Conference PaperPDF Available

Sequenzbasierte Distanzmaße für Kongruenz im Holland-Modell der Interessen- und Umweltorientierungen

Authors:

Abstract

In der Theorie Hollands (1997) besagt die Kongruenz-Hypothese, dass die Passung von individuellen Interessen und beruflichen Umwelten positive Auswirkungen auf das Wohlbefinden, die Mitarbeiterzufriedenheit und Leistung hat. Diese Hypothese findet allerdings unterschiedliche empirische Evidenz. Eine Erklärung der uneinheitlichen Befundlage liegt auch in der unterschiedlichen Operationalisierung von Kongruenz. In der vorliegenden Arbeit werden im Rahmen der Methodik der Sequenzdatenanalyse verschiedene Operationalisierungen für Kongruenz vorgeschlagen, entwickelt und mit etablierten Kongruenzindizes (Tarnai, Langmeyer, & Guglhör-Rudan, 2008), verglichen. In Anlehnung an die Definition einer Sequenz nach Abbott (1995) werden die dreistelligen Holland-Codes als Sequenzen verstanden, deren geordnete Elemente die drei dominanten Interessen Orientierungen darstellen. Als Distanzmaße werden zunächst eine substitutionskostensensitive Hamming-Distanz (Hamming, 1950) und die in den Sozialwissenschaften implementierte Levenshtein-Distanz (Levenshtein, 1966) zugrunde gelegt. Es wird gezeigt, wie die substitutionskostensensitive Hamming-Distanz durch Berücksichtigung von so genannten indel-operationen (löschen & einfügen) zu einer kostensensitiven Levenshtein-Distanz generalisiert werden kann. Dabei werden die Kosten in Abhängigkeit von der Position in der Sequenz, an der gelöscht bzw. eingefügt wird, mit unterschiedlichen Gewichten versehen. Ziel ist die Entwicklung eines allgemeinen Kongruenzmaßes welches positionssensitiv die substitutions- und indel-Kostenstruktur angemessen berücksichtigt. Grundlage der Analysen bildet der vollständige paarweise Vergleich der jeweils 120 (6!/3!) möglichen, dreistelligen Holland-Codes. Für jeden Kongruenzindex resultieren so symmetrische Distanzmatrizen (120 x 120), welche hinsichtlich ihrer circumplexen Struktur mit ordinaler multidimensionaler Skalierung analysiert werden. Erste Ergebnisse deuten darauf hin, dass die Gewichtung erheblichen Einfluss auf die Anpassung an die Circumplex-Struktur hat.
Sequenzbasierte Distanzmaße f¨
ur Kongruenz
im Holland-Modell der Interessen- und
Umweltorientierungen
J¨
org–Henrik Heine1, Heinz Leitg¨
ob2,
Florian G. Hartmann3& Christian Tarnai3
1Technische Universit¨
at M¨
unchen
2Katholische Universit¨
at Eichst¨
att-Ingolstadt
3Universit¨
at der Bundeswehr Neubiberg (bei M¨
unchen)
21. September 2016
Heine (TU M¨
unchen) Sequenzbasierte Distanzmaße 21. September 2016 1 / 14
Hintergrund
1Theorie der Pers¨
onlichkeits- und Interessenorientierungen:
(Holland, 1997)
Sechs Pers¨
onlichkeits- und Interessenorientierungen (R I A S E C)
Hexagonale Anordnung der sechs Dimensionen repr¨
asentiert
psychologische N¨
ahe (Calculus–Hypothese)
2Kongruenz-Hypothese:
Passung von individuellen Interessen und beruflichen Umwelten
Positive Auswirkungen auf das Wohlbefinden, die
Mitarbeiterzufriedenheit und Leistung
Unterschiedliche empirische Evidenz
3Holland–Code:
Bestimmung in der Regel mit Tests: z.B. AIST–R – Bergmann und
Eder (2005) oder EXPLORIX – J¨
orin, Stoll, Bergmann und Eder (2003)
St¨
arkste Interessen Orientierung: Individueller Pers¨
onlichkeitstyp z.B.:
R–Typ, I–Typ, A–Typ,....
