Role of Need-Supportive Family Behaviours on Purpose in Life and Depressive Feelings of French Older People : A Self-Determination Theory Perspective

Grounded in a self-determination theory framework, this study aimed to examine the relationships between satisfaction and frustration of basic psychological needs (BPN), need-supportive family behaviours, social support, purpose in life, and depressive feelings among older people living at home. We used a partial least squares path modelling approach among 118 older people living at home (Mage = 74.42 + 8.96). Results showed that: (1) Availability of social support and family support for competence negatively predicted BPN frustration significantly; (2) family support for autonomy and competence positively predicted BPN satisfaction significantly; (3) BPN satisfaction for competence positively predicted purpose in life significantly; and (4) BPN frustration for competence and relatedness positively predicted depressive feelings significantly and negatively predicted purpose in life significantly. This study provided evidence for broadening the concept of social support by taking into account need-supportive family behaviours when one wants to study welland ill-being in older people.


Introduction
Self-determination theory (SDT) [1] is a macrotheory of human motivation, development, well-being, and personality.According to SDT, basic psychological needs (BPN) for autonomy (feeling fully volitional and free to engage in an activity), competence (feeling effective and masterful), and relatedness (feeling connected and secure in relationships with others) are considered psychological nutriments which, when satisfied, allow individuals to grow, thrive, and experience well-being.When these needs are not satisfied, individuals are likely to experience psychological suffering [1].The social environment has an impact on the extent to which BPN are satisfied [1].Previous SDT studies have characterized three types of need-supportive behaviours fostering BPN satisfaction and well-being [2]: (1) Autonomy supportive behaviour provides choice and opportunity for individuals to use the information to solve a problem in their own way; (2) competence supportive behaviour refers to the individuals' social network that supports and enhances their feelings of competence and mastery using positive feedback and acknowledging improvements and the capacity to achieve the goals; and (3) relatedness supportive behaviour refers to the individual's opportunity to develop secure and emotional relationships with others [1].
SDT researchers have also shown that BPN for autonomy, competence, and relatedness can also be frustrated [1].Indeed, a negative experiential state (need thwarting) can occur when individuals perceive their BPN to be actively undermined via interactions with significant others [3].This experience of need thwarting can promote an individual's vulnerability in developing a personal experience of need frustration for autonomy (feelings of being externally and internally controlled and pressured), competence (doubts about one's efficacy, feelings of failure and inferiority), and relatedness (experienced relational exclusion and loneliness) [4].Indeed, low scores of BPN satisfaction have negatively predicted well-being, whereas BPN frustration has positively predicted ill-being [3,5].The distinction between BPN satisfaction and frustration is of primary importance.For example, an older adult may feel low relatedness to family members and may thus have less vitality.However, one can also be actively rejected by family members.This means that the lack of fulfilment of the BPN and the experience of BPN frustration should be clearly distinguished in both research and practice [5].Although BPN satisfaction and frustration are generally weakly but significantly negatively correlated, the relationship between the lack of fulfilment of the BPN and the experience of BPN frustration seems asymmetrical [5].Low BPN satisfaction does not necessarily involve BPN frustration, whereas BPN frustration by definition involves low BPN satisfaction.In consequence, BPN satisfaction and frustration can co-occur in a given context but differentially contribute to the prediction of specific outcomes.
The present study focuses on older adults living at home.Social networks are considered to be an important determinant of well-being of older people in helping them to stay healthy and cope with stressful environments or difficult life experiences [6].Nevertheless, they decrease with advanced age except within family relationships, which is the main source of social support at old age [7,8].Social support generally refers to various forms of aid and assistance given by network members [9].Several approaches to measure social support have been put forward, and a distinction has been made between its structural (quantity) and functional (quality) aspects [10].A large body of data exists linking these different aspects of social networks with well-being in old age, and most studies have concluded that quality (rather than quantity) of older adults' networks has had a strong influence on well-being [8].Another distinction has also been made between received and perceived social support [11].Perceived social support refers to an individual's belief that social support is available and that it provides what the individual considers necessary [12].This notion of perceived social support implies that only one person can evaluate the quality of his/her relations with others.Research has recognized that perceived support has provided some insight when evaluating the association between social support and well-being [8,10].Whatever the approaches, literature on older people agrees that social support can have not only a beneficial influence on purpose in life but also has the capacity to reduce the risks for geriatric syndromes, such as depressive feelings [13].
Although social support is a variable that is strongly studied in well-being literature, the SDT approach could provide new insights into the role of supportive behaviours on well-and ill-being outcomes.Deci and Ryan [1] have argued that supportive behaviours had a functional significance determined by the perceived relation between the offering of support and one's psychological needs.As such, the aim of this study was to examine the effects of need-supportive family behaviours and two distinct dimensions of perceived social support (availability and satisfaction of social support) on BPN satisfaction and frustration, and in turn on purpose in life and depressive feelings identified as indicators of psychological well-and ill-being in a sample of older adults (see Figure 1 for an overview of the tested model).Purpose in life has been identified as a core component of psychological well-being in older people [14] and is widely recognised as an integral component of eudaimonia [15].Eudaimonia is a concept focusing on using and developing the best in oneself [16] and it integrates a broader vision of well-being than a hedonic approach, which focuses on the quest for pleasure, satisfaction, and subjective happiness [17].Otherwise, some SDT research has used depressive symptoms as indicators of ill-being among older people [14].Based on SDT postulates [1] and previous studies [2,5], we hypothesised that: (1) Availability and satisfaction of social support, and family support for autonomy, competence, and relatedness would positively predict BPN satisfaction and negatively predict BPN frustration; (2) BPN satisfaction would positively predict purpose in life and negatively predict depressive feelings; and (3) BPN frustration would positively predict depressive feelings and negatively predict purpose in life.

