The present study represents the first attempt to identify potential psychological predictors of change in interpretation bias and state worry following a single session of cognitive bias modification of interpretations (CBM-I), which is a computerised training to modify interpretation bias, using two samples of worry-prone young adults.
Methods
High worriers with a history of anxiety and/or depression (Study 1; N = 83) and worry-prone individuals (Study 2; N = 146) completed a single session of CBM-I training. Pre-training, they completed a battery of self-report measures and tasks that assessed key moderators of CBM responses: attentional control, cognitive flexibility, sensitivity to reward, and imagery ability. Levels of interpretation bias and state worry were also assessed pre and post training to index CBM-related changes.
Results
In study 1 a greater ability to imagine positive events and lower levels of cognitive flexibility at baseline were associated with a greater increase in positive interpretation bias. Lower levels of cognitive flexibility pre training were associated with greater reduction in state worry post training. In study 2 higher levels of cognitive flexibility and lower levels of responses to positive affect at baseline had greater increase in positive interpretation bias, but not reductions in worry post training.
Conclusions
In both studies, attentional control was not a significant predictor of change in interpretation bias or state worry following a single session of CBM-I training. There were differences in the role of cognitive flexibility, emotion-focused rumination and positive mental imagery in the two samples. Given non-replications, individual differences that predict change in near and far transfer outcomes require further research. Nevertheless, the present findings provide insights to improve the outcome of CBM-I. For instance, incorporating a longer imagery training or cognitive flexibility training may be helpful.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Worry is a chain of negative and relatively uncontrollable thoughts (Borkovec et al., 1983) that is typically associated with concerns about future events and outcomes (Jing et al., 2016). While worry may be a normal part of life, it becomes problematic when worry is excessive and uncontrollable (Hayes et al., 2010). Worry is the core symptom of generalised anxiety disorder (GAD) and a transdiagnostic feature that is common in almost all psychological disorders including depression (Hayes et al., 2010).
A core feature of worry is the tendency to make negative interpretations of ambiguous, potentially threatening, situations (Eysenck et al., 1991). This negative interpretation bias is a transdiagnostic process that underlies a wide range of emotional disorders (Hirsch et al., 2016). Worry-prone individuals also lack the positive interpretation bias of ambiguous situations, evident in less worried individuals (Krahé et al., 2019). Through modifying interpretation biases, research has identified that interpretation bias has a causal role in maintaining worry (Hayes et al., 2010; Hirsch et al., 2009). The cognitive model of pathological worry (Hirsch & Mathews, 2012) proposes that worry-prone individuals have involuntary (i.e., stimulus driven) “bottom-up” processes that include the habit of drawing negative interpretations from ambiguous information. In addition, they have impaired voluntary (i.e., goal directed) “top-down” control that restricts their ability to shift focus away from worry and potential threat (Berggren & Derakshan, 2013; Hayes et al., 2008; Leigh & Hirsch, 2011; Stefanopoulou et al., 2014). Given the causal role of interpretation biases in worry, computerised training paradigms have been developed to promote benign interpretations (Grey & Mathews, 2000; Mathews & Mackintosh, 2000). This paradigm is known as cognitive bias modification.
Cognitive bias modification for interpretation (CBM-I) has been predominantly used with anxious or depressed individuals (where worry is a key feature) to target negative interpretation bias related to anxiety and depression (MacLeod & Mathews, 2012). The original lab-based paradigm involved reading ambiguous scenarios and then completing a word fragment that facilitates disambiguation of the scenario in a positive way (Mathews & Mackintosh, 2000). Combining different training methods, Hirsch et al. (2020) developed an enhanced CBM-I training to enhancing the effects of CBM-I training on promoting positive interpretations (Feng et al., 2020). This enhanced CBM-I training (see Method section) has been shown to effectively promote a more positive interpretation bias (i.e., near transfer; whether training generalizes to other tasks that assess the mechanism being trained, in this case interpretation bias), reduce negative thought intrusions, as well as trait rumination (i.e., far transfer; whether training transfers to more distal contexts such as symptoms, in this case worry) compared with the standard CBM-I training in populations with high levels of worry and/or rumination (Hirsch et al., 2020).
Although individuals tend to acquire a more positive interpretation bias following CBM-I training (Menne-Lothmann et al., 2014), individuals vary in the extent of change in positive interpretation bias and mood state. Therefore, an understanding of which factors underlie those individual differences is critical when considering who could benefit from CBM-I. While some research has been done on predicting the outcome of established interventions such as cognitive behavioural therapy (CBT; Fisher & Newman, 2016; Klumpp et al., 2014; Lundh & Öst, 2001), very few studies have focused on predictors of change in CBM-I. As with any intervention, there will be variation in the extent to which people benefit. Therefore, identifying the key individual differences that predict change in near transfer and far transfer effects (i.e., whether training transfers to more distal contexts such as symptoms, in this case worry) following CBM-I is important for two reasons. First, since CBM-I training could at some point become a standalone intervention for worry, the findings of this research may potentially inform patient choice regarding taking up this intervention in the future. Second, by knowing what factors are associated with poorer outcome, researchers could seek to identify who do not benefit and consider why this maybe the case and then make adaptations or alternative interventions for these people.
Past research has found that individuals with more severe symptoms (e.g., more anxious, greater negative interpretation bias) at baseline are more likely to benefit from CBM-I training (Muris et al., 2008; Salemink & Wiers, 2011). Although we acknowledge that sample characteristics are of importance when it comes to predicting the outcome of an intervention, we seek to identify other, theoretically driven, moderators of the impact CBM-I. Given that little research has examined how individual differences in modifiable psychological processes influence the outcome of CBM-I, we sought to identify potential predictors of outcome from psychological intervention research on anxiety/depression, more broadly. Attentional control, cognitive flexibility, sensitivity to reward, and imagery ability are proposed here to be potential factors that could be relevant to near and far transfer effects of CBM-I training.