Differenziertere Beschreibung: Hinzunahme der zweit- und drittgr¨
oßten
Interessen –>3 Letter-Code z.B. RIE–Typ, ESC–Typ, . . . 120
Kombinationsm¨
oglichkeiten
Heine (TU M¨
unchen) Sequenzbasierte Distanzmaße 21. September 2016 2 / 14
Distanzmaße (’klassische’) f¨
ur Kongruenz
Brown–C–Index:
(Brown & Gore, 1994)
. . . Specifically, a value of C is obtai-
ned for individuals by first sequenti-
ally comparing the first, second, and
third letters in an individual’s three
letter person and environment codes,
and then calculating C by the followi-
ng formula: C= 3×(Xi)+2(Xi)+(Xi)
where Xiare scores (3, 2, 1, and 0)
assigned to each comparison on the
basis of hexagonal distance between
the letters (3 = identical P and E
letters, 2 =adjacent hexagonal positi-
on, 1 = alternate hexagonal positions,
0 = opposite hexagonal positions).
Iachan–Index: (Iachan, 1984)
Heine (TU M¨
unchen) Sequenzbasierte Distanzmaße 21. September 2016 3 / 14
Distanzmaße aus der Sequenzdatenanalyse
Sequenz:
By sequence I mean an ordered list of elements
(Abbott, 1995, S. 94).
Holland-3-Letter-Codes: Sequenzen der L¨
ange 3 mit den
Orientierungen als nicht-repetitive Elemente.
Ordnung: Reihung nach abnehmender Ausgepr¨
agtheit der drei
dominierenden Orientierungen
>Stellen–Gewichtung.
Sequenzdatenanalyse: Feststellung der (Un-)¨
Ahnlichkeit zwischen
zwei Sequenzen mittels geeignetem Distanzmaß
>hier: Hamming- und Levenshtein-Distanz.
Erweiterung: Im Rahmen der Hexagon Hypothese unterschiedliche
¨
Ahnlichkeiten der sechs Orientierung
>Kostensensitivit¨
at.
Heine (TU M¨
unchen) Sequenzbasierte Distanzmaße 21. September 2016 4 / 14
Distanzmaße . . . formale Darstellung
Hamming Distanz dH(a,b):Vergleich der Sequenzen aund bder
l¨
ange ran jeder Stelle pund bestimmung der Substitutionskosten csp
csp=(0|ap=bp
1|ap6=bp
;dH(a,b) =
r
X
p=1
csp(1)
Levenshtein Distanz dL(a,b):Neben Substitution (csp) Einf¨
uhrung
von zus¨
atzlichen Operationen (1) Einf¨
ugen mit den Kosten cipund
(2) L¨
oschen mit den Kosten cdp; finden des Minimums ...
dL(ap,bp) =
r
X
p=1
min
d(ap1,bp) + cdp
d(ap1,bp
1) + csp
d(ap1,bp
1) + cip
(2)
Heine (TU M¨
unchen) Sequenzbasierte Distanzmaße 21. September 2016 5 / 14
Erweiterungen der Distanzmaße aus der
Sequenzdatenanalyse
1Kostensensitivit¨
at
R I A S E C
R 0.00 0.25 0.50 0.75 0.50 0.25
I 0.25 0.00 0.25 0.50 0.75 0.50
A 0.50 0.25 0.00 0.25 0.50 0.75
S 0.75 0.50 0.03 0.00 0.25 0.50
E 0.50 0.75 0.50 0.25 0.00 0.25
C 0.25 0.50 0.75 0.50 0.25 0.00
2Stellen–Gewichtung
Vergabe von Gewichten wpnach Position pinnerhalb der
3-Letter-Codes
z.B.: 1. Stelle: w1= 1.5; 2. Stelle: w2= 1.25; und 3. Stelle: w3= 1
Heine (TU M¨
unchen) Sequenzbasierte Distanzmaße 21. September 2016 6 / 14
Fragestellung
1K¨
onnen die 120 m¨
oglichen Typen in der hexagonalen Struktur
des Holland-Modells abgebildet werden?
Wie schneiden unterschiedliche Kongruenzmaße ab?
2Ist die Abbildung des Hexagons durch einen Circumplex
m¨
oglich?
