Which psychological interventions hold promise for improving well-being? We explored this question using survey data to look for promising psychological interventions that might improve people's lives.
Summary of Our Research
We ran three online surveys with a total of 780 U.S. participants in order to measure the association between eight psychological challenges (or factors) and three main outcome measures: subjective life quality, depression, and anxiety. We found no promising intervention targets for anxiety, so did not investigate it further. We measured life quality by using pre-written scales (e.g. ‘satisfaction with life’ and ‘5 factors of happiness’) and assessed each person's sense of optimism and positivity related to their own life and life goals. We also asked participants the degree to which they agreed with statements such as ‘My life has a purpose’ and ‘I am a happy person’ etc.
Being (a) correlational in nature and (b) based on self-reports, the results of the studies we describe in this article can only suggest hypotheses for what interventions may improve people's lives effectively, but cannot be used to draw firm causal conclusions. These limitations notwithstanding, our findings point to particular areas that could be targeted to significantly improve life quality and depression.
First, we measured the degree to which participants agreed to statements about each of the eight tested psychological factors (i.e., asking to what extent they experienced challenges related to each psychological factor) and whether they believed improving any of these would lead to significant changes in their lives:
Across eight psychological factors, participants agreed most with statements related to difficulties of coping with emotional consequences from interpersonal conflict (factor: interpersonal conflict) and to feeling emotionally isolated from others when experiencing difficult situations (factor: emotional isolation).
Participants reported that improving any of the eight psychological factors would lead to improvements in their lives. Considering these scores together with the agreement scores (about how often people experienced each challenge), interventions targeting interpersonal conflict and emotional isolation might be most promising.
Next, we investigated the relevance of each psychological factor to depression and life quality by examining pairwise correlations. All factors correlated significantly with life quality and depression, supporting their potential relevance to depression and life quality. Nevertheless, some factors correlated stronger with depression and life quality than others:
For depression: participants who engaged more often in negative self-talk (e.g. ‘I often tell myself that I am going to fail’) also reported especially severe depression.
For life quality: participants who have a more negative self-image (e.g. ‘I wish I were a better person’) also reported especially low life quality.
Finally, we included all relevant psychological factors into a multiple linear regression to determine their unique impact on depression and life quality. Across two studies, we replicated two different factors respectively being predictive of depression and life quality:
For depression: participants who engaged more often in negative self-talk and report being more emotionally overwhelmed in difficult situations reported more severe depression.
For life quality: participants who had a more negative self-image and felt more emotionally isolated reported lower life quality.
In sum, the current data set suggests the hypothesis that interventions targeting self-image and emotional isolation may be promising for improving life quality, while improving self-talk and feelings of emotional overwhelm might decrease depression (see Figure 1 for an overview of our main results).
Figure 1. Main results. Diagrams represent the main results when testing which psychological factors across three categories ("self"," social", and "emotion" related) were predictive of life quality and/or depression. Low life quality was predicted by a low negative self-image and high emotional isolation. Items that were most predictive of life quality across these two psychological factors are shown below the colored boxes. Depression was predicted most strongly by negative self-talk and by high emotional overwhelm when being in intense situations (right panel). Items that predicted depression most strongly across these two psychological factors are described below the colored boxes. (right panel). Regression coefficients (indicated with ‘β’) and p-values are shown for the association of each psychological factor and their most predictive items.
Our three studies can be viewed just as participants saw them using the following preview links. In the appendix, you can find links to the raw data of each study and analysis code, offering the possibility to adjust and/or run the studies yourself.
Given these findings, we wanted to test which psychological strategies might be useful in improving either depression and/or life quality. We found that two strategies seemed particularly promising for improving both: reminder of personal strength and gratitude for positive things in life. In addition, spending time doing healthy activities (like exercise) and relying on other people was correlated with an increase in life quality.
Our Approach to Assessing Well-Being
We measured participants' subjective life quality by drawing mainly on two questionnaires – ‘5 factors of happiness’ and ‘satisfaction with life’ – and asking participants to rate additional statements like ‘I lead a worthy and fruitful life’ and ‘I am a happy person’. These questions capture a person’s sense of optimism and positivity related to their own life and life goals (see the appendix, below, for details on precisely how this and all the other major scales we used were constructed). We measured participants’ tendency to experience symptoms of depression by using the standard ‘phq9’ questionnaire. This questionnaire asked participants to rate the frequency that they experienced problems such as: ‘Little interest or pleasure in doing things’, ‘Feeling down, depressed or hopeless’ and ‘Thoughts that you would be better off dead or of hurting yourself in some way’.
We were interested in how these two outcome measures were related to the ratings of statements about eight psychological factors which we classified into three distinct categories: "self", "emotion" and "social." For example, we asked participants to rate statements related to their self-image ("self" category), their ability to regulate emotions ("emotion" category) and their experiences during interpersonal conflict ("social" category). By using these ratings, we measured which psychological factors were associated with depressive symptoms and/or a low life quality.
