrmed for every single pair of lateral preferences (handfoot: 4.7%
[4.45.2%] vs. 1.9% [1.62.2%]; handeye: 21.9% [21.122.6%] vs.
2.3% [2.02.5%]; handear: 27.0% [26.127.9%] vs. 2.7% [2.43.0%];
footeye: 20.1% [19.320.9%] vs. 4.0% [3.64.4%]; footear: 22.6%
[21.723.5%] vs. 3.6% [3.24.0%]; eyeear: 17.3% [16.518.1%] vs.
14.3% [13.615.1%]).

4. Discussion
Using three independent large samples, and applying a variety of
latent variable analyses, a number of replicable results were obtained,
as discussed point-by-point in the following.

Table 7
Handedness: multinomial regression analysis (n 15,139).
Variable

Mixed

Left

Sex (male)
Age
Mixed-footed
Left-footed
Mixed-eyed
Left-eyed
Mixed-eared
Left-eared
Model t 2(df) Nagelkerke R2

.90 [.60, 1.33]

1.00 [.99, 1.01]
9.56 [6.78, 13.49]nnn
19.32 [12.67, 29.46]nnn
2.86 [2.18, 3.75]nnn
2.12 [1.56, 2.89]nnn
1.40 [1.07, 1.84]n
1.26 [.90, 1.75]
3574.18 (22), p o.001 .392

1.36 [1.06, 1.74]na

1.00 [1.00, 1.01]
3.02 [2.48, 3.67]nnn
42.59 [35.20, 51.53]nnn
1.87 [1.51, 2.31]nnn
4.48 [3.77, 5.32]nnn
1.33 [1.10, 1.61]nn
1.74 [1.45, 2.09]nnn

Note. Using right-handed as common comparator for outcome and right preference as common
comparator for predictors of lateral preference. Effect estimates with highest precision and reliability
(po .001) are printed boldface. Main effects of, and interactions with, sample were controlled in the
analyses (not shown).
a
With regard to sample interactions, the effect was absent in the calibration sample and
comparison sample 1 (OR .93, [.77, 1.13]).

228

U.S. Tran et al. / Neuropsychologia 62 (2014) 220232

increase the quality of assessment.

We strongly encourage the rigorous psychometric testing of lateral
preference scales and the use of psychometrically validated self-report
questionnaires in applied research. Rened lateral preference scales
developed here may be utilized as is in future research. However,
results indicated that there is potential for further improvement: e.g., a
bimanual task (hammer) proved to be the most informative indicator
for handedness. Bimanual tasks have already been assumed before to
confer a higher reliability to classication (Papadatou-Pastou et al.,
2008). However, it may require tasks that need accuracy and which

4.1. Assessment
Results corroborated that widely-used self-report measures of
lateral preferences contain items that distort assessment and
therefore need to be excluded (Dragovic, 2004; Dragovic &
Hammond, 2007; Milenkovic & Dragovic, 2013; Veale, 2013).
Furthermore, SEM and IRT analyses suggested that three response
categories as in the LPI, consisting of right, left, and no
preference, are sufcient for the assessment of lateral preferences.
Further differentiating within right and left options did not

Table 8
Footedness, eyedness, earedness: multinomial regression analyses (n 15,139 each).
Variable

Footedness

Eyedness

Mixed

Left
nnn

Earedness

Mixed
nnn

Left
n

Sex (male)
Age
Mixed-handed

1.25 [1.15, 1.35]

.99 [.99, .99]nnn
9.74 [6.89, 13.77]nnn

1.43 [1.23, 1.65]

.91 [.83, .99]
1.00 [.99, 1.00]
.99 [.99, 1.00]nnn
nnn
19.77 [12.97, 30.14]
2.82 [2.14, 3.72]nnn

Left-handed

3.04 [2.50, 3.70]nnn

43.08 [35.60,
52.13]nnn

.82 [.75, .89]

.99 [.99, 1.00]nn
2.09 [1.53,
2.85]nnn
4.49 [3.78,
5.33]nnn
1.22 [1.10, 1.35]nnn
2.44 [2.07,
2.89]nnn

1.87 [1.51, 2.31]nnn

2.65 [2.40, 2.92]nnn
1.81 [1.47, 2.25]nnn

Mixed-footed
Left-footed
Mixed-eyed
Left-eyed

2.66 [2.41, 2.93]nnn

1.23 [1.11, 1.36]nnn

1.77 [1.42, 2.19]nnn

2.46 [2.08, 2.90]nnn

Mixed-eared
Left-eared

3.98 [3.54, 4.47]nnna

1.56 [1.40, 1.74]nnn

2.64 [2.05, 3.41]nnnb

3.50 [2.94, 4.17]nnn

Model t 2(df) Nagelkerke

R2

5373.32 (22), p o.001

.363

Mixed
nnn

Left
nnn

1.21 [1.11, 1.31]

.99 [.98, .99]nnn
1.37 [1.04, 1.81]n

1.19 [1.08, 1.30]nnn

1.01 [1.00, 1.01]nnn
1.27 [.91, 1.77]

