Validating a real‐time, single‐item subjective well‐being measure with the Positive and Negative Affect Schedule (PANAS) in US and Indian samples

Abstract

The real‐time, single‐item subjective well‐being (SWB) question could reflect the participants’ positive and negative psychological states in our experimental setting. In other words, the single‐item SWB measure exhibited sufficient construct validity for both major subcomponents of the Positive and Negative Affect Schedule.

“How do you feel right now?” or an equivalent question is frequently used to begin and guide both everyday conversations and dialogues in clinical settings. Although such single questions measure real‐time subjective well‐being (SWB), ¹ their validation has yet to be explored. Instead, assessments of individuals’ mental health states and emotional well‐being employ multi‐item questionnaires such as the Kessler‐6 and Psychological Distress Scale (six questions) ² , ³ or the Positive and Negative Affect Schedule (PANAS, 10 or 20 questions). ⁴ , ⁵ To address this gap, we aimed to validate a real‐time, single‐item SWB measure on the positive affect scale (PAS) and negative affect scale (NAS) of the 20‐item PANAS and evaluate the predictive validity of each individual PANAS item.

We analyzed secondary data from our past online experiments that evaluated economic decision‐making and its mental health consequences, consisting of 99 different sessions with study participants recruited through Amazon mTurk (N = 1047). Data were collected between July 2017 and December 2018. The mean age of participants was 33.0 years (standard deviation [SD] 9.3 years, range 19–75 years), 36.6% of participants were female, 69.6% of participants were from the United States, and 30.4% were from India. ⁶ All experiments were conducted in English.

The primary outcome variable was participants’ response to a modified version of an existing SWB measure, ¹ converted to be asked in the present tense (i.e., ‘How do you feel right now?’). Participants were asked to respond with very bad, bad, neutral, good, or very good (this was converted into a score from −2 to 2). Participants also completed the 20‐item PANAS. ⁴ The questions were modified for the present tense and had five possible choices: very slightly or not at all, a little, moderately, quite a bit, or extremely (converted into a score ranging from 1 to 5; see x axis, Figure 1c). The 20 PANAS items were summed into the 10‐item PAS or NAS. ³ Scores for these subscales ranged from 10 (“very slightly or not at all” to all items) to 50 (“extremely” to all items). ³

Figure 1.

Relationships of positive affect scale (PAS), negative affect scale (NAS), and each of the 20‐item Positive and Negative Affect Schedule (PANAS) questions among 1120 mTurk study participants. Boxplot and scatterplot (jittered) of PAS (a) or NAS (b) against SWB. The data from US and Indian (IN) samples is shown separately. (c) The association strength between each of the 20 PANAS items and SWB for US and IN samples. Bars represent β coefficients. Lines indicate the 95% confidence intervals. (d) Internal consistency measured by Cronbach's α (95% confidence intervals in parentheses) for the PAS and NAS was high among participants from both the United States and India. (e) β coefficients for the association of PAS and NAS for the single‐item SWB measure. Standard errors of β coefficients are given in parentheses. The random intercepts were included to account for the effect of clustering of players across individual game sessions. *P < 0.001. CI, confidence interval.

Since each experimental session included multiple participants, we used a random intercept linear regression model. Results for US and Indian participants were analyzed separately to avoid introduction of bias based on participant origin. All statistical analyses were conducted using R version 4.5.0.

Results show that among the US participants, the mean SWB was 0.87 (SD = 1.12), the mean PAS was 29.72 (SD = 10.57), and the mean NAS was 14.49 (SD = 6.10). Among the Indian sample, the mean SWB was 1.06 (SD = 0.97), the mean PAS was 33.46 (SD = 10.45), and the mean NAS was 21.33 (SD = 8.86); these scores were consistent with those from other studies evaluating the PANAS in different populations. ⁴ , ⁷ , ⁸ In the US sample, the regression model found that SWB was positively associated with PAS (β = 0.030, 95% confidence interval [CI] 0.024–0.037, P < 0.001; Figure 1a,e) and negatively associated with NAS (β = −0.048, 95% CI −0.060–−0.036, P < 0.001; Figure 1b,e). In the Indian sample, we did not identify associations between SWB and PAS (β = 0.002, 95% CI −0.0089–0.012, P = 0.769; Figure 1a,e) or between PAS and NAS (β = 0.005, 95% CI −0.0089–0.017, P = 0.466; Figure 1b,e). All 10 positive and negative affect items were associated with SWB in the US sample, while only the “enthusiastic” positive affect item was associated with SWB in the Indian sample (Figure 1c). Internal consistency was high for both PAS and NAS across participants from both US and India (α _PAS,US = 0.93, α _NAS,US = 0.88, α _PAS,IN = 0.92, α _NAS,IN = 0.90) (Figure 1d).

In conclusion, the real‐time, single‐item SWB could reflect participants’ positive and negative psychological states, especially among US participants. The single‐item SWB measure closely reflects PAS and NAS among the US sample as it exhibited sufficient construct validity for both the positive and negative affect components of the PANAS. Our findings suggest that short‐form versions of the PANAS (such as the I‐PANAS‐SF) ⁵ that reduce the number of questions can be used to improve efficiency. This is supported by Crawford and Henry's analysis ⁷ identifying redundancies among the PANAS terms, especially among the NAS. However, only one out of the 20 PANAS terms (enthusiastic) was associated with SWB in the Indian sample. The combination of high internal consistency with a lack of associations between SWB and the PAS, NAS, or with most individual PANAS terms suggests that cultural constructs of the affects may affect PANAS validity when applied to non‐Western or culturally diverse populations. Future work should explore development of PANAS‐like instruments or different measures of SWB that consider the cultural or social differences between different populations.

AUTHOR CONTRIBUTIONS

George Dewey and Akihiro Nishi designed the project and wrote the manuscript. Akihiro Nishi secured funding. George Dewey, Hiroyasu Ando, and Akihiro Nishi conducted the statistical analyses. All authors analyzed the findings and approved the manuscript.

CONFLICT OF INTEREST STATEMENT

Akihiro Nishi is a consultant to Vacan, Inc. and obtained an honorarium from Taisho Pharmaceutical Co., Ltd., which had no role in the project. The remaining authors declare no conflicts of interest.

ETHICS APPROVAL STATEMENT

This study was approved by the UCLA Office of Research Administration (UCLAIRB #16‐001920).

PATIENT CONSENT STATEMENT

Informed consent was obtained online from all participants.

CLINICAL TRIAL REGISTRATION

N/A.

Supporting information

panas validation analysis.

panas data main.

DATA AVAILABILITY STATEMENT

The data and the replication code are available in the Supporting Information.