Findings for Life Quality
Amongst in total eight broad psychological factors, we found that two were particularly predictive of life quality (Figure 1 above, left panel): self-reported emotional isolation and self-image.
A person feels emotional isolation when they believe negative emotional experiences are unique to their own lives and not experienced by others. To measure the degree of emotional isolation we asked participants to rate statements such as ‘When I fail at something that is important to me, I feel alone in my failure’ or ‘When I think about my flaws, it tends to make me feel more separate and cut off from the rest of the world’. We found that life quality was higher when emotional isolation was higher, and more specifically, lower life quality was associated particularly strongly (i.e. highest regression coefficient) with one item: ‘When I am feeling down, I tend to feel like most other people must be having a better time’.
Higher life quality was also related to how participants perceive themselves, i.e. their self-image. Life quality was higher for people with a more positive self-image, and was particularly predicted by the degree to which participants were proud of themselves (as measured by the statements ‘I am proud of myself’ and 'Regardless of whether I achieve a lot or little I think I am an exceptional person’) (Figure 1, right panel).
Findings for Depression
We found two psychological factors (which were not the same as for life quality) that predicted the severity of depression: self-reported negative self-talk and emotional overwhelm (Figure 1 above, right panel). During negative self-talk, participants tell themselves repeatedly that they are less worthy or that they will not be successful. Three items were most predictive of changes in depression: ‘I often tell myself that no one cares about me’, ‘I don’t deserve to be happy or loved’ and ‘I am going to fail’.
Greater depression was additionally predicted by the degree to which participants reported being able to regulate their emotions during emotionally intense situations (i.e., ‘emotional overwhelm’). An increase in depression was associated with increasing difficulties in regulating emotions as shown by the most predictive item: ‘I don’t know how to take control over my emotions’.
The psychological factors mentioned above (both those related to life quality and those related to depression) were found to have substantial associations even when controlling for other psychological factors. We also replicated these findings in a confirmatory follow up study on new study participants.
Important note: These results are based on correlational relations between an array of self-reported psychological factors and outcome measures; we present correlational results rather than making conclusions about the causality between certain scales and outcome measures. Future randomised controlled studies would be needed to determine with confidence whether changes in these factors cause changes in life quality or depression. In sum, the current data set proposes that interventions targeting self-image and emotional isolation may be promising targets for improving life quality, while improving self-talk and emotional control might decrease depression.
Details of Our Studies
We conducted three online surveys (Figure 2)
study 1 (N=474) to develop hypotheses;
study 2 (N=58), a qualitative investigation of question wordings;
study 3 (N=250), a confirmatory study to verify our main findings.
As discussed above, these studies aimed to test the relation between a variety of psychological factors, life quality, and depression (Figure 2). Across the three studies, we narrowed down the psychological factors that the data suggested may have the highest probability to improve people’s lives. All studies were conducted on a U.S. population, and participants were recruited using our online recruitment platform ‘Positly’.
Figure 2. Study Overviews
Psychological factors
Before running our studies, we conducted a rapid literature review to assess which psychological factors might be causal predictors of changes in life quality, depression or anxiety. We identified eight broad psychological factors, which can be classified into three main categories: "self," "emotion," and "social" factors. Please refer to the Appendix for a detailed description of all items within these scales. All factors were assessed by asking participants how much they agree with different statements. Two example statements for each scale are shown below:
Category 1, self-factors:
Negative self-image: e.g., ‘I wish I were a better person’ or ‘I am proud of myself’;
Negative self-talk: e.g., ‘I often tell myself that I am going to fail’ or ‘ I often tell myself that no one cares about me’;
Self-compassion: e.g., ‘When times are really difficult, I tend to be tough on myself’ or ‘When I feel upset, it should be ignored as it is not important’.
Category 2: emotion-factors:
Emotional isolation: e.g., ‘When I fail at something that is important to me, I feel alone in my failure’ or ‘When things are going badly for me, I see the difficulties as part of life that everyone goes through’.
Emotional overwhelm: e.g., ‘When I am upset, that feeling takes over completely’ or ‘When I am upset, I feel that I have no control over my emotions or reactions.’
Emotional expression: e.g., ‘I don’t know how to express my emotions such that others understand me’ or ‘I find it difficult to express to people around me how I feel about them’;
Emotional clarity: e.g., ‘It is difficult for me to find the right words for my emotions’ or ‘I often don’t know how to get insight into my emotions’.
Category 3: social factors:
Interpersonal conflict: e.g., ‘When having a conflict with others, I know how to handle it effectively’, ‘Having a conflict with another person disrupts my daily routine’.
