You are on page 1of 5

Personality and Individual Differences 123 (2018) 171–175

Contents lists available at ScienceDirect

Personality and Individual Differences


journal homepage: www.elsevier.com/locate/paid

Short Communication

Evidence for the reliability and factor solution of the CFCS-14 in Spanish: A T
multi-method validation in Spain and Uruguay

Alejandro Vásquez-Echeverríaa, , Mirko Antinob,c, Lucía Alvarez-Nuñeza,
Alfredo Rodríguez-Muñozb
a
Universidad de la República, Uruguay
b
Universidad Complutense de Madrid, Spain
c
Instituto Universitário de Lisboa (ISCTE-IUL)

A R T I C L E I N F O A B S T R A C T

Keywords: The factor structure of the Consideration of Future Consequences Scale (CFCS-14) is currently subject to debate
Consideration of future consequences scale and little is known about its temporal stability, or indeed whether temporal changes affect the factor structure.
Two-factor solution This study examined the factor structure of a Spanish version of the CFCS-14 in Uruguayan and Spanish samples.
Daily method In Spain data were collected from a sample that completed the CFCS-14 on five consecutive days, which enabled
Multilevel confirmatory factor analysis
separate analysis of the within-and between-subject factor structure. Simple and multilevel confirmatory factor
analysis showed that a two-correlated factor structure provided the best fit to data from both samples and both
the within and between-person levels. Further analyses revealed the expected pattern of correlations between
CFCS-14 and ZTPI subscales. The results suggest that the Spanish version of the CFCS-14 is reliable and has a
robust, two-factor structure. They also suggest that CFCS-14 scores fluctuate over short periods of time.

The Consideration of Future Consequences Scale (CFCS) was in- Sherman, & Funder, 2015). We attempted to address this issue by
troduced by Strathman, Gleicher, Boninger, and Edwards (1994). It was asking participants to complete a daily diary to provide repeated-
designed to assess the extent to which individuals are influenced by the measures data.
anticipated immediate and distant outcomes of their behaviour and is Third, debate about the factor structure of the CFCS continues, al-
used to help determine how people resolve dilemmas involving conflict though the majority of evidence suggests it has a two-correlated-factors
between short- and long-term outcomes. It has been shown to be a re- structure. Initially Strathman et al. (1994) suggested the scale was one-
liable, valid and stable measure of individual differences (Joireman, dimensional, but other studies showed that a two-factor solution pro-
Shaffer, Balliet, & Strathman, 2012; Strathman et al., 1994) that pre- vided a better fit (Joireman, Balliet, Sprott, Spangenberg, & Schultz,
dicts cognitive, affective and behavioural outcomes. CFCS score is ne- 2008; Vásquez Echeverría, Esteves, Gomes, & Ortuño, 2015), with one
gatively related to impulsivity and sensation seeking and positively factor representing consideration of immediate outcomes (CFC-I) and
related to ability to delay gratification and future orientation other representing assessment of distant outcomes (CFC-F). Joireman
(Joireman, Strathman, & Balliet, 2006). The convergent validity of the et al. (2012) introduced the CFCS-14 to improve the reliability of the
CFCS was determined in relation with time perspective measures (Van CFC-F subscale. Other studies have explored alternative factor struc-
Beek, Handgraaf, & Antonides, 2017). tures, for instance Zhang, Kong, Zhang, and Li (2015) suggested a four-
There remain shortcomings in the literature on the CFCS. First, al- factor solution and McKay, Perry, Percy, and Cole (2016) tested bi-
though it is used worldwide, the current version, the CFCS-14, has not factor and two-correlated-factor structures, with the latter having the
been validated in a Spanish-speaking country. Second, there is no better fit.
published assessment of the measurement properties of the CFCS Our first objective was to establish the psychometric proprieties of a
adapted to an everyday experience method. This is important, as there Spanish version of the CFCS-14. Our second objective was to test a
is now debate in the personality and consideration of future con- CFCS-14 scale adapted for repeated use. Our third objective was to test
sequences (CFC) literature about the temporal stability of constructs different factorial models at the within- and between-subjects levels.
were once regarded as stable personality traits, but may be dependent We addressed these objectives through two studies. In the first we
on situational characteristics (Joireman & King, 2016; Rauthmann, analysed cross-sectional data from a Uruguayan sample; in the second


