You are on page 1of 5

Editorial

States and Traits


https://econtent.hogrefe.com/doi/pdf/10.1027/1015-5759/a000413 - Wednesday, June 02, 2021 7:19:19 PM - IP Address:110.138.92.205

Theories, Models, and Assessment


Christian Geiser,1 Thomas Götz,2,3 Franzis Preckel,4
and Philipp Alexander Freund5
1
Department of Psychology, Utah State University, Logan, UT, USA, 2Educational Science, University of
Konstanz, Germany, 3Thurgau University of Teacher Education, 4Department of Psychology,
University of Trier, Germany, 5Department of Psychology, Leuphana University Lüneburg, Germany

The question of whether characteristics of persons (stable methodological developments and substantive findings in
personality dispositions that are often referred to as traits), the context of state and trait analysis.
characteristics of situations, or Person · Situation interac-
tions are more relevant to psychological behavior has been
debated for decades (e.g., Donnellan, Lucas, & Fleeson,
2009; Epstein, 1983; Fleeson, 2001, 2004; Funder, 1991, States and Traits in Psychology
2008; Funder, Guillaume, Kumagai, Kawamoto, & Sato,
2012; Kenrick & Funder, 1988; Mischel, 1968). In psycho- Past studies have shown that most psychological variables
logical measurement, the distinction between trait (enduring contain both state and trait components to varying degrees
or stable) components and state residual (variable or fluctu- (e.g., Deinzer et al., 1995). On the one extreme, measure-
ating) components has been of ever-increasing interest to ments of hormone levels (e.g., Kirschbaum et al., 1990)
researchers since the 1980s and 1990s when more sophisti- and mood (Eid, Schneider, & Schwenkmezger, 1999) show
cated theoretical and statistical approaches became rather low amounts of person-specific trait variance and can
available (Steyer, Ferring, & Schmitt, 1992; Steyer, Majcen, thus be considered more ‘‘state-like’’ constructs. On the
Schwenkmezger, & Buchner, 1989). Furthermore, new other extreme, individual differences in intelligence and
methodological approaches have recently been developed other ability scores have been found to be highly stable
to better conceptualize and understand psychological (‘‘trait-like’’) across time, reflecting only negligible
situations (e.g., Rauthmann, Sherman, & Funder, 2015). amounts of situation-specific variance (Danner, Hagemann,
To illustrate the increasing interest in person-situation Schankin, Hager, & Funke, 2011). However, most psycho-
research, a search in the international database PsychINFO logical variables appear to be somewhere ‘‘in between,’’
using the term ‘‘state-trait’’ (carried out in November 2016) containing both substantial trait and substantial state resid-
returned 491 hits for the period between 1970 and 1979, ual components. Therefore, it seems important to consider
1,538 hits between 1980 and 1989, 1,989 hits between both aspects as well as potential interactions between traits
1990 and 1999, 6,277 hits between 2000 and 2009, and and situations (e.g., Schmitt & Baumert, 2010).
9,044 hits between 2010 and 2016 (19,339 results total).
Moreover, the increase in publication rates related to ‘‘states
and traits’’ was stronger relative to the overall increase in
scientific publications in psychology as shown in Figure 1. Latent State-Trait Theory
Publication rates are presented as the number of ‘‘state-
trait’’ papers per 100,000 publications in the PsychINFO The high interest in psychological states and traits
database. Figure 1 illustrates that the scientific interest has resulted not only in many substantive contributions to
in the ‘‘state-trait’’ topic has strongly increased since the field, but also in the development of more and more
1970. In this special issue, our contributors present new sophisticated methods for measuring states and traits.

