Measuring basic needs satisfaction: Evaluating previous research and conducting new psychometric evaluations of the Basic Needs Satisfaction in General Scale

Abstract

Self-Determination Theory specifies the existence of three basic psychological needs: autonomy, competence, and relatedness. The current set of studies (a) provides a narrative review of past research on the Basic Needs Satisfaction in General Scale, (b) examines its dimensionality which has been assumed but not empirically studied, and (c) gathers external validity evidence. Confirmatory factor analysis was used to test the existence of a one- and a three-factor solution; neither model fit the data. After patterns of misfit were examined across three independent samples, a reduced, 16-item three-factor model with a negative-worded method effect was championed. External validity evidence, collected by examining the differential relationships between the three needs and measures of well-being and worry, supported the distinctiveness of the three needs. Although the results are promising, future research is needed to examine the generalizability of the psychometric properties of the modified scale.

Introduction

During the past century the theory that humans have basic needs has been developed and expanded by several different theorists. For example, some researchers have theorized needs are innate in humans (e.g., Deci and Ryan, 2000, Hull, 1943), whereas other researchers have theorized needs are learned over time (e.g., McClelland, 1965, Murray, 1938). Furthermore, researchers have differed in regards to what constitutes a need. Some researchers have proposed needs are psychological in nature (e.g., dominance; Murray, 1938), whereas other researchers have proposed needs are physiological in nature (e.g., food; Hull, 1943). In contrast, some theorists have proposed needs are a combination of the both (e.g., Maslow, 1970). Self-Determination Theory (SDT) defines needs as innate, psychological, and essential for well-being (Deci & Ryan, 2000).

SDT postulates the existence of three basic needs: autonomy, competence, and relatedness.¹ Autonomy refers to the need to feel that one’s behavior and resulting outcomes are self-determined, or self-caused, as opposed to being influenced or controlled by outside forces (deCharms, 1968, Deci and Ryan, 1985, Deci and Ryan, 2000). Competence refers to the need to feel effective and capable of performing tasks at varying levels of difficulty (Harter, 1978, Ryan and Deci, 2002, White, 1959). Relatedness refers to the need to feel connected to, supported by, or cared for by other people (Baumeister and Leary, 1995, Ryan and Deci, 2002). SDT stipulates all three needs must be fulfilled for psychological well-being to occur (Deci & Ryan, 2000). That is, if only one or two of the three needs are fulfilled psychological health will suffer (Deci and Ryan, 2000, Ryan, 1995).

In addition to being important for psychological well-being, according to the sub-theories of SDT, Cognitive Evaluation Theory (CET) and Organismic Integration Theory (OIT), needs satisfaction is also crucial for psychological growth (Deci and Ryan, 2000, Ryan and Deci, 2000a, Ryan and Deci, 2002). According to CET, the fulfillment of basic needs (autonomy and competence in particular) has a direct, positive influence on intrinsic motivation (i.e., performing an activity for inherent interests and pleasure; Deci and Ryan, 1985, Ryan and Deci, 2000a, Ryan and Deci, 2002). That is, feelings of being controlled by external forces or being ineffective regarding the task at hand will undermine levels of intrinsic motivation and will result in being controlled by external criteria, such as a pay check or a superior’s approval (Deci and Ryan, 1985, Deci and Ryan, 2000, Ryan and Deci, 2000a). Moreover, according to OIT, psychological growth (becoming more autonomous in behaviors performed), aids in the creation of a “unified sense of self”, or a person who successfully interacts with the social environment and others (Ryan & Deci, 2002, p. 5). Specifically, growth occurs with the integration and internalization of ideas or behaviors formerly motivated by extrinsic forces, or the acceptance and valuing of behaviors initially performed for external reasons (Ryan, 1995, Ryan and Deci, 2002). However, according to OIT, this will only occur when the needs of autonomy, competence, and relatedness are met (Ryan and Deci, 2000b, Ryan and Deci, 2002).