Corresponding author at: Instituto de Fundamentos y Métodos en Psicología, Facultad de Psicología, Universidad de la República, Tristán Narvaja 1674, Montevideo 11200, Uruguay.
E-mail address: avasquez@psico.edu.uy (A. Vásquez-Echeverría).

https://doi.org/10.1016/j.paid.2017.11.021
Received 14 September 2017; Received in revised form 11 November 2017; Accepted 13 November 2017
0191-8869/ © 2017 Elsevier Ltd. All rights reserved.
A. Vásquez-Echeverría et al. Personality and Individual Differences 123 (2018) 171–175

we analysed longitudinal data from a diary study carried out in Spain. Table 1
Fit indexes of the models tested in Study 1.
1. Study 1
Model χ2 df CFI TLI RMSEA 90% CI SRMR

This study was designed to confirm the factor structure and relia- 1. One factor 392,11 77 0,68 0,62 0,11 0,10–0,12 0,11
bility of a Spanish version of the CFCS-14 and to provide evidence of its 2. Two factors 113,97 75 0,92 0,90 0,05 0,03–0,06 0,06
3. Bifactor 98,03 62 0,96 0,95 0,04 0,03–0,06 0,04
convergent validity.

1.1. Method 0.90; RMSEA ≤ 0.06; SRMR ≤ 0.08 (Hu & Bentler, 1999).
Only Models 2 and 3 had an acceptable fit to the data (see Table 1).
1.1.1. Participants When looking at factor loadings, the somewhat better values for fit
The participants were 679 students (67.3% women, 5.3% did not indices in the case of the two-factor model should be treated with
report their gender) recruited from a public university in Uruguay (age caution. In the two-factor model, all items load on theoretical target
range: 18–59 years, M = 22.96; SD = 6.37). Participants provided factor, whereas the bifactor model produced non-significant loadings
written, informed consent and completed the measures in paper-and- on the general factor (items 2, 6 and 7) and items 3 and 4 did not loaded
pencil format (n = 581) or online (n = 98) during classes. A subset of significantly on their specific factor (CFC-I). Loadings for both models
the sample (n = 345) also completed time perspective measures. are presented in Tables 2 and 3).

1.1.2. Instruments
1.2.3. Convergent validity
1.1.2.1. Sociodemographic questionnaire. We asked about age, gender
Pearson correlations with time perspective measures (Table 4)
and educational level.
showed moderate associations between CFC-I and the present fatalist
and future negative subscales and between CFC-F and the ZTPI-Future
1.1.2.2. Consideration of Future Consequences Scale (CFCS-14). We
subscale.
produced a Spanish version of the CFCS-14 (Joireman et al., 2012)
using the translation and back-translation procedure (see Appendix A).
2. Study 2: longitudinal validation
Responses were given using a seven-point Likert scale.

