The Basic Psychological Need Satisfaction and Frustration Scales Probably Do Not Validly Measure Need Frustration

Abstract

In basic psychological needs theory (BPNT), the separable constructs of need satisfaction and need frustration are theorized as pivotally related to psychopathology and broader aspects of well-being. The Basic Psychological Need Satisfaction and Frustration Scales (BPNSFS; Chen et al., 2015) have rapidly emerged as the dominant self-report measure in the BPNT domain, with translated versions available in a wide range of languages and a plethora of versions adapted for specific populations and life contexts. Through (a) an extended conceptual discussion of the BPNSFS and (b) a collection of complementary data analyses in eight samples, we demonstrate that the BPNSFS probably does not validly measure need frustration. Most importantly, we conclude that the ostensible distinction between need frustration and need satisfaction in the BPNSFS is likely primarily a method artifact caused by different item keying directions, given the way the measure currently assesses the intended constructs. If so, then the use of the BPNSFS may be generating misleading conclusions, obstructing sound investigation of current BPNT. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Full text

The Basic Psychological Need Satisfaction and Frustration Scales
Probably Do Not Validly Measure Need Frustration
Brett A. Murphy
1
, Ashley L. Watts
2
, Zachary G. Baker
3
, Brian P. Don
1
, Tatum A. Jolink
1
, and Sara B. Algoe
1
1
Department of Psychology and Neuroscience, University of North Carolina at Chapel Hill
2
Department of Psychological Sciences, University of Missouri
3
School of Public Health, University of Minnesota
In basic psychological needs theory (BPNT), the separable constructs of need satisfaction and need
frustration are theorized as pivotally related to psychopathology and broader aspects of well-being. The
Basic Psychological Need Satisfaction and Frustration Scales (BPNSFS; Chen et al., 2015) have rapidly
emerged as the dominant self-report measure in the BPNT domain, with translated versions available in
a wide range of languages and a plethora of versions adapted for specific populations and life contexts.
Through (a) an extended conceptual discussion of the BPNSFS and (b) a collection of complementary data
analyses in eight samples, we demonstrate that the BPNSFS probably does not validly measure need
frustration. Most importantly, we conclude that the ostensible distinction between need frustration and
need satisfaction in the BPNSFS is likely primarily a method artifact caused by different item keying
directions, given the way the measure currently assesses the intended constructs. If so, then the use of the
BPNSFS may be generating misleading conclusions, obstructing sound investigation of current BPNT.
Public Significance Statement
The Basic Psychological Need Satisfaction and Frustration Scales are widely used by researchers, in
many different countries, to investigate basic psychological needs and their importance for human
well-being. We raise serious concerns regarding the validity of the measurement model employed by
many researchers using these scales.
Keywords: BPNSFS, basic psychological needs, need satisfaction, need frustration, construct validity
Supplemental materials: https://doi.org/10.1037/pas0001193.supp
Basic psychological needs theory (BPNT), a mini-theory within
self-determination theory (SDT; Deci & Ryan, 1985;Ryan & Deci,
2017), is currently one of the more widely investigated theoretical
frameworks in psychology. At present, BPNT recognizes three
basic psychological needs: (a) autonomy, the feeling of volition and
self-endorsement; (b) competence, feeling capable of achieving goals
and being effective in one’s environment; and (c) relatedness, feeling
genuinely connected to others in relationships. These three psycho-
logical needs are conceived of as: innate, critical aspects of our
evolved psychological structure; imperative for psychological
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This article was published Online First November 28, 2022.
Brett A. Murphy https://orcid.org/0000-0002-2619-9199
Zachary G. Baker https://orcid.org/0000-0001-5345-7643
Brian P. Don https://orcid.org/0000-0002-0086-9377
Tatum A. Jolink https://orcid.org/0000-0003-4160-337X
Sara B. Algoe https://orcid.org/0000-0002-1273-2968
The authors would like to thank Kaitlyn Werner and Thuy-vy Nguyen for
their very helpful suggestions and edits when revising this article. Brett A.
Murphy, Sara B. Algoe, Brian P. Don, and Tatum A. Jolink were supported
by funding from the John Templeton Foundation for some or all of the time
this research was being conducted (Grant 61280 to Sara B. Algoe). Ashley L.
Watts was supported by K99AA028306 (Principal Investigator: Ashley L.
Watts). Zachary G. Baker was supported by K99 AG073463 (Principal
Investigator: Zachary G. Baker).
Brett A. Murphy played lead role in conceptualization, investigation and
writing of original draft and equal role in formal analysis, methodology,
visualization and writing of review and editing. Ashley L. Watts played
supporting role in conceptualization and investigation and equal role in
formal analysis, methodology, visualization and writing of review and
editing. Zachary G. Baker played lead role in data curation and supporting
role in conceptualization, formal analysis, investigation, methodology,
writing of original draft and writing of review and editing. Brian P. Don
played supporting role in conceptualization, formal analysis, investigation,
methodology, writing of original draft and writing of review and editing.
Tatum A. Jolink played supporting role in conceptualization, data curation,
investigation, visualization, writing of original draft and writing of review
and editing. Sara B. Algoe played supporting role in conceptualization,
investigation, methodology, writing of original draft and writing of review
and editing.
This research was not preregistered. Data, codes, and all supplementary
materials referenced in this article are available at the following link: https://
osf.io/6r2pb/.
Correspondence concerning this article should be addressed to Brett A.
Murphy, Department of Psychology and Neuroscience, University of
North Carolina at Chapel Hill, 235 East Cameron Avenue, Chapel Hill,
NC 27599, United States. Email: bmurphy.psych@gmail.com
Psychological Assessment
© 2022 American Psychological Association 2023, Vol. 35, No. 2, 127–139
ISSN: 1040-3590 https://doi.org/10.1037/pas0001193
127

well-being; distinct from one another; and universal, regardless of
gender, culture, and personality (Ryan & Deci, 2017;Vansteenkiste
et al., 2020).
For many years, BPNT focused primarily on effective need
satisfaction and its associations with heightened psychological
well-being. Need satisfaction was often depicted as a unitary
dimension, spanning low-to-high levels of satisfaction (see review
by Vansteenkiste et al., 2020); the term “need frustration”was used
as synonymous with low need satisfaction. More recently, though,
the satisfaction of basic psychological needs has been conceptually
parsed into two connected yet distinguishable dimensions: need
satisfaction and need frustration. In this dual-dimension perspec-
tive, need frustration has been conceptualized as an asymmetric
configuration of (a) low need satisfaction alongside (b) strong,
active, direct thwarting by other individuals or by the social context
itself (Costa et al., 2015;Vansteenkiste & Ryan, 2013). For exam-
ple, a person may feel a lack of relatedness with others in the
workplace simply because of a lack of shared perspectives (low need
satisfaction). In contrast, if others in the workplace are actively
ostracizing or bullying the person, then the need for relatedness
is being “frustrated”(Vansteenkiste & Ryan, 2013, p. 264). An
analogy is to a plant’s need for water: lack of rainfall can cause low
need satisfaction, whereas salting the plant’s soil constitutes active
need frustration (Vansteenkiste & Ryan, 2013). Need satisfaction is
theorized to relate particularly strongly to psychological well-being,
and need frustration particularly strongly to psychopathology
(Vansteenkiste & Ryan, 2013).
The Basic Psychological Need Satisfaction and Frustration Scales
(BPNSFS; Chen et al., 2015) aim to distinguishably measure these
two aspects, satisfaction and frustration, for each of the three basic
need domains (competence, autonomy, relatedness). These scales
have quickly become the dominant self-report measure in BPNT
research, with the original article introducing it has already garnered
more than 1,900 citations by September 2022 (according to Google
Scholar). It contains six scales (see Figure 1a): a satisfaction scale
and a frustration scale for each of the three basic needs domains. The
BPNSFS is currently promoted as a valid assessment of both need
satisfaction and need frustration (e.g., Vansteenkiste et al., 2020).
The latest manual for the BPNSFS (Van der Kaap-Deeder et al., 2020)
lists a remarkable 50+versions already, spanning a wide range of
languages, contexts (workplace, romantic relationships, school,
exercise, etc.), and specific populations (e.g., mothers, astronauts,
sport coaches).
This article elaborates two specific concerns regarding key com-
ponents of the BPNSFS that challenge the validity of the measure-
ment model. The first concern is that the BPNSFS “need frustration”
items do not appear to match the concept of need frustration, based
on the reconceptualization (e.g., Vansteenkiste & Ryan, 2013) that
separates thwarting of needs from low need satisfaction. Instead, the
need frustration items consist of primarily reverse-keyed versions
of need satisfaction items. The second concern is that the two
autonomy scales might be conceptually incommensurable, with the
Autonomy Frustration scale failing to capture self-endorsement
aspects central to current BPNT conceptualization of autonomy. If
these present concerns are justified, the BPNSFS is not valid for
use as a measure of the dual-dimension theory of need frustration
and satisfaction (Vansteenkiste & Ryan, 2013).
Chen (2013) is the earliest publicly available source for the
BPNSFS creation work published in Chen et al. (2015).Chen (2013)
tentatively concluded that the item pool did not support the creation
of separate frustration scales. Instead, Chen (2013) aggregated posi-
tively keyed (satisfaction) and negatively keyed (frustration) items
together, forming only three need satisfaction scales (see Figure 1b).
There are understandable reasons, though, for the different presentation
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Figure 1
Alternative Factor Structures of the BPNSFS
Note. (a) Six-factor structure proposed by Chen et al. (2015). (b) Three-factor structure with item keying factors. (c) Proposed four-factor structure with item
keying factors. BPNSFS =Basic Psychological Need Satisfaction and Frustration Scales.
128 MURPHY ET AL.