Satisfaction of the basic needs for autonomy, competence, and relatedness has been positively related to well-being (e.g., Reis et al., 2000, Sheldon and Niemiec, 2006), satisfaction with life (e.g., Meyer, Enstrom, Harstveit, Bowles, & Beevers, 2007), aspirations (e.g., Niemiec, Ryan, & Deci, 2009), and self-esteem (e.g., Thøgersen-Ntoumani & Ntoumanis, 2007) and has been negatively related to depression (e.g., Wei, Philip, Shaffer, Young, & Zakalik, 2005), and anxiety (e.g., Deci et al., 2001).

Clearly, there has been an extensive amount of research examining needs fulfillment (Deci & Ryan, 2000). However, less attention has been paid to how needs satisfaction is being measured. The satisfaction of needs has been measured using various methods (e.g., self-report measures and diary studies) and in various settings. The satisfaction of basic needs has been primarily measured in context-specific settings such as work (e.g., Deci et al., 2001) and relationships (e.g., La Guardia, Ryan, Couchman, & Deci, 2000). Recently a measure was created to assess basic needs satisfaction in general as opposed to a specific context: the Basic Needs Satisfaction in General Scale (BNSG-S²; Gagné, 2003).

The BNSG-S has been used in several studies to assess general needs satisfaction (Conroy and Coatsworth, 2007a, Conroy and Coatsworth, 2007b, Gagné, 2003, Kashdan et al., 2006, Kashdan et al., 2009, Meyer et al., 2007, Neff, 2003, Niemiec et al., 2009, Thøgersen-Ntoumani and Ntoumanis, 2007, Vansteenkiste et al., 2006, Wei et al., 2005). Given its use, one may assume the BNSG-S has been extensively studied. Interestingly, there has not been any rigorous study of the psychometric properties of the scale. That is, there are no known factor analytic studies of the BNSG-S (E. Deci, personal communication, July 1, 2008; M. Vansteenkiste, personal communication, July 3, 2008). Unfortunately, this makes it difficult to make valid inferences from the scale as it is unknown what the BNSG-S is truly assessing. Thus, studies examining the psychometric properties of this measure are needed.

In order to evaluate the BNSG-S, the framework for validity explained by Benson (1998) will be used to organize existing information about the scale. Benson (1998) outlined three stages required to establish a strong program of construct validity: substantive, structural, and external (see also Benson & Hagtvet, 1996). The first stage (substantive) refers to the process of defining the theoretical and empirical domains of the construct of interest. Specifically, the theoretical domain is representative of all information known about the construct; it is the “scientific theory surrounding the construct” (Benson, 1998, p. 12). The empirical domain is representative of all observed variables that are used to represent the construct. Thus, the theoretical domain should be broadly defined, whereas the empirical domain is more specific in nature but a function of the theoretical domain. The second stage (structural) involves examining the interrelationships between the observed variables (e.g., the items). Correlations, internal consistency, and factor analysis are often assessed at this step. The third stage (external) involves examining if the construct is related to external constructs in theoretically expected ways. This is considered the most essential stage because it provides information regarding the nomological network of the construct. Using Benson’s strong program of construct validation as a framework, the BNSG-S is evaluated below.

The BNSG-S was adapted from the Basic Needs Satisfaction at Work Scale (BNSW-S²) to measure needs satisfaction in a general domain as opposed to the context-specific domain of work (Gagné, 2003). The items from the BNSW-S were slightly modified to measure needs satisfaction in a general context. For example, the BNSW-S item “Most days I feel a sense of accomplishment from working”, which was written to measure the satisfaction of the need for competence at work, was changed to “Most days I feel a sense of accomplishment from what I do” to measure the satisfaction of the need for competence in general (see Appendix A for the BNSG-S items). Given that the BNSG-S was simply an adaptation from the BNSW-S, it is important to review the theoretical and empirical domains of the BNSW-S.

The BNSW-S was created to assess the satisfaction of employees’ basic needs in the workplace (Kasser, Davey, & Ryan, 1992). Specifically, the BNSW-S was constructed by merging some new items based on “theoretical notions” (Kasser et al., 1992, p. 180) with some modified items from the perceived–competence subscale of the Intrinsic Motivation Inventory (Intrinsic Motivation Inventory, n.d., Ryan, 1982).² At the time of the creation of the BNSW-S, the IMI consisted of at least four subscales that included perceived–competence, interest–enjoyment, effort, and pressure–tension. An example of an item from the perceived–competence subscale is: “I think I am pretty good at this activity”. This and other items from the perceived–competence subscale of the IMI were simply reworded and modified to be specific to a work setting for the BNSW-S. Items from the other IMI subscales were not used to create BNSW-S items. The modified items from the IMI and the newly written items based on needs satisfaction theory appear to adequately represent the empirical domain of assessing the satisfaction of needs at work.