The aims of this study were (a) to assess the short-term temporal
1.1.2.3. Zimbardo Time Perspective Inventory (ZTPI) – short version. We
stability of the CFCS-14 using a repeated-measures design and (b) to
used the 15-item version of the ZTPI developed by Košťál, Klicperová-
determine whether the within-subject (WS) factor structure of the
Baker, Lukavská, and Lukavský (2016), to which responses are given
CFCS-14 is different from the between-subject (BS) factor structure.
using a five-point Likert scale. In this sample the omega reliabilities
Multilevel CFA (MCFA) splits the total sample covariance matrix into
were: past negative = 0.71, past positive = 63, present fatalist = 69,
WS and BS covariance matrices. Values at the BS level represent the
present hedonist = 0.71, future = 0.64.
latent means for each participant at the WS level.
1.1.2.4. Future Time Perspective Inventory (Janeiro, 2012). We used only
the four items of the future negative subscale. Responses are given on a 2.1. Method
seven-point Likert scale. In our sample omega reliability was 0.86.
2.1.1. Participants and Procedure
1.1.3. Data analysis In this study 114 Spanish participants (56 men, Mage = 41.80 years,
Participants who left more than ten items unanswered (n = 11) or SD = 10.02) were recruited through the social networks of the re-
used a repetitive response pattern (n = 4) were excluded from analysis. searchers and their students. The majority of the sample (69.3%) had a
No multivariate outliers were found. Analyses were performed using university degree and the participants were employed in a broad range
MPlus 8, using the FIML method of handling missing data. of professions. Participants had to complete a diary survey, consisting
of the CFCS and other questionnaires not analysed here, on five con-
1.2. Results secutive days (Monday to Friday). At the end of each day participants
received email prompting them to complete the questionnaires, which
1.2.1. Internal consistency they did via an electronic survey platform. The mean number of com-
Under all conditions of administration, the coefficients were > pleted diary entries was 4.24 (N = 483 occasions).
0.70. The reliability statistics (α and ω respectively) were as follows,
CFC-I overall: 0.79, 0.74; CFC-I paper: 0.76, 0.81; CFC-I online: 0.79, 2.1.2. Instruments
0.80; CFC-F overall: 74 0.78; CFC-F paper: 0.77, 0.72; CFC-F online: 74, 2.1.2.1. Socio-demographic questionnaire. As in Study 1.
0.77.
2.1.2.2. Consideration of Future Consequences Scale. We slightly
1.2.2. Factor analyses modified the CFCS-14 items to suit the diary context (e.g., “Today I
We split the sample into halves for the factor analyses. We con- thought about…” see Appendix A). We also used a six-point Likert scale
ducted exploratory factor analysis (EFA) with maximum likelihood es- instead of a seven-point scale, to prevent participants choosing the
timation and geomin rotation. Horn's parallel analysis suggested that a central point to give a neutral response. Item 5 was removed from the
maximum of two factors should be retained. The two-factor solution analysis because it was inadvertently worded incorrectly.
showed acceptable loadings: all loadings on the theoretical target factor
were > 0.30 and all secondary loadings were < 0.15. 2.2. Results
We subjected three models to confirmatory factor analysis (CFA): a
one-dimensional model (Model 1); a two-correlated-factor model Intraclass correlations (ICCs) for items were moderate, ranging from
(Model 2) and the bifactor model tested by McKay et al. (2016) (Model 0.29 to 0.43 suggesting variability at the WS-level. MCFA models
3). We specified the model to correlate the error terms of items 13 and showed that the two-correlated-factors model was the only model with
14 and items 3 and 4. We used robust maximum likelihood estimation. broadly acceptable fit at both the BS and WS levels (see Table 5) (the
The following criteria for acceptable fit were used: CFI and TLI close to BS-level SRMR did not meet the acceptability criterion). Factor loadings

172
A. Vásquez-Echeverría et al. Personality and Individual Differences 123 (2018) 171–175

Table 2
Standardized and unstandardized loadings in the bifactor model (Study 1).

Item General factor CFC-I CFC-F R2

β B S.E β B S.E Β B S.E

1 − 0.252(1) − 0.390 0.101 0.512 0.792 0.094 0.325


2 0.000 0.000 (1) 0.118 0.598 1.119 0.118 0.358
3 0.779 1.287 0.170 −0.16 (1)
−0.26 (1)
0.343 0.631
4 0.659 1.143 0.128 0.137(1) 0.24(1) 0.224 0.452
5 0.346 0.621 0.138 0.274⁎ 0.49⁎ 0.161 0.195
6 0.000(1) 0.000(1) 0.123 0.512 0.681 0.120 0.133⁎
7 0.045(1) 0.080(1) 0.105 0.598 0.909 0.103 0.306
8 − 0.176⁎ − 0.30⁎ 0.109 0.512 1.013 0.103 0.385
9 0.355 0.613 0.146 0.319 0.550 0.158 0.228
10 0.392 0.674 0.162 0.371 0.637 0.181 0.292
11 0.554 0.912 0.218 0.641 1.055 0.213 0.718
12 0.294 0.521 0.141 0.253⁎ 0.45⁎ 0.166 0.151
13 − 0.263 − 0.403 0.103 0.445 0.682 0.096 0.267
14 − 0.240 − 0.385 0.103 0.492 0.791 0.099 0.300

Note. All loadings significant at p < 0.001, except for ⁎p < 0.05, and (1)
non significant.