 2017 Hogrefe Publishing European Journal of Psychological Assessment 2017; Vol. 33(4):219–223
DOI: 10.1027/1015-5759/a000413
220 Editorial

800 variance and thus help clarify the question of whether


person or situation/interaction influences determine individ-
700 ual differences in psychological measurements to a larger
extent.
600
In a recently published updated version of LST theory
500 (so-called LST-R theory) the static person concept in the
original theory has been revised to take into account the fact
400 that persons constantly make experiences so that psycho-
logically, the person u at time t is not necessarily exactly
300 the same as the person u at time t + 1 (Steyer, Mayer,
https://econtent.hogrefe.com/doi/pdf/10.1027/1015-5759/a000413 - Wednesday, June 02, 2021 7:19:19 PM - IP Address:110.138.92.205

Geiser, & Cole, 2015). Related to this idea of a more


200
‘‘dynamic’’ person concept, recent methodological studies
100
have clarified connections between pure ‘‘variability’’ (clas-
sical LST) models and models for measuring trait changes
0 (Geiser, Keller, et al., 2015) and have discussed hybrid
1970-1979 1980-1989 1990-1999 2000-2009 2010-2016 approaches that allow researchers to study state, trait, and
trait change processes simultaneously (Bishop, Geiser, &
Figure 1. Number of papers on ‘‘state-trait’’ per 100,000 Cole, 2015; Eid & Hoffmann, 1998; Tisak & Tisak,
publications in PsycINFO databases. 2000). Furthermore, extended LST designs including both
random and fixed situations allow for an analysis of the
potential situation specificity of traits and make it possible
The development of latent state-trait (LST) theory and to study main effects of situations as well as person-situa-
models (Steyer et al., 1989, 1992) represented a significant tion interactions in more detail (Geiser, Litson, et al., 2015).
methodological advancement in the area of person-situation Other recent extensions of LST models have enabled
research in the late 1980s and early 1990s. Formulated on researchers to take into account and analyze systematic
the basis of stochastic measurement theory (Zimmerman, method-specific effects. Method effects are of interest to
1975), LST theory demonstrated how state, trait, and state psychologists in general (e.g., Podsakoff, MacKenzie, &
residual latent variables can be constructively defined based Podsakoff, 2012), because their presence can indicate a lack
on conditional expectations of observed (measured) vari- of validity of psychological measures (Campbell & Fiske,
ables and separated from random measurement error. 1959) or because different methods’ unique perspectives
‘‘Constructively’’ means that latent variables in LST theory can provide special insights (Eid & Diener, 2006). Multi-
are not simply ‘‘assumed to exist,’’ but are given explicit method extensions of LST approaches allow researchers
definitions in terms of (conditional expectations of) to examine the consistency of states and traits across differ-
observed variables. This is helpful to clarify how states ent methods (Courvoisier, Nussbeck, Eid, Geiser, & Cole,
and traits can be operationalized mathematically, to avoid 2008; Eid, 1996; Eid et al., 1999; Geiser & Lockhart,
ambiguities in the labeling and interpretation of latent vari- 2012; Koch, Schultze, Holtmann, Geiser, & Eid, in press;
ables that are common in latent variable modeling, and to Steyer et al., 1992) and our special issue features a novel
derive the properties of latent variables. contribution in this area as discussed below.
According to LST theory, latent state scores reflect
individuals’ true scores within a specific situation, that is,
averages of (hypothetical) intra-individual distributions of Contributions in This Special Issue
measured (e.g., item, test, or questionnaire) scores within
a given situation. Trait scores reflect averages of (hypothet- The present special issue features both methodological and
ical) intra-individual distributions of measured scores substantive advancements related to the analysis of states
across situations. State scores thus reflect persons in situa- and traits. On the methodological side, two areas of state-
tions, whereas trait scores reflect the persons only. State trait modeling that appear to be receiving more and more
residual latent variables are defined as the difference attention in recent years concern (1) the use of item-level
between state and trait latent variables and reflect both data (i.e., categorical observed variables such as binary or
situation effects and potential person-situation interactions ordinal response variables) for measuring categorical or
(Steyer et al., 1992). continuous latent state and trait variables and (2) the ques-
On the applied side, LST theory showed how state- and tion of how method effects can be appropriately modeled in
trait-related variance components can be estimated in statis- longitudinal studies. As part of this special issue, Crayen,
tical measurement models based on repeated measurements Eid, Lischetzke, and Vermunt (2017) present a new
of observed variables. The LST methodology uses latent continuous-time latent Markov modeling approach that uses
variables, allowing researchers to separate systematic categorical observed variables as indicators for measuring
effects of persons and situations from unsystematic effects latent classes (i.e., categorical latent variables). In this
due to random measurement error. LST models make approach, several different state and trait latent class
it possible to determine what proportion of the observed variables can be examined that represent stability and
score variance is due to trait (person-specific), occasion- variability at different levels. Crayen et al.’s new approach
specific (situation or person-situation interaction), and error represents an extension of Eid and Langeheine’s (1999)

