Intelligence differentiation in early childhood

Full text

Page 1

Prediction of self-reported knowledge with over-claiming, fluid and crystallized
intelligence and typical intellectual engagement
Gizem Hülüra,⁎, Oliver Wilhelmb, Stefan Schipolowskia
aHumboldt-University Berlin, Germany
bUlm University, Germany
a b s t r a c ta r t i c l ei n f o
Article history:
Received 19 April 2011
Received in revised form 7 September 2011
Accepted 9 September 2011
Keywords:
Over-Claiming Technique (OCT)
Over-Claiming Questionnaire (OCQ)
Self-reported knowledge
Fluid and crystallized intelligence
Intellectual engagement
We investigated the usefulness of the Over-Claiming Questionnaire (OCQ) as a measure of cognitive abilities. In
OCQs respondents are asked to rate their familiarity with items of academic or everyday knowledge (Paulhus,
Harms, Bruce, & Lysy, 2003). Some items exist in reality (reals), and others do not (foils). We developed four
OCQs, each consisting of 40 reals and 8 foils from the domains of Science, Humanities and Civics. The OCQs
were administered in a longitudinal rotation design to 112 participants who attended the 9th school grade at
the beginning of the study. In latent variable regression analyses 53% of variation in the reals could be explained
by fluid and crystallized intelligence and over-claiming as indicated by responses to foils. Further variation in re-
sponses to reals and foils was explained by intellectual engagement. Our results show that self-reported knowl-
edge, although positively related to measures of ability, to a large extent reflects over-claiming.
© 2011 Elsevier Inc. All rights reserved.
1. Introduction
Self-reports of ability are only low to moderately correlated with
objective measures of ability. A review of 55 studies (Mabe & West,
1982) showed that the correlation between self-reports and objective
measures of abilities is low (mean r=.29) with high variability
(SD=.25). The tendency towards self-enhancement may lead to in-
accurate self-report of abilities (Paulhus, Harms, Bruce, & Lysy,
2003). This tendency can be captured with the over-claiming tech-
nique (OCT). The OCT was introduced by Raubenheimer in 1925
(Raubenheimer, 1925) and has previously been used by Phillips and
Clancy (1972) in order to assess over-claiming in a consumer ques-
tionnaire. Paulhus et al. (2003) first used the OCT as a promising
self-report instrument of abilities. Respondents are asked to rate
their familiarity with larger sets of items of academic and everyday
knowledge usually relying on 7-point-scales ranging from 0 (never
heard of it) to 6 (know it really well). A fraction of the items, fre-
quently around 20% are so called foils, i.e., they do not exist in reality.
The remaining items refer to existing terms and they are labeled reals.
The foils supposedly resemble the reals and therefore seem plausible
to non-experts. The familiarity judgments are assumed to probe the
knowledge within the tested domains. In order to express such
knowledge, valid familiarity claims and invalid claims of familiarity
with foils need to be considered. It is recommended to score OCQs
within signal detection analytic approaches (Paulhus et al., 2003).
Within this approach, each response in the OCQ falls in one of the fol-
lowing categories:
1) Hit: respondent claims to be familiar with an existing item,
2) Miss: respondent claims to be unfamiliar with an existing item,
3) False alarm: respondent claims to be familiar with a non-existing
item,
4) Correct rejection: respondent claims to be unfamiliar with a non-
existing item.
In order to assign the responses to four categories, a dichotomiza-
tion of the responses is necessary (i.e. familiar vs. unfamiliar). From
the relative distributions of the responses to these categories, hit
and false alarm rates are computed. Two indices for accuracy and re-
sponse bias are calculated within the signal detection approach. The
accuracy index, called d′ (“d prime”) is calculated as the difference
between the z-scores of the hit rate and the false alarm rate. The
index for the response bias c (“criterion location”) is calculated as
the sum of the z-scores of the hit and false alarm rates, divided by
two.