Table 3 are shown in Fig. 1.


Standardized and unstandardized loadings in the two-factor model (Study 1).

CFC-I CFC-F R2
3. General discussion
Β B S.E Β B S.E
The goals of this study were (a) to establish the psychometric
1 0.561 0.869 0.101 0.315 properties of a Spanish version of the CFCS-14; (b) to develop and test a
2 0.568 1.063 0.129 0.322
3 0.526 0.869 0.118 0.276
repeated-measures version of the CFCS-14; (c) to compare different
4 0.636 1.104 0.109 0.405 models of the factor structure of the CFCS-14 at the WS and BS levels.
5 0.448 0.805 0.118 0.201 Overall, our results indicate that the Spanish CFCS-14 is reliable and
6 0.352 0.657 0.135 0.124⁎ valid. Reliabilities were high regardless of the conditions of adminis-
7 0.508 0.838 0.120 0.258
tration. Convergent validity analyses showed that both subscales ex-
8 0.624 1.063 0.101 0.390
9 0.472 0.814 0.114 0.222 hibited the expected relationships (Van Beek et al., 2017) with time
10 0.552 0.948 0.105 0.305 perspective subscales. Present time perspective subscales were posi-
11 0.758 1.248 0.089 0.575 tively correlated with CFC-I whereas the ZTPI-future subscale was po-
12 0.397 0.701 0.107 0.157 sitively correlated with CFC-F and negatively correlated with CFC-I. As
13 0.501 0.767 0.123 0.251
in previous studies (Joireman et al., 2008, 2012), we found that spe-
14 0.540 0.868 0.114 0.291
cifying a correlation between the error terms of the ‘new’ items (items
All loadings significant at p < 0.001, except for ⁎p < 0.05. 13 and 14) and two of the original items improved model fit, possibly
due to a method effect because a similar wording and a serial position
Table 4 effect. The new CFC-F items ask directly about the respondent's eva-
Pearson correlations between CFCS-14 and time perspective subscales. luation of future outcomes.
Secondly, we analysed the short-term stability of the CFC-14, using
Past Past Present Present Future Future
responses provided at daily intervals. We found that fit indexes of the
negative positive fatalist hedonist negative
two-factor solution were close to the acceptable cut-off criteria, but
CFC_I 0,26⁎⁎ 0,17⁎⁎ 0,41⁎⁎ 0,22⁎⁎ −0,20⁎⁎ 0,35⁎⁎ weaker if compared to those found in study 1. This can be partly at-
CFC_F 0,06 0,12 − 0,02 0,12 0,34⁎⁎ −0,01 tributed to the smaller sample size and greater complexity of the model.
⁎⁎
Nevertheless, our results offer some interesting insights for CFC re-
p < 0.01.
searchers. The moderate ICCs for the daily data suggest that although
the data were nested within subjects, scores were not stable. For in-
Table 5
stance, in inter-observer agreement, it is generally requested a 0.60 of
Fit indexes for Study 2.
ICC to be good (Cicchetti, 1994). It would be useful to search for si-
Model χ2 Df CFI TLI RMSEA SRMR (WI- tuational variables that could be responsible for these temporal fluc-
BI) tuations and it might also be helpful to take situational characteristics
into account when assessing CFC (Joireman & King, 2016; Rauthmann
1. One factor WP, two 779,23 132 0.616 0.546 0.101 0.178 0.153
BP et al., 2015).
2. Two factors WP, one 446,32 132 0.813 0.779 0.070 0.087 0.335 In both samples we found that the two-correlated-factors model
BP provided the best fit according to several tests, including a parallel
3. Two factors WP, two 341,22 124 0.871 0.838 0.060 0.060 0.182 analysis, size and significance of factor loadings, EFA, CFA and MCFA
BP
with repeated-measures data. Our results resemble those of Joireman
et al. (2012) and McKay et al. (2016). Considering the bifactor model,
in McKay et al. (2016) study and in this sample, some fit indexes were
better compared to those of the two-correlated factor model. However,
the two-factor model appears more robust since all loadings were sig-
nificant and > 0.40 (except for item 6).
This study has limitations. First, our samples differed with respect to