1.40 [1.16, 1.69]nnn

1.75 [1.46, 2.11]nnn

3.65 [3.34, 3.98]nnn

2.01 [1.66, 2.43]nnn

1.56 [1.40, 1.74]nnn

3.49 [2.93,
4.15]nnn
1.34 [1.17, 1.54]nnn
2.57 [2.32,
2.84]nnn

3.30 [2.98, 3.65]nnn

1.17 [1.05, 1.31]nn
3.51 [2.99, 4.12]nnn
1.35 [1.18, 1.55]nnn

.99 [.82, 1.19]c

2.56 [2.31,
2.84]nnn

3230.77 (22), p o .001

.224

3285.11 (20), p o .001

.224

Note. Using right preference as common comparator for outcomes and predictors of lateral preference. Effect estimates with highest precision and reliability (p o .001) that
were also replicable across samples are printed boldface. Main effects of, and interactions with, sample were controlled in the analyses (not shown). With regard to sample
interactions, main effects were:
a
b
c

Smaller in the calibration sample and comparison sample 2 (OR 3.47, 95% CI [3.05, 3.94])
Smaller in the calibration sample and comparison sample 2 (OR 1.41 [1.08, 1.83])
Present in the calibration sample and comparison sample 1 (OR 1.29 [1.12, 1.46]).

Table 9
Sidedness: t of latent class models (collapsing all three samples).
Model

LL

Npar

BIC(LL)

L2

df

Classication error, %

1-cluster
2-cluster
3-cluster
residual eyeear dependence
residual handfoot dependence
4-cluster

 49,339.25
 46,876.99
 45,375.54
 45,208.60
 45,133.42
 45,200.11

8
17
26
30
34
35

98,755.52
93,917.64
91,001.39
90,706.02
90,594.18
90,737.17

8602.01
3677.49
674.59
340.71
190.36
323.72

72
63
54
50
46
45

o .001
o .001
o .001
o .001
o .001
o .001

.00
8.46
13.04
9.00
3.51
21.74

Note. LL log-likelihood; Npar number of parameters; BIC(LL) Bayes information criterion based on LL; L2 likelihood ratio test statistic. Retained solution printed
boldface.

Table 10
Sidedness: class sizes, response probabilities, and probability means in the latent class model (collapsing all three samples).

Class size
Handedness
Footedness
Eyedness
Earedness

Response probabilities of preferences within sidedness classes

Probability means of sidedness within preference classes

Right

Mixed

Left

Right

Mixed

Left

.610
.01/.00/.98
.01/.00/.99
.20/.12/.68
.18/.19/.63

.303
.05/.05/.89
.00/.94/.06
.18/.35/.46
.15/.52/.33

.088
.62/.05/.33
.89/.06/.05
.59/.15/.25
.53/.24/.23

.03/.30/.67
.01/.03/.97
.04/.24/.72
.04/.20/.76

.08/.73/.19
.00/.98/.02
.07/.55/.38
.07/.53/.40

.68/.20/.12
.91/.00/.09
.23/.24/.53
.23/.23/.54

Note. Response probabilities pertain to left/no/right preferences within sidedness classes, probability means to proportions of left/mixed/right sidedness within preference
classes.

U.S. Tran et al. / Neuropsychologia 62 (2014) 220232

may entail negative or hurtful consequences to the performing subject

if wrongly executed. Hammer was previously reported to best
distinguish left- from right-handers (see Annett (2002), p. 3537),
correlating more strongly with hand preference for other tasks than
did writing. This ought to be followed up in future research.
Footedness, eyedness, and earedness were only investigated
with four indicators each. Even though they were found to
constitute reliable and useful scales here, and showed also favorable psychometric properties in other studies (Kalaycolu et al.,
2008; Mikheev et al., 2002; Reiss, 1998; Schneiders et al., 2010;
Suar et al., 2007), self-report inventories other than the LPI and
larger item banks need to be investigated in the future. Furthermore, even though chair refers to an unskilled (balance) activity
and had only low discriminatory power with regard to footedness
classication, it still tted with the skilled activities on a common
scale (cf. Chapman et al. (1987), Kang and Harris (2000), Mikheev
et al. (2002), Schneiders et al. (2010); but see Kalaycolu et al.
(2008)). Previous analyses lacked an adequate statistical approach
to take the skewedness of response distributions into account
which may have introduced spurious results with regard to the
assessment of dimensionality. The current study did not provide
any indication that self-reported handedness is multidimensional
(cf. Healey et al. (1986), Kang and Harris, 2000, Mikheev et al.
2002, Nicholls et al. (2013) and Steenhuis and Bryden (1989)).
With regard to footedness (Kalaycolu et al., 2008; Kang & Harris,
2000; Mikheev et al., 2002; Schneiders et al., 2010), more research
is still needed. Larger item banks need to be investigated and
adequate statistical methods need to be utilized in future psychometric studies to further examine the dimensionality of selfreported lateral preferences.
According to our data and in contrast to footedness, headphone may truly not t on a common earedness scale. With the
widespread use of portable audio players in the last decades,
headphones are likely expected to be used in pairs which may
introduce ambiguity in responding to this item. Earedness is still
the least investigated of all lateral preferences; more research is
needed especially with regard to its assessment.
4.2. Self-reported lateral preferences are discrete and trichotomous
Handedness, footedness, eyedness, and earedness, as assessed
in self-report with items of the EHI and LPI, were found to be
taxonic and discrete, containing three taxa each: right, mixed, and
left. Thus, lateral preferences seem primarily a matter of differences in kind, i.e., qualitative, similar across the various preference
domains, but not of degree (cf. Prichard, Propper, and Christman
(2013)). Even though this does not render a dimensional conceptualization of lateral preferences inappropriate, it highlights
the existence of distinct categories, whose differences are more
than merely dimensional. This study provides prevalence estimates of the three taxa in each domain and empirically derived
cutoffs that may be used in future research.