Attentional control refers to the ability to deliberately shift attention between tasks or to sustain focus on one task when there is distracting information (Eysenck et al., 2007). Past research suggests that worry depletes attentional control capacity among individuals with high levels of worry and/or GAD (Hayes et al., 2008; Stefanopoulou et al., 2014). Impaired attentional control may thus contribute to difficulty in shifting attention to alternative non-worrying information, including benign interpretations of ambiguously threatening situations (Hirsch & Mathews, 2012). In a pilot study of a broad CBT-based psychological therapy service for depression, attentional control has been demonstrated to be a potential predictor of response (Buckman et al., 2019). In another study, it has been demonstrated that individuals with stronger attentional control in the presence of emotional distractors (i.e., emotional attentional control) were more likely to have better treatment response from CBT for GAD (Klumpp et al., 2014). Therefore, we hypothesise that better attentional control (including general attentional control and emotional attentional control) will be a predictor of better response (i.e., acquiring a more positive interpretation bias and greater reductions in worry symptoms) to CBM-I training. However, it is also possible that individuals with lower attentional control would show greater change in interpretation bias. For instance, it has been found that adolescents with lower levels of regulatory control (i.e., a system related to working memory, attentional control, and executive control) benefitted the most from CBM-I training (Salemink & Wiers, 2012). Based on two counter arguments that arise from the existing literature, the present study aims to investigate the role that attentional control plays in predicting the outcome of CBM-I training.
Another potential process is cognitive flexibility, which refers to the ability to think in different ways to adapt to changing environmental stimuli (Dennis & Vander Wal, 2010). Individuals with anxiety disorders tend to have poor cognitive flexibility (Gabrys et al., 2018; Wilson et al., 2018). It is possible that having greater cognitive flexibility would be beneficial during CBM-I training, enabling the adoption of an alternative interpretation more readily. Whereas individuals with lower cognitive flexibility may find it more challenging to generate alternative interpretations of ambiguous situations. Yet, as with attentional control, the alternative may be true; people with lower cognitive flexibility may have more room for improvement to improve their interpretation bias, which has also been suggested to be an underlying mechanism of change in CBM-I training (Edwards et al., 2018; Steinman et al., 2021). In short, we propose that cognitive flexibility is a critical factor in predicting outcome from CBM-I, however, whether higher or lower cognitive flexibility is associated with greater change remains unclear.
A further process that could relate to outcome of CBM-I is sensitivity to reward. This is a transdiagnostic process that is a part of the approach system (Craske et al., 2019), which motivates actions toward rewards (i.e., positive experiences) and goals. Individuals with anxiety and depression sometimes experience anhedonia, which is a loss of interest or enjoyment in pleasurable activities. In addition, they have lower intensity and shorter duration of positive affect (e.g., dampening) in response to reward (Young et al., 2019). Impairments in the approach system are linked to lower chances of recovery, poorer response to pharmacological treatments for depression (McMakin et al., 2012), and poorer outcomes of CBT for social anxiety disorder (Craske et al., 2016). Based on the different overlapping components of the reward system (Der-Avakian & Markou, 2012; Rømer Thomsen et al., 2015), we propose that motivation for reward (i.e., the amount of effort expended to receive/approach positive stimuli), consumption of reward (i.e., the hedonic impact of reward), and learning in relation to reward (i.e., tendency to approach reward based on reinforcement) may be relevant to predicting outcome of CBM-I.
Considering the nature of CBM-I training, we hypothesise that individuals with a low motivation for reward would be less likely to generate vivid and positive outcomes for ambiguous situations, due to a lack of motivation to approach/process positive information. Individuals with deficits in consuming reward would have decreased ability to feel pleasure after experiencing something rewarding (i.e., positive). For example, engaging in positive imagery may be less rewarding; for instance, due to the inability to savour those positive feelings. Likewise, individuals with deficits in reward learning may find the feedback (e.g., ‘well done’) given throughout CBM-I training less impactful in encouraging them to generate positive/vivid outcomes of ambiguous scenarios. Overall, we posit that the more sensitive to reward and the more positive a person feels after experiencing something pleasurable, may be associated with greater change in symptoms following CBM-I training.
Another process that could help predict outcome of CBM-I is the ability to generate images. People with anxiety disorders or depression find it difficult to generate positive imagery about the future compared to those without emotional disorders (Morina et al., 2011). Furthermore, people with GAD generate less imagery when thinking about themselves in positive, personally relevant, future situations (Hirsch et al., 2012), as well as when worrying (Borkovec & Inz, 1990; Hirsch et al., 2012). In this study, participants complete a brief imagery training prior to CBM-I, which is designed to facilitate greater vividness/positivity of mental images. However, given that worry-prone individuals have less imagery in general, the imagery training may not be enough to train individuals to implement imagery during training, especially for those who find it particularly difficult to create vivid mental images to begin with. As such, poor imagery ability at baseline may hinder the degree to which an individual could be trained to make positive interpretations (near transfer), which may also have an impact on reductions in worry symptoms (far transfer).
The Present Research
With worry being a widespread problem in young adults (Pereira et al., 2019), the present study focused on worry-prone young adults between 18 and 25 years old. The predictors we examined were attentional control (assessed using the Emotional Attentional Control Scale and the Attention Network Test), cognitive flexibility (Cognitive Flexibility Inventory), sensitivity to reward (Behavioural Activation System Scale and Responses to Positive Affect Questionnaire), and imagery ability (Prospective Imagery Task). Two studies with complementary designs investigated whether these psychological processes help predict changes in interpretation bias (near transfer; assessed using the Recognition Test) and state worry (far transfer; assessed using a breathing focus task) following a single session of CBM-I training. Study 1 represents an initial (pilot) study, which assessed the predictors in high worriers with a known history of clinical anxiety and/or depression. Study 2 was a larger replication study in a broader worry-prone population that does not limit to those with a past diagnosis of clinical anxiety and/or depression. Both studies were conducted during COVID-19 restrictions, which may serve as a background precipitator of worries (see Table S1).
Hypotheses
Across both studies, we hypothesised that:
1.
Individual differences in attentional control, cognitive flexibility, sensitivity to reward, and imagery ability at baseline would predict change in interpretation bias (near transfer) post CBM-I training.
2.
Individual differences in attentional control, cognitive flexibility, sensitivity to reward, and imagery ability at baseline would predict change in state worry (far transfer) post CBM-I training.
Method
Design
This is a within-subjects (single-arm) study in which all participants received a single session of positive CBM-I training. Pre-training, participants completed a battery of tests that assessed attentional control, cognitive flexibility, sensitivity to reward, and imagery ability. To capture change in near transfer and far transfer outcomes, levels of interpretation bias and state worry were assessed prior to and following CBM-I training.
Self-Report Measure to Assess Trait Worry
The Penn State Worry Questionnaire (PSWQ) is a 16-item self-report measure that assesses trait worry (Meyer et al., 1990). Participants responded to items (e.g., “My worries overwhelm me”) on a scale ranging from 1 (not at all typical of me) to 5 (very typical of me). Higher scores on the PSWQ indicate higher trait worry levels. Participants with a score of 56 or above were classified as high worriers in the current study (Hirsch et al., 2009) and selected as participants. The PSWQ has high test–retest reliability (r = 0.92) and good validity (Meyer et al., 1990). The internal consistency of the PSWQ wasα = 0.76 in Study 1 and in α = 0.72 Study 2.