In the first study (which was exploratory in nature), we asked participants to state the degree to which they agree with statements from each psychological factor. For example, we assessed negative self-talk with the following text: “we'd like you to evaluate *the extent to which you say different things to yourself* (e.g., things you say to yourself in your own mind, or even that you speak out loud to yourself)”. Agreement was measured with a 7-Likert scale from totally disagree (-3) to totally agree (+3) with a midpoint of neutral (0). We averaged agreement scores across items within a given scale for each participant and then averaged the overall score for each scale across participants. In this way, we first identified whether participants experience these psychological factors per se.
In the first study, we also asked participants whether they believed that improving a particular psychological factor would lead to significant improvements in their lives. For the same psychological factor as above, we measured the Improvement score as follows: “If I learned to put myself down less (i.e., engage in less "negative self-talk") that would substantially improve my life.” These statements were also measured with a 7-Likert scale from totally disagree (-3) to totally agree (+3) with a midpoint of neutral (0). Both the self-reported agreement score (how much participants say they already encounter different psychological characteristics) and improvement scores (how much participants believed they would benefit from applying different interventions), allowed us to identify psychological factors that might be promising for potential psychological interventions (insofar as people’s self-reflections on these questions are accurate).
In Study 2, we asked participants to elaborate on the most predictive items from Study 1 in a qualitative manner. We asked them the following: ‘Please explain your answer and the circumstances in which it occurs.' This allowed us to detect nuances within factors that pointed towards different answer categories. Accordingly, we created additional questions for each specific psychological factor that were more sensitive towards these nuances.
We tested whether results from Study 1 could be replicated in Study 3. Study 3 only included a subset of the psychological factors explored in Study 1 as we eliminated those factors that appeared least promising (see Appendix for agreeableness and improvement scores for all psychological factors from Study 1). Study 3 included the additional refined items based on the more detailed answers given by participants in Study 2. Figure 3 displays the results of the agreement scores (Figure 3, left panel) and improvement scores (Figure 3, right panel) averaged separately for Study 1 and 3. The green and blue dots show the average of each psychological factor measured in Study 1 and Study 3, respectively.
Figure 3. Agreement and improvement scores. Scores are averaged across Study 1 and Study 3 for psychological factors that were tested across both studies. Bar graphs represent the average pooled over both studies. The green and blue dots show the average score separately for Study 1 and Study 3, respectively. Here, we show the agreement and improvement scores. We asked participants to rate the degree of agreement to statements of each psychological factor. For example, for negative self-talk we asked: “we'd like you to evaluate *the extent to which you say different things to yourself* (e.g., things you say to yourself in your own mind, or even that you speak out loud to yourself)”. Agreement was measured with a 7-Likert scale from totally disagree (-3) to totally agree (+3) with a midpoint of neutral (0). For the same psychological factor, we measured the Improvement score as follows: “If I learned to put myself down less (i.e., engage in less "negative self-talk") that would substantially improve my life.” These statements were also measured with a 7-Likert scale from totally disagree (-3) to totally agree (+3) with a midpoint of neutral (0). (n=727, error bars are denoted as standard error of the mean (S.E.M.) across participants for the average across both studies and each study individually.)
In general, participants reported that improving any of these psychological factors would lead to substantial improvements in their lives (Figure 3, right panel). Considering these scores together with the agreement scores (Figure 3, left panel), interventions targeting interpersonal conflict and emotional isolation might be most promising. Participants were most likely to agree with statements that were associated with interpersonal conflict (highest positive agreement score). Hence, participants agreed with statements that described experiencing emotional consequences due to conflicts (Figure 3, left (agreeableness): t(726) = 2.35, p=0.019, 95% confidence interval (CI): [0.017 0.19]) and further, reported that reducing interpersonal conflict and being better able to handle emotional consequences from these conflicts could lead to substantial improvement of their lives (Figure 3, right (improvement): t(726)=12.8, p<0.0001, 85% CI: [0.61 0.83]). The second psychological factor with which participants agreed most frequently was emotional isolation. They further indicated that feeling more connected in their emotional experience to others would improve their lives.
The remaining psychological factors showed on average a negative agreement score, indicating that participants do not experience too many of these psychological factors in their daily life. Even though this data suggests which psychological factor might be important, it is measured in isolation from any outcome measure. In other words, even though on average participants might agree less with statements describing negative self-image, variations in negative self-image might still be predictive of changes in life quality or depression. This is what we set out to investigate next.
Psychological Factors, Depression and Life Quality
The relationship between each psychological factor and both outcome measures was first tested with a pairwise correlation (Figure 4). A significant correlation indicates that higher values of the psychological factors are associated with higher values of the outcome measure. For example, a higher score for self-image is associated with a higher life quality.
All psychological factors tested correlated with depression. In other words, higher values of the psychological factors were related to experiencing more depressive symptoms. Even though all intervention significantly correlated with depression, negative self-talk had the highest correlation: participants who engaged more often in negative self-talk also reported more severe depression (Figure 4a, grey-coloured bar in the upper panel). Other interventions that were strongly correlated with depression were emotional overwhelm (i.e., difficulties controlling intense emotions) and negative self-image.