229

Previous estimates of self-reported right-handedness (Coren,

1993; Peters et al., 2006) and of mixed- and left-handedness
(Vuoksimaa, Koskenvuo, Rose, & Kaprio, 2009) could be broadly
replicated in this study. However, it was also found that mixed
preferences need to be taken into consideration with regard to all
investigated domains of lateral preferences. Even though conned
to a small minority in handedness, mixed preferences were far
more frequent in footedness, earedness, and eyedness, where, in
turn, right-preferences were much less frequent than previously
reported, especially in footedness (cf. Bourassa et al. (1996) and
Porac and Coren (1976, 1981)). Rates of mixed foot preference
were, however, comparable to a recent study (Asai, Sugimori, &
Tanno, 2011). Previous studies often neglected mixed preferences
or did not differentiate mixed from left preferences. The present
study shows that mixed preferences were relevant in all investigated domains and could be reliably differentiated from left and
right preferences, even though mixed-handers showed some
similarity to right-handers (see Vuoksimaa et al. (2009)). We
found no evidence, however, of more than three classes (e.g., up
to eight; Annett, 2004) underlying self-reported handedness or
any of the other lateral preferences.
4.3. Interrelations of lateral preferences and associations
with sex and age
Lateral preferences were concordantly interrelated across the
four domains. Interrelations were strongest between handedness
and footedness and weaker between and with the other lateral
domains; however, earedness was more strongly inter-related
with footedness than with handedness. Considering this overall
pattern, we do not consider these results to be inuenced by a
response bias in the form of a carry-over effect, with participants
indiscriminately responding to other lateral preference items in
the same fashion they responded to handedness items before
(Bourassa et al., 1996). Furthermore, results replicated a number of
previous ndings, regarding: (1) a higher rate of left-footedness
among men (Dittmar, 2002); (2) decreasing rates of mixed
preferences among older respondents in footedness (Gabbard &
Iteya, 1996; Porac, Coren, and Duncan 1980) and eyedness
(Dellatolas et al., 1998; Porac et al. 1980); (3) an increasing rate
of left-earedness among older respondents (Porac et al., 1980). In
addition, we found that men were also more likely to be mixedfooted than women and that women were more likely to be lefteyed than men (also apparent, but nonsignicant, in the metaanalysis of Bourassa et al. (1996)).
4.4. Sidedness, genetic models, and cerebral dominance
Finally and most importantly, this study provided converging
evidence that three classes of sidedness, right, mixed, and left,
underlay lateral preferences in self-reported handedness, footedness,
eyedness, and earedness, suggesting, moreover, a far lower rate of

Table 11
Observed and predicted proportions, derived from RS theory, of combinations of left (L) and non-left (non-L) preferences (combining mixed and right preferences) for all
pairs of lateral preferences, collapsing all three samples.

Non-L/Non-L
Non-L/L
L/Non-L
L/L
2

HandFoot

1/3-rule

1/4-rule

HandEye

HandEar

FootEye

FootEar

EyeEar

88.2% (85.2%)
3.7% (6.8%)
3.2% (3.2%)
4.9% (1.8%)
1209.66***

(89.5%)
(2.3%)
(2.1%)
(6.1%)
280.09***

(88.5%)
(3.4%)
(3.1%)
(5.1%)
6.35*

74.0% (72.9%)
17.9% (19.1%)
3.2% (4.1%)
4.9% (3.9%)
80.38***

73.9% (73.8%)
18.1% (18.2%)
4.1% (4.2%)
4.0% (3.8%)
2.19

73.5% (72.6%)
17.9% (18.8%)
3.7% (4.4%)
4.8% (4.2%)
41.77***

74.1% (73.4%)
17.4% (18.0%)
3.9% (4.6%)
4.6% (4.0%)
31.09***

64.3% (64.1%)
12.9% (12.9%)
13.7% (13.9%)
9.1% (9.1%)
.60

Note. 2 chi squared goodness-of-t tests. Predicted values are provided in parentheses. To account for an observed pull to concordance in handfoot lateral preferences,
the 1/3-rule, as suggested by Annett (2000), was applied to the data (see main text); column 1/3-rule displays the respective predicted proportions. As application of the 1/
3-rule did not sufciently improve model t, an exploratory 1/4-rule was also applied to the handfoot data (column 1/4-rule).