The accuracy index is positively correlated with intelligence
(r=.40; Bertsch & Pesta, 2009; r=.50–.59, Paulhus & Harms, 2004),
course grades (r=.32–.36, Paulhus & Harms, 2004), and self
(r=.25–.37, Paulhus & Harms, 2004) and peer ratings of cognitive abil-
ity(r=.32–.33,Paulhus&Harms,2004).Theresponsebiasindexiscor-
related with alternative measures of self-enhancement (r=.22–.35,
Paulhus et al., 2003), but independent from cognitive ability (r=.06,
Bertsch & Pesta, 2009; r=.17, Paulhus et al., 2003). The accuracy
index is positively correlated with typical intellectual engagement
Learning and Individual Differences 21 (2011) 742–746
⁎ Corresponding author at: Humboldt-University Berlin, Department of Psychology,
Unter den Linden 6, D-10099 Berlin, Germany. Tel.: +49 030/2093 9421.
E-mail address: gizem.hueluer@hu-berlin.de (G. Hülür).
1041-6080/$ – see front matter © 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.lindif.2011.09.006
Contents lists available at SciVerse ScienceDirect
Learning and Individual Differences
journal homepage: www.elsevier.com/locate/lindif

Page 2

(TIE) and need for cognition (r=.45 and .29, respectively, Woo, Harms,
&Kuncel,2007).Theresponsebiasindexispositivelyrelatedtosubclin-
ical narcissism (r=.17, Paulhus & Williams,2002; r=.14, Tracy, Cheng,
Robins,&Trzesniewski,2009).Over-claimingonfoilsreflectsindividual
differences in faking (Bing et al., 2011).
In the current study, we administered four sets of OCQ items in a
longitudinal rotation design to 112 participants. Over-claiming is
commonly defined by indices such as “c”, which include ratings of
both reals and foils. In this study, we used an alternative operationali-
zation of over-claiming as familiarity ratings of foils. Self-reported
knowledge was operationalized as familiarity ratings of reals. We
hypothesized that self reported-knowledge would be significantly
predicted by fluid (gf) and crystallized intelligence (gc) and over-
claiming. In latent variable regression analyses, we regressed self-
reported knowledge on gf, gc and over-claiming.
Wealsoexpectedself-reportedknowledgetoberelatedtomeasures
of TIE. TIE captures interindividual differences in engagement in intel-
lectually challenging activities, in the variety of interests, and in the in-
trinsic need to gain a deeper understanding of the world (Goff &
Ackerman, 1992, Wilhelm, Schulze, Schmiedek, & Süß, 2003). In the
currentstudy,weoperationalizedTIEmoreextensively,usingindicators
of self-motivated cognition (Schulze, Roberts, Minsky, & Wilhelm,
2006) and the Big Five facet of openness to new ideas (Costa & McCrae,
1992). We hypothesized that TIE will be related to self-reported
knowledge after controlling for over-claiming, gf and gc.
2. Method
2.1. Participants
The sample consisted of 112 participants who completed a two-
year intensive-longitudinal study with 44 testing sessions on student
achievement and working memory. Self-report measures were also
administered at each session. At the beginning of the study the partic-
ipants attended the 9th grade and had a mean age of 14.7 (SD=.72).
67.9% of the students attended a German Gymnasium (typically pre-
paring for university education), 20.5% of the students attended a
Realschule (typically preparing for vocational education), and 11.6%
of the students attended a Gesamtschule (comprehensive school, pos-
sible to earn different degrees). 64.3% of the students were female.
2.2. Measures
2.2.1. OCQ
Four newly developed sets of OCQs (item sets A to D) were admin-
isteredinalongitudinalrotationdesign.Eachitemsetwasadministered
threetimesduringthecourseofthestudyatdifferentmeasurementoc-
casions. The questionnaire at each testing session consisted of two sets
with each 48 items (40 reals and 8 foils). Participants rated their famil-
iarity on a 5-point scale (1=“Never heard of it before”, 2=“Heard of it
before, but cannot describe it exactly”, 3=“I could describe it roughly”,
4=“I could describe it relatively precisely”, 5=“I could give an
exact description”). The items covered the knowledge domains of Sci-
ence (Biology, Medicine, Physics, Chemistry, Geography, Technology),
Humanities (Literature, Art, Music, Religion, Philosophy), and Civics
(History, Law, Politics, Economy, Finance). Participants were instructed
to give a familiarity rating of 1 (“Never heard of it before”), if they only
heard aboutthis item in thepreviousadministrationsofthesame ques-
tionnaire. One of the supposed foils in the item set C existed in reality
and was therefore discarded from further analyses.