European Journal of Psychological Assessment 2017; Vol. 33(4):219–223  2017 Hogrefe Publishing
Editorial 221

latent class LST approach to intensive longitudinal designs satisfaction across time? Substantively, their paper exami-
with many measurements for each person. The new nes state variability (short-term fluctuations) and trait
approach should be of particular interest to substantive changes (long-term growth or decline) in life satisfaction.
researchers who want to model stability and change based The assessment of family-level and dyadic data is of
on many repeated measurements of the same individuals, increasing interest to psychologists. Consequently, many
for example, through ecological momentary assessments. novel methodological contributions related to the analysis
Thielemann, Sengewald, Kappler, and Steyer (2017) of such data in both cross-sectional and longitudinal studies
also present a methodological innovation for the modeling have been presented (e.g., Ledermann & Macho, 2014).
of state-trait data with categorical observed variables. Their Loncke et al. (2017) fill an important gap in the dyadic data
new latent state item response model for longitudinal data modeling literature by presenting an LST modeling
https://econtent.hogrefe.com/doi/pdf/10.1027/1015-5759/a000413 - Wednesday, June 02, 2021 7:19:19 PM - IP Address:110.138.92.205

enables researchers to study method effects (such as approach for dyadic data. Their new approach allows
response styles) at the level of binary items relative to a researchers to model states and traits with dyadic data.
‘‘gold-standard’’ item using a latent difference score From a substantive perspective, Loncke et al. (2017) show
approach. Thielemann et al.’s approach should be useful that variability in perceived family support can mostly be
for researchers interested in using binary items to measure attributed to individual perceiver effects. However, Loncke
continuous latent state and trait variables and to investiga- et al. (2017) found that the degree to which perceiver effects
tors who want to study method effects at the item level. of support can be attributed to traits rather than states was
Item-level method effects can occur, for example, when smaller for adolescents and siblings as compared to parents.
an item set includes both positively and negatively keyed Furthermore, the relative importance of trait and state
items or when multi-method measurement designs are used components in adolescents’ perceptions changed over time
(e.g., self- and other reports). The approach is also useful whereas supportive family climate remained rather stable.
when different items are not perfectly homogenous in These findings offer new insights into developmental
measuring a single state or trait (i.e., when items capture aspects in relational processes within families and underline
different facets of a construct such as sadness versus the usefulness of state-trait models.
sleeping problems as indicators of depression).. Nett, Bieg, and Keller (2017) present an application
Researchers dealing with measurement designs in which of multiconstruct LST models to academic emotions
measurement occasions are closely spaced in time (e.g., (enjoyment, pride, anger, anxiety, and boredom) in the
multiple measurement occasions per week) often find that context of mathematics classes. Nett et al. found compara-
adjacent measurements are more strongly related to one ble proportions of trait variance and state residual compo-
another than measurements that are spaced out farther in nents for all five emotions. Interestingly, on the one hand
time. LST modeling extensions that take into account such latent trait components of emotions of different valence
additional sources of stability beyond pure trait effects have (i.e., positive vs. negative connotation) were mostly
therefore been of great interest to researchers, especially unrelated to each other. On the other hand, latent state resid-
since Cole, Martin, and Steiger’s (2005) seminal discussion ual components of emotions of different valence were neg-
of autoregressive effects in LST models. However, it has atively correlated. According to Nett et al. (2017, p. 239),
previously been unclear whether models with autoregres- this shows that ‘‘an anxious student can also be a happy
sive effects are in line with the concepts of LST-R theory student’’ with regard to psychological traits, but in a given
(Steyer et al., 2015). In this special issue, Eid, Holtmann, situation (during typical mathematics classes), a student
Santangelo, and Ebner-Priemer (2017) propose a new will feel either anxious or happy, but not experience
model with autoregressive effects and show that this model emotions of different valence simultaneously.
is fully in line with LST-R theory. Eid et al.’s (2017)
approach is innovative in that it represents the first LST
model that explicitly links situational influences to subse- Outlook
quent trait scores. This constitutes a radically different
approach to the modeling of the interplay of traits and states The analysis of state and trait components is of increasing
compared to previous LST approaches. It offers exciting interest to psychological research. We hope that this special
new possibilities for the modeling of intensive longitudinal issue will stimulate further substantive and methodological
data from a more dynamic trait perspective: In Eid et al.’s research in this area. As we have alluded to in this editorial,
(2017) modeling framework, traits can change due to previ- LST theory provides suitable statistical models for studying
ous situational influences such as momentary hassles or the nature of psychological constructs with regard to the
uplifts or critical life events. question if they should be conceptualized and defined as
In an applied paper, Gnambs and Butins (2017) states, traits, or both. From a developmental perspective,
examined longitudinal measurement properties of single the interaction between traits and situations and their rela-
life satisfaction items versus multi-item scales using tive influence at different developmental stages deserves
sophisticated hybrid LST/trait change models. Their contri- further investigation.
bution is relevant from both a methodological and a In addition, LST theory can help researchers justify
substantive point of view. Methodologically, Gnambs and their operationalization of a given construct. For example,
Butins’ paper addresses an important assessment-related LST theory can be useful in exploring one central aspect
question: (When) should researchers use single items versus of construct validity – construct representation (Embretson,
multi-item scales to measure variability and change in life 1983). Construct representation refers to a cognitive theory