230

U.S. Tran et al. / Neuropsychologia 62 (2014) 220232

right-sidedness in the population, around 60%, and a far higher rate

of mixed-sidedness, around 30%, than would have been expected
from handedness rates alone. Yet, sidedness was less dependent on
handedness than on footedness that proved its primary indicator.
Mixed- and left-sidedness were overall more frequent among men,
corroborating previous ndings in children and adults (Greenwood,
Greenwood, McCullagh, Beggs, & Murphy, 2007; Suar et al., 2007),
and in younger participants. Moreover, we observed that inconsistencies in preferences were more frequent in younger respondents,
mixed-siders and left-siders, and that crossed preferences were more
frequent among left- than right-siders. This is consistent with prior
evidence, showing that preferences appear less lateralized among
younger respondents and that consistency in preference across
different domains increases among older respondents (Bourassa et
al., 1996; Dellatolas et al., 1998; Dittmar, 2002; Gabbard & Iteya, 1996;
Greenwood et al. 2007; Porac, 1997; Porac et al., 1980; Suar et al.,
2007), and that crossed preferences are more frequent among leftthan right-handers (Kang & Harris, 2000; see also Elias et al. (1998),
for an overview on studies).
The obtained evidence on a primacy of footedness appears at a
rst glance suggestive of evolutionary postural control theories
that consider footedness as primary to handedness with regard to
cerebral lateralization (Day & MacNeilage, 1996; MacNeilage,
1991). Yet, it appears that specically skilled movements are
indicative of language lateralization (Elias et al., 1998), instead of
unskilled/balancing movements, as predicted by this theory. The
apparent association of footedness with earedness in our data is,
however, in favor of theories positing a dependence of footedness,
and of ensuing motoric lateralization, on the asymmetric prenatal
development of the vestibular organ, set in motion by the fetus inutero position and maternal walking patterns (Previc, 1991). There
is also neuroscientic evidence linking the vestibular system with
handedness (Dieterich et al. 2003).
Probing predictions of single-locus genetic models, our data were
consistent with RS theory (Annett, 2002), but not with the DC model
(McManus, 1985). Even though lateral preferences are in all likelihood
governed by multilocus, rather than single-locus processes (McManus
et al., 2013; Reiss, 1999; Warren et al., 2006), there is some evidence
that differences between these models statistically is only small and
may not easily be detected with classic genetic research designs
(McManus et al., 2013). We found that the bivariate distributions of
lateral preferences could be approximated with reasonable accuracy
with RS theory, thus suggesting that lateral preferences may at least be
described as if they obeyed a single-locus genetic model involving
chance and an RS-allele. Further, RS theory appears also informative
with regard to mixed lateral preferences and mixed-sidedness: a
middle class is implied in RS theory by genotype RS . It is
tempting to speculate that the relatively large class of mixed-siders
observed in the present study in majority could possess the hypothetical RS  genotype. Footedness is likely the best predictor of the RS
genotype on a behavioral level, as it was the best predictor of
sidedness.
In contrast, self-reported handedness appears a likely biased
indicator of sidedness, and specically of mixed-sidedness. In our
study, proportions of mixed preferences in footedness, eyedness,
and earedness were around 1435%. Mixed-handedness was much
less prevalent, around 23%, using the 4-item short scale. With
regard to this, it should be noted, rst, that handedness is subject
to social and cultural pressures (Preti, Sisti, Rocchi, Busca, and
Vellante 2011), favoring right-handedness. This affects not only
left-handers (McManus, Moore, Freegard, & Rawles, 2010), but
likely also mixed-handers. Second, various manual tasks in hand
preference inventories are performed on a regular basis in everyday life. Mixed-handers may develop a preference for one hand
over the other, out of necessity or convenience that is reinforced
by regular practice. In a mostly right-sided world (Coren &