2.2.2. TIE
The TIE questionnaire (Wilhelm et al., 2003) consisted of three sub-
scales: intellectual curiosity, contemplative engagement, and reading.
In the current study only the scales reading (5 items) and intellectual
curiosity (8 items) were used. Two items belonged to both scales. The
responsealternativesrangedfrom1(“stronglydisagree”)to6(“strong-
ly agree”).
2.2.3. Self-motivated cognition
Weusedaquestionnaireofself-motivatedcognitionconsistingof30
items that belong to the five subscales of reasoning, speed, creativity,
memory and knowledge (Schulze et al., 2006). In the current study
only the subscales reasoning (11 items) and knowledge (6 items)
were used. The participants indicated their agreement with the state-
ments on a 6-point scale ranging from 1 (“strong disagreement”) to 6
(“strong agreement”).
2.2.4. Fluid and crystallized intelligence
gf is measured with three subtests for figural, verbal, and numer-
ical content. Each subtest consisted of 16 items. gc is measured with
a declarative knowledge test with 64 items, covering the content do-
mains of Science, Humanities and Civics (Wilhelm & Schipolowski,
2010). The mean accuracy in each subtest and each content domain
was computed as the proportion of correct responses.
2.2.5. Openness to ideas
Openness to ideas was assessed using a subscale of a Big Five per-
sonality questionnaire (Ostendorf & Angleitner, 2004) consisting of 6
items with response alternatives ranging from 1 (“strong disagree-
ment”) to 5 (“strong agreement”).
3. Results
3.1. Signal detection analyses
We computed the signal detection indices of d′ (indicator of accu-
racy) and c (criterion location, indicator of response bias) separately
for each item set in order to illustrate their relations to gf and gc,
and to familiarity ratings of reals and foils. The responses were dichot-
omized at the cutoff between 1 (“Never heard of it before”) and 2
(“Heard of it before, but cannot describe it exactly”). Table 1 shows
the mean values and standard deviations for signal detection indices
and familiarity ratings in each item set. The internal consistencies of
the composites ranged from .90 to .93 for reals and from .69 to .88
for foils. The retest reliabilities ranged from .69 to .84 for reals and
from .38 to .65 for foils.
The correlations of the signal detection and familiarity ratings for
reals and foils are shown in Table 2. Replicating previous results (Paulhus
& Harms, 2004), the signal detection indices d′ and c were negatively
Table 1
Means and standard deviations of familiarity ratings for reals and foils and for signal
detection indices.
Item setMean (SD)
d′
c realsfoils
A1
A3
A4
B1
B2
B5
C2
C3
C6
D4
D5
D6
1.36 (.80)
1.49 (.93)
1.55 (.88)
1.44 (.75)
1.47 (.68)
1.58 (.92)
1.30 (.83)
1.47 (.95)
1.56 (.91)
1.64 (.91)
1.68 (.86)
1.69 (89)
−.22 (.72)
−.12 (.82)
.09 (.77)
−.30 (.69)
−.16 (.65)
−.02 (.74)
−.20 (.78)
−.18 (.79)
−.04 (.75)
−.18 (.78)
−.10 (.78)
.07 (.84)
2.70 (.69)
2.73 (.60)
2.94 (.62)
2.66 (.65)
2.84 (.59)
2.95 (.65)
2.68 (.62)
2.79 (.59)
3.12 (.61)
2.74 (.63)
2.85 (.65)
2.95 (.66)
1.45 (.53)
1.40 (.49)
1.48 (.49)
1.39 (.48)
1.39 (.42)
1.43 (.51)
1.46 (.53)
1.43 (.49)
1.54 (.55)
1.37 (.55)
1.42 (.54)
1.44 (.48)
Note. Familiarityratings for reals and foils were given on a 5-point scale: 1=“Never heard
ofitbefore”,2=“Heardofitbefore,butcannotdescribeitexactly”,3=“Icoulddescribe it
roughly”, 4=“I could describe it relatively precisely”, 5=“I could give an exact
description”.