Measures to Assess the Predictor Variables
Emotional Attentional Control Scale
The Emotional Attentional Control Scale (eACS; Barry et al., 2013) assesses voluntary attentional control in the presence of emotion. Participants rated 14 items (e.g., “I get distracted by my feelings”) on a 1 (almost never) to 4 (always) scale. Higher scores on the eACS indicate good voluntary attentional control. The eACS has a strong positive correlation with general attentional control. In addition, it has strong negative correlations with trait anxiety and depressive symptoms (Barry et al., 2013). The internal consistency of the eACS was α = 0.77 in Study 1 and α = 0.84 in Study 2.
Attention Network Test
The Attention Network Test (ANT; Fan et al., 2002, 2005) is a behavioural measure of general attentional control. In each trial, participants were presented with five congruent (→→→→→) or incongruent stimuli (→→←→→). Then, they were asked to determine the direction of the central stimulus. A total of 96 trials were spread in four blocks. Participants who scored 2.5 SD below the mean accuracy were removed from analysis. Levels of attentional control were determined by calculating a difference score between the reaction time for congruent and incongruent trials. Higher difference scores indicate lower levels of attentional control.
Cognitive Flexibility Inventory
Cognitive flexibility was assessed using the 20-item Cognitive Flexibility Inventory (CFI; Dennis & Vander Wal, 2010). The CFI comprises two subscales. The Alternatives subscale (CFI-A) measures the ability to generate multiple alternative solutions to a situation (e.g., “I consider multiple options before making a decision”). The Control subscale (CFI-C) measures the tendency to perceive difficult situations as controllable (e.g., “I am capable of overcoming the difficulties in life that I face”). Participants rated the items on a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). Higher scores indicate greater cognitive flexibility. The CFI has a high test–retest correlation coefficient (0.83) and adequate levels of concurrent validity (Dennis & Vander Wal, 2010). The internal consistency ranged from α = 0.77 to 0.90 in Study 1 and α = 0.80 to 0.88 in Study 2.
Behavioural Activation System Scale
The Behavioural Activation System Scale (BAS; Carver & White, 1994) measures individuals’ sensitivity to approach reward: BAS Drive (BAS-D; e.g., “I go out of my way to get things I want”), BAS Fun Seeking (BAS-F; e.g., “I’m always willing to try something new if I think it will be fun”), and BAS Reward Responsiveness (BAS-R; e.g., “When good things happen to me, it affects me strongly”). Participants rated 13 items on a 4-point scale ranging from 1 (very true for me) to 4 (very false for me). After reverse scoring, higher scores indicate greater tendency to approach reward. The BAS Scale has adequate convergent validity (Vandeweghe et al., 2016). The internal consistency ranged fromα = 0.71 to 0.81 in Study 1 and α = 0.71 to 0.78 in Study 2.
Responses to Positive Affect Questionnaire
The Responses to Positive Affect Questionnaire (RPA; Feldman et al., 2008) is a 17-item self-report measure that assesses different cognitive responses to positive affect. The three subscales are: Emotion-Focused Positive Rumination (RPA-E; e.g., “I notice how I feel full of energy”), Self-Focused Positive Rumination (RPA-S; e.g., “I think about how proud I am of myself”), and Dampening (RPA-D; e.g., “I remind myself theses feelings won’t last”). Participants rated the items on a scale ranging from 1 (almost never) to 4 (almost always). Higher scores indicate greater responses to positive affect. The RPA has acceptable convergent validity with measures of self-esteem and mood (Feldman et al., 2008). The internal consistency ranged from α = 0.72 to 0.73 in Study 1 and α = 0.74 to 0.80 in Study 2.
Prospective Imagery Task
The Prospective Imagery Task (PIT; Holmes et al., 2008; MacLeod et al., 1996; Stöber, 2000) has been used to assess the ability to create prospective imagery of positive or negative situations (Morina et al., 2011). Participants were asked to form mental images of 10 positive (PIT-P; e.g., “You will do well on your course”) and 10 negative (PIT-N; e.g., “You will have a serious disagreement with your friend”) potential future events and to rate the vividness of the image on a scale ranging from 1 (no image at all) to 5 (very vivid). Higher scores indicate better imagery ability. The internal consistency ranged from α = 0.83 to 0.86 in Study 1 and α = 0.78 to 0.84 in Study 2.
Filler Task
The Speed of Comprehension Test (Version A; Baddeley et al., 1992) was used as a filler task to reduce the likelihood of potential mood effects resulting from the CBM-I task (Black & Grisham, 2018), which could otherwise confound interpretation of the data in relation to changes in interpretation bias as compared to an effect of mood induction (Lothmann et al., 2011). The task required participants to determine whether the statements (e.g., “Dragonflies have wings”) are true or false. To minimise anxiety, participants were instructed that the speed of response was not important (Hirsch et al., 2009).
Measures to Assess the Outcome Variables
Recognition Test
The recognition test (RT) was used to assess interpretation bias (based on Mathews & Mackintosh, 2000; with materials adapted by Hirsch et al., 2018). Participants were asked to read 10 ambiguous scenarios and answer comprehension questions. Then, participants were presented with the title of each scenario, with four statements (one positive, one negative, and two foil interpretations of the scenario). They were asked to rate the similarity of each statement with the original scenario on a scale ranging from 1 (very different in meaning) to 4 (very similar in meaning). A recognition test index (RT index) was computed by subtracting the mean ratings of negative targets from the mean ratings of positive targets. Higher scores indicate a more positive interpretation bias.
Breathing Focus Task
State worry was assessed using a 5-min breathing focus task (Eagleson et al., 2016; Hirsch et al., 2020). At randomly cued intervals, participants were asked to indicate (i.e., by selecting one of the options that appeared on the screen) whether they were focusing on their breathing or experiencing a negative, positive, or neutral thought intrusion. Higher frequency of negative intrusions is taken as an index of higher state worry.