 2017 Hogrefe Publishing European Journal of Psychological Assessment 2017; Vol. 33(4):219–223
222 Editorial

that explains response behavior for that measure. LST Embretson, S. E. (1983). Construct validity: Construct repre-
models can be used to investigate the impact of situations sentation versus nomothetic span. Psychological Bulletin,
on response behavior. Therefore, we like to suggest that 93, 179–197.
Epstein, S. (1983). The stability of confusion: Reply to Mischel
LST studies could routinely be implemented in the develop- and Peake. Psychological Review, 90, 179–184.
ment of new psychometric measures and help test develop- Fleeson, W. (2001). Toward a structure- and process-integrated
ers gain a better understanding of the relationship between view of personality: Traits as density distributions of states.
the constructs they target and the tests they devise to Journal of Personality and Social Psychology, 80,
measure them. If the measurement of a construct does not 1011–1027.
work in the intended way, it is theoretically useless in the Fleeson, W. (2004). Moving personality beyond the person-
situation debate: The challenge and the opportunity of
given situation, be it a research study or a practical context.
https://econtent.hogrefe.com/doi/pdf/10.1027/1015-5759/a000413 - Wednesday, June 02, 2021 7:19:19 PM - IP Address:110.138.92.205

within-person variability. Current Directions in Psycholog-


Working toward a better understanding of the nature of ical Science, 13, 83–87.
constructs should actually be an important research goal Funder, D. C. (1991). Global traits: A Neo-Allportian approach
in itself, especially in the light of the ever-growing number to personality. Psychological Science, 2, 31–39.
of constructs encountered in published studies. We argue Funder, D. C. (2008). Persons, situations, and person-situation
that such endeavors will ultimately improve the quality of interactions. In O. P. John, R. Robins, & L. Pervin (Eds.),
psychological research. Handbook of personality (3rd ed, pp. 568–580). New York,
NY: Guilford Press.
Funder, D. C., Guillaume, E., Kumagai, S., Kawamoto, S., &
Sato, T. (2012). The person-situation debate and the
assessment of situations. Japanese Journal of Personality,
References 21, 1–11.
Geiser, C., Keller, B. T., Lockhart, G., Eid, M., Cole, D. A., &
Bishop, J., Geiser, C., & Cole, D. A. (2015). Modeling latent Koch, T. (2015). Distinguishing state variability from trait
growth with multiple indicators: A comparison of three change in longitudinal data: The role of measurement
approaches. Psychological Methods, 20, 43–62. (non)invariance in latent state-trait analyses. Behavior
Courvoisier, D. S., Eid, M., & Nussbeck, F. W. (2007). Mixture Research Methods, 47, 172–203.
distribution latent state-trait analysis: Basic ideas and Geiser, C., Litson, K., Bishop, J., Keller, B. T., Burns, G. L.,
applications. Psychological Methods, 12, 80–104. Servera, M., & Shiffman, S. (2015). Analyzing person,
Crayen, C., Eid, M., Lischetzke, T., & Vermunt, J. K. (2017). A situation, and person · situation interaction effects: Latent
continuous-time mixture latent state-trait Markov model for state-trait models for the combination of random and fixed
experience sampling data: Application and evaluation. situations. Psychological Methods, 20, 165–192.
European Journal of Psychological Assessment, 33, Geiser, C., & Lockhart, G. (2012). A comparison of four
296–311. doi: 10.1027/1015-5759/a000418 approaches to account for method effects in latent state trait
