Despite the centrality of time to the aging process (Hendricks, Reference Hendricks, McFadden and Atchley2001; Schroots & Birren, Reference Schroots, Birren, Birren and Warner Schaie1990), little coordinated empirical research has been conducted in this area. Isolated archipelagos of conceptual knowledge from antiquity to the present exist, yet systematic methodological, theoretical, and empirical foundations are lacking in gerontology. This is an important and pressing omission given the integral nature of time in developmental processes, as well as the increasing proportion of elderly adults worldwide. Indeed, as McFadden and Atchley (Reference McFadden, Atchley, McFadden and Atchley2001) observed, “[a]ll gerontologists must take time seriously, and yet this difficult construct is rarely addressed by those who would understand age and aging” (preface).

Various facets of time, from a non-developmental perspective (e.g., perceived duration, temporal orientation, subjective judgements of affective correlates, extension, and density) have been investigated, including the concept of time perspective (e.g., Boniwell & Zimbardo, Reference Boniwell, Zimbardo, Linley and Joseph2004). Time perspective refers to a relatively stable individual difference variable in which individuals express a preferred mode of temporal focus (i.e., past, present, or future). An important trend in this area is the notion of a balanced time perspective, in which orientations to more than one temporal focus are flexibly employed to potentially enhance well-being and facilitate problem-solving. However, this is a nascent research area. For example, a Psychology Abstracts database search using the keywords balanced time perspective produced only nine results. When the keyword aging was introduced, the resultant search produced zero articles.

It has been suggested that time perspective in general (e.g., Carstensen, Reference Carstensen2006) and a balanced time perspective in particular (e.g., Boniwell, Osin, Linley, & Ivanchencko, Reference Boniwell, Osin, Linley and Ivanchencko2010) play a powerful role in many psychological outcomes, including those related to motivation, mental health, interpersonal relationships, sense of self, and addictive behaviors, among many others. However, according to literature reviews (e.g., Boniwell, Reference Boniwell and Lopez2009; Kazakina, Reference Kazakina1999), early research was hampered by lack of psychometrically sound measures and by a focus on a single temporal orientation, typically the future. Moreover, investigation of age differences has not been the focus of most research in this area. Finally, as Boniwell et al. (Reference Boniwell, Osin, Linley and Ivanchencko2010) noted, “One of the unanswered questions with regard to time perspective concerns the relationship between different temporal orientation profiles with well-being” (p. 26).

Like Janus, the Roman god of portals, most of us at least occasionally reflect on our past and imagine our future. There is a large literature on reminiscence that suggests that focusing on the past can have negative implications, such as rumination (Nolen-Hoeksema, Reference Nolen-Hoeksema1998), bitterness revival (Webster, Reference Webster1993), intrusive thoughts as in trauma recollections (Tedeschi & Calhoun, Reference Tedeschi and Calhoun2004), and escapism (Watt & Wong, Reference Watt and Wong1991). These types of processes have been associated with increased anxiety, depression, neuroticism, and hopelessness (Webster, Bohlmeijer, & Westerhof, Reference Webster, Bohlmeijer and Westerhof2010).

Similarly, anticipating the future can also involve maladaptive outcomes. For instance, imagining the type of person we most fear becoming, or feared selves (Markus & Nurius, Reference Markus and Nurius1986), can trigger an array of anxiety-provoking thoughts, as can anticipating performance inadequacies, existential concerns about life’s meaning, death anxiety, and endings which Kennedy, Fung, and Carstensen (Reference Kennedy, Fung, Carstensen, McFadden and Atchley2001) stated can provoke “dread and sadness” (p. 51).

In contrast, there is also good evidence that focusing on the past and future can have very positive outcomes as well. We know from reminiscence and life review research that accessing positive memories of our past can enhance happiness, increase self-esteem, bolster our sense of purpose and meaning in life, and potentially contribute to longevity (see Webster et al., Reference Webster, Bohlmeijer and Westerhof2010, for review). For example, Stones, Rattenbury, and Kozma (Reference Stones, Rattenbury, Kozma, Haight and Webster1995) found decreased mortality rates at an 18-month follow-up for a community-based reminiscence group. Similarly, Zimbardo and Boyd (Reference Zimbardo and Boyd1999) have found that a future time perspective can be associated with many adaptive outcomes, including optimism, internal locus of control, achievement orientation, and delay of gratification, among others. As with the past, there is some evidence that a future time perspective also contributes to significantly increased longevity (e.g., Fry & Debats, Reference Fry and Debats2011). Finally, Carstensen (Reference Carstensen2006) and colleagues’ (e.g., Charles & Carstensen, Reference Charles, Carstensen, Reis and Sprecher2009; Lang & Carstensen, Reference Lang and Carstensen2002) well-known socio-emotional selectivity theory outlines the complex relationship among emotional, motivational, and future orientation variables. Findings derived from this theory generally indicate that an open-ended future has implications for both informational and emotional goal pursuits, and that, on average, younger adults score higher on future time perspective (FTP) relative to older adults. On the other hand, older adults, who are less future oriented, are motivated to pursue emotional satisfaction.