Halpern, 1991), the preferred hand is likely the right hand again.
There is also evidence of a higher accident proneness among
mixed-handers compared to right-handers (Hicks, Pass, Freeman,
Bautista, & Johnson, 1993), that generalizes to mixed preferences
in footedness, eyedness or earedness only to a lesser extent
(Mandal, Sabharwal, Misra, Suman, & Suar, 2012). Continuous
use of one hand could allow mixed-handers to adapt better to a
right-sided world and could also benet their motor coordination,
which may be diminished among mixed-handers (Hicks, Inman,
Deharo, & Hicks, 1999). Both effects likely diminish the accident
proneness of mixed-handers, which may be another driving factor
for an acquired hand preference among mixed-handers. Third,
regularity of preference of one hand over the other varied in our
data with regard to examined tasks: the proportion of mixedhanders was larger (around 7%) when including tasks with an
overall higher propensity of equal hand preference for classication, like using a y ap.
From this perspective, self-reported handedness may be an
unreliable indicator of cerebral dominance, specically with
regard to the diffuse cerebral dominance that is implied by mixed
preferences (Rodriguez & Waldenstrm, 2008), and the underlying
biological and genetic processes. Instead, footedness may need to
be investigated more closely. There is indeed evidence suggesting
an association between non-right-footedness and schizophrenia
(Schiffman et al. 2005). Results with regard to schizotypy and
footedness are currently inconclusive (Kelley & Coursey, 1992;
Nicholls, Orr, & Lindell, 2005). Future research may benet from
using psychometrically validated instruments (Rodriguez et al.,
2010), but also from simultaneously assessing footedness besides
handedness. This may help in disentangling and dening developmental trajectories of neurobehavioral pathology, but may also
clarify the status of the other domains of laterality and of
sidedness as a risk factor.
Even though consistent with our data, our results suggest that RS
theory needs to be supplemented with further genetic, prenatal,
developmental, and environmental factors in order to arrive at a valid
overall model of lateral preferences. First, inconsistent and crossed
preferences were frequently observed in our data in handear and
handeye pairings, requiring the modeling of residual interdependencies of handedness and footedness, and of eyedness and earedness
over and above sidedness. In terms of RS theory, there was a pull to
concordance in handfoot preferences which may stem from shared
sensori-motor control systems (Annett, 2000; Annett, 2002). This is
direct evidence of a complex interrelationship of lateral preferences
overall. Second, self-reported lateral preferences were subject to sex
and age effects, being suggestive of further genetic, prenatal, and
developmental inuences (Annett, 2004). However, whatever the
denitive overall model of lateral preferences, it appears likely that
footedness, not handedness, plays a prominent role in it, given the
evidence accumulated here. We recommend the use of a trichotomy
(right vs. mixed vs. left) for the classication of lateral preferences in
future studies, which may specically benet genetic studies. The
present line of research should also be followed up, using family and
twin study designs, in order to examine predictions and the validity of
RS theory further.

5. Limitations
Limitations of our study pertain to the utilized measures of
lateral preferences. We examined only self-reported lateral preferences, using items of only two specic, but widely-used,
inventories. Results may thus not generalize to inventories other
than the EHI and the LPI or to measures of ability or performance.
Moreover, participants were not asked to actually perform the
tasks. This may have prompted higher numbers of no preference

U.S. Tran et al. / Neuropsychologia 62 (2014) 220232

responses (see Byrne et al. (2004)) which may have impacted on

results.

6. Conclusions
This study provided converging evidence on the taxonity of
self-reported handedness, footedness, eyedness, and earedness,
highlighting that it is primarily qualitative differences that differentiate these lateral preferences. Across domains, lateral preferences were overall concordant. Three classes of sidedness were
found to underlie, but could not fully explain, the interrelations of
lateral preferences. The observed interrelations were further consistent with predictions of right shift theory. In combination with a
number of additional ndings regarding associations of lateral
preferences with sex and age, results were in sum suggestive of a
complex and multifactorial model of lateral preferences. The
primacy of footedness with regard to overall sidedness needs to
be investigated in more detail in future research. More generally,
laterality research may benet from using psychometrically
validated self-report measures and utilizing a trichotomy in
classication.

Acknowledgments
Thanks are extended to Ingrid Koller, Ingo W. Nader, Jakob
Pietschnig, Anne H. E. Schild, and Elisabeth L. Zeilinger for
supporting the data collection with regard to the two comparison
samples analyzed in this study.

Appendix A. Supporting information

Supplementary data associated with this article can be found in
the online version at http://dx.doi.org/10.1016/j.neuropsychologia.
2014.07.027.
References
Annett, M. (2000). Predicting combinations of left and right asymmetries. Cortex,
36, 485505.
Annett M., Handedness and brain asymmetry: the right shift theory, 2002,
Psychology Press; Hove, UK.
Annett, M. (2004). Hand preference observed in large healthy samples: classication, norms and interpretations of increased non-right-handedness by the right
shift theory. British Journal of Psychology, 95, 339353.
Asai, T., Sugimori, E., & Tanno, Y. (2011). A psychometric approach to the relationship between handfoot preference and auditory hallucinations in the general
population: atypical cerebral lateralization may cause an abnormal sense of
agency. Psychiatry Research, 189, 220227.
Asendorpf, J. B., Conner, M., De Fruyt, F., De Houwer, J., Denissen, J. J., Fiedler, K.,
et al. (2013). Recommendations for increasing replicability in psychology.
European Journal of Personality, 27, 108119.
Beaton, A. A. (2008). Handedness assessment in studies of seasonal anisotropy.
Cortex, 44, 9798.
Beauducel, A., & Herzberg, P. Y. (2006). On the performance of maximum likelihood
versus means and variance adjusted weighted least squares estimation in CFA.
Structural Equation Modeling, 13, 186203.
Bernstein, I. H., & Teng, G. (1989). Factoring items and factoring scales are different:
spurious evidence for multidimensionality due to item categorization. Psychological Bulletin, 105, 467477.
Bourassa, D. C., McManus, I. C., & Bryden, M. P. (1996). Handedness and eyedominance: a meta-analysis of their relationship. Laterality, 1, 534.
Byrne, M., Clafferty, R. A., Cosway, R., Grant, E., Hodges, A., Lawrie, S. M., et al.
(2004). Measurement of lateral preferences and schizophrenia: results of the
Edinburgh high-risk study and methodological issues. Psychiatry Research, 125,
205217.
Carey, D. P., Smith, D. T., Martin, D., Smith, G., Skriver, J., Rutland, A., et al. (2009).
The bi-pedal ape: plasticity and asymmetry in footedness. Cortex, 45, 650661.
Chapman, J. P., Chapman, L. J., & Allen, J. J. (1987). The measurement of foot
preference. Neuropsychologia, 25, 579584.
Collins, L. M., & Lanza, S. T. (2010). Latent class and latent transition analysis for the
social, behavioral, and health sciences. New York: Wiley.