Next, we measured the relationship between each psychological factor and life quality. A pairwise correlation showed that self-image strongly correlated with life quality: participants who see themselves more negatively are also those who experience lower life quality (Figure 4b, grey-coloured bar in the upper panel). Other psychological factors that were predictive of changes in life quality were emotional isolation (i.e. a lack of understanding that others might have very similar emotional experiences) and negative self-talk.
Figure 4. Pairwise correlations between life quality, depression and psychological factors. (A) Correlation between depression and psychological characteristics. The upper left panel illustrates a bar graph of each pairwise Pearson correlation. The lower left panel illustrates a correlation matrix, additionally showing the correlation across psychological factors. Depression was most correlated with self-reported negative self-talk: participants who engaged more frequently in negative self-talk also experienced higher scores on depression. (B) Correlation between life quality and all psychological factors are represented in the upper right panel, while correlations across factors are additionally represented in a correlation matrix in the lower right panel. The strongest relationship can be seen between negative self-image and life quality: participants who see themselves more negatively also indicate lower life quality. All correlations presented above were calculated from data in Study 1.
More broadly, these results suggest that the included psychological factors are relevant in predicting both life quality and depression. Furthermore, they hint at which psychological factors might be promising to target for potential effective interventions.
Our next task was to disentangle which psychological factor was most predictive of changes in either depression or life quality when controlling for other factors.
Unique Contribution of Each Psychological Factor to Life Quality and Depression
So far, we have shown that higher levels of some psychological factors are related to higher values of life quality and depression. However, many of these psychological factors are correlated with each other (see correlation matrices for the psychological factors, i.e., the off-diagonal elements in Figure 4) – this means participants who engage more often in negative self-talk are also more likely to have a more negative self-image. Next, we investigated the unique effect of each of these psychological factors by controlling for the effects of alternative factors. We used a multiple linear regression to predict each outcome measure using as independent variables each study participant's score for each psychological factor.
Figure 5. Multiple regression: depression predicted by all psychological factors. We predicted changes in depression across participants in Study 1 (a) and Study 3 (b) by including each participant's score on each psychological factor as independent variables. Both dependent and independent variables were normalised by subtracting the mean and dividing by the standard deviation. When including all psychological factors into the model, the model predicted depression with an R-squared of 0.58 (adjusted R-square of 0.57) in Study 1, while when reducing the amount of psychological factors, the model predicted depression with an R-squared value of 0.66 (adjusted R-square value of 0.65) in Study 3.
Figure 5 illustrates the results of a multiple regression analysis from Study 1 (left, exploratory study) and Study 3 (right, verification study). Across both studies, the strongest predictor of depression was negative self-talk: participants who engaged more often in negative self-talk also scored higher on depression [study 1: t(465) = 6.5, p<0.0001; study 3: t(247) = 8.1, p<0.0001]. A second predictor that replicated across both studies was emotional overwhelm: participants who were less able to control their emotions during emotionally intense situations scored higher on depression [study 1: t(465) = 2.1, p=0.04; study 3: t(247) = 2.35, p=0.009]. Note that we are now accounting for the shared variance across negative self-talk and negative self-image (Pearson correlation r>0.8) by including both factors in the same multiple regression analysis. In such a way, negative self-image no longer predicted depression when controlling for negative self-talk. Both negative self-talk and emotional overwhelm were the only psychological factors predicting depression that replicated across both Study 1 and Study 3 and thereby, are the most promising potential target interventions to improve depression. However, due to the design of this study, we cannot be certain that an intervention that were to improve these psychological factors would necessarily cause a reduction in depression.
Figure 6. Multiple regression: life quality predicted by all psychological factors. We predicted changes in life quality scores across participants in Study 1 (a) and Study 3 (b) by including as independent variables the scores for each participant on each psychological factor. Both dependent and independent variables were normalised to have mean 0 and standard deviation 1. When including all psychological factors into the model, the model predicted life quality with an R-squared of 0.51 (adjusted R-square of 0.5) in Study 1, while when reducing the amount of psychological factors, the model predicted life quality with an R-squared value of 0.38 (adjusted R-square value of 0.36) in Study 3.
In a second separate analysis, we tested which psychological factor (when controlling for all other factors) predicted changes in life quality (Figure 6). Across both Study 1 and Study 3, the strongest predictor of life quality which replicated across studies was negative self-image[study 1: t(465) = -6.1, p<0.0001; study 3: t(247) = -5.83, p<0.0001]: participants who viewed themselves more negatively also experienced lower quality of life. The only other predictor that replicated across both studies was the impact of emotional isolation on life quality [study 1: t(465) = -5.2, p=0.0004; study 3: t(247) = -3.51, p=0.0005]: participants who felt isolated in their emotional experience also experienced lower life quality. Both negative self-image and emotional isolation were the only psychological factors that replicated across both studies.