743
G. Hülür et al. / Learning and Individual Differences 21 (2011) 742–746

Page 3

correlated. The accuracy index d′ was also negatively correlated with fa-
miliarity ratings of foils. The correlations between d′ and familiarity rat-
ings for reals ranged from −.32 to .05 and were less strong than the
correlations between d′ and familiarity ratings for foils, as the ratings for
reals not exclusively reflect over-claiming. The response bias index c
was positively correlated with familiarity ratings for both reals and foils.
This result indicates that both familiarity ratings are affected by over-
claiming. The correlations between the familiarity ratings for foils and
reals were also positive.
Table 3 shows the correlations of the signal detection indices and
familiarity ratings with the gf and gc scores. gf correlated positively
with d′ and with familiarity ratings for reals. gf did not significantly
correlate with c. The correlations between gf and familiarity ratings
for foils were negative. The signal detection indices d′ and c, as well
as familiarity ratings for reals were positively correlated with gc.
The correlations between familiarity ratings for foils and gc ranged
from −.09 to .02. A significant positive correlation between familiar-
ity ratings for foils and gf and gc would implicate that participants
with higher gf and gc mistake foils for similar sounding knowledge
content. The non-significant correlations between gf and gc and fa-
miliarity ratings for foils show that the foils function as intended.
3.2. Latent variable regression models
The latent variable explaining the familiarity ratings for reals was
regressed on the latent factors reflecting gf and gc and on the latent
factor explaining the familiarity ratings for foils. The regression
model was computed separately for each of the item sets A to D. In
Fig. 1, the model is illustrated for the item set A. As each item set
was administered three times during the study, three mean familiar-
ity ratings for foils and three mean familiarity ratings for reals were
available for each item set. The familiarity ratings for foils loaded on
a latent factor reflecting over-claiming. Familiarity ratings for reals
loaded on a latent factor reflecting self-reported knowledge. The re-
siduals of the familiarity ratings for reals and foils from the same ques-
tionnaire were allowed to correlate (i.e. A1-reals with A1-foils, A3-
reals with A3-foils and so forth). The gf subtest scores (verbal, numer-
ic and figural), as well as the gc scores in the content domains of Sci-
ence, Humanities and Civics loaded on a latent factor of gf. The gc
scores additionally loaded on a nested factor of gc. The latent factor
representing over-claiming was regressed on the gf and gc factors.
The latent factor representing self-reported knowledge was regressed
on the over-claiming and gf and gc factors. The fit of the models was
evaluated with the fit indices of CFI (Comparative Fit Index; Bentler,
1990), RMSEA (Root Mean Square Error of Approximation; Steiger,
1990) and SRMR (Standardized Root Mean Square Residual; Fan &
Sivo, 2007). Table 4 shows the model fit, standardized regression
weights, and proportion of explained variance in self-reported
knowledge and over-claiming in each of the item sets. The model
fitted the data well in all item sets. We assessed the reliability of the
latent factors using the reliability coefficient ω (McDonald, 1999).
The reliability of self-reported knowledge ranged from .91 to .92
across the four item sets. The over-claiming factors had reliabilities
ranging from .71 to .79. A large amount of variance (44 to 57%) of
self-reported knowledge could be explained by accounting for over-
claiming and gf and gc. gf and gc explained only a small proportion
of variance (3 to 9%) of over-claiming.
Nextweappliedthelatentvariableregressionmodeltoallitemsets.