Enhanced CBM-I
Participants received mental imagery training prior to the start of CBM-I training (see enhanced CBM-I condition in Hirsch et al., 2020, 2021) to facilitate vivid and positive mental images via practice. Tailored feedback was provided to encourage the formation of a positive/vivid image. In the actual CBM-I training, participants were asked to listen to and vividly imagine 40 scenarios pertaining to worry-related themes that were ambiguous (see Feng et al., 2020; Hirsch et al., 2020). The training was presented in two blocks, with a 5-min optional break in between each block. Corresponding to Hirsch et al.’s (2020) study, 50% of trials were provided with a positive resolution to the scenario. The other 50% of trials were ambiguous (i.e., not resolved positively or negatively), in which participants were required to generate their own positive resolution. After participants heard each scenario, they generated a vivid/positive image for 7 s and then completed a yes/no comprehension question designed to reinforce the positive interpretation of the scenario. Feedback (i.e., a green tick mark for correct or a red cross sign for incorrect) was given on the accuracy of their responses. Participants were also asked to rate the vividness (50% of trials) or positivity (50% of trials) of the image on a visual analogue scale from 0 (not at all vivid/positive) to 100 (extremely vivid/positive) after each trial. Positive feedback (e.g., good effort) and prompts (e.g., keep trying to generate a vivid image) were given throughout the training to reinforce the formation of vivid/positive images.
Procedure
Potential participants completed a brief online screening questionnaire. Participants were eligible to participate if they were: 1) aged between 18 to 25 years old, had high levels of worry (had a score of 56 or above on the PSWQ; Eysenck et al., 1991), and 3) had normal or corrected hearing. Individuals who did not have normal or corrected hearing were excluded from participation as the CBM-I training involved listening to audio clips.
At baseline, participants completed a battery of tests in the following order: CFI, eACS, BAS, RPA, ANT, and PIT. Within 48 h from the baseline session, participants completed the recognition test and a breathing focus task, followed by a single session of CBM-I training. Then, participants completed a filler task before interpretation bias and state worry were re-assessed. The entire study took 1 to 1.5 h to complete. Participants who completed the study were offered to be entered into a prize draw or to receive research credits. The study was conducted online via Gorilla software (Anwyl-Irvine et al., 2019).
Data Analysis
Prior to analysis, participants who had a score of 2.5 SD below the mean for accuracy on the ANT trials (n = 3 in Study 1 and n = 8 in Study 2) and RT comprehension questions (n = 3 in Study 1 and n = 8 in Study 2) were excluded from the respective ANT and/or RT analysis. Data from participants with less than 75% accuracy on the attention-check questions (i.e., questions that explicitly asked participants to choose a response such as “strongly agree” to ensure they were paying attention to each item/question) were also excluded (n = 1 in Study 2). Bivariate correlations were computed for all variables. The assumptions of independent errors and no multicollinearity were tested using Durbin-Watson tests and variation inflation factors.1 Given all the variance inflation factors (VIF) values were below 10 and the tolerance statistics were above 0.2, the results indicated that multicollinearity was not a concern (see Supplementary Materials for full details). Two multiple regression analyses (forced entry method) were conducted on each data set to examine whether levels of attentional control, cognitive flexibility, sensitivity to reward, and imagery ability predicted changes in interpretation bias (near transfer) and worry (far transfer) following a single session of CBM-I training.2 Baseline trait worry was entered as a covariate for all regression models. All data analyses were performed using R version 4.0.4.
STUDY 1
Participants
Worry-prone community volunteers (N = 83; Mage = 23.04; SD = 2.17) were recruited from Genetic Links to Anxiety and Depression (GLAD) study (Davies et al., 2019) which comprises of a large cohort of participants with a history of clinical anxiety and/or depression. Based on self-report at the start of the current study, all participants had a clinical history of both anxiety and depression, except one who had a history of anxiety but not depression. Participants’ diagnostic status for lifetime incidence of anxiety and/or depressive disorder was assessed prior to the start of the study, hence we refer to them as individuals with a clinical history of anxiety and/or depression, but do not know about their current clinical status at the time of the study. Participants were predominately White British (87%); a minority were from Other white background (5%), Asia (3%), or Mixed background (5%). Most participants were women (80%), with a small proportion of men (13%) and other (4% non-binary, 2% transgender, 1% prefer not to say). Given this was a pilot study for the subsequent study, a power analysis was not conducted.
Study 1 Results
Descriptive Statistics
Means and standard deviations are provided in Table 1. To visualize the relationship between different variables, bivariate correlations are presented using a heatmap in Fig. 1.
Table 1
Descriptive Statistics (Study 1)
Variable
M
SD
Penn State Worry Questionnaire
68.58
6.04
Cognitive Flexibility Inventory
Alternatives Subscale
64.23
11.58
Control Subscale
21.41
6.38
Emotional Attentional Control Scale
23.28
4.87
Behavioural Activation System Scale
Drive
9.71
2.43
Fun Seeking
10.40
2.64
Reward Responsiveness
16.19
2.62
Responses to Positive Affect Questionnaire
Emotion-Focused Positive Rumination
11.94
3.22
Self-Focused Positive Rumination
7.42
2.50
Dampening
22.96
4.86
Attention Network Task
43.53
45.24
Prospective Imagery Task
Positive
28.37
7.32
Negative
38.66
7.00
Baseline Breathing Focus Task
2.89
2.28
Post Breathing Focus Task
2.46
2.91
Baseline Recognition Test Index
−0.70
0.70
Post Recognition Test Index
0.34
0.71
Fig. 1
A Heatmap of Correlation Between Variables (Study 1). PSWQ Penn State Worry Questionnaire, CFI-A Cognitive Flexibility Inventory–Alternatives subscale, CFI-C Control subscale, eACS emotional Attentional Control Scale, BAS-D Behaviour Activation System Scale–Drive subscale, BAS-F Fun seeking subscale, BAS-R Reward Responsiveness Subscale, RPA-E Responses to Positive Affect Questionnaire–Emotion-focused positive rumination subscale, RPA-S Self-focused positive rumination subscale, RPA-D Dampening subscale, ANT Attention Network Test, PIT-P Prospective Imagery Task–Positive Subscale, PIT-N Negative Subscale, BF Breathing Focus Task, RT Recognition Test
×
Predicting Change in Interpretation Bias (Near Transfer)
A multiple regression analysis3 was conducted to predict change in RT index following a single session of CBM-I training from a set of baseline predictor variables. The results (see Table 2) indicated that together the set of predictors did not account for a significant amount of the variability in RT index (explained 21% of the variance), R2 = 0.21, adjusted R2 = 0.03, F(14, 61) = 1.16, p = 0.33, though the effects were moderate (Cohen’s f2 = 0.27). When examining the individual contribution of each predictor variable, scores on the CFI-A were a significant predictor of change in RT index (while accounting for all other variables in the model), β = −0.29, t(61) = −2.06, p = 0.04, indicating that individuals with lower levels of cognitive flexibility in generating alternative solutions to different situations tended to have greater increase in positive interpretation bias following a single session of CBM-I training. In addition, scores on the PIT-P were also a significant predictor of change in RT index, β = 0.30, t(61) = 2.13, p = 0.04, which suggests that individuals with more vivid imagery of positive scenarios had greater increase in positive interpretation bias (while accounting for all other variables in the model) following a single session CBM-I training.