173
A. Vásquez-Echeverría et al. Personality and Individual Differences 123 (2018) 171–175

Fig. 1. Standardized loadings for the two-factor model at


the within person level and two factors at the between
person level. All loadings are significant at p < 0.001.

age and working status. The Uruguayan sample consisted of university Funding
students, whereas the Spanish sample consisted of working adults. In
addition, we relied on self-reported data. Future research should re- This research was supported partially by grants provided by the
plicate these results in more representative samples and would benefit Research Commission of the University of the Republic to AVE, and a
from considering other sources of information. master student scholarship from the National Research Agency of
Uruguay to LAN.
Acknowledgements

We thank Alexas Murnikovas and Daniela Vilaró for their assistance Conflicts of interest
in collecting data. We discussed some ideas reflected on this manuscript
with Jeff Joireman. None.

Appendix A

Item CFCS 14 CFCS 14 (version adapted for daily diary measurement)


nr.

1 Considero cómo las cosas pueden ser en el futuro, y trato de Hoy he considerado cómo las cosas podrían ser en el futuro, y he
influenciar esas cosas con mi comportamiento, día a día. tratado de influenciar esas cosas con mi comportamiento.
2 A menudo me empeño en un comportamiento en particular para Hoy me he empeñado en un comportamiento en particular para
obtener resultados que podrían no suceder por varios años. obtener resultados que podrían no suceder por varios años.
3 Actúo solamente para satisfacer asuntos inmediatos, imaginando que Hoy he actuado solamente para satisfacer deseos inmediatos,
el futuro se ocupará de sí mismo. imaginando que el futuro se ocupará de sí mismo.

174
A. Vásquez-Echeverría et al. Personality and Individual Differences 123 (2018) 171–175

4 Mi comportamiento solamente está influenciado por los resultados Hoy mi comportamiento solo ha estado influenciado por los
inmediatos (por ej., en cuestión de días o semanas) de mis acciones. resultados inmediatos (por ej., en cuestión de días o semanas) de
mis acciones.
5 Me guio por la ley del mínimo esfuerzo. Hoy me he guiado por la ley del mínimo esfuerzo
6 Estoy dispuesto a sacrificar mi felicidad o bienestar inmediato para Hoy he estado dispuesto a sacrificar mi felicidad o bienestar
lograr resultados a futuro. inmediato para lograr resultados en el futuro.
7 Creo que es importante tomar seriamente las advertencias sobre Hoy he creído que es importante tomar seriamente las advertencias
resultados negativos, incluso si esos resultados no van a ocurrir en sobre resultados negativos, incluso si esos resultados no van a
varios años. ocurrir en varios años.
8 Creo que es más importante realizar un comportamiento con Hoy he creído que es más valioso realizar un comportamiento con
consecuencias importantes a largo plazo que un comportamiento con consecuencias importantes a largo plazo que un comportamiento
consecuencias inmediatas menos importantes. con consecuencias inmediatas menos importantes.
9 Generalmente ignoro las advertencias acerca de posibles problemas Hoy he ignorado las advertencias acerca de posibles problemas
futuros porque creo que los problemas se resolverán antes de que futuros porque creo que los problemas se resolverán antes de que
alcancen un nivel de crisis. alcancen un nivel de crisis.
10 Creo que sacrificarse ahora es por lo general innecesario dado que los Hoy he creído que sacrificarse ahora es por lo general innecesario,
resultados futuros se pueden tratar más adelante. dado que los resultados futuros se pueden tratar más adelante.
11 Actúo solamente para satisfacer temas inmediatos, imaginando que Hoy he actuado solamente para satisfacer temas inmediatos,
me ocuparé de futuros problemas que pudieran surgir más adelante. imaginando que me ocuparé de futuros problemas que pudieran
surgir más adelante.
12 Dado que mi trabajo cotidiano tiene resultados específicos, es más Dado que mi trabajo cotidiano tiene resultados específicos, es más
importante para mí que el comportamiento que tiene resultados a importante para mí que el comportamiento que tiene resultados a
largo plazo. largo plazo.
13 Cuando tomo una decisión pienso de qué manera me afectará en el Hoy he tomado decisiones pensando de qué manera me afectará en
futuro. el futuro.
14 En general, mi comportamiento está influenciado por las Hoy mi comportamiento ha estado influenciado por las
consecuencias que tendrá en el futuro. consecuencias que tendrá en el futuro.