in Chen et al. (2015); as we will explain further, it is difficult to make
a completely dispositive conclusion as to which presentation is
more valid. Nonetheless, on balance, we believe the earlier conclusions
of Chen (2013) are better supported than those in the later publication
by Chen et al. (2015). To a substantial extent, the arguments of the
present article are corroborations of Chen’s (2013) insightful con-
clusions; some of the present analyses are replications of analyses
conducted by Chen (2013) but using new samples.
Concerns With the BPNSFS Need Frustration Scales
Given their asymmetric conceptual relationship, generating assess-
ment items that effectively distinguish between low need satisfaction
and need frustration will inherently be a challenging task for a test
developer. Nevertheless, the original description of item development
does not appear to have intended to tackle this specifictask(Chen,
2013), where the central empirical question was “whether the
association between need satisfaction and well-being would be
higher for those who have higher desire for the need”(pp. 65–66). To
investigate this question, Chen (2013) created a revised measure of
need satisfaction with measurement invariance across different cul-
tures, witha balance of negatively and positivelykeyed items for each
need domain. In sum, Chen’s (2013) approach was consistent with
the older, unitary dimension view, in which “need frustration”is a
synonym for “low need satisfaction.”
1
Chen’s (2013) alignment with the older unitary view is evident
in the items themselves. None of the three BPNSFS “frustration”
scales include substantial active thwarting content, which is central
to the construct’s newer definition as reconceptualized by Vansteenkiste
and Ryan (2013). For instance, the four Competence Frustration items
do not reference any kind of thwarting by other individuals or one’s
social environment. Instead, two items (e.g., “I have serious doubts
about whether I can do things well”and “I feel insecure about my
abilities”) are reverse-worded versions of conceptually near-identical
items on the Competence Satisfaction scale (e.g., “Ifeelconfident that
I can do things well”and “I feel capable at what I do”). The other two
Competence Frustration items both relate to past or present failures,
which are not paralleled in the Competence Satisfaction items (e.g., “I
feel disappointed with many of my performance [sic]”and “I feel like
a failure because of the mistakes I make”). Feeling negatively about
one’s failures seems far removed from the kind of threatening
thwarting force that might undermine one’s needs. The Relatedness
and Autonomy frustration and satisfaction dimensions are less obvi-
ously opposite wording versions of one another. Still, those Frustra-
tion scales also largely lack active thwarting content (e.g., “Ifeelthat
the relationships I have are just superficial”and “Most of the things
I do I feel like ‘I have to’”).
2
Item Coding Direction: The Elephant in the Room
In arguing for the validity of the BPNSFS, researchers have
pointed to the fact that the need frustration and satisfaction items
tend to load on different factors, with a six-factor model generally
demonstrating a better fit than a three-factor model (e.g., Chen et al.,
2015;Cordeiro et al., 2016;Šakan, 2020). The superior fit of a six-
factor over a three-factor model, though, may be primarily due to
methodological artifact and other construct-irrelevant variance.
Except for the autonomy domain (which we will return to later),
the BPNSFS frustration and satisfaction scales appear to measure
the same construct but with opposite item-keying directions.
In psychological research, scale creators have frequently em-
ployed a mix of both positively and negatively keyed items of a
single construct within their questionnaires, partly out of a desire
to combat response acquiescence or other response distortions
(e.g., Nunnally, 1978). One long-standing idea is that switches in
the directionality of the items operate as “speed bumps,”forcing
the respondent to exert more careful, effortful responding (e.g.,
Podsakoff et al., 2003). At the same time, this methodological
approach will also typically lead to some degree of inconsistent
responding between positively and negatively keyed items as
participants—out of carelessness, confusion, or distraction—blow
through a few of these speed bumps. Even a small amount of careless
responding by participants will often cause item-keying method
factors to manifest (e.g., Woods, 2006), with unidimensional con-
structs splitting into two item-keying direction factors. Although there
are sound reasons for using a mix of item-keying directions in
scales, one must keep these complications in mind.
In many cases, such as with the BPNSFS, item-keying direction is
also equivalent to the general positive or negative valence of the
items: favorable versus unfavorable and good versus bad (see
Kam&Meyer,2015, for fuller explanation). In such cases, item
coding direction factors may relate more to construct-irrelevant
substantive domains such as defensiveness, self-esteem, social desir-
ability, or behavioral inhibition (e.g., DiStefano & Motl, 2009;Quilty
et al., 2007) than to the specific, intended substantive domain (e.g.,
need satisfaction or frustration). Given the emotional primacy of
the basic psychological needs, the BPNSFS may be particularly
vulnerable to this kind of systematic response distortion.
Based on the content of the items themselves, our interpretation
of the BPNSFS scales is that the apparent distinction between the
Satisfaction and Frustration scales is likely primarily driven by item-
keying direction, not by substantive content distinguishing con-
structs of need satisfaction and need frustration. Chen (2013) came
to this same conclusion following an exploratory factor analysis
(EFA) of the total item pool, finding only two large factors, one for
the negatively keyed items and one for the positively keyed items.
Chen (2013) interpreted this result as likely being driven by
methodological artifact. To account for this “methodologically
induced bias”(Chen, 2013, p. 75), Chen (2013) modeled positive
and negative items as different methods to assess the same underly-
ing need satisfaction construct. Accounting for this method bias,
Chen (2013) found that a three-factor model fits well and created
one scale for each of the three basic needs, with “frustration”
items serving as negatively keyed satisfaction items.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1
For example, “self-reported desire for a need may partly reflect previous
experiences of need frustration. That is, when a person has experienced less
satisfaction of a basic need, the desire for it may become more salient.”
(Chen, 2013, p. 89).
2
The BPNSFS’s lack of face validity is apparent when comparing it to the
Psychological Need Thwarting Scale (PNTS; Bartholomew et al., 2011),
which was initially created to assess need frustration in the context of sports
but which has since been modified for general use (e.g., Costa et al., 2015).
The PNTS measures being thwarted far more explicitly than does the
BPNSFS. For example, in Costa et al. (2015, emphasis added), items include:
“I feel inadequate because I am not given opportunities to fulfill my
potential,”“There are times when I am told things that make me feel
incompetent,”“I feel forced to follow decisions made for me,”and “There
are situations where I am made to feel inadequate.”
VALIDITY OF NEED FRUSTRATION SCALES 129