Table 2
Summary of Study 1 Results (Model 1: predicting change in interpretation bias)
Predictor
R2
B
SE B
β
p
.21
33
Constant
1.14
1.80
.53
PSWQ
0.01
0.02
.08
.61
CFI-A
−0.02
0.01
−.29*
.04
CFI-C
0.01
0.02
.10
.47
eACS
0.00
0.02
.005
.98
BAS-D
0.01
0.04
.03
.81
BAS-F
−0.01
0.04
−.04
.82
BAS-R
−0.03
0.04
−.12
.41
RPA-E
0.001
0.03
.01
.97
RPA-S
0.01
0.04
.04
.77
RPA-D
−0.01
0.02
−.05
.71
ANT
−0.00
0.00
−.06
.64
PIT-P
0.03
0.01
.30*
.04
PIT-N
−0.02
0.01
−.15
.30
Baseline RT Index
0.18
0.14
.17
.21
PSWQ Penn State Worry Questionnaire, CFI-A Cognitive Flexibility Inventory – Alternatives subscale, CFI-C Control subscale, eACS Emotional Attentional Control Scale, BAS-D Behavioural Activation System Scale – Drive subscale, BAS-F Fun seeking subscale, BAS-R Reward responsiveness Subscale, RPA-E Responses to Positive Affect Questionnaire – Emotion-focused positive rumination subscale, RPA-S Self-focused positive rumination subscale, RPA-D Dampening subscale, ANT Attention Network Test, PIT-P Prospective Imagery Task – Positive Subscale, PIT-N Negative Subscale, RT Recognition Test
*p < .05
Predicting Change in State Worry (Far Transfer)
Another multiple regression analysis was conducted to predict change in state worry following a single session of CBM-I training from a set of baseline predictor variables. The results (see Table 3) indicated that the set of predictors accounted for a significant amount of the variability (explained 38% of the variance) in change in state worry, R2 = 0.38, adjusted R2 = 0.24, F(14, 65) = 2.82, p = 0.002, with a large effect size (Cohen’s f2 = 0.61). When examining the individual contribution of each predictor variable, we found that CFI-A [β = 0.27, t(65) = 2.14, p = 0.04] and baseline state worry [β = 0.42, t(65) = 3.52, p = < 0.001] were significant predictors of change in state worry. The results suggest that individuals with lower levels of cognitive flexibility in generating alternative solutions as well as lower state worry at baseline (while accounting for all other variables in the model) were more likely to have greater reduction in state worry following a single session of CBM-I training.
Table 3
Summary of Study 1 Results (Model 2: predicting change in state worry)
Predictor
R2
B
SE B
β
p
.38
.002
Constant
−8.39
6.07
.17
PSWQ
0.06
0.06
.13
.31
CFI-A
0.07
0.03
.27*
.04
CFI-C
0.09
0.06
.20
.12
eACS
−0.09
0.08
−.16
.25
BAS-D
0.13
0.15
.11
.37
BAS-F
0.08
0.14
.08
.57
BAS-R
−0.06
0.14
−.05
.68
RPA-E
−0.06
0.11
−.07
.60
RPA-S
0.14
0.15
.13
.33
RPA-D
0.08
0.07
.14
.25
ANT
−0.01
0.01
−.13
.23
PIT-P
−0.07
0.05
−.18
.14
PIT-N
0.01
0.05
.02
.88
Baseline BF
0.55
0.15
.42***
< .001
PSWQ Penn State Worry Questionnaire, CFI-A Cognitive Flexibility Inventory – Alternatives subscale, CFI-C Control subscale, eACS Emotional Attentional Control Scale, BAS-D Behavioural Activation System Scale – Drive subscale, BAS-F Fun seeking subscale, BAS-R Reward responsiveness Subscale, RPA-E Responses to Positive Affect Questionnaire – Emotion-focused positive rumination subscale, RPA-S Self-focused positive rumination subscale, RPA-D Dampening subscale, ANT Attention Network Test, PIT-P Prospective Imagery Task – Positive Subscale, PIT-N Negative Subscale, BF Breathing Focus Task
*p < .05. *** p < .001
Study 1 Discussion
The first study examined the potential psychological factors that predict near and far transfer effects from CBM-I training using a sample of high worriers with a known history of clinical anxiety and/or depression. The results from this first study indicated that lower cognitive flexibility (in generating alternative solutions) at baseline predicted greater increase in positive interpretation bias amongst high worriers with a history of clinical anxiety and/or depression. Furthermore, in the current study we found that having a more vivid imagery of positive events at baseline was associated with greater increase in positive interpretation bias post CBM-I training. It is possible that positive mental imagery is a pertinent skill to elicit change in interpretation bias among high worriers with a history of clinical anxiety and/or depression, but this requires further research, given the small sample size.
The results of Study 1 supported the second hypothesis that attentional control, cognitive flexibility, sensitivity to reward, and imagery ability at baseline would predict change in state worry post CBM-I training. Together, the combination of predictors accounted for a significant amount of the variability in change in state worry. Furthermore, we found that lower state worry and lower cognitive flexibility at baseline (while accounting for all other variables in the model) was associated with lower state worry following a single session of CBM-I training. In sum, the findings of Study 1 suggest that individual differences in cognitive flexibility, positive mental imagery, and baseline worry severity were associated with change in symptoms post CBM-I training in high worriers with a known history of clinical anxiety and/or depression. However, replication is needed to understand the reliability and generalisability of the findings given the current study was a pilot study with a small sample size. Thus, the results are likely to be underpowered and need to be interpreted with caution.
STUDY 2
Participants
Study 2 aims to replicate Study 1 using a larger sample of high worriers, regardless of past diagnosis of clinical anxiety and/or depression. Based on Green’s (1991) rule-of-thumb for testing the contribution of the individual predictors (N ≥ 104 + number of individual predictors), the present study needed more than 116 participants to obtain a reliable regression model.4 A total of 146 young adults (Mage = 20.85, SD = 2.13) participated in the study from across the United Kingdom recruited through university/college circular emails, posting of advertisements on mental health websites, social media (e.g., Facebook, Twitter), as well as university research participation platforms. The sample was ethnically diverse (23% White British, 29% Other white background, 33% Asian, 12% Mixed background, and 3% Other) and predominantly women (90%; 9% men and 1% non-binary).
Study 2 Results
Descriptive Statistics
Means and standard deviations are provided in Table 4. A heatmap of bivariate correlations are provided in Fig. 2.