Danner, D., Hagemann, D., Schankin, A., Hager, M., & Funke, J. analysis. Psychological Methods, 17, 255–283.
(2011). Beyond IQ: A latent state-trait analysis of general Gnambs, T., & Butins, K. (2017). The measurement of
intelligence, dynamic decision making, and implicit learning. variability and change in life satisfaction: A comparison
Intelligence, 39, 323–334. doi: 10.1016/j.intell.2011.06.004 of single-item and multi-item instruments. European Jour-
Deinzer, R., Steyer, R., Eid, M., Notz, P., Schwenkmezger, P., nal of Psychological Assessment, 33, 224–238. doi: 10.1027/
Ostendorf, F., & Neubauer, A. (1995). Situational effects in 1015-5759/a000414
trait assessment: The FPI, NEOFFI and EPI questionnaires. Hertzog, C., & Nesselroade, J. R. (1987). Beyond autoregressive
European Journal of Personality, 9, 1–23. models: Some implications of the trait-state distinction for
Donnellan, M. B., Lucas, R. E., & Fleeson, W. (2009). Editorial: the structural modeling of developmental change. Child
Introduction to personality and assessment at age 40: Development, 58, 93–109.
Reflections on the legacy of the person-situation debate Kenny, D. A., & Zautra, A. (1995). The trait-state-error model
and the future of person-situation integration. Journal of for multiwave data. Journal of Consulting and Clinical
Research in Personality, 43, 117–119. Psychology, 63, 52–59.
Eid, M. (1996). Longitudinal confirmatory factor analysis for Kenrick, D. T., & Funder, D. C. (1988). Profiting from contro-
polytomous item responses: Model definition and model versy: Lessons from the person-situation debate. The
selection on the basis of stochastic measurement theory. American Psychologist, 43, 23–34.
Methods of Psychological Research – Online, 1, 65–85. Kirschbaum, C., Steyer, R., Eid, M., Patalla, U.,
Eid, M., & Hoffmann, L. (1998). Measuring variability and Schwenkmezger, P., & Hellhammer, D. H. (1990). Cortisol
change with an item response model for polytomous and behavior: 2 Application of a latent state-trait model to
variables. Journal of Educational and Behavioral Statistics, salivary cortisol. Psychoneuroendocrinology, 15, 297–307.
23, 193–215. doi: 10.1016/0306-4530(90)90080-S
Eid, M., Holtmann, J., Santangelo, P., & Ebner-Priemer, U. Koch, T., Schultze, M., Holtmann, J., Geiser, C., & Eid, M. (in
(2017). On the definition of latent state-trait models press). A multimethod latent state-trait model for
with autoregressive effects: Insights from LST-R theory. structurally different and interchangeable methods.
European Journal of Psychological Assessment, 33, 285–295. Psychometrika.
doi: 10.1027/1015-5759/a000435 Ledermann, T., & Macho, S. (2014). Analyzing change at the
Eid, M., & Langeheine, R. (1999). Measuring consistency dyadic level: The common fate growth model. Journal of
and occasion specificity with latent class models: A new Family Studies, 28, 204–213.
model and its application to the measurement of affect. Loncke, J., Mayer, A., Eichelsheim, V., Branje, S., Meeus, W.,
Psychological Methods, 4, 100–116. Koot, H., . . . Loeys, T. (2017). Latent state-trait models
Eid, M., Schneider, C., & Schwenkmezger, P. (1999). Do you for longitudinal family data: Investigating consistency
feel better or worse? The validity of perceived deviations in perceived support. European Journal of Psychologi-
of mood states from mood traits. European Journal of cal Assessment, 33, 256–270. doi: 10.1027/1015-5759/
Personality, 13, 283–306. a000415