Heine (TU M¨
unchen) Sequenzbasierte Distanzmaße 21. September 2016 7 / 14
Datenbasis
’Simulierte’ ¨
Ahnlichkeits- bzw. Distanzmatrizen f¨
ur jedes
untersuchte Kongruenzmaß
F¨
ur alle m¨
oglichen 3-Letter-Code Kombinationen
Dimension: 120 x 120
Symetrische Matrizen
jede Kombination ist genau einmal bzw. bei vernachl¨
assigung der
Reihenfolge zweimal enthalten
...
.
.
.
Heine (TU M¨
unchen) Sequenzbasierte Distanzmaße 21. September 2016 8 / 14
Programm – Software
1R(R Core Team, 2015) R version 3.2.2 (2015-08-14)
2Selbst entwickelte R-Funktionen f¨
ur folgende Kongruenzmaße
Funktion iachan.holland() zur Berechnung des Iachan Index (Iachan,
1984).
Funktion brown.c.holland() zur Berechnung des C Index nach Brown
und Gore (1994).
Funktion hamming.holland() zur Berechnung der Stellen–gewichteten,
kostensensitiven Hammig Distanz (Hamming, 1950).
Funktion levenshtein.holland() zur Berechnung der
Stellen–gewichteten, kostensensitiven Levenshtein Distanz (Levenshtein,
1966).
3Package: smacof version 1.8-13 (Mair, Leeuw, Borg & Groenen,
2016)
Berechnung der 2 dimensionalen, ordinalen MDS L¨
osung
Funktion: smacofSym() mit den folgenden Argumenten:
delta = ’simulierte datenmatrizen’
ndim = 2
itmax = 10000
type = ’ordinal’
Heine (TU M¨
unchen) Sequenzbasierte Distanzmaße 21. September 2016 9 / 14
Konfigurationen - klassische Indices
−1.0 −0.5 0.0 0.5 1.0
−0.5 0.0 0.5
Konfiguration iachan (Index)
Dimension 1
Dimension 2
RAC
RAE
RAI
RAS
RCA
RCE
RCI
RCS
REA
REC
REI
RES
RIA
RIC
RIE RIS
RSA
RSC
RSE RSI
IAC
IAE
IAR
IAS
ICA
ICE
ICR
ICS
IEA
IEC IER
IES
IRA
IRC
IRE
IRS ISA
ISC
ISE
ISR
ACE
ACI
ACR
ACS
AEC
AEI
AER
AES
AIC
AIE
AIR
AIS
ARC
ARE
ARI
ARS
ASC
ASE
ASI
ASR
SAC
SAE
SAI
SAR
SCA
SCE
SCI
SCR
SEA
SEC SEI
SER
SIA
SIC
SIE
SIR
SRA
SRC
SRE
SRI
EAC
EAI
EAR
EAS
ECA
ECI
ECR
ECS
EIA
EIC
EIR
EIS
ERA
ERC
ERI
ERS
ESA
ESC
ESI
ESR
CAE
CAI
CAR
CAS
CEA
CEI
CER
CES
CIA
CIE
CIR
CIS
CRA
CRE
CRI CRS
CSA
CSE
CSI
CSR
(a) Stress = 0.3912; Iteration = 427;
iachan.holland(a,b)
−1.0 −0.5 0.0 0.5 1.0
−0.5 0.0 0.5
Konfiguration brown (Index)
Dimension 1
Dimension 2
RAC
RAE
RAI
RAS
RCA
RCE
RCI
RCS
REA
REC REI
RES
RIA
RIC
RIE
RIS
RSA
RSC
RSE
RSI
IAC
IAE
IAR
IAS
ICA ICE
ICR
ICS
IEA
IEC
IER
IES IRA
IRC
IRE
IRS
ISA
ISC
ISE
ISR
ACE
ACI
ACR
ACS
AEC
AEI
AER
AES
AIC
AIE
AIR
AIS
ARC ARE
ARI
ARS
ASC
ASE
ASI
ASR
SAC
SAE
SAI
SAR
SCA
SCE
SCI
SCR
SEA
SEC
SEI
SER
SIA SIC
SIE
SIR SRA
SRC
SRE SRI
EAC
EAI
EAR
EAS
ECA
ECI
ECR
ECS
EIA
EIC
EIR
EIS
ERA
ERC
ERI
ERS