So far, we have seen that accounting for the shared variance across psychological factors allowed us to identify and replicate two factors that were most predictive of life quality (in one analysis) and depression (in a different analysis). Life quality was predicted by negative self-image and emotional isolation, while depression was predicted by negative self-talk and emotional overwhelm.
Intervention Targets for Depression and Life Quality
Our findings provide the first step towards considering which broad psychological factors are potentially useful intervention targets to improve depression and life quality. In the next analysis, we investigated which items within these scales were most relevant to predict changes in depression and (separately) life quality.
These results might in turn allow us to design interventions that are targeting narrower constructs from specific items rather than general psychological factors. For example, we could test which items from the negative self-image scale predict life quality: whether an item such as ‘I am proud of myself’ is more or less predictive of life quality compared to an item ‘I don’t believe I deserve to be loved’. Such a comparison provides evidence of whether life quality may be most improved by becoming proud of oneself or by increasing self-love/love from others.
Notably, these items are not unrelated to each other. For this reason, we used a multiple regression analysis to test the unique impact of items within each relevant psychological factor to predict life quality and depression. All items (i.e., individual statements shown to participants) across both factors were included into the same linear regression to determine their unique impact on the outcome measure. As noted previously, depression was predicted by negative self-talk and emotional overwhelm. Three items of negative self-talk and one item of emotional overwhelm explained unique variance in depression even when controlling for all remaining items (Figure 1, right panel):
Negative Self-talk: I often tell myself, I am going to fail.
Negative Self-talk: I often tell myself, no one cares about me.
Negative Self-talk: I often tell myself, I don’t deserve to be happy or loved.
Emotional overwhelm: I don’t know how to take control over my emotions.
Life quality was predicted by items across negative self-image and emotional isolation factors. Two items from the self-image factor and one item of the emotional isolation factor predicted unique variance of life quality (Figure 1, left panel):
Self-image: I am proud of myself.
Self-image: Regardless of whether I achieve a lot or a little, I think I am an exceptional person.
Emotional isolation: When I am feeling down, I tend to feel like most other people must be having a better time.
Potential Strategies to Improve Life Quality and Depression
Our research has tried to single out specific items within relevant psychological factors that might be most predictive of scores on depression and life quality. These items might represent starting points when considering potential interventions as they describe which concepts to intervene on. A complementary approach when thinking about potential interventions is to consider strategies that people use to handle difficult circumstances. For example, when faced with difficult life circumstances, a person might engage in a specific strategy – such as gratitude journaling – in order to successfully handle their situation.
Importantly, we did not assume that the eight psychological factors paired exclusively with particular psychological strategies; in other words, we asked participants to select as many strategies as they wanted which they found helpful. In study 1, we presented participants with a number of strategies that could be used to handle difficult or negative life circumstances. We asked participants about the frequency of using certain strategies and how effective they are (or imagine how effective they would be, if they were using them) when handling difficult life situations. An example statement presented to participants would be: “I remind myself of all the things that are still good in my life (despite the negative situation) so that I feel grateful for the things that I do still have” (gratitude strategy; see Appendix Information 4 for a description of all strategies).
Figure 7. Psychological strategies: frequency and subjective effectiveness. In Study 1, we asked participants to indicate how often they use a particular strategy during difficult and negative life situations (Frequency, left panel) and how effective this strategy is (or would be if they would be using it) in helping them to overcome the situations (subjective effectiveness, right panel). Frequency was rated on a 5-point scale from ‘always or almost always’ (2) to ‘sometimes’ (0) to ‘never or almost never’ (-2). Subjective effectiveness was rated on a 7-point scale from ‘completely’ (6) to moderately (3) to not at all (0). See Appendix Information 4 for descriptions of each step within the scale. Participants indicated to use strategies of Gratitude, Acceptance and Planning during difficult life situations. These strategies partially overlapped with the most highly rated strategies on subjective effectiveness; here, participants believed that strategies directed towards gratitude, planning, relying on other people and investing time in healthy activities are most effective in handling difficult life situations.(Study 1, n=474, error bars are denoted as standard error of the mean (S.E.M.))
First, we tested how often each strategy was used, how effective each strategy was rated and whether both of these measures converged. The frequency scores are represented in Figure 7 (left panel) and indicate that a majority of strategies might be relevant as most of them score positively. However, strategies that seemed to be used most often focused on gratitude, acceptance (“I try to fully and completely accept that this negative event occurred and what its consequences are, and accept that this new situation is now the new reality”) and planning (“I develop a plan of what actions I will take to reduce the negative impact of the situation”). These results only partially overlapped with the most highly rated strategies on subjective effectiveness. Participants believed that strategies directed towards gratitude, planning, relying on other people and investing time in healthy activities were most effective in handling difficult life circumstances (Figure 7, right panel).