231

Corballis, M. C., & Morgan, M. J. (1978). On the biological basis of human laterality: I.
Evidence for a maturational leftright gradient. Behavioural and Brain Sciences,
2, 261269.
Coren, S. (1993). The left-hander syndrome: the causes and consequences of lefthandedness. New York: Vintage Books.
Coren, S., & Halpern, D. F. (1991). Left-handedness: a marker for decreased survival
tness. Psychological Bulletin, 109, 90106.
Crow, T. J. (2013). The XY gene hypothesis of psychosis: origins and current status.
American Journal of Medical Genetics B, 162, 800824.
Day, L. B., & MacNeilage, P. H. (1996). Postural asymmetries and language
lateralisation in humans (Homo sapiens). Journal of Comparative Psychology,
110, 8896.
Dellatolas, G., Curt, F., Dargent-Par, C., & De Agostini, M. (1998). Eye dominance in
children: a longitudinal study. Behavior Genetics, 28, 187195.
Dieterich, M., Bense, S., Lutz, S., Drzezga, A., Stephan, T., Bartenstein, P., et al. (2003).
Dominance for vestibular cortical function in the non-dominant hemisphere.
Cerebral Cortex, 13, 9941007.
Dittmar, M. (2002). Functional and postural lateral preferences in humans:
interrelations and life-span age differences. Human Biology, 74, 569585.
Dragovic, M. (2004). Towards an improved measure of the Edinburgh handedness
inventory: a one-factor congeneric measurement model using conrmatory
factor analysis. Laterality, 9, 411419.
Dragovic, M., & Hammond, G. (2005). Handedness in schizophrenia: a quantitative
review of evidence. Acta Psychiatrica Scandinavica, 111, 410419.
Dragovic, M., & Hammond, G. (2007). A classication of handedness using the
Annett Hand Preference Questionnaire. British Journal of Psychology, 98,
375387.
Dragovic, M., Milenkovic, S., & Hammond, G. (2008). The distribution of hand
preference is discrete: a taxometric examination. British Journal of Psychology,
99, 445459.
Elias, L. J., & Bryden, M. P. (1998). Footedness is a better predictor of language
lateralisation than handedness. Laterality, 3, 4151.
Elias, L. J., Bryden, M. P., & Bulman-Fleming, M. B. (1998). Footedness is a better
predictor than is handedness of emotional lateralization. Neuropsychologia, 36,
3743.
Gabbard, C., & Iteya, M. (1996). Foot laterality in children, adolescents and adults.
Laterality, 1, 199205.
Golembo-Smith, S., Schiffman, J., Kline, E., Srensen, H. J., Mortensen, E. L.,
Stapleton, L., et al. (2012). Premorbid multivariate markers of neurodevelopmental instability in the prediction of adult schizophrenia-spectrum disorder: a
high-risk prospective investigation. Schizophrenia Research, 139, 129135.
Greenwood, J. G., Greenwood, J. J. D., McCullagh, J. F., Beggs, J., & Murphy, C. A.
(2007). A survey of sidedness in Northern Irish schoolchildren: the interaction
of sex, age, and task. Laterality, 12, 118.
Haslam, N., Holland, E., & Kuppens, P. (2012). Categories versus dimensions in
personality and psychopathology: a quantitative review of taxometric research.
Psychological Medicine, 42, 903920.
Healey, J. M., Liederman, J., & Geschwind, N. (1986). Handedness is not a
unidimensional trait. Cortex, 22, 3353.
Hicks, R. A., Inman, G., Deharo, D., & Hicks, G. J. (1999). Consistency of hand use and
frequent falls. Perceptual and Motor Skills, 88, 11071110.
Hicks, R. A., Pass, K., Freeman, H., Bautista, J., & Johnson, C. (1993). Handedness and
accidents with injury. Perceptual and Motor Skills, 77, 11191122.
Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for t indexes in covariance structure
analysis: conventional criteria versus new alternatives. Structural Equation
Modeling, 6, 155.
Kalaycolu, C., Kara, C., Atbaolu, C., & Nalac, E. (2008). Aspects of foot
preference: differential relationships of skilled and unskilled foot movements
with motor asymmetry. Laterality, 13, 124142.
Kang, Y., & Harris, L. J. (2000). Handedness and footedness in Korean
college students. Brain and Cognition, 43, 268274.
Kelley, M. P. (2012). Lateral preference and schizotypy revisited: comparison of
handedness measurement and classication methods. Laterality, 17, 150168.
Kelley, M. P., & Coursey, R. D. (1992). Lateral preference and neuropsychological
correlates of schizotypy. Psychiatry Research, 41, 115135.
Kenny, D. A., Kaniskan, B., & McCoach, D. B. (2014). The performance of RMSEA in
models with small degrees of freedom. Sociological Methods & Research.
Advance online publication. http://dx.doi.org/10.1177/0049124114543236.
MacNeilage, P. F. (1991). The postural origins theory of primate neurobiological
symmetries. In: N. A. Krasnegor, D. M. Rumbaugh, R. L. Schiefelbusch, &
M. Studdert-Kennedy (Eds.), Biological and behavioural determinants of language
development (pp. 165188). Hillsdale, NJ: Erlbaum.
Mandal, M. K., Sabharwal, A., Misra, I., Suman, S., & Suar, D. (2012). Mixed-sided
individuals with neuroticism sustain more unintentional injuries in India.
International Journal of Psychology, 47, 296304.
McCarthy, M. I., Abecasis, G. R., Cardon, L. R., Goldstein, D. B., Little, J., Ioannidis, J. P.
A., et al. (2008). Genome-wide association studies for complex traits: consensus, uncertainty and challenges. Nature Reviews Genetics, 9, 356369.
McManus, I. C. (1985). Handedness, language dominance and aphasia: A genetic
model. Psychological Medicine, Monograph Suppl 8. Cambridge: Cambridge
University Press.
McManus, I. C., Davison, A., & Armour, A. L. (2013). Multilocus genetic models of
handedness closely resemble single-locus models in explaining family data and
are compatible with genome-wide association studies. Annals of the New York
Academy of Sciences, 1288, 4858.