Given that such psychological benefits occur for a positive past and positive future orientation separately (at least when the focus is on affective quality), the question arises as to whether a joint, or balanced, time perspective would result in even stronger outcomes. As noted, less than a handful of studies have investigated this question and have used instruments not originally intended to measure a balanced time perspective. The most prominent of these measures is the Zimbardo Time Perspective Inventory (ZTPI; Zimbardo & Boyd, Reference Zimbardo and Boyd1999). Researchers have employed various exploratory coding schemes, including a 33rd and 66th percentile cut-off range (Zhang, Howell, & Stolarski Reference Zhang, Howell and Stolarski2013), a hierarchical clustering technique (Boniwell et al., Reference Boniwell, Osin, Linley and Ivanchencko2010), and deviation from supposed optimal subscale scores (Stolarski, Bitner, & Zimbardo, Reference Stolarski, Bitner and Zimbardo2011). The latter approach assumes there is an optimal score for each of the five subscales of the ZTPI. Obtained scores for each participant are subtracted from the optimal score for each subscale, squared to eliminate negative values, and then the five subscale results are summed. Finally, the square root of this total score is calculated to return to the original metric.

Each of these approaches has advantages and disadvantages, and currently opinions concerning the best scoring approach remain open to further empirical evaluation. In our study discussed here, we have replicated the original cut-off approach reported by Webster (Reference Webster2011), but additionally, as suggested by a reviewer, for this article we also report results based upon the deviation from optimal subscale scores method as described by Stolarski et al. (Reference Stolarski, Bitner and Zimbardo2011).^{Footnote 1}

Recently, Webster (Reference Webster2011) developed the Balanced Time Perspective Scale (BTPS) which assesses a person’s positive, subjective evaluation of both their past and future on a 28-item scale. Webster (Reference Webster2011) purposely omitted a present-orientation subscale, citing philosophical, conceptual, and pragmatic reasons for doing so. For instance, he noted that both remembering the past and imagining the future rely on similar brain regions (see Webster, Reference Webster2011, for further explication of similar issues). Results from this initial scale construction indicated that the BTPS had very good to excellent psychometric properties, including internal consistency, convergent validity (with the Zimbardo Time Perspective Inventory), discriminant validity in regard to a measure of social desirability, and construct validity as regards measures of happiness and satisfaction with life. Webster (Reference Webster2011) suggested that important areas of future research include replication of factor structure and well-being results, as well as the inclusion of middle-aged and older adults.

The purpose of our present study was to address all three of these recommendations. In order to do so, we conducted a direct replication, an important and frequently neglected aspect of psychological science (Roediger, Reference Roediger2012). We extended the replication results by including younger, middle-aged, and older adults as participants, as well as evaluating a new scoring method as already described. We made seven predictions; prediction 1 was that the factor structure of the original BTPS would be replicated; prediction 2 was that the relationship between the BTPS and the two well-being measures would be replicated as well. Further, based on previous studies (e.g., Diener, Suh, Lucas, & Smith, Reference Diener, Suh, Lucas and Smith1999) in which mean levels of life satisfaction stayed the same or increased in old age, prediction 3 was that older adults would report the same or higher scores on the two well-being measures than younger adults and middle-aged adults.

In terms of the relationship between age and BTPS, we relied on recent conference findings (Webster, Bohlmeijer, & Westerhof, Reference Webster, Bohlmeijer and Westerhof2011), given the lack of published data, to make several additional predictions. In prediction 4, the percentage of younger adults would be highest in the BTPS futurists category (a description of the four possible BTPS categories follows in the Results section) and lowest in the reminiscers category; in prediction 5, older adults would be lowest in the futurists category and highest in the reminiscers category; in prediction 6, middle-aged adults would be highest in the time-expansive category. Finally, in prediction 7 a balanced time perspective (as the Results section explains, we measured this using the deviation from optimal subscale scores method) would predict both happiness and life satisfaction scores above and beyond demographic variables in a regression analysis.

Results

First, we report the replication of the factor structure and well-being results reported by Webster (Reference Webster2011) using the same criteria as the original study. Next, we report on the age differences and their potential implications for theory and empirical work. Finally, we report the regression analysis results using the deviation from optimal subscale score method.

The 28 items of the BTPS were submitted to principal components analysis (PCA). The number of factors was set a priori at two, and varimax rotation was employed to facilitate the interpretation of factors. Results, supporting prediction 1, strongly replicate our initial findings, and can be seen in Table 1. All Past items loaded on a single factor and had relatively low cross-loadings. Loadings ranged from .60 to .82 (M _Loading = .72). Similarly, all Future items loaded on a single factor with low cross-loadings. Loadings ranged from .55 to .82 (M _Loading = .75).