References Košťál, J., Klicperová-Baker, M., Lukavská, K., & Lukavský, J. (2016). Short version of the
Zimbardo Time Perspective Inventory (ZTPI–short) with and without the Future-
Negative scale, verified on nationally representative samples. Time & Society, 25(2),
Cicchetti, D. V. (1994). Guidelines, criteria, and rules of thumb for evaluating normed and 169–192.
standardized assessment instruments in psychology. Psychological Assessment, 6, McKay, M. T., Perry, J. L., Percy, A., & Cole, J. C. (2016). Evidence for the reliability and
284–290. validity, but not the practical utility of the two-factor Consideration of Future
Hu, L., & Bentler, P. M. (1999). Cut-off criteria for fit indexes in covariance structure Consequences Scale-14. Personality and Individual Differences, 98, 133–136.
analysis: Conventional criteria versus new alternatives. Structural Equation Modelling, Rauthmann, J. F., Sherman, R. A., & Funder, D. C. (2015). Principles of situation research:
6, 1–55. Towards a better understanding of psychological situations. European Journal of
Janeiro, I. N. (2012). O Inventário de Perspectiva Temporal: Estudo de Validação. Revista Personality, 29, 363–381.
Iberoamericana de Diagnóstico y Evaluación Psicológica. 34. Revista Iberoamericana de Strathman, A., Gleicher, F., Boninger, D. S., & Edwards, C. S. (1994). The consideration of
Diagnóstico y Evaluación Psicológica (pp. 117–133). future consequences: Weighing immediate and distant outcomes of behavior. Journal
Joireman, J., Balliet, D., Sprott, D., Spangenberg, E., & Schultz, J. (2008). Consideration of Personality and Social Psychology, 66, 742–752. http://dx.doi.org/10.1037/0022-
of future consequences, ego-depletion, and self-control: Support for distinguishing 3514.66.4.742.
between CFC-immediate and CFC-future sub-scales. Personality and Individual Van Beek, J., Handgraaf, M. J., & Antonides, G. (2017). Time orientiation effects on
Differences, 45, 15–21. https://doi.org/10.1016/j.paid.2008.02.011. health behaviour. In M. Altman (Ed.). Handbook of behavioural economics and smart
Joireman, J., & King, S. (2016). Individual differences in the consideration of future and decision-making: Rational decision-making within the bounds of reason (pp. 413–428).
(more) immediate consequences: A review and directions for future research. Social Northampton, MA: Edwing Elgar Inc.
and Personality Psychology Compass, 10, 313–326. Vásquez Echeverría, A., Esteves, C., Gomes, C., & Ortuño, V. (2015). Portuguese valida-
Joireman, J., Shaffer, M. J., Balliet, D., & Strathman, A. (2012). Promotion orientation tion of the consideration of future consequences scale. The Spanish Journal of
explains why future-oriented people exercise and eat healthy evidence from the two- Psychology, 18, e7.
factor consideration of future consequences-14 scale. Personality and Social Psychology Zhang, Z., Kong, M., Zhang, L., & Li, Z. (2015). Consideration of future consequences:
Bulletin, 38, 1272–1287. http://dx.doi.org/10.1177/0146167212449362. Preliminary evidences for a four-factor distinction. Personality and Individual
Joireman, J., Strathman, A., & Balliet, D. (2006). Considering future consequences: An Differences, 87, 99–104.
integrative model. In L. Sanna, & E. Chang (Eds.). Judgments over time: The interplay of
thoughts, feelings and behaviours (pp. 82–99). Oxford: OUP.

175

You might also like