232

U.S. Tran et al. / Neuropsychologia 62 (2014) 220232

McManus, I. C., Moore, J., Freegard, M., & Rawles, R. (2010). Science in the making:
right hand, left hand. III: estimating historical rates of left-handedness.
Laterality, 15, 186208.
McManus, I. C., Porac, C., Bryden, M. P., & Boucher, R. (1999). Eye-dominance,
writing hand, and throwing hand. Laterality, 4, 173192.
Meehl, P. E. (2004). What's in a taxon? Journal of Abnormal Psychology, 113, 3943.
Meehl, P. E., & Yonce, L. J. (1994). Taxometric analysis: I. Detecting taxonicity with
two quantitative indicators using means above and below a sliding cut
(MAMBAC procedure). Psychological Reports, 74, 10591274.
Merni, F., Di Michele, R., & Soffritti, G. (2013). Assessment of handedness using
latent class factor analysis. Laterality, 19, 405423.
Mikheev, M., Mohr, C., Afanasiev, S., Landis, T., & Thut, G. (2002). Motor control and
cerebral hemispheric specialization in highly qualied judo wrestlers. Neuropsychologia, 40, 12091219.
Milenkovic, S., & Dragovic, M. (2013). Modication of the Edinburgh handedness
inventory: a replication study. Laterality, 18, 340348.
Muthn, L. K., & Muthn, B. O. (2008). Mplus users guide. Los Angeles, CA: Muthn &
Muthn.
Nicholls, M. E., Orr, C. A., & Lindell, A. K. (2005). Magical ideation and its relation to
lateral preference. Laterality, 10, 503515.
Nicholls, M. E. R., Thomas, N. A., Loetscher, T., & Grimshaw, G. M. (2013). The
Flinders Handedness survey (FLANDERS): a brief measure of skilled hand
preference. Cortex, 49, 29142926.
Noonan, M., & Axelrod, S. (1981). Earedness (ear choice in monaural tasks): its
measurement and relationship to other lateral preferences. Journal of Auditory
Research, 21, 263277.
Oldeld, R. C. (1971). The assessment and analysis of handedness: the Edinburgh
inventory. Neuropsychologia, 9, 97113.
Papadatou-Pastou, M., Martin, M., Munafm, M. R., & Jones, G. V. (2008). Sex
differences in left-handedness: a meta-analysis of 144 studies. Psychological
Bulletin, 134, 677699.
Peters, M., Reimers, S., & Manning, J. T. (2006). Hand preference for writing and
associations with selected demographic and behavioral variables in 255,100
subjects: the BBC Internet study. Brain and Cognition, 62, 177189.
Porac, C. (1997). Eye preference patterns among left-handed adults. Laterality
, 2, 305316.
Porac, C., & Coren, S. (1976). The dominant eye. Psychological Bulletin, 83,
880897.
Porac, C., & Coren, S. (1981). Lateral preferences and human behavior. New York:
Springer.
Porac, C., Coren, S., & Duncan, P. (1980). Life-span age trends in laterality. Journal of
Gerontology, 35, 715721.
Preti, A., Sisti, D., Rocchi, M. B., Busca, M., & Vellante, M. (2011). Male-female
differences in left-handedness in Sardinia, Italy. Laterality, 16, 737752.
Previc, F. H. (1991). A general theory concerning the prenatal origins of cerebral
lateralisation in humans. Psychological Review, 98, 299334.
Prichard, E., Propper, R. E., & Christman, S. D. (2013). Degree of handedness, but not
direction, is a systematic predictor of cognitive performance. Frontiers
in Psychology, 4, 9. http://dx.doi.org/10.3389/fpsyg.2013.00009.
Reiss, M. (1998). Current investigations of earedness. Folia Phoniatrica et Logopaedica, 50, 1927.
Reiss, M. (1999). Genetic associations between lateral signs. Anthropologischer
Anzeiger, 57, 6168.
Reiss, M., Tymnik, G., Kgler, P., Kgler, W., & Reiss, G. (1999). Laterality of hand,
foot, eye, and ear in twins. Laterality, 4, 287297.
Reiss, M., & Reiss, G. (1999). Earedness and handedness: distribution in a German
sample with some family data. Cortex, 35, 403412.
Rodriguez, A., Kaakinen, M., Moilanen, I., Taanila, A., McGough, J. J., Loo, S., et al.
(2010). Mixed-handedness is linked to mental health problems in children
and adolescents. Pediatrics, 125, e340.