Participants
A total of 118 French older adults living at home (80 women, Mage = 74.42,SD = 8.96, age range: 60-95) voluntarily participated in the study.Among them, 61% were still married, 25% were widows or widowers, 13% were divorced/separated, and 1% was single.With respect to participants' education level, 20% of participants had a certificate of elementary education, 52% had a high school education, and 28% were university graduates.They had few chronic illnesses (M = 0.39, SD = 0.49).Hence, participants of the present study were relatively healthy for old age.

Measures
The short form of the social support questionnaire [2] was used to assess availability and satisfaction of social support.The French version of this self-report questionnaire [18] is composed of 6 items.Each item requires a two-part answer.The first part measures the availability of social support; participants reported the number of perceived social support sources from none to nine individuals.The second part assesses the satisfaction of social support; respondents reported how satisfied they are with the received support on a 6-point Likert scale ranging from 1 (very dissatisfied) to 6 (very satisfied).In the present study, Cronbach's alphas were 0.93 and 0.92 for availability and satisfaction of social support.
An adaptation for older adults of the interpersonal behaviours scale [19] was used to measure perceived family support for BPN.It is composed of three subscales assessing family support for autonomy (4 items), competence (4 items), and relatedness ( 4

Participants
A total of 118 French older adults living at home (80 women, Mage = 74.42,SD = 8.96, age range: 60-95) voluntarily participated in the study.Among them, 61% were still married, 25% were widows or widowers, 13% were divorced/separated, and 1% was single.With respect to participants' education level, 20% of participants had a certificate of elementary education, 52% had a high school education, and 28% were university graduates.They had few chronic illnesses (M = 0.39, SD = 0.49).Hence, participants of the present study were relatively healthy for old age.

Measures
The short form of the social support questionnaire [2] was used to assess availability and satisfaction of social support.The French version of this self-report questionnaire [18] is composed of 6 items.Each item requires a two-part answer.The first part measures the availability of social support; participants reported the number of perceived social support sources from none to nine individuals.The second part assesses the satisfaction of social support; respondents reported how satisfied they are with the received support on a 6-point Likert scale ranging from 1 (very dissatisfied) to 6 (very satisfied).In the present study, Cronbach's alphas were 0.93 and 0.92 for availability and satisfaction of social support.
An adaptation for older adults of the interpersonal behaviours scale [19] was used to measure perceived family support for BPN.It is composed of three subscales assessing family support for autonomy (4 items), competence (4 items), and relatedness (4 items).Participants responded on a 5-point Likert scale ranging from 1 (never) to 5 (always).Cronbach's alphas were of 0.78, 0.69 and 0.68 for family support for autonomy, competence, and relatedness.
The psychological need frustration scale for older adults [20,21] was used to assess BPN frustration for autonomy, competence, and relatedness.The French version of this questionnaire is composed of three four-item subscales [20].Responses were reported on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree).Cronbach's alphas of BPN frustration for autonomy, competence, and relatedness were 0.66, 0.72, and 0.72.
Following the suggestions of previous studies [14,20,21], three different French questionnaires were used to assess satisfaction of BPN for autonomy, competence, and relatedness.Responses were reported on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree).To assess BPN satisfaction for autonomy, five items collated by Standage, Duda, and Ntoumanis [22] were used.BPN satisfaction for competence was assessed using five items from the perceived competence subscale of the intrinsic motivation inventory [23].Finally, BPN satisfaction for relatedness was assessed using the six items from the need for relatedness scale [24].All the items were preceded by the phrase "In my daily life . . .". Cronbach's alphas of BPN satisfaction for autonomy, competence, and relatedness were 0.64, 0.78, and 0.83.
The French version [25] of the 15-item geriatric depression scale was used to assess depressive feelings.It is a self-report measure of specific manifestations of depression in older people.Participants responded in a "Yes/No" format.A sum of scores between 0 and 15 was computed with higher values indicating more depressive feelings.Scores of 0-4 are considered normal; 5-8 indicates mild depression; 9-11 indicates moderate depression; and 12-15 indicates severe depression [25].Cronbach's alpha of depressive symptoms was 0.74.
The French version [26] of the psychological well-being scale was used to assess purpose in life.The purpose in life scale is composed of 14 items and indicates whether the respondent has goals and an orientation toward the future and gives meaning to its present and its past.Participants were asked to rate their agreement with each statement using a 6-point Likert scale ranging from 1 (totally disagree) to 6 (totally agree).Cronbach's alpha was 0.79 for purpose in life.