European Journal of Psychological Assessment 2017; Vol. 33(4):219–223  2017 Hogrefe Publishing
Editorial 223

Mischel, W. (1968). Personality and assessment. New York, NY: to determine consistency and specificity coefficients.
Wiley. Anxiety Research, 1, 281–299.
Nett, U. E., Bieg, M., & Keller, M. M. (2017). How much trait Steyer, R., Mayer, A., Geiser, C., & Cole, D. A. (2015). A theory
variance is captured by measures of academic state of states and traits – revised. Annual Review of Clinical
emotions? A latent state-trait-analysis. European Journal Psychology, 11, 71–98. doi: 10.1146/annurev-clinpsy-
of Psychological Assessment, 33, 239–255. doi: 10.1027/ 032813-153719
1015-5759/a000416 Steyer, R., Schmitt, M., & Eid, M. (1999). Latent state-trait
Podsakoff, P. M., MacKenzie, S. B., & Podsakoff, N. P. (2012). theory and research in personality and individual differ-
Sources of method bias in social science research and ences. European Journal of Personality, 13, 389–408.
recommendations on how to control it. Annual Review of Thielemann, D., Sengewald, M.-A., Kappler, G., & Steyer, R.
Psychology, 65, 539–569. (2017). A probit latent state IRT model with latent item-
https://econtent.hogrefe.com/doi/pdf/10.1027/1015-5759/a000413 - Wednesday, June 02, 2021 7:19:19 PM - IP Address:110.138.92.205

Rauthmann, J. F., Sherman, R. A., & Funder, D. C. (2015). effect variables. European Journal of Psychological Assess-
Principles of situation research: Towards a better under- ment, 33, 271–284. doi: 10.1027/1015-5759/a000417
standing of psychological situations. European Journal of Tisak, J., & Tisak, M. S. (2000). Permanency and ephemerality
Personality, 29, 363–381. doi: 10.1002/per.1994 of psychological measures with application to organizational
Roberts, B. W. (2009). Back to the future: Personality and commitment. Psychological Methods, 5, 175–198.
assessment and personality development. Journal of Zimmerman, D. W. (1975). Probability spaces, Hilbert spaces,
Research in Personality, 43, 137–145. and the axioms of test theory. Psychometrika, 40, 395–412.
Schmitt, M., & Baumert, A. (2010). On the diversity of dynamic doi: 10.1007/BF02291765
person · situation interactions. European Journal of
Personality, 24, 497–500.
Steyer, R., Ferring, D., & Schmitt, M. J. (1992). States and Christian Geiser
traits in psychological assessment. European Journal of
Psychological Assessment, 8, 79–98. Department of Psychology
Steyer, R., Geiser, C., & Fiege, C. (2012). Latent state- Utah State University
trait models. In H. Cooper (Ed.), Handbook of research 2810 Old Main Hill
methods in psychology (pp. 291–308). Washington, DC: Logan, UT 84322-2810
APA. USA
Steyer, R., Majcen, A.-M., Schwenkmezger, P., & Buchner, A. E-mail christian.geiser@usu.edu
(1989). A latent state-trait anxiety model and its application

 2017 Hogrefe Publishing European Journal of Psychological Assessment 2017; Vol. 33(4):219–223

You might also like