It was interesting to find out which strategies participants were employing and which they regarded as effective. However, as it was our aim to determine which of these strategies might be most objectively useful to improve depression and life quality, we tested the relationship between each strategy and outcome measure with a pairwise correlation (Figure 8). We additionally compared correlation coefficients across different pairwise correlations to inspect which strategy might be most effective when trying to improve depression and life quality.
Figure 8. Pairwise correlation between depression, life quality and all psychological strategies. The number in each box represents the pairwise Pearson correlation coefficient. The main correlations of interest are between the outcome measures (depression and life quality) and each strategy; the respective rows are highlighted with a black rectangle. It can be observed that four strategies correlate positively with Life Quality with Pearson correlation coefficient above r =0.45 (see Figure 7). These strategies focus on gratitude for good things, reminder of being strong, reliance on other people and time investment into healthy activities. Two of these strategies, gratitude and reminder of being strong, have the strongest negative correlation with depression. Interestingly, an increase in using strategies such as ‘distraction’ and ‘repeated Exposure’ was correlated with an increase in depression (Study 1, n=474).
Four strategies positively correlated with Life Quality with a Pearson correlation coefficient of at least r=0.45. These strategies focused on gratitude, relying on others, investing in healthy time and reminder of personal strength (“I remind myself that I am strong and that I am capable of getting through difficult situations like this one.”) (Figure 8, second row). Two of these strategies, gratitude and reminder of personal strength, had the strongest negative correlation with depression amongst all other strategies with depression (Figure 8, first row).
Mediation Analysis: Psychological Factors, Strategies and Outcome Variables
So far we have shown that some psychological factors and some strategies for combating these psychological factors predict changes in life quality and depression – and that the latter could be used to improve both depression and life quality. In a final investigation, we used a mediation analysis to understand in more detail the relationship between the three different types of variables: the relevant psychological factors, the relevant psychological strategies, and the outcome measures life quality and depression.
A mediation analysis tests whether the initial relationship between two types of variables, such as a specific psychological strategy and a specific outcome measure, is explained away by a third type of variable such as psychological factor (the mediator). As an example, one could test whether the initial negative correlation between depression and the strategy ‘gratitude’ is mediated by the psychological factor ‘negative self-talk’. In case of a mediation effect, using a strategy such as gratitude might be useful as it provides a mechanism for changing the psychological factor ‘negative self-talk’. Hence, disentangling the relationships between strategies, psychological factors, and outcome measures would allow us to fine-tune the impact of interventions.
In a mediation analysis, multiple effects can be tested:
Path a: the relationship between the initial two variables, for example psychological strategy and outcome measure;
Path b: the relationship between the psychological factor and the outcome measure;
Path c: the relationship between the psychological factor and the strategy.
Mediation: a mediation occurs when the effect of a psychological strategy on the outcome measure can be explained by an indirect pathway, for example a psychological factor (mediator effect) (see schema in Figure 9a).
We used a mediation analysis to test the relation between each of the outcome measures – depression and life quality – and the set of respectively relevant psychological factors and strategies. We used the psychological factors and strategies that were relevant for predicting depression and/or life quality. We first tested whether there was a mediating effect of psychological factors on the relationship between strategy and outcome measures. These models were calculated separately for life quality and depression. Only two analyses showed a clear mediation effect: in both analyses, the psychological factor ‘negative self-talk’ mediated the relationship between depression and respectively the strategies ‘gratitude’ and ‘reminder of personal strength’ (Figure 8 b,c). Hence, the implementation of strategies such as ‘gratitude’ or ‘reminder of personal strength’ might be useful as they might provide a mechanism for changing the psychological factor of negative self-talk.
This clear pattern was not observed for life quality. When including psychological factors ‘negative self-image’ or ‘emotional isolation’ into a mediation analysis with psychological strategies – including ‘gratitude’, ‘reminder of personal strength’, ‘rely on other people,’ and ‘healthy time’ – they did not fully mediate the relationship between either of these strategies and life quality. In other words, intervening on either of these strategies would not necessarily change the psychological factor.
Figure 9. Mediation analysis between psychological factors, strategies and depression. (a) We used a mediation model to test the relationship amongst the psychological factors, strategies and depression: these analyses were only run for the factors and strategies that have been previously indicated as relevant. (b) Negative self-talk mediated the effect of each strategy (gratitude, panel b; reminder of personal strength, panel c) on depression. Numbers on each arrow indicate regression coefficient. Thick arrows indicate a significant effect, thin arrows indicate a non-significant effect (Study 1, n=474).