Thelatentfactors reflectingself-reported knowledge andover-claiming
hadeachfourindicators:Themeansofthreeindicatorsbelongingtothe
same item set were used as indicators. The model fitted the data very
well (χ2=82.3, df=65, p=.07; CFI=.99, RMSEA=.05, SRMR=.04).
53% of the variance in self-reported knowledge could be explained by
over-claiming, gf and gc, with the regression weights of .66 (pb.001),
.41 (pb.001) and .13 (p=.107), respectively. Only 5% of the variation
in over-claiming could be explained by gf and gc, with the regression
weights of −.21 (p=.055) and .10 (p=.378), respectively.
3.3. Relations to TIE
To explain further variation in self-reported knowledge and over-
claiming, the disturbance terms of the latent factors were regressed
on a latent variable of TIE. The disturbance terms reflect the systematic
variationinlatentvariablesofself-reportedknowledgeandover-claim-
ingthatcouldnotbeexplainedbygfandgc.Thereadingandintellectual
curiosity subscales of the TIE questionnaire, reasoning and knowledge
subscales of the questionnaire of self-motivated cognition, and the
NEO-PI-R openness to ideas scale loaded on a latent factor reflecting
TIE. Subscales from the same questionnaire were allowed to correlate.
The disturbance terms of the self-reported knowledge and over-
claimingfactors wereregressedontheTIEfactor. The model isillustrat-
ed inFig. 2. Themodelfit wasacceptable(χ2=284.2, df=186,p=.00;
CFI=.95, RMSEA=.07, SRMR=.06). The reliabilities of the latent fac-
tors were assessed using the reliability coefficient ω (McDonald,
1999). The self-reported knowledge factor had a reliability of .98 and
the over-claiming factor had a reliability of .93. 14% additional variance
in self-reported knowledge and 10% of additional variance in over-
claiming could be explained by TIE with regression weights of .37 and
Table 2
Correlation of the signal detection indices of d′ (indicating accuracy) and c (criterion
location, indicating response bias) with familiarity ratings for foils and reals in each
item set.
Item set Correlations
d′ with cd′ with
reals
d′ with
foils
c with
reals
c with
foils
reals with
foils
A1
A3
A4
B1
B2
B5
C2
C3
C6
D4
D5
D6
−.51⁎⁎
−.63⁎⁎
−.52⁎⁎
−.57⁎⁎
−.52⁎⁎
−.59⁎⁎
−.68⁎⁎
−.71⁎⁎
−.72⁎⁎
−.64⁎⁎
−.57⁎⁎
−.63⁎⁎
.02
−.58⁎⁎
−.75⁎⁎
−.67⁎⁎
−.65⁎⁎
−.61⁎⁎
−.73⁎⁎
−.74⁎⁎
−.74⁎⁎
−.78⁎⁎
−.74⁎⁎
−.68⁎⁎
−.79⁎⁎
.68⁎⁎
.66⁎⁎
.61⁎⁎
.70⁎⁎
.70⁎⁎
.61⁎⁎
.78⁎⁎
.65⁎⁎
.62⁎⁎
.66⁎⁎
.64⁎⁎
.68⁎⁎
.82⁎⁎
.83⁎⁎
.79⁎⁎
.80⁎⁎
.82⁎⁎
.79⁎⁎
.84⁎⁎
.78⁎⁎
.77⁎⁎
.81⁎⁎
.82⁎⁎
.85⁎⁎
.59⁎⁎
.56⁎⁎
.51⁎⁎
.57⁎⁎
.56⁎⁎
.38⁎⁎
.66⁎⁎
.51⁎⁎
.53⁎⁎
.46⁎⁎
.48⁎⁎
.54⁎⁎
−.10
−.05
−.05
−.04
.05
−.32⁎⁎
−.24⁎
−.22⁎
−.07
−.01
−.17
Notes.
⁎⁎ pb.01.
⁎ pb.05.
Table 3
Correlation of the signal detection indices of d′ (indicating accuracy) and c (criterion
location, indicating response bias) and familiarity ratings for foils and reals with fluid
and crystallized intelligence scores.