Can we dispositively determine whether the BPNSFS scales
primarily reflect item-keying method bias or substantive constructs
of frustration/satisfaction? Unfortunately, when purported substan-
tive variance is fully confounded with item-keying direction, factor
analyses are not able to provide information to fully adjudicate
between these possibilities (Kam & Meyer, 2015;seeNaragon-
Gainey & DeMarree, 2017, for discussion in this journal). For
instance, even if a six-factor structure is the best-fitting model of
the item pool, one could not determine whether the factors represent
item-keying factors or the purported substantive variance factors.
Similarly, when substantive variance is fully confounded with
item-keying direction, findings of nomological network differences
are also poorly equipped to adjudicate the item-keying artifact
versus substantive variance question (Kam & Meyer, 2015;Naragon-
Gainey & DeMarree, 2017). In multiple studies, the BPNSFS
“frustration”and “satisfaction”scales have appeared to exhibit
modestly different nomological networks, with the frustration scales
exhibiting stronger associations with ill-being (e.g., Chen et al., 2015;
Liga et al., 2020). Yet, the same construct-irrelevant causes of splits
between positively and negatively keyed items onto different factors
will also often cause them to demonstrate different nomological
networks (Credéet al., 2009;DiStefano & Motl, 2009;Kam &
Meyer, 2015;Ray et al., 2016). For instance, diverging associations
with external variables may be partly driven by the extent to which
those external variables are assessed using negative or positive
items (Naragon-Gainey & DeMarree, 2017). Similarly, if item-
keying direction splits partly reflect general defensiveness, self-
esteem, social desirability, or other effects tied to general “item
valence”(Kam & Meyer, 2015), item-keying “artifactors”will
tend to be differentially associated with other measures affected
by such variables. For example, negatively and positively valenced
items have been found to be differentially related to depression,
negative affectivity, emotional stability, or self-enhancement (e.g.,
Lindwall et al., 2012;Michaelides et al., 2016;Quilty et al., 2006),
constructs included in, or similar to, domains of ill-being and well-
being at the focus of much BPNT research.
In the BPNSFS, as in many other psychological measures (e.g.,
in this journal; cf. Crasta et al., 2021;Neubauer et al., 2022), item-
keying and purported substantive variance are fully confounded.
Though this is not necessarily a flaw itself, it does mean it is inherently
difficult to empirically determine the underlying meaning of the
Satisfaction versus Frustration scales. In such cases, generally, the
most that is possible using existing data sets is to (a) subjectively
evaluate the face validity of the items themselves and (b) tentatively
investigate whether associations with external variables are partly
driven by negative versus positive items in those external variable
scales.
Building on the work of prior assessment scholars, we argue that,
first, when all positively coded and all negatively coded items split
into separate factors, despite having very similar item contents, a test
designer or user is on uncertain ground if they conclude that this
factor analytic split supports treating the two factors as conceptually
distinct measurement dimensions (e.g., see Böckenholt, 2019;
Hankins, 2008;Kam & Meyer, 2012;Tomas & Oliver, 1999).
Second, if the negatively and positively valenced items of a poten-
tially unidimensional construct are preferentially related to similarly
positively or negatively valenced items from external variable scales,
there should be a rebuttable presumption that item-keying drives
much of any apparent nomological network differences between the
scales (see Kam & Meyer, 2015, for an exemplar of this analytical
approach).
In the specific case of the BPNSFS, we can make somewhat more
dispositive conclusions about item-keying method variance in only
one respect: In the full item pool, item covariance associated with
item keying direction should be less than the covariance associated
with the three basic psychological need domains themselves, con-
sistent with arguments made in the multitrait–multimethod model-
ing literature (Campbell & Fiske, 1959). In BPNT, the three basic
psychological need domains are conceptualized as fundamentally
and universally distinct, but frustration of any particular need is
conceptualized, by definition, as inherently covarying with need
satisfaction of that need. Thus, the conceptual theory presumes that
need domains (competence, autonomy, relatedness) follow a higher
order structure, with lower order satisfaction and frustration domains
(e.g., competence satisfaction, competence frustration) explained
by higher order need domains (e.g., competence). In contrast, if the
BPNSFS is best explained by a higher order structure with lower
order satisfaction and frustration need domains and higher order
satisfaction and frustration dimensions, that would suggest that
item-keying may more strongly influence the structure of the BPNSFS
than does the intended substantive variance.
Reduce the BPNSFS to a Three-Factor Model?:
The Complications of the Autonomy Scales
If the “frustration”and “satisfaction”items are primarily different
keying methods of assessing the same substantive dimensions, the
practical decision might seem to be that all users of the BPNSFS
could simply revert to the three-factor model originally proposed
by Chen (2013), only measuring need satisfaction for each of the
three domains (competence, relatedness, autonomy) and treating the
“frustration”items as reverse-scored satisfaction items. This may,
however, be an inappropriate decision. Whereas the frustration and
satisfaction scales for Competence and Relatedness likely can be
collapsed together into single satisfaction scales (reverse-scoring
the previously named “frustration”items to represent low satisfac-
tion), this might not be the case for the Autonomy Frustration and
Autonomy Satisfaction scales.
Autonomy has long been the most conceptually difficult and
debated of the three BPNT need domains (e.g., see excellent discus-
sions in Chen, 2013;Chen et al., 2013). For instance, although some
scholars have described “autonomy”in individualistic terms, such as
being independent of others or not relying upon others, defining it as
such puts it in conflict with the basic psychological need of relatedness
and implies that autonomy is less satisfied or less universally essential
for people in more collectivist cultural contexts. The BPNT definition
of autonomy, in contrast, emphasizes feelings of self-endorsement of
one’s actions, “such that one’s actions are grounded in authentic values
and interests”(Chen et al., 2013, p. 1186) and “one experiences
a sense of integrity as when one’s actions, thoughts, and feelings
are self-endorsed and authentic”(Vansteenkiste et al., 2020,p. 3).
Such feelings of autonomous self-endorsement can be present
regardless of the independence or interdependence of one’s func-
tioning. In other words, even if one’s actions are heavily influenced
by the needs of others, or by perceived obligations to others, one
can still feel that one’s actions are consistent with one’s authentic
values and interests, and thus one’s autonomy needs can still be
satisfied.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
130 MURPHY ET AL.

As with the other frustration scales, the Autonomy Frustration
scale lacks clear thwarting content. Yet, the Autonomy Satisfaction
and Frustration scales are not mirror opposites of one another.
Most of the Autonomy Satisfaction items reflect feelings of self-
endorsement (e.g., “I feel my choices express who I really am,”
“I feel that my decisions reflect what I really want,”“I feel I have
been doing what really interests me”). In contrast, the Autonomy
Frustration items largely do not appear to reflect such content;
instead, they may reflect feeling that one’slifeisbusywithtoo
many things to do or with many obligations (“I feel pressured to do
too many things,”“Most of the things I do feel like ‘I have to,’” and
“My daily activities feel like a chain of obligations”; but see “I feel
forced to do many things I wouldn’t choose to do”). Thus, the
Autonomy Frustration and Satisfaction scales may conceptually
diverge: A person may feel that their life is too busy with things
they would rather not do, yet also feel that their choices and decisions
reflect their authentic values and interests. If so, then the Autonomy
Satisfaction items may closely align with the self-endorsement
content central to autonomy in current BPNT conceptualization,
but the Autonomy Frustration items may not. Thus, aggregating the
two scales together may fail to produce scores that reliably reflect
the BPNT definition of autonomy.
This conceptual divergence between the two Autonomy scales is
hinted at in Table 1 of Chen (2013). In a factor analysis with three
need satisfaction factors and controlling for item-keying direction,
all of the Competence and Relatedness items loaded above .5
(the loading threshold benchmark used by Chen et al., 2015),
whereas only one Autonomy Frustration item (“I feel forced to
do many things I wouldn’tchoosetodo”) loaded strongly on the
Autonomy factor (loading =.51).
The imperfect alignment of the Autonomy Frustration and
Satisfaction items may be partly a side effect of the item-keying
complications described above. Faced with the difficult challenge
of working around the item-keying bias in the item pool, Chen
(2013) broke the positive and negative items into separate item
pools, conducting a three-factor EFA with the positive items and a
three-factor EFA with the negative items. Chen (2013) reduced the
item pool for final scales in that manner, separately for positively and
negatively keyed items. This is an understandable approach, one
that has been entertained by many test developers (including us)
when trying to work around item-keying direction bias. With this
approach, though, there is no guarantee that, for instance, the
“autonomy”factor emerging from EFA of the positive coded
items will be equivalent to the “autonomy”factor emerging from
EFA of the negative coded items. We theorize that, in the case of
the BPNSFS, this approach inadvertently resulted in a positively
keyed factor heavily reflecting the critical concept of self-
endorsement and a negatively keyed factor that did not reflect
self-endorsement.
The facial content of the scales, and the results in Chen (2013),
lead us to hypothesize that the best measurement model for the
BPNSFS items may actually be a four substantive-factor model,
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Table 1
Exploratory Factor Analyses of BPNSFS, Extracting 2–4 Factors (Combined Sample N =3,692)
Item Item keying
Two-factor model Three-factor model Four-factor model
F1 F2 F1 F2 F3 F1 F2 F3 F4
Item 1 (autonomy satisfaction) +.65 .03 .67 .03 .00 .00 .59 −.04 .31
Item 2 (autonomy satisfaction) +.68 −.08 .68 −.08 .02 .01 .59 .00 .16
Item 3 (autonomy satisfaction) +.71 −.07 .71 −.06 .02 .03 .62 .01 .15
Item 4 (autonomy satisfaction) +.64 −.02 .65 −.01 .02 −.03 .57 −.02 .28
Item 5 (autonomy frustration) −.01 .71 .01 .69 .03 .02 .03 .00 .79
Item 6 (autonomy frustration) −−.03 .74 −.04 .73 .08 .22 −.02 .15 .51
Item 7 (autonomy frustration) −−.05 .74 −.03 .72 −.01 .18 −.01 .02 .61
Item 8 (autonomy frustration) −−.01 .73 −.01 .71 .07 .04 .01 .06 .75
Item 9 (relatedness satisfaction) +.67 .10 .59 .09 .40 .09 .49 .52 −.06
Item 10 (relatedness satisfaction) +.79 −.01 .71 −.03 .45 −.02 .60 .53 −.03
Item 11 (relatedness satisfaction) +.76 .05 .68 .03 .43 −.02 .58 .50 .04
Item 12 (relatedness satisfaction) +.73 .00 .65 −.01 .40 −.09 .56 .43 .10
Item 13 (relatedness frustration) −.09 .68 .03 .69 .23 .44 .01 .44 .07
Item 14 (relatedness frustration) −.08 .65 −.01 .66 .31 .42 −.03 .54 .00
Item 15 (relatedness frustration) −.08 .69 .00 .70 .25 .46 −.02 .48 .03
Item 16 (relatedness frustration) −.04 .61 −.02 .60 .25 .21 −.01 .37 .25
Item 17 (competence satisfaction) +.67 .11 .73 .11 −.14 .41 .61 −.04 −.02
Item 18 (competence satisfaction) +.70 .07 .74 .07 −.08 .37 .62 .01 −.05
Item 19 (competence satisfaction) +.71 −.02 .73 −.01 −.01 .25 .61 .05 −.04
Item 20 (competence satisfaction) +.65 .03 .70 .03 −.10 .28 .59 −.04 .01
Item 21 (competence frustration) −.03 .72 .09 .74 −.20 .67 .06 −.02 .16
Item 22 (competence frustration) −.02 .70 .06 .70 −.14 .65 .03 .04 .09
Item 23 (competence frustration) −.01 .70 .06 .72 −.19 .74 .03 .01 .04
Item 24 (competence frustration) −−.06 .70 −.01 .71 −.18 .69 −.03 .00 .06
Factor intercorrelations F2 F1 F2 F1 F2 F3
F1 .59 F1 F1
F2 .57 F2 .28
F3 .06 .13 F3 .35 .17
F4 .65 .23 .49
Note. Bolded =factor loading above .5. BPNSFS =Basic Psychological Need Satisfaction and Frustration Scales.
VALIDITY OF NEED FRUSTRATION SCALES 131