Table 4
Descriptive Statistics (Study 2)
Measure
M
SD
Penn State Worry Questionnaire
66.39
6.10
Cognitive Flexibility Inventory
Alternatives
68.08
10.77
Control
25.82
7.44
Emotional Attentional Control Scale
25.42
6.48
Attention Network Test
37.66
52.97
Behavioural Activation System Scale
Drive
10.13
2.43
Fun Seeking
11.01
2.61
Reward Responsiveness
16.80
2.77
Responses to Positive Affect Questionnaire
Emotion-Focused Positive Rumination
12.97
3.18
Self-Focused Positive Rumination
7.99
2.74
Dampening
22.48
4.60
Prospective Imagery Task
Positive
33.31
6.91
Negative
36.73
6.51
Baseline Breathing Focus Task
2.54
2.31
Post Breathing Focus Task
1.98
2.63
Baseline Recognition Test Index
−0.46
0.74
Post Recognition Test Index
0.23
0.77
Fig. 2
A Heatmap of Correlation Between Variables (Study 2). PSWQ Penn State Worry Questionnaire, CFI-A Cognitive Flexibility Inventory–Alternatives subscale, CFI-C Control subscale, eACS emotional Attentional Control Scale, BAS-D Behaviour Activation System Scale–Drive subscale, BAS-F Fun seeking subscale, BAS-R Reward Responsiveness Subscale, RPA-E Responses to Positive Affect Questionnaire–Emotion-focused positive rumination subscale, RPA-S Self-focused positive rumination subscale, RPA-D Dampening subscale, ANT Attention Network Test, PIT-P Prospective Imagery Task–Positive Subscale, PIT-N Negative Subscale, BF Breathing Focus Task, RT Recognition Test
×
Predicting Change in Interpretation Bias (Near Transfer)
A multiple regression analysis was conducted to predict change in RT index following a single session of CBM-I training from a set of baseline predictor variables. The results (see Table 5) indicated that together the set of predictors did not account for a significant amount of the variability in change in RT index (explained 16% of the variance), R2 = 0.16, adjusted R2 = 0.04, F(14, 100) = 1.32, p = 0.18,5 though the effects were moderate (Cohen’s f2 = 0.19). When examining the individual contribution of the predictors within the model (while controlling for the effects of other variables), scores on the CFI-A were a significant predictor of change in RT index, β = 0.22, t(100) = 1.99, p = 0.05, indicating that individuals with higher levels of cognitive flexibility in generating alternative solutions to different situations tended to have greater increase in positive interpretations following a single session of CBM-I training. In addition, scores on the RPA-E were a significant predictor of change in RT index, β = −0.33, t(100) = −2.50, p = 0.01, which suggests that lower levels of emotion-focused rumination were associated with greater increase in positive interpretation bias following CBM-I training.
Table 5
Summary of Study 2 Results (Model 3: predicting change in interpretation bias)
Predictor
R2
B
SE B
β
p
.16
.21
Constant
1.60
1.26
.21
PSWQ
−0.01
0.01
−.05
.63
CFI-A
0.01
0.01
.22*
.05
CFI-C
−0.02
0.01
−.16
.22
eACS
−0.00
0.01
−.03
.78
ANT
−0.00
0.00
−.01
.90
BAS-D
−0.00
0.03
−.00
.97
BAS-F
−0.01
0.03
−.03
.82
BAS-R
0.03
0.03
.10
.35
RPA-E
−0.07
0.03
−.33*
.01
RPA-S
0.03
0.03
.12
.40
RPA-D
−0.02
0.02
−.10
.42
PIT-P
0.01
0.01
.08
.48
PIT-N
−0.02
0.01
−.20
.10
Baseline RT Index
0.12
0.10
.12
.22
PSWQ Penn State Worry Questionnaire, CFI-A Cognitive Flexibility Inventory – Alternatives subscale, CFI-C Control subscale, eACS Emotional Attentional Control Scale, ANT Attention Network Test, BAS-D Behavioural Activation System Scale – Drive subscale, BAS-F Fun Seeking subscale, BAS-R Reward Responsiveness subscale, RPA-E Responses to Positive Affect Questionnaire – Emotion-Focused Positive Rumination subscale, RPA-S Self-Focused Positive Rumination subscale, RPA-D Dampening subscale, PIT-P Prospective Imagery Task – Positive subscale, PIT-N Negative subscale, RT Recognition Test
*p < .05
Predicting Change in State Worry (Far Transfer)
Another multiple regression analysis was conducted to predict change in state worry following a single session of CBM-I training from a set of baseline predictor variables. The results (see Table 6) indicated that the set of predictors together accounted for a significant amount of the variability (explained 37% of the variance) in change in state worry, R2 = 0.37, adjusted R2 = 0.29, F(14, 107) = 4.50, p < 0.001, with a large effect size (Cohen’s f2 = 0.59). When examining the individual contribution of each predictor variable, none of the predictors were predicting unique statistically significant variance except for state worry assessed at baseline, β = 0.52, t(107) = 6.09, p < 0.001. The results suggest that individuals with lower state worry (breathing focus task) at baseline (while accounting for all other variables in the model) tended to have lower state worry following a single session of CBM-I training.
Table 6
Summary of Study 2 Results (Model 4: predicting change in state worry)
Predictor
R2
B
SE B
β
p
.37
< .001
Constant
5.99
3.85
.12
PSWQ
−0.02
0.04
−.05
.57
CFI-A
−0.03
0.02
−.12
.17
CFI-C
0.03
0.04
.09
.38
eACS
−0.04
0.04
−.08
.38
ANT
−0.01
0.00
−.13
.14
BAS-D
0.18
0.10
.16
.08
BAS-F
−0.06
0.10
−.06
.53
BAS- R
−0.02
0.09
−.02
.85
RPA -E
0.01
0.09
.01
.89
RPA-S
−0.15
0.11
−.16
.16
RPA-D
−0.04
0.06
−.06
.52
PIT-P
−0.05
0.04
−.13
.22
PIT-N
0.02
0.04
.07
.51
Baseline BF
0.62
0.10
.52***
< .001
PSWQ Penn State Worry Questionnaire, CFI-A Cognitive Flexibility Inventory – Alternatives subscale, CFI-C Control subscale, eACS Emotional Attentional Control Scale, ANT Attention Network Test, BAS-D Behavioural Activation System Scale – Drive subscale, BAS-F Fun Seeking subscale, BAS-R Reward Responsiveness subscale, RPA-E Responses to Positive Affect Questionnaire – Emotion-Focused Positive Rumination subscale, RPA-S Self-Focused Positive Rumination subscale, RPA-D Dampening subscale, PIT-P Prospective Imagery Task – Positive subscale, PIT-N Negative subscale, BF Breathing Focus Task
***p < .001
Study 2 Discussion
The aim of the second study was to replicate Study 1 using a fully powered sample of high worriers. The results indicated that higher cognitive flexibility at baseline (while accounting for all other variables in the model) significantly predicted greater increase in positive interpretation bias. It is possible that that high worry students with greater cognitive flexibility were more capable of generate positive interpretations of emotionally ambiguous situations during CBM-I training, which led to greater increase in positive interpretation bias. Based on this finding, future research could seek to investigate whether incorporating cognitive flexibility training prior to CBM-I to support high worry students with low levels of cognitive flexibility is helpful. Contrary to our expectation, we found that high worry students who typically focus less on positive emotions (as measured by the RPA-E) had greater increase in positive interpretation bias (while accounting for all other variables in the model) following CBM-I in study 1. This may suggest that the ability to ruminate on positive emotions is not a required skill to learn the techniques in CBM-I. In fact, our findings could be taken to support a diminished learning curve such that individuals with a more positive emotion-focused rumination style at baseline may be more likely to reach a learning plateau.