Procedure
This study was conducted in accordance with the Declaration of Helsinki 1975 revised in 2013 and was approved by the University Ethics Committee.Written informed consent was obtained prior to data collection, and anonymity was ensured by attributing an identification number to each participant.Data were collected using a convenience sample.Hence, the recruitment of participants took place based on simple volunteering.Questionnaires were administered during individual sessions or in groups of 5 participants maximum by one of the research team members.Once the questionnaires were completed and returned by the participants, we verified that each item was well filled and our sincere thanks were addressed to them.

Data Analysis
A partial least squares path modelling approach (PLS-PM) was used in the present study.PLS-PM is a variance-based structural equation modelling technique.We chose this statistical tool because it is not constrained by distributional assumptions (significant deviations from normality were observed on several variables, Table 1) and can be used with relatively small sample sizes.In PLS-PM, each causal subsystem sequence of paths is estimated separately.As such, a rule of thumb suggests that the sample size in PLS-PM be equal to ten times: (1) The number of indicators of the scale with the largest number of manifest indicators, or (2) the largest number of structural paths directed at a particular construct in the inner path model [27].Thus, the sample size of the present study was acceptable.PLS-PM analyses were conducted with the R package labelled PLS-PM [28].A bootstrap procedure with 100 replications allowed to compute 95% bias-corrected percentile confidence intervals for assessing the significance of the parameter estimates.The items of availability and satisfaction of social support, perceived family support for BPNs, and satisfaction and frustration of BPNs were used as manifest variables.In order to retain a reasonable number of manifest variables for the last two constructs, depressive feelings (15 items) and purpose in life (14 items) were measured by three random aggregates of items.A two-step modelling approach was used to better identify the sources of poor overall model fit [27].The first step (outer model) allowed us to focus on the factor structure underlying the items and/or parcels of each construct.This analysis also allowed us to examine the psychometric properties of each of the constructs.Standardised factor loadings, composite reliability values (ρ), average variance extracted (AVE) values and an eigenvalue analysis of the correlation matrix of each set of manifest variables were used to assess the quality of the measurement model [27][28][29].An AVE value of 0.50 or greater and a ρ value of 0.70 or greater indicated acceptable reliability of the scores [28].The first eigenvalue should be larger than 1, whereas the second eigenvalue should be smaller than 1 [28].Then, the second step (inner model) consisted of simultaneously testing the structural and measurement models and allowed us to focus on conceptual connections among the latent factors [27].

Descriptive Statistics
Descriptive statistics and the correlation matrix are shown in Table 1.Based on previous research [20,21] or on cut-points for the geriatric depression scale [25], participants reported: (a) Low levels of depressive feelings and BPN frustration; (b) moderate scores of availability and satisfaction of social support; and (c) high scores of family support and BPN satisfaction for autonomy, competence and relatedness, and purpose in life.The correlations among the study variables did not indicate multicollinearity, as they ranged from −0.52 to 0.63.

Partial Least Squares-Path Modelling
The results of the PLS-PM measurement model (outer model) provided evidence for the reliability and validity of all the constructs examined in the present study, as indicated by the loadings, ρ values, AVE values, and first and second eigenvalues (Table 2).The results from the structural (inner) model showed that: (a) Satisfaction of social support and family support for competence negatively predicted BPN frustration for competence (β = −0.41 and −0.21, respectively) and relatedness (β = −0.21and −0.41) significantly; (b) family support for competence positively predicted BPN satisfaction for competence (β = 0.21), relatedness (β = 0.22) and competence (β = 0.32) significantly and negatively predicted BPN frustration for autonomy (β = −0.31)significantly; (c) family support for autonomy positively predicted BPN satisfaction for autonomy (β = 0.27) and relatedness (β = 0.24) significantly; (d) BPN frustration for relatedness and competence positively predicted depressive feelings (β = 0.26 and 0.29 respectively) significantly and negatively predicted purpose in life (β = 0.22 and 0.36) significantly; and (e) BPN satisfaction for competence positively predicted purpose in life (β = 0.19) significantly (Table 3).