Unfortunately, similar to the BNSG-S, there are no known studies that have examined the factor structure of the BNSW-S. As such, there is no empirical evidence of the dimensionality of either measure and, in turn, there appears to be confusion regarding how to score the responses to the measures. Specific to the BNSG-S, some researchers have used a total score of general needs satisfaction (e.g., Gagné, 2003), whereas others have computed three subscales to represent three distinct needs (e.g., Niemiec et al., 2009). Similarly, specific to the BNSW-S, some researchers have created a total score of needs satisfaction at work (e.g., Deci et al., 2001), whereas other researchers have created three subscales scores to represent the three distinct needs of autonomy, competence, and relatedness (e.g., Ghorbani & Watson, 2006). Furthermore, for both the BNSG-S and the BNSW-S, some researchers who have created a total score of needs satisfaction have also created individual subscale scores to represent the three distinct needs (e.g., Gagné, 2003). Researchers have assumed validity, versatility, and flexibility of the scale, and have stated it can be scored either way (E. Deci, personal communication, June 12, 2008). These beliefs appear to be due in part to researchers erroneously citing others studies to support their use and scoring of the BNSG-S. For example, authors have claimed the BNSG-S has “excellent psychometric properties” and cited Gagné (2003) in support (Kashdan et al., 2006, p. 565). Although, Gagné (2003) used the BNSG-S to study general needs, she did not examine the psychometric properties of the BNSG-S. In addition, researchers have also erroneously cited studies which used the BNSW-S to provide psychometric evidence of the properties of the BNSG-S. For example, Neff (2003) cited two studies (i.e., Ilardi et al., 1993, Kasser et al., 1992) as providing good psychometric evidence of the BNSG-S; however both studies cited used older versions of the BNSW-S, not the BNSG-S. This is problematic because the BNSW-S and the BNSG-S are two different measures and do not provide evidence of dimensionality for one another.

For the BNSG-S, the only sources of information regarding the structural component of the validity process are in the form of reliability coefficients and subscale correlations, not in the form of factor analytic studies, which according to Benson and Hagtvet (1996), “would provide evidence whether the observables (e.g., the items) behave according to theory” (p. 91). Reported measures of internal consistency associated with a total needs satisfaction score ranged from .84 to .90 (Gagné, 2003, Meyer et al., 2007, Vansteenkiste et al., 2006, Wei et al., 2005). With respect to the three subscales computed from the BNSG-S scores, researchers reported values of internal consistency ranging from .61 to .81 for the autonomy subscale, .60 to .86 for the competence subscale, and .61 to .90 for the relatedness subscale (Conroy and Coatsworth, 2007a, Conroy and Coatsworth, 2007b, Gagné, 2003, Kashdan et al., 2006, Kashdan et al., 2009, Meyer et al., 2007, Niemiec et al., 2009, Thøgersen-Ntoumani and Ntoumanis, 2007, Vansteenkiste et al., 2006, Wei et al., 2005). It is important to note that Cronbach’s coefficient alpha is only appropriate to interpret when researchers are confident there is only one construct or phenomenon being measured; that is, the scores are unidimensional (McDonald, 1999). Therefore, it is confusing when researchers (i.e., Gagné, 2003, Meyer et al., 2007, Wei et al., 2005) report measures of internal consistency associated with a total needs scale (and use a total score in the analyses) when it was assumed by the same researchers that the scale measures three separate constructs (autonomy, competence, and relatedness). This practice may simply reflect the confusion associated with the dimensionality and scoring of the measure.