Rodriguez, A., & Waldenstrm, U. (2008). Fetal origins of child non-right-handedness

and mental health. Journal of Child Psychology and Psychiatry, 49, 967976.
Ruscio, J. (2012). Taxometric Programs for the R Computing Environment: Users
Manual [computer software and manual]. http://www.tcnj.edu/  ruscio/taxo
metrics.html (Retrieved from 28.06.13).
Ruscio, J., & Kaczetow, W. (2009). Differentiating categories and dimensions:
evaluating the robustness of taxometric analyses. Multivariate Behavioral
Research, 44, 259280.
Ruscio, J., Walters, G. D., Marcus, D. K., & Kaczetow, W. (2010). Comparing the
relative t of categorical and dimensional latent variable models using
consistency tests. Psychological Assessment, 22, 521.
Sadeghi, H., Allard, P., Prince, F., & Labelle, H. (2000). Symmetry and limb
dominance in able-bodied gait: a review. Gait Posture, 12, 3445.
Samejima, F. (1969). Estimation of latent ability using a response pattern of graded
scores. Psychometric Monograph No. 17. Richmond, VA: Psychometric Society.
Schiffman, J., Pestle, S., Mednick, S., Ekstrom, M., Sorensen, H., & Mednick, S. (2005).
Childhood laterality and adult schizophrenia spectrum disorders: a prospective
investigation. Schizophrenia Research, 72, 151160.
Schneiders, A. G., Sullivan, J., OMalley, K. J., Clarke, S. V., Knappstein, S. A., & Taylor,
L. J. (2010). A valid and reliable clinical determination of footedness. Physical
Medicine and Rehabilitation, 2, 835841.
Searleman, A. (1980). Subject variables and cerebral organization for language.
Cortex, 16, 239254.
Somers, M., Sommer, I., Boks, M., & Kahn, R. (2009). Hand preference and
population schizotypy: a meta-analysis. Schizophrenia Research, 108, 2532.
Sommer, I., Ramsey, N., Kahn, S., Aleman, A., & Bouma, A. (2001). Handedness,
language lateralisation and anatomical asymmetry in schizophrenia: metaanalysis. British Journal of Psychiatry, 178, 344351.
Steenhuis, R. E., & Bryden, M. P. (1989). Different dimensions of hand preference
that relate to skilled and unskilled activities. Cortex, 25, 289304.
Strauss, E. (1986). Hand, foot, eye and ear preferences and performance on a
dichotic listening test. Cortex, 22, 475482.
Suar, D., Mandal, M. K., Misra, I., & Suman, S. (2007). Lifespan trends of side bias in
India. Laterality, 12, 302320.
Szaarski, J. P., Binder, J. R., Possing, E. T., McKiernan, K. A., Ward, B. D., & Hammeke,
T. A. (2002). Language lateralization in left-handed and ambidextrous people:
fMRI data. Neurology, 59, 238244.
Szaarski, J. P., Holland, S. K., Schmithorst, V. J., & Byars, A. W. (2006). fMRI study of
language lateralization in children and adults. Human Brain Mapping, 27,
202212.
Veale, J. F. (2013). Edinburgh handedness inventoryshort form: a revised version
based on conrmatory factor analysis. Laterality, 19, 164177. http://dx.doi.org/
10.1080/1357650X.2013.783045.
Vuoksimaa, E., Koskenvuo, M., Rose, R. J., & Kaprio, J. (2009). Origins of handedness:
a nationwide study of 30,161 adults. Neuropsychologia, 47, 12941301.
Waller, N. G., & Meehl, P. E. (1998). Multivariate taxometric procedures: distinguishing
types from continua. Thousand Oaks, CA: Sage.
Warren, D. M., Stern, M., Duggirala, R., Dyer, T. D., & Almasy, L. (2006). Heritability
and linkage analysis of hand, foot, and eye preference in Mexican Americans.
Laterality, 11, 508524.
Willems, R. M., Van der Haegen, L., Fisher, S. E., & Francks, C. (2014). On the other
hand: including left-handers in cognitive neuroscience and neurogenetics.
Nature Reviews Neuroscience, 15, 193201.