Discussion
The present study examined the relationships between family support for BPNs, perceived social support (availability and satisfaction of social support), BPN satisfaction and frustration, purpose in life, and depressive feelings.Perceived availability of social support (quantitative aspect of the network) did not come into play in the prediction of BPN satisfaction and frustration.This result is interesting insofar as this study takes into account both networks' quantity and quality (satisfaction of social support) and that it is the participants themselves who identified the members of their network they considered to be important.Satisfaction (and not availability) of social support negatively predicted BPN frustration for competence and relatedness.
As a whole, participants reported high scores on family support for autonomy, competence, and relatedness, emphasising a positive perception of social support from their family.Although Fiori, Antonucci, and Cortina [30] indicated that high-quality relationships were often associated with positive outcomes, no relationship was found in the present study between family support for relatedness and BPN satisfaction.Support for relatedness refers to the degree of network involvement and to the way network communicates its involvement in terms of warmth, responsiveness, security, and emotional support.One possible explanation of this unexpected result could be that participants considered it normal that they were connected with their family.Regarding the two other family supports, results showed that family support for autonomy positively predicted BPN satisfaction for autonomy and relatedness, whereas family support for competence positively predicted BPN satisfaction for autonomy, competence, and relatedness and negatively predicted BPN frustration for autonomy, competence, and relatedness.As a whole, such findings are promising because they suggest that the nature of the support (autonomy, competence, relatedness support) could differently predict BPN satisfaction and/or frustration, and in turn purpose in life and depressive feelings.
PLS-PM findings also showed that BPN satisfaction for competence was especially salient among older people living at home, as it was the strongest predictor of purpose in life, which is considered an important resource for ageing well [31].As indicated by previous research [32], BPN satisfaction for competence is not only related to the individual's history and beliefs in capabilities, but also to perceived aspects of the social environment.However, PLS-PM results also showed that BPN frustration for competence and relatedness positively predicted depressive feelings.When the environment undermines these needs and individuals perceive that BPN for competence and relatedness are frustrated, depressive symptoms can emerge.BPN satisfaction and frustration can be considered distinct constructs promoting different outcomes and are dynamic processes which are highly sensitive to environment [1,3].
Otherwise, the present PLS-PM findings are in the same line that family literature [33,34].In particular, our results are consistent with other key family influences across the life span, such as parenting socialisation and its developmental consequences in childhood, adolescence, and adulthood [35,36].In this sense, support of parents in the family context (parental acceptance, involvement and promotion of psychological autonomy) is positively related to a child's sense of competence and child optimal development, as suggest studies analysing short-term [33,35,37] and long-term socialisation outcomes [34][35][36].
Some limitations of the present study should be addressed.First, this study is cross-sectional.However, BPN satisfaction fluctuates on a daily basis, suggesting that the perceptions of social support from family among older people could also fluctuate across the course of time [38].Hence, a longitudinal study is needed to further explore the relationships between need-supportive behaviours, BPN satisfaction and frustration, purpose in life, and depressive feelings.Second, as both positive and negative social exchanges exist in family networks, the use of a scale assessing need-supportive and need-thwarting family support might provide additional information on potential pressure.Third, the fact that the sample size is rather small and from a volunteer sample limits the generalisability of the findings to a specific segment of the older people population.For instance, participants of this study were mainly older women who declared themselves relatively healthy.Because women are considered to be more embedded in their network support than men, it is possible that this led to a response bias.Moreover, because perceived social support could differ based on living alone versus with someone, future studies involving a larger sample size could examine this issue in more depth.
In conclusion, SDT gives another overview of how supportive behaviours can influence particular indicators of well-and ill-being, such as purpose in life and depressive feelings.From a practical point of view, the concept of need-supportive behaviours for autonomy, competence, and relatedness provides a way of understanding why some social behaviours can promote well-being whereas others do not [1].Given the loss or disruption of family and friend relationships in older people in later life, this topic is particularly relevant among older adults.This study suggests that there is a need to broaden the concept of social support by taking into account the relationships between the offering of support and older people's BPN, when one wants to study well-and ill-being in the older people.

Figure 1 .
Figure 1.Hypothesised structural model of the partial least square path modelling.

Figure 1 .
Figure 1.Hypothesised structural model of the partial least square path modelling.

Table 1 .
Descriptive statistics and correlation matrix.

Table 2 .
Psychometric properties of the outer model of the partial least square path modelling.

Table 3 .
Inner model of the partial least square path modelling.For simplicity, the nonsignificant paths are omitted from Table 3 (all the structural results are available upon request from the first author).