ESA
ESC
ESI
ESR
CAE
CAI
CAR
CAS
CEA
CEI
CER
CES
CIA
CIE
CIR
CIS
CRA
CRE
CRI
CRS
CSA
CSE
CSI
CSR
(b) Stress = 0.4004; Iteration = 846;
brown.c.holland(a,b)
Heine (TU M¨
unchen) Sequenzbasierte Distanzmaße 21. September 2016 10 / 14
Konfigurationen - neue Indices ungewichtet
−0.5 0.0 0.5 1.0
−0.5 0.0 0.5
Konfiguration hammingUngew (Index)
Dimension 1
Dimension 2
RAC
RAE
RAI
RAS
RCA
RCE
RCI
RCS
REA
REC
REI
RES
RIA
RIC
RIE
RIS
RSA
RSC
RSE
RSI
IAC
IAE
IAR
IAS
ICA
ICE
ICR
ICS
IEA
IEC
IER
IES
IRA
IRC
IRE
IRS
ISA
ISC
ISE
ISR
ACE
ACI
ACR
ACS
AEC
AEI
AER
AES
AIC
AIE
AIR
AIS
ARC
ARE
ARI
ARS
ASC
ASE
ASI
ASRSAC
SAE
SAI
SAR
SCA
SCE
SCI
SCR
SEA
SEC
SEI
SER
SIA
SIC
SIE
SIR
SRA
SRC
SRE
SRI
EAC
EAI
EAR
EAS
ECA
ECI
ECR
ECS
EIA
EIC
EIR
EIS
ERA
ERC
ERI
ERS
ESA
ESC
ESI
ESR
CAE
CAI
CAR
CAS
CEA
CEI
CER
CES
CIA
CIE
CIR
CIS
CRA
CRE
CRI
CRS
CSA
CSE
CSI
CSR
(c) Stress = 0.2785; Iteration = 53;
hamming.holland( a, b, costs = "hexa", weights =
c(1,1,1))
−1.0 −0.5 0.0 0.5 1.0
−0.5 0.0 0.5
Konfiguration levenshteinUngew (Index)
Dimension 1
Dimension 2
RAC
RAE
RAI
RAS
RCA
RCE
RCI
RCS
REA
REC
REI
RES
RIA
RIC
RIE
RIS
RSA
RSC
RSE
RSI
IAC
IAE IAR
IAS
ICA
ICE ICR
ICS
IEA
IEC
IER
IES
IRA
IRC
IRE
IRS
ISA
ISC
ISE
ISR
ACE
ACI
ACR
ACS AEC
AEI AER
AES AIC
AIE
AIR
AIS
ARC
ARE
ARI
ARS ASC
ASE
ASI ASR
SAC
SAE
SAI
SAR
SCA
SCE
SCI
SCR
SEA
SEC
SEI
SER
SIA
SIC
SIE
SIR
SRA
SRC
SRE
SRI
EAC
EAI EAR
EAS
ECA
ECI ECR
ECS
EIA EIC
EIR
EIS
ERA ERC
ERI
ERS
ESA ESC
ESI ESR
CAE
CAI
CAR
CAS
CEA
CEI
CER
CES
CIA
CIE
CIR
CIS
CRA
CRE
CRI
CRS
CSA
CSE
CSI
CSR
(d) Stress = 0.2096; Iteration = 25;
levenshtein.holland( a, b, costs = "hexa", weights =
c(1,1,1))
Heine (TU M¨
unchen) Sequenzbasierte Distanzmaße 21. September 2016 11 / 14
Konfigurationen - neue Indices
−1.0 −0.5 0.0 0.5 1.0
−0.5 0.0 0.5
Konfiguration hamming (Index)
Dimension 1
Dimension 2
RAC
RAE
RAI
RAS
RCA
RCE
RCI
RCS
REA
REC
REI
RES
RIA RIC
RIE
RIS
RSA
RSC
RSE
RSI
IAC
IAE
IAR
IAS
ICA
ICEICR
ICS
IEA
IEC
IER
IES
IRA
IRC
IRE
IRS
ISAISC
ISE
ISR
ACE
ACI
ACR
ACS
AEC
AEI
AER
AES
AIC
AIE
AIR
AIS ARC
ARE
ARI
ARS
ASC
ASE
ASI
ASR
SAC
SAE
SAI
SAR
SCA
SCE
SCI
SCR
SEA
SEC
SEISER
SIA
SIC
SIE
SIR
SRA
SRC
SRE
SRI
EAC
EAI
EAR
EAS ECAECI
ECR
ECS
EIA EIC
EIR
EIS ERA ERC
ERI
ERS
ESA
ESC
ESI
ESR
CAE
CAI
CAR
CAS
CEA
CEI
CER
CES
CIA
CIE
CIR
CIS
CRA
CRE
CRI
CRS
CSA
CSE
CSI
CSR
(e) Stress = 0.2728; Iteration = 28;
hamming.holland( a, b, costs = "hexa", weights =
c(1.5,1.25,1))
−1.0 −0.5 0.0 0.5 1.0
−0.5 0.0 0.5