with factors reflecting relatedness, competence, and two different
autonomy dimensions, along with two method factors, one each for
positively and negatively keyed items (see Figure 1c). To the best
of our knowledge, this model of the BPNSFS has never been tested
in the existing published literature, whereas two-factor, three-factor,
and six-factor confirmatory factor analysis (CFA) solutions have
been repeatedly tested (e.g., Frielink et al., 2019;Kuźma et al.,
2020;Nishimura & Suzuki, 2016).
The Present Study
We believe the conceptual issues we describe above are compelling
on their own, yet empirical investigation would be helpful. In this
study, we combined data from eight different samples (N=3,692;
individual sample ns ranged from 153 to 926) and conducted a
series of exploratory and confirmatory factor analyses, as well
as external validation of the factor models, to further address our
concerns.
Issue A: Are the Satisfaction and Frustration scales more likely to
reflect construct-relevant substantive variance or reflect construct-
irrelevant variance associated with item-keying direction? As dis-
cussed above, when purported construct-relevant substantive variance
is completely confounded with item-keying direction, as is the case
with the BPNSFS, standard analytic techniques are incapable of
dispositively adjudicating this kind of issue. As a result, this present
article looks for evidence that is suggestive but not fully dispositive.
The main such evidence is in the BPNSFS item contents themselves:
the “frustration”items appear to lack thwarting content, with some
clearly being mirror-opposite versions of satisfaction items (e.g.,
“Ifeelconfident that I can do things well”and “I have serious
doubts about whether I can do things well”).
There are a few suggestive empirical analyses that we can conduct.
First, if the positively keyed (positively valenced) BPNSFS items are
disproportionately correlated with the positively valenced items of
external variable scales and/or the negatively keyed (negatively
valenced) BPNSFS items are disproportionately associated with
the negatively valenced items of external variable scales, that would
be suggestive evidence that apparent differences between the
BPNSFS Frustration and Satisfaction scales are at least partly due to
item-keying effects. For an exemplar of this approach to investi-
gating item valence method effects, see Kam and Meyer (2015),
who observed that the positive items of the Rosenberg Self-Esteem
(RSE; Rosenberg, 1965) scale were disproportionately associated
with the positively valenced items of a variety of external variable
measures, and vice versa for the negative items of the RSE and
negative items of those external variable measures. In three samples,
we had data for other scales, related to well-being, that are commonly
viewed as unidimensional but which some past studies have observed
to split into two keying direction factors. We partitioned each of these
other unidimensional well-being scales into the two item keying
factors, specifically as observed in prior studies, with one scale for
positively valenced items (indicating higher well-being; e.g., from
the RSE, “On the whole, I am satisfied with myself”) and one for
negatively valenced items (indicating lower well-being; e.g., from
the RSE, “All in all, I am inclined to think that I am a failure”).
Then, we examined whether they were differentially associated
with the BPNSFS Satisfaction and Frustration scales. We predicted
that the BPNSFS satisfaction items (relative to the frustration items)
would be more strongly associated with mini-scales of positively
valenced items (higher well-being) and the frustration items would
be more strongly associated with mini-scales of negatively valenced
items (lower well-being).
Finally, and potentially more dispositive, Chen’s (2013) finding
that the BPNSFS item pool was characterized by two item-keying
factors (positive items and negative items) suggests that the variance
of the items does not reflect the specific conceptual model of basic
needs theory: competence, relatedness, and autonomy should be
expected to be the highest order factors, not the item-keying factors.
Here, we will attempt to replicate this finding. We conducted
exploratory and confirmatory factor analyses of the BPNSFS to
see whether we observe that item-keying direction factors emerge
more strongly than do factors of the three basic needs dimensions.
We also compare two higher order models: one with three higher
order substantive factors (competence, relatedness, and autonomy)
and one with two higher order method factors (positively and
negatively keyed items).
Issue B: Are the two Autonomy factors incommensurable? Even if
we find evidence that the BPNSFS “frustration”items are primarily
a different keying method of assessing satisfaction, that does not
necessarily mean that the 3-scale collection proposed initially by
Chen (2013) is valid. As discussed earlier, our own investigation of
the results presented by Chen (2013), alongside our own conceptual
interpretation of the items, leads us to believe that the two Autonomy
factors perhaps should not be combined into a single scale: Instead,
they may represent two incommensurable autonomy satisfaction
dimensions, one of which fails to capture the self-endorsement
content critical to current BPNT conceptualization of autonomy.
To investigate this hypothesis, we conduct two sets of analyses.
First, we attempt to use new data to investigate Chen’s(2013)main
confirmatory factor analytic model: the three substantive need
domain factors, accounting for item-keying direction with method
factors. Based on the data we observed in Chen (2013), we predicted
that this model would reveal that all of the Competence and Related-
ness items would have strong loadings (>.5) on their respective
factors, but that many of the Autonomy items would not load as
well on an Autonomy factor.
Second, extending this line of analysis, we compare confirmatory
models with either three substantive factors (Competence, Related-
ness, and Autonomy) or with four substantive factors (Competence,
Relatedness, and two different Autonomy dimensions), both models
also include two method factors (positive and negative items). We
predicted that the model parsing Autonomy into two separate dimen-
sions would yield more robust factors.
Third, and finally, our interpretation of the items led us to theorize
that the Autonomy Satisfaction items would strongly relate to self-
endorsement, but the Autonomy Frustration items would not
strongly relate to such content. In two samples, we identified scales
that have been previously used in the literature to measure aspects of
self-endorsement, namely the Index of Autonomous Functioning
(Weinstein et al., 2012), which contains an Authorship/Self-Con-
gruence subscale that aims to capture the self-endorsement aspects
of autonomy, and the Meaning in Life questionnaire (Steger et al.,
2006), which contains a Presence subscale that is heavily related to
Authorship/Self-Congruence (Weinstein et al., 2012). We predicted
that the Autonomy Satisfaction scale would correlate much more
strongly with these external validators than would the Autonomy
Frustration scale.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
132 MURPHY ET AL.