In regard to state worry after CBM-I training, the present results are largely consistent with the results found in Study 1. Unsurprisingly, we found that lower state worry at baseline (while accounting for all other variables in the model) was associated with lower state worry following a single session of CBM-I training. This pattern of results is consistent with the previous literature which suggests that those with lower pretreatment symptom severity is associated lower symptom severity following cognitive therapy (Hamilton & Dobson, 2002; Stiles-Shields et al., 2015).
When comparing the current findings to Study 1, we found inconsistencies. While in Study 1 we found that lower cognitive flexibility (in generating alternative solutions) at baseline predicted greater increase in positive interpretation bias amongst high worriers with a history of clinical anxiety and/or depression, the present study has shown that high worriers with higher cognitive flexibility (in generating alternative solutions) at baseline predicted greater increase in positive interpretation bias following CBM-I training. Furthermore, while in Study 1 we found that having a more vivid imagery of positive events at baseline was associated with greater increase in positive interpretation bias post CBM-I training, however, the same association was not found in Study 2. It is possible that positive mental imagery is a more pertinent skill to elicit change in interpretation bias among high worriers with a history of clinical anxiety and/or depression, but not high worriers in general, but this requires further research. Importantly, the sample in Study 1 was small given the nature of a pilot study. Thus, the findings may not be comparable to Study 2.
Nevertheless, the inconsistent findings across studies warrant further research. In particular, the predictive value of cognitive flexibility (alternatives aspect) in relation to change in interpretation bias and state worry is questionable. The current findings may imply that cognitive flexibility is an unstable predictor of change following CBM-I training. Alternatively, it is possible that the findings are due to unreliable data (i.e., due to the small sample size in Study 1) or that there were confounding factors (e.g., comorbid symptoms) that we failed to account for.
General Discussion
Across two samples of worriers, the present paper examined whether individual differences in attentional control, cognitive flexibility, sensitivity to reward, and imagery ability predicted changes in interpretation bias and state worry following a single session of CBM-I training. To our knowledge, this is the first study to investigate a range of cognitive predictors of outcome for CBM-I and to conduct an internal replication across two studies. However, given the exploratory nature of this study, with the inclusion of a large number of predictor variables which may unintentionally inflate type I error, the results need to be interpreted with caution. Furthermore, given Study 1 was an initial pilot study which recruited only a small number of participants, the inconsistencies across Study 1 and Study 2 may reflect differences in sample characteristics or unreliable findings due to a lack of power. It is important to point out that whist we adopted Green’s rule of thumb which indicated a higher sample size than the two other approaches that we also tested, there are other sample size calculation approaches that may have indicated a larger sample size was required.
There were some similarities and differences in findings across Study 1 and 2. In both studies, attentional control was not a significant predictor of change in interpretation bias or state worry following a single session of CBM-I training. This finding is in contrast to Salemink and Wiers’s (2012) study which found that adolescents with lower regulatory control (a concept that encompasses working memory, attentional control, and executive control) benefitted more from CBM-I training. However, since Salemink and Wiers (2012) used an unselected adolescent sample between 14 and 16 years old, the range of scores may be less restricted leading to greater potential to observe individual differences compared with our sample of high worriers. It is also possible that attentional (or regulatory) control may play a role in predicting the outcome of CBM-I in young adolescents, perhaps due to less mature prefrontal cortex (Arain et al., 2013), but not in young adults (18–25 years old) who are in the later stages of brain maturation. Future research should seek to replicate and extend the current findings.
Across both studies, cognitive flexibility (the alternatives aspect of CFI, more specifically) appeared to be a significant predictor of change in interpretation bias following a single session of CBM-I training. Importantly, however, the direction of effects differed between studies. In Study 1, we found that for high worriers with a history of clinical anxiety and/or depression, lower cognitive flexibility predicted greater change in interpretation bias, whereas in Study 2, we found that for general high worriers, higher cognitive flexibility predicted greater change in interpretation bias, whereas. Based on previous research that examined the role of cognitive flexibility in predicting treatment response from CBT, it has been suggested that greater cognitive flexibility may provide a better foundation for learning how to restructure/interpret situations in different ways (Johnco et al., 2013). This may potentially explain why for student high worriers (Study 2) with higher cognitive flexibility at baseline, this predicted greater increase in positive interpretations. However, while poorer cognitive flexibility has been associated with poorer cognitive restructuring skills, it did not impact the overall outcome of CBT for depression (Johnco et al., 2014; Lindner et al., 2016). In fact, the findings of Study 1 demonstrated that individual with high levels of worry who have a lifetime history of anxiety or depression with lower cognitive flexibility at baseline had superior effects than those with higher cognitive flexibility following CBM-I training. The inconsistent findings suggest that cognitive flexibility is an unreliable predictor of outcome. The predictive value of cognitive flexibility in the context of psychological interventions warrants further research.