Konfiguration levenshtein (Index)
Dimension 1
Dimension 2
RAC
RAE
RAI
RAS
RCA
RCE
RCI
RCS
REA
REC
REI
RES
RIA
RIC
RIE
RIS
RSA
RSC
RSE
RSI
IAC
IAE IAR
IAS
ICA
ICE ICR
ICS
IEA
IEC
IER
IES
IRA
IRC
IRE
IRS
ISA ISC
ISE ISR
ACE
ACI
ACR
ACS AEC
AEIAER
AES
AIC
AIE
AIR
AIS
ARC
ARE
ARI
ARS
ASC
ASE
ASI
ASR
SAC
SAE
SAI
SAR
SCA
SCE
SCI
SCR
SEA
SEC
SEI
SER
SIA
SIC
SIE
SIR
SRA
SRC
SRE
SRI
EAC
EAI EAR
EAS
ECA ECI ECR
ECS
EIA EIC
EIR
EIS
ERA ERC
ERI
ERS
ESA ESC
ESI ESR
CAE
CAI
CAR
CAS
CEA
CEI
CER
CES
CIA
CIE
CIR
CIS
CRA
CRE
CRI
CRS
CSA
CSE
CSI
CSR
(f) Stress = 0.1747; Iteration = 18
levenshtein.holland( a, b, costs = "hexa", weights =
c(1.5,1.25,1))
Heine (TU M¨
unchen) Sequenzbasierte Distanzmaße 21. September 2016 12 / 14
Zusammenfassung – Diskussion
1Vergleich der Kongruenzmaße: Die Sequenzbasierten
Kongruenzmaße schneiden besser ab
Geringere Stress Werte bei der 2 Dimensionalen, ordinalen MDS
Konfigurationen lassen Hexagonale Anordnung deutlicher erkennen
2Gewichtung: Die Stelle-Gewichtung hat eine Einfluss auf die
Anpassung an die Circumplex–Struktur
f¨
ur die kostensensitive Hamming und Levenshtein Distanz
Geringere Stress Werte mit Stellen-Gewichtung
Frage: Stellen-Gewichte letztlich arbitr¨
ar?
Heine (TU M¨
unchen) Sequenzbasierte Distanzmaße 21. September 2016 13 / 14
Literatur
Abbott, A. (1995, August). Sequence Analysis: New Methods for Old Ideas. Annual Review
of Sociology,21, 93–113.
Bergmann, C. & Eder, F. (2005). AIST-R Allgemeiner Interessen-Struktur-Test mit
Umwelt-Struktur-Test (UST-R) - Revision. G¨
ottingen: Beltz Test.
Brown, S. D. & Gore, P. A. (1994, Dezember). An Evaluation of Interest Congruence Indices:
Distribution Characteristics and Measurement Properties. Journal of Vocational
Behavior,45 (3), 310–327.
Hamming, R. (1950, April). Error detecting and error correcting codes. Bell System Technical
Journal, The,29 (2), 147–160.
Holland, J. L. (1997). Making vocational choices: A theory of vocational personalities and
work environments (3rd ed.) (Bd. xiv). Odessa, FL, US: Psychological Assessment
Resources.
Iachan, R. (1984). A measure of agreement for use with the Holland classification system.
Journal of Vocational Behavior,24 (2), 133–141.
J¨
orin, S., Stoll, F., Bergmann, C. & Eder, F. (2003). EXPLORIX – Das Werkzeug zur
Berufswahl und Laufbahnplanung. Bern: Huber.
Levenshtein, V. I. (1966, Februar). Binary Codes Capable of Correcting Deletions, Insertions
and Reversals. Soviet Physics Doklady,10, 707.