Conclusion and Limitations
Across three studies, we tested the relationship between eight psychological factors, divided into the three categories of "social", "emotion," and "self"-related factors, and two outcome measures, depression and life quality. The aim was to disentangle which of these psychological factors was related to changes in these wellbeing outcomes, thereby exploring what type of psychological interventions might be most promising for improving depression and subjective life quality. Amongst a broad set of psychological factors, we showed that life quality and depression were each predicted by two different psychological factors (i.e., two different factors in each change). Life quality was predicted by a person’s self-image and their tendency to feel emotionally isolated. Depression was predicted by negative self-talk and emotional overwhelm. Further, we showed which items within these factors were most predictive of each outcome measure, thereby providing specificity on the potential nature of intervention to target either psychological factor.
It's important to note that we can only draw conclusions about the correlational (rather than the causal) relationships between these factors and the outcome measures. For example, it is unclear whether negative self-image is due to low life quality or whether low life quality leads to negative self-image; randomised controlled studies would be needed to determine with confidence whether changes in these factors cause changes in life quality or depression. Nevertheless, these results offer a starting point for the development of potential interventions for improving mental wellbeing which could then be tested experimentally.
Appendix
Appendix Information 1
Items included into the ‘life quality’ factor after factor analysis.
Note that the analyses conducted above were replicated when using the sum score over two questionnaires: satisfaction with life and five factors of happiness. Most of the items below are in fact derived from these questionnaires.
Life Quality Questions
My life has a purpose.
I know I am on the right path to achieve my objectives.
I get so engaged in what I am doing that I lose track of time.
I feel my health is good.
I am often cheerful.
There are people that I can count on when I need them.
I often meet my personal goals.
I lead a worthy and fruitful life.
I am satisfied with my life.
I am a happy person.
When having conflict with others I know how to handle them effectively.
I talk to someone about how I am feeling.
In most ways, my life is close to ideal.
The conditions of my life are excellent.
So far I have gotten the important things I want in life.
If I could live my Life over, I would change almost nothing.
Appendix Information 2
Depression factor. Depression was measured by using the ‘phq9’ questionnaire:
Over the last two weeks, how often have you been bothered by the following problem?
Little interest or pleasure in doing things.
Feeling down, depressed or hopeless.
Trouble falling or staying asleep, or sleeping too much.
Feeling tired or having little energy.
Poor appetite or overeating.
Feeling bad about yourself - or that you are a failure or have let yourself or your family down.
Trouble concentrating on things, such as reading the newspaper or watching television.
Moving or speaking noticeably slower than usual or the opposite - faster than usual.
Thoughts that you would be better off dead or of hurting yourself in some way.
Appendix Information 3
New items derived from qualitative reports in Study 2.
Original items are shown in italic. The new items were added into the set of questions for Study 3.
Emotional isolation:
When I fail at something that is important to me, I feel alone in my failure.
I know that failing is a normal part of everyone’s life which makes it not so bad when I fail.
When I fail at something that is important to me, I have good support from others.
Self-image:
I am proud of myself.
I am proud of myself when I achieve something that is important to me.
I feel less proud than other people do when I have the same achievement as they have.
I am proud of myself. [original item]
Regardless of whether I achieve a little or a lot, I think I am an exceptional person.
Emotional isolation:
When I am feeling down, I tend like most other people must be having a better time.
When I am feeling bad, I keep in perspective that others also struggle sometimes.
When I am feeling bad, nobody can understand what I am going through.
When I am feeling down, I tend like most other people must be having a better time. [original item]
Emotion overwhelm:
I often struggle with my emotions.
I often struggle with my emotions. [original item]
I usually have insight into my emotional state that helps me cope with my emotions.
Negative self-talk:
I often tell myself that I am not going to make it.
I often tell myself that if I work hard enough, I will succeed.
I am optimistic about my future.
I often tell myself that I am going to fail.
Negative self-talk:
I often tell myself that no one cares about me or likes me.
I have meaningful connections to others.
I feel excluded by other people around me.
I often tell myself that no one cares about me. [original item]
I often tell myself that people don’t like me.
Negative self-image:
I feel positive and confident about myself.
I feel positive about myself.
I feel confident about myself.
I know I can rely on myself.
Interpersonal conflict:
Having a conflict with another person causes me distress.
Having a conflict with another person causes me distress. [original item]
Once I find myself in a conflict, I try to resolve it immediately.
I avoid conflicts with others at all costs.
I rarely have conflict with others.
I think it is not a big deal to sometimes get into irritations with others.
Appendix Information 4
Psychological strategies (Study 1)
We asked participants about the frequency of using certain strategies when being faced with negative life events and how effective they are (or imagine how effective they would be, if they were using them) in helping to go through difficult life situations.
Frequency question: To what extent do you use this strategy when you are faced with a difficult or negative life circumstance?