Correlations among the three subscales have been examined by researchers and shed some limited light on the distinctiveness of the three needs as operationalized by the BNSG-S. Specifically, correlations between the subscales ranged between r = .46 and .72 for autonomy and competence, r = .33–.79 for autonomy and relatedness, and r = .27–.80 for relatedness and competence (Conroy and Coatsworth, 2007a, Gagné, 2003, Kashdan et al., 2009, Meyer et al., 2007, Thøgersen-Ntoumani and Ntoumanis, 2007, Vansteenkiste et al., 2006, Wei et al., 2005). Given the lack of factor analytic studies, researchers may be tempted to infer that the wide range of subscale correlations is evidence to suggest the BNSG-S is not measuring one overall construct of needs satisfaction, but instead a measure of three different constructs. Furthermore, the differential relationships exhibited between the subscales and external criteria described below could be used as additional evidence to suggest the BNSG-S may be representing multiple constructs.

Because some studies have scored the BNSG-S as both a total measure of needs satisfaction and three individual needs, whereas other studies have only scored the BNSG-S as three individual needs, and given the individual subscales have exhibited similar and differential relationships with external variables, the following section examines the relationships between the BNSG-S and external variables in three parts. First, the relationships between the total needs satisfaction scores and external variables are examined. Second, the relationships between the individual subscales (autonomy, competence, and relatedness) and external variables of similar strength are reviewed. Finally, the relationships between the individual subscales and external variables of differential strength are reviewed.³

The total BNSG-S score has been positively related to well-being among Chinese college students studying abroad in Denmark (Vansteenkiste et al., 2006), happiness and self-actualization in women who worked as models in Great Britain, happiness, self-actualization, and life satisfaction in women who did not work as models in Great Britain (Meyer et al., 2007), pro-social behavior (i.e., volunteering) and parental support (Gagné, 2003). In contrast, the satisfaction of needs when represented by a total needs score has been negatively related to depression in Chinese college students studying abroad (Vansteenkiste et al., 2006) and anxiety, depression, and loss of confidence in women working as models in Great Britain (Meyer et al., 2007). Thus, the external relationships associated with a total needs satisfaction score appear to support SDT’s definition of needs satisfaction (i.e., higher levels of needs satisfaction are positively related to measures of well-being and negatively related to measures of ill-being).

Relationships of similar strengths have emerged between the three subscales (autonomy, competence, and relatedness) and external variables that serve as proxy measures of well-being. Specifically, satisfaction of the needs for autonomy, competence, and relatedness related positively and at approximately the same magnitude to life satisfaction (r = .50, r = .59, r = .57), self-esteem (r = .50, r = .53, r = .41) and positive affect in American college graduates (r = .58, r = .58, r = .46; Niemiec et al., 2009); life satisfaction (r = .51, r = .49, r = .31) and happiness in British women (r = .44, r = .43, r = .36; Meyer et al., 2007); and psychological well-being (r = .56, r = .58, r = .60) and vitality in Chinese college students (r = .51, r = .58, and r = .61, respectively; Vansteenkiste et al., 2006). In contrast, the individual subscales of autonomy, competence, and relatedness related negatively and at approximately the same magnitude to depression in American college students (r = −.63, r = −.63, r = −.62; Wei et al., 2005); negative affect in male (r = −.43, r = −.55, r = −.59) and female (r = −.50, r = −.50, r = −.37) college students (Kashdan et al., 2009); negative affect (r = −.46, r = −.39, r = −.34) and anxiety (r = −.47, r = −.40, r = −.35) in college graduates (Niemiec et al., 2009), and the drive for thinness (r = −.51, r = −.46, r = −.39) and introjected regulation in British aerobics instructors (r = −.29, r = −.36, r = −.26, respectively; Thøgersen-Ntoumani & Ntoumanis, 2007). The correlations of similar magnitude between the individual subscales and external variables may be incorrectly used to suggest that the three subscales may not be very distinct from one another. Obviously, the three needs may have equal predictive utility in some situations or for some external variables; thus, findings of this nature do not imply that non-differential relationships will always occur or that theoretically the needs are not distinct.