Table 1: Factor loadings, descriptive statistics, and communalities for the Balanced Time Perspective Scale^a

^ah ² = extracted communalities; factor loadings are bolded.

M = mean

SD = standard deviation

Following Webster (Reference Webster2011), we performed a median split on both the past and future orientation subscales and then used this cut-off to produce four categories: (a) persons scoring below the median on both the past and future were termed time restrictive; (b) persons scoring below the median on the future but above the median on the past were termed reminiscers; (c) persons scoring above the median on the future but below the median on the past were termed futurists; and (d) persons scoring above the median on both the future and the past were termed time expansive.

We performed multivariate tests using BTPS categories and age groups as independent variables and Satisfaction With Life Scale and Oxford Happiness Questionnaire as dependent variables. For the Satisfaction With Life Scale, results demonstrated a significant main effect for BTPS category (supporting prediction 2), F(3, 217) = 8.75, p <.001, partial η ² = .11 and age group, F(2, 217) = 3.52, p < .05, partial η ² = .03. The BTPS by age group interaction was not significant. Post hoc analysis for the BTPS category using Tukey HSD showed that the time-expansive category (M = 27.91, SD = 4.50) was significantly higher than both the time-restrictive (M = 23.05, SD = 6.83) and the futurists (M = 23.80, SD = 6.73) categories. No other categories were significantly different. For the age group, no subgroups were significantly different (supporting prediction 3).

For the Oxford Happiness Questionnaire, the study results demonstrated a main effect of the BTPS category, F(3, 225) = 12.25, p < .001, partial η ² = .15 and age group, F(2, 217) = 7.32, p < .01, partial η ² = .06. The BTPS by age group interaction was not significant. Post hoc analysis for the BTPS category using Tukey HSD showed that the time-expansive category (M = 36.37, SD = 5.00) was significantly higher than both the time-restrictive (M = 30.48, SD = 6.26) and the futurists (M = 33.03, SD = 5.26) categories. The reminiscers category (M = 34.79, SD = 6.00) was significantly higher than the time-restrictive category. No other categories were significantly different (supporting prediction 2). Post hoc analysis for the age group revealed that the older age group (M = 35.86, SD = 5.75) was significantly higher in happiness than both the younger (M = 32.75, SD = 6.77) and mid-life (M = 31.81, SD = 5.01) age groups (supporting prediction 3). No other subgroups were significantly different.

Next, we examined the relationship between age and BTPS category. Figure 1 shows the relative percentage of persons in each age group falling within the four BTPS categories. The chi-square test of independence was significant: χ² (6, n = 228) = 50.41, p < .001, Cramér’s V = .33, p < .001. Standardized adjusted residuals indicated that the percentage of younger adults was less, and the percentage of middle-aged adults was greater, than expected in the time-restrictive category; the percentage of younger adults was less, and the percentage of older adults was greater, than expected in the reminiscers category (supporting prediction 4 and 5); the percentage of younger adults was greater, and the percentage of older adults was less (in fact, no older adults fell in this category), than expected in the futurists category (supporting prediction 4 and 5); and the percentage of middle-aged adults was less than expected in the time-expansive category, therefore failing to support prediction 6.

Figure 1: Percentage of young, middle-aged, and older adults within BTPS categories

Finally, we constructed a balanced time perspective score following suggestions by Stolarski et al. (Reference Stolarski, Bitner and Zimbardo2011). Specifically, the optimal score for each item of both the past and future orientation subscales of the BTPS is the maximum score available (i.e., 6) indicating strong agreement with each BTPS item. Recall that each item expresses a positive orientation to either the past or future, so the maximum score of six is theoretically optimal. Given this, the formula for calculating a deviation from the optimal score is (6 – M _Past) + (6 – M _Future). Note that the squaring/square root procedures used in calculating ZTPI scores are not required here as it is not possible to obtain a negative value because the optimal score is also the maximum possible score (i.e., 6).^{Footnote 2} We then used this balanced time perspective score in regression analyses, as reported in Tables 2 and 3.

Table 2: Hierarchic regression on happiness

** = p < .001

BTP = deviation from balanced time perspective score

F = Fisher’s F ratio

R = multiple correlation

R ² = multiple correlation squared

Table 3: Hierarchic regression on Satisfaction With Life Scale

** = p < .001

BTP = deviation from balanced time perspective score

F = Fisher’s F ratio

R = multiple correlation

R ² = multiple correlation squared

As Table 2 shows, a balanced time perspective explains an additional 8.8 per cent of the variance in happiness scores after accounting for the demographic variables of gender, age, education, and health. Similarly, as can be seen in Table 3, a balanced time perspective explains an additional 7.4 per cent of the variance in satisfaction with life scores after accounting for the same set of demographic variables (supporting prediction 7).