Answer 5-point scale: always or almost always (2), frequently (1), sometimes (0), rarely (-1), never or almost never (-2)
Effectiveness question: How effective is this strategy (or how effective do you think it would be if you used it) for helping you when you are going through a difficult life circumstance?
Answer 7-point scale: Completely (6), very much (5), quite a bit (4), moderately (3), somewhat (2), slightly (1), not at all (0).
Strategies
Gratitude: I remind myself of all the things that are still good in my life (despite the negative situation) so that I feel grateful for the things that I do still have.
Acceptance: I try to fully and completely accept that this negative event occurred and what its consequences are, and accept that this new situation is now the new reality.
Planning: I develop a plan of what actions I will take to reduce the negative impact of the situation.
Reminder of Strength: I remind myself that I am strong and that I am capable of getting through difficult situations like this one.
Future Fading: I remind myself that the impact of the situation will fade over time and that it will eventually bother me less or improve with time.
Distraction: I distract myself with things I enjoy whenever I am reminded of the negative situation.
Silver Lining: I try to look for any good things that have happened as a result of the negative situation (i.e., I focus on any silver linings).
Rely on People: I rely on people I feel close to as a source of strength, comfort and advice, to help me cope with the negative situation.
Humour: I try to view the negative situation in a humorous light or laugh at the absurdity of the situation to help me cope.
Healthy time: I add more time than usual to engage in activities that I know are generally healthy for me, for example doing exercise, meditating, eating healthy foods, spending time in nature, or taking extra time to relax.
Repeated Exposure: I repeatedly expose myself to the truth or the reality of the situation, for instance by talking about it repeatedly or by imagining in my mind over and over, until I get used to it.
Distortions: I identify distortions or exaggerations in my thinking that may be amplifying my negative feelings about the situation, and then try to correct for them to see the situation more accurately.
Compare to Worse: I mentally contrast the negative situation with even worse possible situations that could have happened but didn't, to try to make myself feel better about the negative event that actually did happen.
Express Emotions: I express my emotions about the situation, for example by writing them into a journal or by talking to others about my feelings.
Appendix Figures
Appendix Figure 1. Factor analysis across items from all psychological factors.
A factor analysis was conducted to test the dis-/similarity between items across psychological factors. The result identified five main factors. Crosses above each item correspond to the highest loading across factors. The type of factors resemble a pattern previously observed when calculating pairwise correlations across psychological factors (Figure 4, off-diagonal). The factors can be summarised approximately as follows: Factor 1: negative self-image and negative self-talk; Factor 2: self-compassion; Factor 3: difficulties with emotional comprehension; Factor 4: coping strategy (not discussed here); Factor 5: emotional overwhelm.
Appendix Figure 2. Linear regression analysis predicting depression with individual items from most predictive psychological factors, negative self-talk and emotional overwhelm.
We predicted depression with all items included in the most predictive psychological factor, negative self-talk (A) and emotional overwhelm (B). Items that predicted depression with a p<0.05 were included into a merged analysis (C) where depression was predicted by items across scales of negative self-talk and emotional overwhelm. Blue filled bars show significant regression weights. The final items depicted in (C) are summarised in Figure 1, and denote the most predictive items of depression. Note that these final analyses were conducted on data from Study 3, after refining items according to the qualitative data from Study 2. Items that correlated above r=0.7 with each other were removed prior to the regression analysis shown in (A) and (B). Both dependent and independent variables are normalised. Both dependent and independent variables were normalised by subtracting the mean and dividing by the standard deviation.
Appendix Figure 3. Linear regression analysis predicting life quality with individual items from most predictive psychological factors, emotional isolation and negative self-image.
We predicted life quality with all items included in the most predictive psychological factor, emotional isolation (A) and negative self-image (B). Items that predicted life quality with a p<0.05 were included into a merged analysis (C) where life quality was predicted by items across scales of emotional isolation and negative self-image. Blue filled bars show significant regression weights. The final items depicted in (C) are summarised in Figure 1, and denote the most predictive items of life quality. Note that these final analyses were conducted on data from Study 3, after refining items according to the qualitative data from Study 2. Items that correlated above r=0.7 with each other were removed prior to the regression analysis shown in (A) and (B). Both dependent and independent variables were normalised by subtracting the mean and dividing by the standard deviation.
Appendix Code
These links will direct you to the code of each study. You will be able to see all questions that were include into each study. Studies were run on the GuidedTrack platform. The code allows you to edit and/or run each study.
Code for Study 1: https://www.guidedtrack.com/programs/15059/edit
Code for Study 2: https://www.guidedtrack.com/programs/16620/edit
Code for Study 3: https://www.guidedtrack.com/programs/16904/edit
The raw data and analysis code can be accessed via this zip file. Note that the raw data is separated into three folders, corresponding to each of the studies. You can also find a README file for further instructions. The analysis code produces all figures that are shown in the maintext. If you have any questions, please contact nadeschatrudel@gmail.com.