When scored individually, the three separate subscales of the BNSG-S have exhibited differential relationships with external variables. For instance, the autonomy subscale exhibited a stronger relationship with well-being in British women (r = .73) than the need for competence (r = .52), although relationships of similar strength emerged between the three subscales and well-being in Chinese college students (Meyer et al., 2007, Vansteenkiste et al., 2006). The need for competence subscale (r = .60) exhibited a stronger relationship with positive affect in female college students than the need for autonomy or relatedness subscales (r = .44, r = .39, respectively; Kashdan et al., 2009). In addition, the need for competence subscale exhibited a stronger relationship with pro-social engagement (i.e., volunteering; r = .40) than the autonomy subscale (r = .19), but was not different from the relatedness subscale (r = .26; Gagné, 2003). The need for relatedness subscale exhibited a stronger negative relationship with depression in Chinese college students (r = −.61) than did the need for competence subscale (r = −.46; Vansteenkiste et al., 2006). In contrast, the need for relatedness subscale was not correlated with depression in British models although depression was negatively correlated with the need for autonomy and competence subscales (Meyer et al., 2007). Furthermore, the need for relatedness exhibited a weaker relationship with social physique anxiety among British aerobics instructors (r = −.36) than the needs for autonomy or competence (r = −.57, r = −.52, respectively; Thøgersen-Ntoumani & Ntoumanis, 2007).

Researchers who believe the BNSG-S is a unidimensional measure of needs satisfaction may turn to the studies reviewed above that exhibit theoretically-expected relationships between the total score and external variables or similar relationships between the three subscales and external variables as evidence of the unidimensional nature of the BNSG-S (e.g., Vansteenkiste et al., 2006). As noted above, finding non-differential relationships for one external variable does not imply the needs would not function differentially for a different external variable. Researchers who believe the BNSG-S is a measure of three distinct needs may be tempted to use the studies that exhibited differential relationships between the three subscales and external variables as evidence of the multidimensional nature of the scale (e.g., Neff, 2003). However, it is extremely important to note that the similar and different relationships exhibited between the individual subscale scores and the external variables could be simply due to the differential reliabilities of the autonomy, competence, and relatedness subscales. That is, reliability coefficients impact the magnitude of relationships between the three subscales and external variables, which could result in relationships appearing more similar or more different than they actually are. In addition, the differential correlations exhibited could be due to sampling error or interactions with different populations and contexts being compared. Thus, the relationships exhibited between the subscale scores and external criteria should not be solely relied onto make judgments regarding the dimensionality of a measure.

In sum, although the external component stage is considered the most important stage (Benson, 1998), it should not be examined until the factor structure of the scale has been studied, because it is unclear how the scores should be computed. If the scale was multidimensional, the use of a total score would mask the differential relationships exhibited between the individual subscale scores and external variables.

Given the importance of needs satisfaction in regards to psychological well-being and growth as defined by SDT, the use of the BNSG-S to assess the satisfaction of needs in general, the confusion regarding how to score the measure, and the dearth of psychometric study of the measure, the purpose of the current study was to investigate the construct validity of the BNSG-S by (a) examining the factor structure of the BNSG-S and (b) investigating the external validity of the scale by estimating theoretically-expected relationships with external variables.

The first purpose involved testing the following models using confirmatory factor analysis (CFA): (a) a one-factor model assessing the unidimensionality of the BNSG-S and (b) a three-factor model representing autonomy, competence, and relatedness (see Fig. 1). It was hypothesized a three-factor model would fit the data significantly and practically better than a one-factor model given (a) SDT stipulates three distinct needs and emphasizes that all three needs must be individually met for wellness and (b) three distinct needs appeared to have been operationally defined when the BNSG-S items were created.

The second purpose focused on the relationships between needs satisfaction and theoretically-related variables: (a) psychological well-being, (b) the motive to avoid failure, and (c) anxiety. These variables were chosen based on congruence with SDT’s perspective of well-being. Given the three-factor model was hypothesized to be championed, differential relationships were hypothesized between the three needs and external measures; these hypotheses are described below.