Item
set
Correlations
d′ with
gf
d′ with
gc
c with
gf
c with
gc
reals
with gf
reals
with gc
foils
with gf
foils
with gc
A1
A3
A4
B1
B2
B5
C2
C3
C6
D4
D5
D6
.30⁎⁎
.38⁎⁎
.31⁎⁎
.28⁎⁎
.33⁎⁎
.33⁎⁎
.15
.28⁎⁎
.17
.25⁎⁎
.35⁎⁎
.22*
.28⁎⁎
.27⁎⁎
.14
.19⁎
.33⁎⁎
.16
.14
.18
.07
.28⁎⁎
.18
.18
.09
.01
.10
.08
.12
.05
.17
.05
.17
.06
.06
.21⁎
.07
.10
.15
.10
.15
.16
.22⁎
.18
.24⁎
.06
.20⁎
.23⁎
.18
.17
.16
.18
.24⁎
.25⁎⁎
.20⁎
.16
.26⁎⁎
.15
.24⁎
.30⁎⁎
.22⁎
.21⁎
.17
.24⁎
.29⁎⁎
.27⁎⁎
.29⁎⁎
.23⁎
.24⁎
.21⁎
.30⁎⁎
.27⁎⁎
−.13
−.17
−.12
−.12
−.12
−.18
−.09
−.15
−.02
−.06
−.18
−.03
−.05
−.07
−.08
.02
−.04
.00
.01
−.06
.00
−.09
.01
.01
Notes.
⁎⁎ pb.01.
⁎ pb.05.
744
G. Hülür et al. / Learning and Individual Differences 21 (2011) 742–746

Page 4

.32, respectively (pb.01). TIE did not correlate significantly with gf
(r=.09) and gc (r=.17).
4. Discussion
In the current study, we investigated covariates of self-reported
knowledge. We developed four OCQ item sets that were administered
to 112 participants in a longitudinal rotation design. We operationa-
lized over-claiming as familiarity ratings of foils. Self-reported knowl-
edge was operationalized as familiarity ratings of reals. We showed
that 53% of variation in self-reported knowledge can be accounted
for by gf and gc and over-claiming. Further 14% variation could be
explained by a broad operationalization of TIE. Only 5% of the varia-
tion in over-claiming could be explained by gf and gc. Additional
10% of variation could be explained by TIE.
OCQ measures are recommended to be scored within signal detec-
tion analytic approaches (Paulhus et al., 2003). In order to compute
signal detection indices, participants' responses on a continuous
scale are typically dichotomized which leads to an information loss.
In signal detection analyses usually two indices are computed based
on the hit rate and the false alarm rate: accuracy and response bias.
However, the two indices are not stochastically independent vari-
ables, as the same information underlies the computation of both in-
dices (Wixted & Stretch, 2004). Therefore, in the current study, mean
familiarity ratings for foils and reals were used as indicators of over-
claiming and self-reported knowledge, respectively. Nevertheless,
the signal detection method can be a useful tool to further explore
unwanted contributions of over-claiming to self-reported knowledge.
In addition to measures of gf and gc and over-claiming, TIE was
incrementally related to self-reported knowledge. In other words,
participants who were more intellectually engaged rated their knowl-
edge with existing items higher than participants with comparable
intelligence and over-claiming levels. The Big-Five openness/intellect
scale is significantly related to confidence (Pallier et al., 2002;
Schaefer, Williams, Goodie, & Campbell, 2004). Therefore, partici-
pants who are more intellectually interested and engaged, might
also show greater confidence in their knowledge and rate their
knowledge accordingly in the OCQ.
In the latent regression analyses, only a small amount of variation
(5%) in over-claiming could be explained by gf and gc. TIE explained
more variation (10%) in over-claiming. Measures of intellectual en-
gagement possibly also reflect to some extent the tendency to present
oneself as knowledgeable and intellectual. As the non-significant re-
gression weights of gf and gc show, this tendency seems to be unre-
lated to actual knowledge.