Another inconsistency across studies was that positive emotion-focused rumination was a significant predictor of change in interpretation bias in Study 2, but not in Study 1. Similarly, positive mental imagery was a significant predictor of change in interpretation bias in Study 1, but not in Study 2. While more research is needed, these differences may be related to differences in the two samples such as, chronicity of symptoms, life history of anxiety or depression diagnosis and or potential comorbidity, even though participants were all recruited on the basis of the same trait worry level (PSWQ), and current diagnosis was not assessed. For example, Study 1 participants may have potentially included more individuals with a current diagnosis of GAD since this is a chronic condition and so they may have still had GAD at the time of the study if they had it at the original diagnostic assessment. Diagnosis of GAD is associated with greater negative beliefs about uncontrollability of worrying and greater perceived inability to control worry compared to non-diagnosed high worriers with similar levels of worry (Wells & Carter, 2001; Hirsch et al., 2013). Therefore, although participants in Study 1 and 2 were quantitively similar in terms of levels of worry, they may be qualitatively different in characteristics of worry in these populations. We do not know however if they had lifetime incidence of GAD or depression since this information from the earlier diagnostic assessment was not available.
The findings of the present study should be interpreted in light of the following limitations. Firstly, the classification of high worriers based on the cut-off score of 56 or above on the PSWQ may not be appropriate. Despite the fact that the same cut-off score has been widely used in previous research (Goodwin et al., 2017; Hayes et al., 2008; Hirsch et al., 2009), Davey et al. (2021) suggested that a higher cut-off score (e.g., PSWQ ≥ 58) may be needed for university students due to an increase in worry in this population over the past twenty years. Although our samples reported a relatively high mean PSWQ score (M = 68.58 for Study 1 and M = 66.39 for Study 2) that is well above the cut-off for 86% sensitivity for GAD (i.e., PSWQ ≥ 62; Behar et al., 2003), a small proportion of the participants had a PSWQ score lower than 58 (2% in Study 1 and 7% in Study 2). Secondly, we failed to assess other related symptoms (e.g., anxiety, depression) or to gather information on the current diagnosis and treatment status of the participants. Thus, we could not control for the effects of potential confounding variables that might influence the reliability of the findings. Thirdly, the reliance on self-report measures to assess cognitive processes may be a potential limitation, due to low ecological validity. Nevertheless, we employed self-report measures because they allow for: (1) the understanding of long-term cognitive function, as opposed to cognitive function at a specific point in time (Snyder et al., 2021), and (2) the measurement of thoughts and behaviours under a broader range of real-world situations beyond the typical emotional/non-emotional conditions in task-based measures (Friedman & Gustavson, 2022), which are more relevant to the present study. Another limitation of the study was the under-powered sample size in Study 1, due to it being an initial pilot study. Considering the large number of predictor variables we examined, future research should explore other approaches to sample size calculation; for instance, increasing the number of cases per variable from 10 to 15. It is also important to note that the study was conducted under the influence of COVID-19. Thus, the generalisability of the findings to post-COVID contexts requires further research. A further limitation is that we did not pre-register this research, conducted in 2020, which is now considered good science practice improving transparency and credibility of research findings. While the research design, sample size, hypotheses, and analyses were planned prior to the start of data collection, nevertheless, we regard this to be an exploratory study to identify potential predictors of outcome. To determine if the findings of the present study can be replicated or generalised to other populations, further pre-registered research is needed.
The present study represents the first attempt to identify potential psychological predictors of change in interpretation bias and state worry following a single session of CBM-I using two samples of worry-prone young adults. Given that multi-session CBM-I training has promising clinical utility and therapeutic value when being applied as an intervention (MacLeod & Mathews, 2012), future research should seek to examine whether predictors of change within a single session of CBM-I is useful in predicting change following multi-session CBM-I training. Furthermore, future research should also explore other potential predictors of outcome that were not examined in the present study (e.g., working memory capacity). Beyond predicting the outcome of CBM-I, future research could also examine the predictive value of the mechanisms we assessed in predicting responsiveness to other cognitive or mechanism-focused interventions. It is possible that these variables are common factors to intervention outcomes. Finally, while the current studies focused on the outcome of worry, future research could focus on predicting the outcome of anxiety since the CBT literature we drew upon was mainly based on treatment for anxiety.
In summary, the present findings provide insights to improve the outcome of CBM-I. Given our findings suggested that general high worriers with less cognitive flexibility had less change in interpretation bias, future research may consider incorporating cognitive flexibility training so that participants are more prepared to generate positive interpretations of ambiguous scenarios during CBM-I training. Furthermore, given that a less vivid imagery of positive scenarios in high worriers with a clinical history of anxiety and/or depression was associated with less change in interpretation bias following CBM-I, a longer imagery training may be beneficial to initiate vivid and positive mental image during CBM-I training. Together, the present findings generate an array of new research questions for future research to explore.
Acknowledgements
The authors would like to thank the NIHR BioResource Centre Maudsley and the NIHR BioResource volunteers (Genetic Links to Anxiety and Depression Study) for their participation in Study 1, and gratefully acknowledge NIHR BioResource centres, NHS Trusts and staff for their contribution. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health and Social Care. The authors would like to thank Shagun Agrawal and Ryhana Ali for their help in data collection. Colette Hirsch, Katherine Young, and Alicia Hughes had salary support from the National Institute for Health Research (NIHR), Mental Health Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and King’s College London.
Declarations
Conflict of Interest
Yun-Lin Wang, Katherine S. Young, Jennifer Y. F. Lau, Alicia M. Hughes and Colette R. Hirsch declare that they have no conflict of interest.
Ethical Approval
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. The study was approved by the PNM Research Ethics Subcommittee at King’s College London (HR-19/20-17692; MOD-19/20-17692).
Consent to Participate
Informed consent was obtained from all participants included in the study.
Consent for Publication
Not applicable.
Animal Rights
No animal studies were carried out by the authors for this article.
Informed Consent
Informed consent was obtained from all individual participants included in the study.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
share
DELEN
Deel dit onderdeel of sectie (kopieer de link)
Optie A:
Klik op de rechtermuisknop op de link en selecteer de optie “linkadres kopiëren”
Met BSL Psychologie Totaal blijf je als professional steeds op de hoogte van de nieuwste ontwikkelingen binnen jouw vak. Met het online abonnement heb je toegang tot een groot aantal boeken, protocollen, vaktijdschriften en e-learnings op het gebied van psychologie en psychiatrie. Zo kun je op je gemak en wanneer het jou het beste uitkomt verdiepen in jouw vakgebied.
Another commonly used rule-of-thumb for multiple regression suggests a minimum of 10 cases per predictor (i.e., 120 participants; Field et al., 2012). Still another a priori power calculation for multiple regression (based on f2 = 0.15, power = 0.8, and 12 predictors) suggests a minimum sample size of 127 (Soper, 2022). Together, these different approaches for sample size calculation displayed similar results.
Note that the primary focus of this study was to test the individual predictors with the model, rather than the overall fit of the model. Thus, while the combination of predictors did not significantly predict the outcome, the contribution of each individual predictors was interpreted.