Method
Data
The BPNSFS (Chen et al., 2015) was developed exclusively in
undergraduate samples. All eight of our present samples contain
data from university students, with the exception of Sample 6, which
was collected in an online community sample. In the case of
Samples 6–8, data were also collected from romantic partners or
friends (some of whom were university students and some of whom
were not). Table 2 summarizes: (a) the sample sizes and basic
demographics, (b) the version of the BPNSFS administered, and
(c) the non-BPNSFS measures used in our analyses. Institutional
review board approval was obtained from the University of North
Carolina at Chapel Hill for Sample 1 and from the University of
Houston for Samples 2–8.
Measures
BPNSFS
All instances of the BPNSFS were collected at baseline except in
Sample 2, in which the BPNSFS was administered after either a
supraliminal, authenticity-focused prime, or control condition. One
Competence Satisfaction item was accidentally omitted from data
collection in Sample 3. In Samples 1–6, the BPNSFS was rated on
a 5-point Likert scale; in Samples 7 and 8, a 7-point scale was used.
Some data from Sample 6 have been previously published in (Baker
et al., 2020).
To harmonize data across samples, we first divided items by the
maximum possible score on the scale for each sample (either 5 or 7),
in turn converting items into the proportion of maximum possible
response. Then, we combined data across samples and treated items
as continuous given that there were more than five possible response
options. Sample 2 administered the BPNSFS within the context of a
social priming condition to half of the participants. We accounted for
this study feature by categorizing the data in terms of two groups:
(a) all samples with the exception of sample 2 +participants from
sample that did not receive the prime and (b) conditions from Sample
2 that received the prime. We used this variable as a cluster variable
to account for nonindependence within each of the subgroups.
External Validators
In our data sets, we had measures of multiple well-being con-
structs which (a) are typically treated as unidimensional measures
but (b) have been observed in various past validity investigations to
split into two item-keying direction factors. Thus, for investigating
Issue A, we used the following measures of well-being, broken into
a priori, rational scales with only “positive-valence”items and only
“negative-valence”items based on evidence from past studies: RSE
scale (Rosenberg, 1965; item valence split observed in Kam &
Meyer, 2015; positive items, α=.81; negative items, α=.83; one
negative item was inadvertently missing from our study protocol),
Center for Epidemiological Studies–Depression scale (Radloff,
1977; item valence split observed in Miller et al., 1997; positive
items, α=.84; negative items, α=.84), Perceived Stress Scale
(Cohen et al., 1983; item valence split observed in Taylor, 2015;
positive items for Samples 1 and 3, αs=.73 and .74; negative
items, αs=.75 and .86), Self-Compassion Scale–Short Form
(Raes et al., 2011; item valence split observed in Hayes et al.,
2016; positive items, α=.78; negative items, α=.87). We also had a
version of the Need Satisfaction Scale (La Guardia et al., 2000)
modified to reference one’s romantic relationship; prior work has
not investigated item valence in the factor structure of this specific
version of the scale, but we conducted a two-factor EFA and
observed that the two-factor solution produced two factors based
on item keying direction, so we broke it apart into positive and
negative scales on that basis (positive items aggregated across need
domains, α=.93; negative items across need domains, α=.82).
Other than being measures of well-being and having at least three
items for each coding direction, these scales were not chosen for any
specific substantive reasons; our item keying direction perspective
would theoretically apply to any such well-being measures. For ease
of interpretation, we scored or reverse-scored all items so that higher
ratings always indicated higher well-being. Similarly, we reverse-
scored all the BPNSFS frustration scale items, so that all BPNSFS
items indicated higher need satisfaction.
For testing Issue B, we used the Presence subscale of the Meaning
in Life scale (Steger et al., 2006;α=.90), which assesses the degree
to which respondents feel they have clear meaning/purpose in their
lives, and the Authorship Self-Congruence scale of the Index of
Autonomous Functioning (Weinstein et al., 2012;α=.84), which
assesses the degree to which respondents feel their actions are
congruent with their identities and desires.
Data Analyses
We conducted all EFAs in R using the psych package (Revelle,
2021) and with maximum likelihood and geomin rotation. We
conducted all CFAs with Mplus Version 7.31 and with the robust
maximum likelihood estimator. This study was not preregistered.
Data and code are available at: https://osf.io/6r2pb/.
Results
Issue A
Testing Item Coding Direction Effect on Nomological
Networks
Here, Satisfaction refers to all Satisfaction items (i.e., positively
keyed items) from Autonomy, Competence, and Relatedness
domains, whereas Frustration refers to all Frustration items (i.e.,
negatively keyed items) from Autonomy, Competence, and Related-
ness domains.
For self-compassion, Satisfaction was more strongly correlated
with the positively valenced items of the SCS-Short than was
Frustration (rs were .35 and .09, respectively; Steiger’sz=4.22),
but Frustration was more strongly correlated with the negatively
valenced items (rs were .67 and .34, respectively; Steiger’sz=
6.50).
3
For depression, Satisfaction and Frustration did not differ
in their correlations with the positive items of the Center for
Epidemiologic Studies Depression (CESD; rs were .59 and .56,
respectively; Steiger’sz=.49), but Frustration was slightly more
strongly correlated with the negative items of the CESD (r=.62)
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
3
As a reminder, we scored or reverse-scored all items so that higher
ratings always indicated higher well-being. Similarly, we reverse-scored all
the BPNSFS frustration scale items, so that all BPNSFS items indicated
higher need satisfaction.
VALIDITY OF NEED FRUSTRATION SCALES 133

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Table 2
Eight Samples, Measures, and Demographics
Sample BPNSFS version Other measures
Demographics
Age M(SD) Sex % Race/ethnicity %
Sample 1 (N=153) Standard (6 subscale αs from .74 to .90) •Rosenberg self-esteem
•CESD depression
•Perceived stress
•Meaning in life
18.15 (.59) 66.9% Female White/Caucasian: 56.2%; Black/African American:
19.5% East Asian: 11.8%; Latino: 8.3%
Hispanic: 7.7%; South Asian: 4.1%; American
Indian or Alaskan Native: 0.6%; Pacific Islander
or Native Hawaiian: 0.6%; Other: 2.4%
Sample 2 (N=569) Standard, but after either an authenticity prime
or control (αs from .74 to .86)
21.72 (4.24) 78.98% Female Hispanic/Latino: 31.50%; Asian: 27.96%;
Caucasian: 22.48%; African American: 12.39%;
Middle Eastern: 3.89%; Other: 1.77%
Sample 3 (N=263) Referenced last 2 weeks (αs from .64 to .85) •Perceived stress
•Self-compassion
•IAF authorship
22.45 (5.72) 63.50% Female Hispanic/Latino: 36.50%; White/Caucasian:
36.50%; Asian: 26.62%; Black/African
American: 11.79%; Multirace: 5.70%; Native
American/American Indian: 2.66%; Other
ethnicity: 16.73%
Sample 4 (N=379) Standard (αs from .81 to .91) 23.38 (6.30) 86.31% Female White/Caucasian: 37.43%; Asian: 23.74%;
Multiethnic: 10.61%; Black/African American:
7.82%; Native Hawaiian/Pacific Islander: .84%;
Native American/American Indian: .56%; Other:
18.99%
Sample 5 (N=294) Romantic relationship (αs from .78 to .86) 22.91 (7.60) 51.36% Female White/Caucasian: 35.03%; Asian: 19.73%; Black/
African American: 12.59%; Multiethnic: 5.78%;
Native American/American Indian: 1.70%;
Native Hawaiian/Pacific Islander: 0.34%; Other:
24.83%
Sample 6 (N=678) Romantic relationship (αs from .80 to .90) •Need Satisfaction
(La Guardia scale)
43.71 (14.56) 70.80% Female White/Caucasian: 76.47%; Black/African American:
11.32%; Hispanic/Latino: 8.86%; Asian: 5.59%;
Multiethnic or Other: 3.82%; Native American/
American Indian: 2.35%; Native Hawaiian/Pacific
Islander: 0.44%
Sample 7 (N=926) Romantic relationship (αs from .78 to .90) 21.45 (3.89) 83.78% Female Hispanic/Latino: 30.24%; Asian: 25.95%; White/
Caucasian: 23.96%; Black/African American:
12.54%; Multiethnic: 4.14%; Native American/
American Indian: 0.56%; Native Hawaiian/Pacific
Islander: 0.22%; Other: 2.35%
Sample 8 (N=432) Friendships (αs from .79 to .90) 23.10 (5.23) 30.45% Female Hispanic/Latino: 34.56%; White/Caucasian:
23.11%; Asian: 20.30%; Black/African
American: 13.17%; Multiethnic: 4.32%; Native
Hawaiian/Pacific Islander: 0.86%; Native
American/American Indian: 0.43%; Other: 3.24%
Note. CESD =Center for Epidemiological Studies–Depression scale; IAF Authorship =Authorship/Self Congruence subscale of the Index of Autonomous Functioning; BPNSFS =Basic
Psychological Need Satisfaction and Frustration Scales.
134 MURPHY ET AL.