Paulhus and Harms (2004) recommend the use of familiarity ac-
curacy in OCQ as a measure of cognitive ability in some contexts. In
our study, the correlations of the accuracy index with gc were small
in all item sets (from .07 to .33). The correlations of the accuracy
index with gf were also only weak to moderate (.15 to .38). In our
latent variable regression analyses, the association of self-reported
knowledge with gc was not significant. The mechanisms underlying
familiarity judgments as well as the constructs measured with this
technique are not well understood. Although we were able to explain
a significant amount of variance in familiarity ratings for reals, a non-
trivial amount of variance could not be explained through measures
of ability and interests. As supposed measures of knowledge, the fa-
miliarity ratings for reals, as well as the accuracy index, should be
more closely related to gc than to gf. This was however not the case
in the current study. Therefore, we do not necessarily view familiarity
accuracy in the OCQ as a measure of ability. Further research is need-
ed in order to understand the constructs assessed with the OCQ.
The current study furthers the understanding of self-reports of
knowledge. 53% of the variation in self-reported knowledge is
explained by over-claiming and gf and gc. Further 14% variation is
explained by TIE. A limitation of the current study is the low familiar-
ity ratings of foils (see Table 1). Therefore, the relations between self-
reported knowledge, gf, gc and TIE require a replication with a differ-
ent item set. A further extension to the current study could be a direct
assessment of self-reported knowledge by asking participants to rate
their knowledge in a specific domain (Ackerman, Beier, & Bowen,
2002). These self-reports could then be linked to familiarity ratings
Table 4
Model fit and parameters of the latent variable regression models.
Item set Model fit CFIRMSEASRMR Regression weights for SRKRegression weights OC
χ2/df (p) OCgf gcgf gcR2SRKR2OC
A
B
C
D
30.7/43 (.92)
39.9/43 (.60)
42.7/43 (.49)
48.8/43 (.25)
1.00
1.00
1.00
.99
.00
.00
.00
.04
.04
.04
.06
.05
.68⁎⁎
.70⁎⁎
.81⁎⁎
.56⁎⁎
.38⁎⁎
.45⁎⁎
.18⁎⁎
.38⁎⁎
.12
.10
.08
.16
−.25⁎
−.24⁎⁎
−.08
−.14
.05
.18
.04
.08
.50
.57
.55
.44
.06
.09
.04
.03
Notes. SRK = self-reported knowledge, OC = over-claiming, gf = fluid intelligence, gc = crystallized intelligence.
⁎⁎ pb.001.
⁎ pb.05
Fig. 1. Illustration of the latent variable regression model for the item set A.
745
G. Hülür et al. / Learning and Individual Differences 21 (2011) 742–746

Page 5

of knowledge items, over-claiming, gf and gc, and measures of intel-
lectual engagement.
Acknowledgments
This research was supported by the Institute for Educational Pro-
gress, Humboldt University Berlin and by a grant from the Deutsche
Forschungsgemeinschaft (WI 2667/7-1) to Oliver Wilhelm and Alex-
ander Robitzsch. Gizem Hülür was supported by a predoctoral fellow-
ship of the International Max Planck Research School “The Life
Course: Evolutionary and Ontogenetic Dynamics (LIFE)”.
References
Ackerman, P. L., Beier, M. E., & Bowen, K. R. (2002). What we really know about our
abilities and our knowledge. Personality and Individual Differences, 33, 587–605.
Bentler, P. M. (1990). Comparative fit indexes in structural models. Psychological Bulle-
tin, 107, 238–246.
Bertsch, S., & Pesta, B. (2009). The Wonderlic Personnel Test and elementary cognitive
tasks as predictors of religious sectarianism, scriptural acceptance and religious
questioning. Intelligence, 37, 231–237.
Bing, M. N., Kluemper, D., Davison, H. K., Taylor, S., & Novicevic, M. (2011). Over-claim-
ing as a measure of faking. Organizational Behavior and Human Decision Processes,
116, 148–162.