Researchers often use several different measures to represent well-being (e.g., positive affect, negative affect, and life satisfaction; Diener, Emmons, Larsen, & Griffin, 1985). When measured as a combination of life satisfaction, happiness, and self-actualization, well-being has been positively related to satisfaction of the needs for autonomy and competence in British models, and autonomy, competence, and relatedness in non-models (Meyer et al., 2007). For the current study, three of the six well-being dimensions conceptualized by Ryff (1989) were used: autonomy (AU; i.e., not being influenced by others), environmental mastery (EM; i.e., the ability to shape or create environments that are in line with personal needs), and positive relations with others (PR; i.e., the ability to be in close relationships with others). These dimensions of well-being were chosen because they align with SDT’s eudaimonic, as opposed to hedonic, perspective of well-being (Ryan and Deci, 2000c, Ryan and Deci, 2002). That is, well-being is defined “in terms of a fully functioning person” (Ryan & Deci, 2000c, p. 323), as opposed to a person displaying individual pieces of evidence such as “a subjective experience of affect positivity” (Deci and Ryan, 2000, Ryan and Deci, 2000c, Ryan and Deci, 2002).

In line with SDT, AU, EM, and PR were expected to be positively related to satisfaction of the needs for autonomy, competence, and relatedness. In addition, AU was expected to exhibit a stronger positive relationship with the satisfaction of the need for autonomy than the satisfaction of the need for competence or relatedness. This differential relationship was expected because AU and the satisfaction of the need for autonomy both assess feelings and attitudes towards autonomy, whereas the satisfaction of the need for competence and relatedness do not. Furthermore, EM was expected to exhibit a stronger positive relationship with the satisfaction of the need for competence than the satisfaction of the need for autonomy or relatedness because as people’s confidence in their ability to handle and control surrounding environments increases (i.e., environmental mastery increases), the extent to which they feel capable, or confident (i.e., satisfaction of the need for competence) should also increase. Also, PR was expected to exhibit a stronger positive relationship with satisfaction of the need for relatedness than the need for autonomy or competence because PR and satisfaction of the need for relatedness both address the experience, the need to develop, and the desire to have personal relationships with other people, whereas the satisfaction of the needs for autonomy and competence do not.

The motive to avoid failure (MAF) refers to the extent to which people will avoid situations that induce feelings that failure is a possibility (Hagtvet & Benson, 1997). The MAF has been positively related to measures of worry, emotionalism, anxiety, test-irrelevant thinking, and tension and has been negatively related to competiveness and self-determination (Elliot and McGregor, 2001, Hagtvet and Benson, 1997). For the current study, Hagtvet and Benson’s (1997) unidimensional conceptualization of the MAF was used. The MAF was expected to be negatively related to the satisfaction of the needs for autonomy, competence, and relatedness. As individuals decrease in the degree to which their needs are satisfied, their desire to avoid failure increases. More specifically, it was hypothesized that the MAF would exhibit a stronger negative relationship with satisfaction of the need for competence than with the needs for autonomy or relatedness. That is, people who feel competent at performing tasks should also not feel the need to avoid situations in which failure is a possibility.

Anxiety refers to feelings of worry and restlessness. When measured in general and with respect to one’s physical appearance, anxiety has been negatively related to satisfaction of the needs for autonomy, competence, and relatedness (Niemiec et al., 2009, Thøgersen-Ntoumani and Ntoumanis, 2007). For the current study, three of the six dimensions of worry as conceptualized by Osman et al. (2001) were used to assess anxiety and worry: general anxiety symptoms (GAS), worrisome thinking (WST), and social adequacy concern (SAC). The three subscales of autonomy, competence, and relatedness were expected to be negatively related to the GAS, WST, and SAC subscales. In addition, the GAS and WST subscales were expected to exhibit a stronger negative relationship with the satisfaction of the need for autonomy subscale than the satisfaction of the need for competence or relatedness subscales because both constructs reflect a feeling of being controlled by anxiety (e.g., “I feel physically tired and exhausted when worrying about things”, “No matter how hard I try, I cannot stop or control worrying about something.”). SAC was expected to exhibit a stronger negative relationship with satisfaction of the need for relatedness than to the need for autonomy or competence. This differential relationship was expected because SAC is specific to worrying about relationships and behavior around other people (e.g., “I worry about making a fool of myself around other people.”); if people felt cared for and supported by other people (i.e., need for relatedness was satisfied), they should not worry about their social behavior.