than was Satisfaction (r=.53; Steiger’sz=1.63). For self-esteem,
Satisfaction and Frustration did not differ in their associations with
the positive items of the RSE (rs were .59 and .57, respectively;
Steiger’sz=.33), but Frustration was more strongly correlated with
thenegativeRSEitems(r=.67) than was Satisfaction (r=.57,
Steiger’sz=1.90).
We had the Perceived Stress Scale in two data sets and present the
mean rs here. BPNSFS Satisfaction and Frustration were roughly
equally correlated with the positive items (mean rs were .48 and
.53, respectively; mean Steiger’sz=−.66), but Frustration was
much more strongly correlated with the negative items than was
Satisfaction (mean rs were .65 and .46, respectively; mean Steiger’s
z=3.58). Also, Satisfaction was substantially more strongly corre-
lated with the positive items of the La Guardia Need Satisfaction total
score than was Frustration (rs were .78 and .58, respectively; Steiger’s
z=7.80), but Frustration was much more strongly correlated with
the negative items than was Satisfaction (rs were .75 and .45,
respectively; Steiger’sz=11.16).
Factor Analyses
Parallel analysis indicated that the BPNSFS contained six factors
and three components, whereas the MAP test indicated four factors.
As shown in Table 1, an EFA extracting two factors from the BPNSFS
items produced a factor with all positively keyed items and a factor
with all negatively keyed items (see also Chen, 2013). When we
extracted three factors, the first two factors continued to be entirely
characterized by item-keying direction; some Relatedness items
loaded on the third factor, but none as strongly as on their respective
item-keying factors. These findings suggest that the first major
dimensions that emerge in factor analysis are due to item keying.
When we extracted four factors, we found: (a) a Frustration factor
composed of Competence and Relatedness Frustration items, (b) a
Satisfaction factor composed of Autonomy and Relatedness Satisfac-
tion items, (c) a Relatedness dimension composed of both Frustration
and Satisfaction items, and (d) an Autonomy Frustration factor. These
findings highlight the intertwined nature of method and substantive
variance in the BPNSFS.
When we extracted six factors in either an EFA or CFA framework
(see Supplemental Materials, for details), each factor corresponded
to the factors reported by Chen et al. (2015). That said, the patterning
of factor intercorrelations was revealing. The magnitudes of factor
correlations (a) among negatively keyed (frustration) dimensions
(EFA rs ranged from .64 to .70), on the one hand, and (b) among
positively keyed (satisfaction) dimensions (rs ranged from .66 to .75),
on the other, were much stronger than were the factor correlations
between shared substantive dimensions (rs were .17 [competence
satisfaction–competence frustration], .34 [relatedness satisfaction–
relatedness frustration], and .38 [autonomy satisfaction–autonomy
frustration]). This finding highlights the predominant role of method
variance on the BPNSFS structure.
As a more direct test of the role of method variance on the
BPNSFS structure, we tested a set of higher order models using
CFA. The first model contained six lower order factors like in Chen
et al. (2015), but with three higher order dimensions, one for each
substantive dimension (e.g., autonomy). We tested this model
because implicit in the conceptual model of frustration and satisfac-
tion in BPNT is that the substantive need domains should reflect
a higher order construct grouping than would satisfaction versus
frustration across all the different need domains. This model did not fit
the data well (root-mean-square error of approximation [RMSEA] =
.064; comparative fit index [CFI] =.924; Tucker–Lewis index [TLI]
=.915; Akaike information criterion [AIC] =−29,218; Bayesian
information criterion [BIC] =−28,734). Further, the higher order
factors’correlations exceeded 1 in most cases (rs ranged from .94 to
1.11), hinting that keying method variance inflated the correlations
among substantive dimensions. In contrast, a model with six lower
order factors and two higher order factors for negatively and
positively keyed items, respectively, fits the data reasonably well
(RMSEA =.052; CFI =.950; TLI =.944; AIC =−30,116; BIC =
−29,631). These models (see Supplemental Tables S1 and S2; also
see the Open Science Framework page for all model outputs)
suggest that higher order dimensions in the BPNSFS are more
attributable to method variance than they are to substantive
variance.
Issue B
Factor Analyses
As shown in Table 3, a CFA with three substantive factors
(Competence, Relatedness, Autonomy) and two item-keying fac-
tors (Positive and Negative) fitwell(RMSEA=.033; CFI =.981;
TLI =.977). There, none of the Autonomy Satisfaction items
loaded above .50 (the factor loading threshold adopted in Chen et al.,
2015) on the Autonomy factor, whereas all other items loaded
above .5 on their respective factors, with the exception of one
Competence item (loading =.48; see also Chen, 2013). The CFA with
four substantive factors (Competence, Relatedness, two different
Autonomy dimensions) and two item-keying factors fit nearly
equivalently (RMSEA =.033; CFI =.982; TLI =.978; AIC =
−32,575; BIC =−31,941). In that CFA, though, two of the four
Autonomy Frustration items still did not load above .5 on their
substantive factor, indicating that our four-factor model also does
not have adequately robust fit. Also, the two Autonomy factors
were highly correlated (r=.82), which could indicate they are less
distinguishable than we predicted. Nonetheless, the factor loadings
for almost all the Autonomy items were generally slightly stronger
in that model, indicating that a model with four substantive factors
may be somewhat more robust, even if not adequately so.
Conceptual Incommensurability of the
Two Autonomy Scales
In Sample 3, Autonomy Satisfaction was substantially related to
the Authorship/Self-congruence scale of the Index of Autonomous
Functioning (r=.39, p<.001), whereas Autonomy Frustration
was not (r=−.08, p=.20). In Sample 1, Autonomy Satisfaction
was strongly related to perceiving the presence of meaning in one’s
life (r=.52, p<.001), whereas Autonomy Frustration was not
(r=.15, p=.06).
Discussion
The conceptual and empirical evidence we have offered sug-
gests, but cannot dispositively conclude, that the BPNSFS is
not valid for use as a 6-scale collection. Most critically, the
BPNSFS probably does not validly measure the dual-dimension
theory of need frustration and need satisfaction introduced by
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
VALIDITY OF NEED FRUSTRATION SCALES 135

Vansteenkiste and Ryan (2013), and its use for that purpose might
be hindering sound exploration of this interesting theoretical
domain. For instance, many published findings regarding well-
being that rely on the BPNSFS may be at least partly driven by
methodological artifact. Our hope is that this article will encourage
efforts to develop new measures that validly assess the distinction
between need frustration and need satisfaction.
Issue A: The Confounding of Item-Keying Direction
With Purported Substantive Variance
The largest concern is that legions of users and adapters of the
BPNSFS (e.g., see the BPNSFS manual: Van der Kaap-Deeder et al.,
2020) may be mistaking item-keying direction method factors
for substantive frustration and satisfaction factors. The BPNSFS
“frustration”items lack face validity as measuring a construct
distinguished from low need satisfaction in current BPNT (e.g.,
Vansteenkiste et al., 2020): They have very little active, thwarting
content. Moreover, we observed that item-keying direction was an
overwhelming source of the covariance in the BPNSFS item pool,
more so than the three basic psychological needs dimensions them-
selves. This result is inconsistent with the conceptual framework
of BPNT: The three needs domains (Competence, Relatedness,
Autonomy) need to emerge as the highest order factors. We interpret
this result as strongly suggestive of item-keying method bias over-
whelming substantive variance.
Divergences in the nomological networks of the satisfaction and
“frustration”items are likely at least partially attributed to item keying
direction aswell. Our analyses demonstrated that the frustration scales
more strongly relate to other negatively valenced items from a variety
of measures of well-being (depression, self-compassion, self-esteem,
perceived stress, a different measure of basic need satisfaction),
whereas the satisfaction items, though not consistently, related more
strongly to other positively valenced items of two of those measures.
This observation could reasonably explain the nomological network
differences between satisfaction and frustration observed by Chen
et al. (2015),findings that have since been widely referenced as
demonstrating the distinct importance of need frustration. Chen
et al. (2015) observed that need satisfaction was particularly
associated with the Satisfaction with Life Scale (Diener et al.,
1985), which has no negatively valenced items, and the Subjective
Vitality Scale (Ryan & Frederick, 1997), which has one negatively
valenced item. In contrast, Chen et al. (2015) found that need
frustration was particularly associated with the Center for Epi-
demiological Studies–Depression Scale (Radloff, 1977), which
consists primarily of negatively valenced items. Given that most
measures of positive well-being are disproportionately measured
by positively valenced items, and most measures of ill-being are
disproportionately measured by negatively valenced items, we
would expect the BPNSFS need satisfaction and need frustration
scales to demonstrate this kind of nomological network divergence
due to method bias alone, especially in the case of ill-being variables.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Table 3
Confirmatory Factor Analyses of BPNSFS, 3–4 Substantive Factors and 2 Item Keying Factors (Combined Sample N =3,692)
Item Item keying
Three substantive factors +keying factors Four substantive factors +keying factors
ARC+−AS AF R C +−
Item 1 (autonomy satisfaction) +.49 .54 .57 .48
Item 2 (autonomy satisfaction) +.34 .61 .44 .54
Item 3 (autonomy satisfaction) +.36 .64 .46 .57
Item 4 (autonomy satisfaction) +.42 .56 .51 .49
Item 5 (autonomy frustration) −.76 .26 .77 .25
Item 6 (autonomy frustration) −.62 .41 .62 .44
Item 7 (autonomy frustration) −.66 .36 .66 .36
Item 8 (autonomy frustration) −.75 .27 .76 .26
Item 9 (relatedness satisfaction) +.67 .39 .67 .41
Item 10 (relatedness satisfaction) +.69 .47 .67 .51
Item 11 (relatedness satisfaction) +.73 .42 .71 .46
Item 12 (relatedness satisfaction) +.64 .43 .62 .46
Item 13 (relatedness frustration) −.58 .55 .60 .52
Item 14 (relatedness frustration) −.58 .53 .60 .51
Item 15 (relatedness frustration) −.59 .56 .61 .52
Item 16 (relatedness frustration) −.55 .38 .57 .35
Item 17 (competence satisfaction) +.62 .51 .64 .49
Item 18 (competence satisfaction) +.58 .53 .61 .52
Item 19 (competence satisfaction) +.48 .54 .50 .55
Item 20 (competence satisfaction) +.52 .49 .53 .49
Item 21 (competence frustration) −.70 .40 .74 .33
Item 22 (competence frustration) −.58 .46 .62 .41
Item 23 (competence frustration) −.63 .43 .68 .37
Item 24 (competence frustration) −.53 .47 .56 .43
Factor intercorrelations A R AS AF R
.72 .82
.75 .67 .66 .72
.78 .72 .69
Note. Bolded =factor loading above .5. A =autonomy; R =relatedness; AS =autonomy satisfaction; AF =autonomy frustration; +=positively keyed;
−=negatively keyed; BPNSFS =Basic Psychological Need Satisfaction and Frustration Scales.
136 MURPHY ET AL.