Costa, P. T., & McCrae, R. R. (1992). NEO PI-R. Professional manual. Odessa, FL: Psycho-
logical Assessment Resources, Inc.
Fan, X., & Sivo, S. (2007). Sensitivity of fit indexes to misspecified structural or mea-
surement model components: Rationale of two-index strategy revisited. Structural
Equation Modeling, 12, 343–367.
Goff, M., & Ackerman, P. J. (1992). Personality–intelligence relations: Assessment of
typical intellectual engagement. Journal of Educational Psychology, 84, 537–552.
Mabe, P. A., & West, S. G. (1982). Validity of self-evaluation of ability: A review and
meta-analysis. The Journal of Applied Psychology, 67, 280–296.
McDonald, R. P. (1999). Test theory: A unified treatment. Mahwah, NJ: Erlbaum.
Ostendorf, F., & Angleitner, A. (2004). NEO-Persönlichkeitsinventar nach Costa und
McCrae: NEO-PIR [NEO personality inventory according to Costa and McCrae: NEO-
PIR]. Göttingen: Hogrefe.
Pallier, G., Wilkinson, R., Danthir, V., Kleitman, S., Knezevic, G., Stankov, L., et al. (2002).
The role of individual differences in the accuracy of confidence judgments. The
Journal of General Psychology, 129, 257–299.
Paulhus, D. L., Harms, P. D., Bruce, M. N., & Lysy, D. C. (2003). The over-claiming tech-
nique: Measuring bias independent of accuracy. Journal of Personality and Social
Psychology, 84, 681–693.
Paulhus, D. L., & Harms, P. D. (2004). Measuring cognitive ability with the over-
claiming technique. Intelligence, 32, 297–314.
Paulhus, D. L., & Williams, K. (2002). The Dark Triad of personality: Narcissism, Machi-
avellianism, and psychopathy. Journal of Research in Personality, 36, 556–568.
Phillips, D. L., & Clancy, K. J. (1972). Some effects of “social desirability” in survey stud-
ies. The American Journal of Sociology, 77, 921–940.
Raubenheimer, A. S. (1925). An experimental study of some behavioral traits of the po-
tentially delinquent boy. Psychological Monographs, 159, 1–107.
Schaefer, P. S., Williams, C. C., Goodie, A. S., & Campbell, W. K. (2004). Overconfidence
and the Big Five. Journal of Research in Personality, 38, 473–480.
Schulze, R., Roberts, R. D., Minsky, J., & Wilhelm, O. (2006). The concept of self-motivat-
ed cognition and its measurement. Paper presented at the 2006 Annual Meeting of
the American Educational Research Association San Francisco, CA, USA.
Steiger, J. H. (1990). Structural model evaluation and modification: An interval estima-
tion approach. Multivariate Behavioural Research, 25, 173–180.
Tracy, J. L., Cheng, J., Robins, R. W., & Trzesniewski, K. (2009). Authentic and hubristic
pride: The affective core of self-esteem and narcissism. Self and Identity, 8,
196–213.
Wilhelm, O., & Schipolowski, S. (2010). Intelligence assessment in educational psychol-
ogy. In G. L. Huber (Ed.), Encyclopedia Pedagogy Online. Area educational psychology.
Weinheim: Juventa.
Wilhelm, O., Schulze, R., Schmiedek, F., & Süß, H. -M. (2003). Interindividuelle
Unterschiede im typischen intellektuellen Engagement [Interindividual differ-
ences in typical intellectual engagement]. Diagnostica, 49, 49–60.
Wixted, J. T., & Stretch, V. (2004). In defense of the signal-detection interpretation of
remember/know judgments. Psychonomic Bulletin & Review, 11, 616–641.
Woo, S. E., Harms, P. D., & Kuncel, N. R. (2007). Integrating personality and intelligence:
Typical intellectual engagement and need for cognition. Personality and Individual
Differences, 43, 1635–1639.
Fig. 2. Illustration of the latent variable regression model with covariates.
746
G. Hülür et al. / Learning and Individual Differences 21 (2011) 742–746