Mair, P., Leeuw, J. D., Borg, I. & Groenen, P. J. F. (2016, April). smacof: Multidimensional
Scaling. Zugriff am 2016-09-16 auf https://CRAN.R-project.org/package=smacof
R Core Team, V. (2015). R: A Language and Environment for Statistical Computing. Vienna,
Austria. Zugriff auf http://www.R-project.org/ (ISBN 3-900051-07-0)
Heine (TU M¨
unchen) Sequenzbasierte Distanzmaße 21. September 2016 14 / 14
... According to Abbott [71], such sequences are defined as ordered lists of elements from a finite selection. This general definition of a sequence or of sequence data also applies to Holland letter codes so that methods from sequence data analysis are an interesting approach for the investigation of the congruence of two Holland letter codes (see [72], [for theoretical overview and empirical investigation]). In the Holland package, two sequence-based distance measures are currently implemented as indices for congruence. ...
... For these cases, the R package holland provides well known and established measures, as well as some interesting new measures, that have rarely been considered in the vocational literature such as the Hamming distance [61] or the Levensthein distance [62]. These measures seem to be promising in terms of depicting the hexagonal structure of vocational interests [72]; however, proof of the criterion validity still has to be provided. ...
Article
Full-text available
John L. Holland’s theory of vocational choice is one of the most prominent career theories and is used by both researchers and practitioners around the world. The theory states that people should seek work environments that fit their vocational interests in order to be satisfied and successful. Its application in research and practice requires the determination of coefficients, which quantify its core concepts such as person-environment fit. The recently released R package holland aims at providing a holistic collection of the references, descriptions and calculations of the most important coefficients. The current paper presents the package and examines it in terms of its application for research and practice. For this purpose, the functions of the package are applied and discussed. Furthermore, recommendations are made in the case of multiple coefficients for the same theoretical concept and features that future releases should include are discussed. The R package holland is a promising computational environment providing multiple coefficients for Holland’s most important theoretical concepts.
Article
The author was led to the study given in this paper from a consideration of large scale computing machines in which a large number of operations must be performed without a single error in the end result. This problem of “doing things right” on a large scale is not essentially new; in a telephone central office, for example, a very large number of operations are performed while the errors leading to wrong numbers are kept well under control, though they have not been completely eliminated. This has been achieved, in part, through the use of self-checking circuits. The occasional failure that escapes routine checking is still detected by the customer and will, if it persists, result in customer complaint, while if it is transient it will produce only occasional wrong numbers. At the same time the rest of the central office functions satisfactorily. In a digital computer, on the other hand, a single failure usually means the complete failure, in the sense that if it is detected no more computing can be done until the failure is located and corrected, while if it escapes detection then it invalidates all subsequent operations of the machine. Put in other words, in a telephone central office there are a number of parallel paths which are more or less independent of each other; in a digital machine there is usually a single long path which passes through the same piece of equipment many, many times before the answer is obtained.
Article
This article first presents research that compares 10 extant measures of interest congruence, derived from Holland′s (1985a) theory of career choice, for their underlying distribution characteristics and their abilities to discriminate among subtle but important differences in person-environment congruence. Using simulated data, we show that (a) none of the three-letter code measures are sensitive to differences among persons across the entire possible range of congruence scores, and (b) with one exception (Kwak & Pulvino, 1982), they are incapable of making fine distinctions among persons with like, but out of order, three-letter person and environment codes. In a second study, we present a new measure of interest congruence that retains the advantages of the Kwak and Pulvino measure, but is easier to calculate. Results are discussed in terms of how users can choose among measures, and needed areas for future research on the new measure and on congruence measurement in general.
Article
A measure of agreement for partially ordered data is suggested that makes use of the k categories with the highest ranks, the ranks of the remaining categories being disregarded. The proposed measure is applied to the Self-Directed Search in order to evaluate its agreement with occupational preference. Other forms of agreement in the Holland classification system can also be assessed by the proposed measure. Examples are congruence and husband-wife and counselor-client agreement.
Article
A wide variety of work in social science concerns sequences of events or phenomena. This essay reviews concepts of sequence and methods for analyzing sequences. After a brief definitional discussion, I consider sequence literatures from various areas. I then discuss recent methodologies for sequence analysis. I review stepwise approaches like Markovian and event history analysis as well as whole sequence approaches resting on new developments in biology and other fields.