Issue B: The Conceptual Incommensurability of the
Two Autonomy Factors
Chen (2013) concluded that the “frustration”items for each
need domain should be understood as reflecting the same substan-
tive dimension as the “satisfaction”items, presenting a three-scale
collection, not a 6-scale collection. Given the present arguments
and analyses arguing that the apparent distinction between the
frustration and satisfaction items is likely primarily due to item-
keying method bias, the natural assumption might be that the field
needs to simply collapse them together into the three satisfaction
scales (Competence, Relatedness, Autonomy) originally proposed
by Chen (2013). Our analyses, though, raise some concerns that
this kind of aggregation should only be done for the Competence
and Relatedness domains: the “Autonomy Satisfaction”and
“Autonomy Frustration”scales possibly should not be combined
into a single scale. Across our samples, we replicated the data we
observed in Chen (2013): accounting for item-keying direction, all
but one of the Competence and Relatedness items loaded well on
their respective factors, whereas none of the Autonomy Frustration
items loaded well on an Autonomy factor.
Moreover, our conceptual interpretation of the Autonomy Sat-
isfaction and Frustration items led us to hypothesize that they may
markedly diverge in their conceptual content. Specifically, we
hypothesized that the Autonomy Satisfaction items assess aspects
of self-endorsement, which is critical content in the BPNT concep-
tualization of autonomy, whereas the Autonomy Frustration items
do not assess such content. This prediction was supported when
we examined the respective correlations of these two scales with
external variables that reflect self-endorsement (i.e., Presence subscale
of the Meaning in Life Questionnaire, Authorship/Self-Congruence
subscale of the Index of Autonomous Functioning). Still, we encour-
age further research into the potential differences between Autonomy
Satisfaction and Autonomy Frustration, as our data were limited by
having relevant external variables in only two samples and also by the
fact that these external variables were assessed largely by positively
valenced items, inherently limiting our ability to evaluate the substan-
tive versus item-keying divergences between the autonomy scales
in this regard.
We do not challenge the theoretical and empirical tradition within
SDT, which has established autonomy as a singular overarching basic
need. That is, we are not suggesting there are actually two different
needs for autonomy. Instead, the present analysis of these particular
scales suggests that, because of the way the BPNSFS was constructed:
(a) the Autonomy Frustration items may not align with the BPNT
conceptualization of autonomy, which emphasizes self-endorsement
and (b) aggregating the Autonomy Satisfaction and Autonomy
Frustration items together into a single scale might not result in a
robust unitary measurement dimension. That said, although pars-
ing the two autonomy dimensions apart in our model with four
substantive factors did lead to more robust factors, it did not offer
meaningful improvement in model fit, and the two autonomy
dimensions were very strongly correlated. Thus, even though there
are grounds for concern regarding the commensurability of the two
autonomy scales and the construct validity of the autonomy frustra-
tion items, it is not clear that our hypothesized four substantive factor
model is adequate to overcome these concerns.
In sum, we conclude a three-factor (plus method factors) model is
better justified than a six-factor model. We also observed some
evidence that a four-factor (plus method factors) model might be
most appropriate, both theoretically and psychometrically, yet the
results of the present analyses were not consistent and compelling
enough for us to confidently assert the superior usefulness of such a
model. Further research is needed to more effectively evaluate the
construct validity of the autonomy items in the BPNSFS, as our data
suggest current uses of the BPNSFS might be inadvertently mis-
aligned with the current BPNT conceptualization of autonomy.
Limitations and Future Directions
The main limitation of our empirical investigation is that our eight
samples were not originally collected with the aim of evaluating the
validity of the BPNSFS. Although we report results for all relevant
samples/measures to which we had access, only a few of our samples
were amenable to investigating (a) the effects ofitem coding direction
on the nomological networks of the BPNSFS scales and (b) the
substantive incommensurability of the two Autonomy scales. More
importantly, in these eight samples (and perhaps in all of the many
existing data sets with the BPNSFS), item-keying direction was
completely confounded with the purported substantive distinction
between need satisfaction and need frustration. As explained earlier,
this complete confounding prevents us and other researchers from
being able to dispositively determine how much item covariance
is due to item-keying method bias versus due to the substantive
distinction between satisfaction and frustration. In future research,
one way to potentially ameliorate this inherent limitation is to craft
reverse-worded twins for each of the 24 BPNSFS items and then
use that doubled item pool in a series of studies evaluating the
validity of the BPNSFS (for exemplar of this kind of approach, see
Naragon-Gainey & DeMarree, 2017; also, for a newly developed
exploratory graph analysis approach to be used with twinned items,
see Garcia-Pardina et al., 2022).
Another notable limitation of our empirical investigation is that
our samples included only a few of the different versions of the
BPNSFS, all in English and almost all in (demographically diverse)
university student samples. Given the extensive variety of transla-
tions and adaptations of the BPNSFS and its wide usage in many
different types of participant samples, it would be particularly valu-
able to assess the extent to which our findings generalize across a
more diverse range of study designs. We would, though, encourage
readers to consult Chen’s (2013), which focused on international
samples of university students and presented analyses that tended
to corroborate the findings we present here.
On balance, further scrutiny of the BPNSFS is warranted. If our
conclusions are corroborated in future work, the BPNSFS needs to
be substantially revised or a new measure construction needs to be
initiated so that the innovatively reconceptualized construct of need
frustration can be validly assessed.
References
Baker, Z. G., Watlington, E. M., & Knee, C. R. (2020). The role of rapport in
satisfying one’s basic psychological needs. Motivation and Emotion,
44(2), 329-343. https://doi.org/10.1007/s11031-020-09819-5
Bartholomew, K. J., Ntoumanis, N., Ryan, R. M., & Thøgersen-Ntoumani,
C. (2011). Psychological need thwarting in the sport context: Assessing
the darker side of athletic experience. Journal of Sport & Exercise
Psychology,33(1), 75–102. https://doi.org/10.1123/jsep.33.1.75
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
VALIDITY OF NEED FRUSTRATION SCALES 137

Böckenholt, U. (2019). Contextual responses to affirmative and/or reversed-
worded items. Psychometrika,84(4), 986–999. https://doi.org/10.1007/
s11336-019-09680-7
Campbell, D. T., & Fiske, D. W. (1959). Convergent and discriminant
validation by the multitrait-multimethod matrix. Psychological Bulletin,
56(2), 81–105. https://doi.org/10.1037/h0046016
Chen, B. (2013). The universal and culture specific character of basic
psychological need satisfaction [Doctoral dissertation]. Universiteit
Ghent.
Chen, B., Vanst eenkiste, M., Beyers, W., Boo ne, L., Deci, E. L., Duriez, B.,
Lens, W., Matos, L., Mouratidis, A., Ryan, R. M., Sheldon, K. M.,
Soenens, B., Petegem, S. V., & Verstuyf, J. (2015). Basic psychological
need satisfaction, need frustration, and need st rength across four cultures.
Motivation and Emotion,39(2), 216–236. https://doi.org/10.1007/s11031-
014-9450-1
Chen, B., Vansteenkiste, M., Beyers, W., Soenens, B., & Van Petegem, S.
(2013). Autonomy in family decision making for Chinese adolescents:
Disentangling the dual meaning of autonomy. Journal of Cross-Cultural
Psychology,44(7), 1184–1209. https://doi.org/10.1177/0022022113480038
Cohen, S., Kamarck, T., & Mermelstein, R. (1983). A global measure of
perceived stress. Journal of Health and Social Behavior,24(4), 385–396.
https://doi.org/10.2307/2136404
Cordeiro, P., Paixão, P., Lens, W., Lacante, M., & Luyckx, K. (2016). The
Portuguese validation of the basic psychological need satisfaction and
frustration scale: Concurrent and longitudinal relations to well-being and
ill-being. Psychologica Belgica,56(3), 193–209. https://doi.org/10.5334/
pb.252
Costa, S., Ntoumanis, N., & Bartholomew, K. J. (2015). Predicting the
brighter and darker sides of interpersonal relationships: Does psychological
need thwarting matter? Motivation and Emotion,39(1), 11–24. https://
doi.org/10.1007/s11031-014-9427-0
Crasta, D., Rogge, R. D., Maniaci, M. R., & Reis, H. T. (2021). Toward an
optimized measure of perceived partner responsiveness: Development
and validation of the perceived responsiveness and insensitivity scale.
Psychological Assessment,33(4), 338–355. https://doi.org/10.1037/pa
s0000986
Credé, M., Chernyshenko, O. S., Bagraim, J., & Sully, M. (2009). Contextual
performance and the job satisfaction–dissatisfaction distinction: Exam-
ining artifacts and utility. Human Performance,22(3), 246–272. https://
doi.org/10.1080/08959280902970427
Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-
determination in human behavior. Plenum Publishing. https://doi.org/10
.1007/978-1-4899-2271-7
Diener,E.,Emmons,R.A.,Larsen,R.J.,&Griffin, S. (1985). The satisfaction
with life scale. Journal of Personality Assessment,49(1), 71–75. https://
doi.org/10.1207/s15327752jpa4901_13
DiStefano, C., & Motl, R. W. (2009). Personality correlates of method effects
due to negatively worded items on the Rosenberg self-esteem scale.
Personality and Individual Differences,46(3), 309–313. https://doi.org/
10.1016/j.paid.2008.10.020
Frielink, N., Schuengel, C., & Embregts, P. J. (2019). Psychometric
properties of the basic psychological need satisfaction and frustration
scale–intellectual disability (BPNSFS-ID). European Journal of Psy-
chological Assessment,35(1), 37–45. https://doi.org/10.1027/1015-
5759/a000366
Garcia-Pardina, A., Abad, F. J., Christensen, A. P., Golino, H., & Garrido, L.
(2022, May 14). Dimensionality assessment in the presence of wording
effects: A network psychometric and factorial approach. PsyArXiv. https://
doi.org/10.31234/osf.io/7yqau
Hankins, M. (2008). The factor structure of the twelve item General Health
Questionnaire (GHQ-12): The result of negative phrasing? Clinical Practice
and Epidemiology in Mental Healt h,4(1), Article 10. https://doi.org/10
.1186/1745-0179-4-10
Hayes, J. A., Lockard, A. J., Janis, R. A., & Locke, B. D. (2016). Construct
validity of the Self-Compassion Scale-Short Form among psychotherapy
clients. Counselling Psychology Quarterly,29(4), 405–422. https://doi.org/
10.1080/09515070.2016.1138397
Kam, C., & Meyer, J. P. (2012). Do optimism and pessimism have different
relationships with personality dimensions? A re-examination. Personality
and Individual Differences,52(2), 123–127. https://doi.org/10.1016/j.paid
.2011.09.011
Kam, C. C., & Meyer, J. P. (2015). Implications of item keying and item
valence for the investigation of construct dimensionality. Multivariate
Behavioral Research,50(4), 457–469. https://doi.org/10.1080/00273171
.2015.1022640
Kuźma, B., Szulawski, M., Vansteenkiste, M., & Cantarero, K. (2020).
Polish adaptation of the basic psychological need satisfaction and frustra-
tion scale. Frontiers in Psychology,10, Article 3034. https://doi.org/10
.3389/fpsyg.2019.03034
La Guardia, J. G., Ryan, R. M., Couchman, C. E., & Deci, E. L. (2000).
Within-person variation in security of attachment: A self-determination
theory perspective on attachment, need fulfillment, and well-being. Journal
of Personality and Social Psychology,79(3), 367–384. https://doi.org/10
.1037/0022-3514.79.3.367
Liga, F., Ingoglia, S., Cuzzocrea, F., Inguglia, C., Costa, S., Lo Coco, A., &
Larcan, R. (2020). The basic psychological need satisfaction and frustra-
tion scale: Construct and predictive validity in the Italian context. Journal
of Personality Assessment,102(1), 102–112. https://doi.org/10.1080/
00223891.2018.1504053
Lindwall, M., Barkoukis, V., Grano, C., Lucidi, F., Raudsepp, L., Liukkonen,
J., & Thøgersen-Ntoumani, C. (2012). Method effects: The problem with
negatively versus positively keyed items. Journal of Personality Asse ssment,
94(2), 196–204. https://doi.org/10.1080/00223891.2011.645936
Michaelides, M. P., Zenger, M., Koutsogiorgi, C., Brähler, E., Stöbel-
Richter, Y., & Berth, H. (2016). Personality correlates and gender
invariance of wording effects in the German version of the Rosenberg
self-esteem scale. Personality and Individual Differences,97,13–18.
https://doi.org/10.1016/j.paid.2016.03.011
Miller, T. Q., Markides, K. S., & Black, S. A. (1997). The factor structure of
the CES-D in two surveys of elderly Mexican Americans. The Journals of
Gerontology Series B: Psychological Sciences and Social Sciences,52(5),
S259–S269. https://doi.org/10.1093/geronb/52B.5.S259
Naragon-Gainey, K., & DeMarree, K. G. (2017). Structure and validity of
measures of decentering and defusion. Psychological Assessment,29(7),
935–954. https://doi.org/10.1037/pas0000405
Neubauer, A. B., Kramer, A. C., & Schmiedek, F. (2022). Assessing
domain-general need fulfillment in children and adults: Introducing
the General Need Satisfaction and Frustration scale. Psychological
Assessment. Advance online publication. https://doi.org/10.1037/pas
0001169
Nishimura, T., & Suzuki, T. (2016). Basic psychological need satisfaction
and frustration in Japan: Controlling for the big five personality traits.
Japanese Psychological Research,58(4), 320–331. https://doi.org/10
.1111/jpr.12131
Nunnally, J. C. (1978). Psychometric theory (2nd ed.). McGraw-Hill.
Podsakoff, P. M., MacKenzie, S. B., Lee, J. Y., & Podsakoff, N. P. (2003).
Common method biases in behavioral research: A critical review of the
literature and recommended remedies. Journal of Applied Psychology,
88(5), 879–903. https://doi.org/10.1037/0021-9010.88.5.879
Quilty,L.C.,Oakman,J.M.,&Farvolden,P.(2007).Behavioural
inhibition, behavioural activation, and the preference for familiarity.
Personality and Individual Differences,42(2), 291–303. https://doi.org/
10.1016/j.paid.2006.07.007
Quilty, L. C., Oakman, J. M., & Risko, E. (2006). Correlates of the Rosenberg
self-esteem scale method effects. Structural Equation Modeling,13(1),
99–117. https://doi.org/10.1207/s15328007sem1301_5
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
138 MURPHY ET AL.

Radloff, L. S. (1977). The CES-D scale: A self-report depression scale for
research in the general population. Applied Psychological Measurement,
1(3), 385–401. https://doi.org/10.1177/014662167700100306
Raes, F., Pommier, E., Neff, K. D., & Van Gucht, D. (2011). Construction
and factorialvalidation of a short form of the self-compassionscale. Clinical
Psychology & Psychotherapy,18(3), 250–255. https://doi.org/10.1002/
cpp.702
Ray,J.V.,Frick,P.J.,Thornton,L.C.,Steinberg,L.,&Cauffman,E.(2016).
Positive and negative item wording and its influence on the assessment of
callous-unemotional traits. Psychological Assessment,28(4), 394–404.
https://doi.org/10.1037/pas0000183
Revelle, W. (2021). psych: Procedures for personality and psychological
research (Version 2.1.9) [Computer software] . https://CRAN.R-project
.org/package=psych
Rosenberg, M. (1965). Society and adolescent self-image. Princeton
University. https://doi.org/10.1515/9781400876136
Ryan, R. M., & Deci, E. L. (2017). Self-determination theory: Basic
psychological needs in motivation, development, and wellness. Guilford
Publishing. https://doi.org/10.1521/978.14625/28806
Ryan, R. M., & Frederick, C. (1997). On energy, personality, and health:
Subjective vitality as a dynamic reflection of well-being. Journal of
Personality,65(3), 529–565. https://doi.org/10.1111/j.1467-6494.1997
.tb00326.x
Šakan, D. (2020). Validation of the basic psychological need satisfaction and
frustration scale (BPNSFS) on adolescents in Serbia. Current Psychology,
41(4), 2227–2240. https://doi.org/10.1007/s12144-020-00742-z
Steger, M. F., Frazier, P., Oishi, S., & Kaler, M. (2006). The meaning in life
questionnaire: Assessing the presence of and search for meaning in life.
Journal of Counseling Psychology,53(1), 80–93. https://doi.org/10.1037/
0022-0167.53.1.80
Taylor, J. M. (2015). Psychometric analysis of the ten-item perceived stress
scale. Psychological Assessment,27(1), 90–101. https://doi.org/10.1037/
a0038100
Tomas, J. M., & Oliver, A. (1999). Rosenberg’s self-esteem scale: Two
factors or method effects. Structural Equation Modeling,6(1), 84–98.
https://doi.org/10.1080/10705519909540120
Van der Kaap-Deeder, J., Soenens, B., Ryan, R. M., & Vansteenkiste, M.
(2020). Manual of the basic psychological need satisfaction and frustration
scale (BPNSFS). Ghent University.
Vansteenkiste, M., & Ryan, R. M. (2013). On psychological growth and
vulnerability: Basic psychological need satisfaction and need frustration
as an unifying principle. Journal of Psychotherapy Integration,3(3),
263–280. https://doi.org/10.1037/a0032359
Vansteenkiste, M., Ryan, R. M., & Soenens, B. (2020). Basic psychological
need theory: Advancements, critical themes, and future directions. Moti-
vation and Emotion,44(1), 1–31. https://doi.org/10.1007/s11031-019-
09818-1
Weinstein, N., Przybylski, A. K., & Ryan, R. M. (2012). The index of
autonomous functioning: Development of a scale of human autonomy.
Journal of Research in Personality,46(4), 397–413. https://doi.org/10
.1016/j.jrp.2012.03.007
Woods, C. M. (2006). Careless responding to reverse-worded items: Im-
plications for confirmatory factor analysis. Journal of Psychopathology
and Behavioral Assessment,28(3), 189–194. https://doi.org/10.1007/
s10862-005-9004-7
Received September 24, 2021
Revision received September 23, 2022
Accepted September 28, 2022 ▪
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
VALIDITY OF NEED FRUSTRATION SCALES 139