TABLE 3.
Summary of key study effect sizes
| Author (year) | Predictor | Outcome | Analysis | Unadjusted effect | Adjusted effect |
|---|---|---|---|---|---|
| Ammerman et al. (2017) | Negative affect (range: 0–40, greater values mean greater negative affect) | Daily NSSI (binary) | Mixed‐model, concurrent association | B = 5.46* | None |
| Anestis et al. (2012) | Negative affect (average across ESM period) | Number of NSSI episodes | Between‐person, concurrent association | r = .04, n.s. | r part = .02, n.s. (adjusting for: affect instability, depression, disordered eating, suicide attempt history, interaction of suicide attempt history and affective lability) |
| Anestis et al. (2012) | Affective instability (across ESM period) | Number of NSSI episodes | Between‐participants, concurrent association | r = .17, n.s. | r part = .01, n.s. (adjusting for: negative affect, depression, disordered eating, suicide attempt history, interaction of suicide attempt history and affective lability) |
| Briones‐Buixassa et al. (2021) | Anger (range: 0–100, greater values mean greater anger) | NSSI (binary) | Between‐person, concurrent association | None | OR = 0.92* (adjusting for: other negative affect variables, person‐level affective problems, Borderline Personality symptoms, metacognitive skill). Several significant interactions are reported |
| Briones‐Buixassa et al. (2021) | Sadness (range: 0–100, greater values mean greater sadness) | NSSI (binary) | Between‐person, concurrent association | None | OR = 0.98, n.s. (adjusting for: other negative affect variables, person‐level affective problems, Borderline Personality symptoms, metacognitive skill). Several significant interactions are reported |
| Briones‐Buixassa et al. (2021) | Guilt (range: 0–100, greater values mean greater guilt) | NSSI (binary) | Between‐person, concurrent association | None | OR = 0.98, n.s. (adjusting for: other negative affect variables, person‐level affective problems, Borderline Personality symptoms, metacognitive skill). Several significant interactions are reported |
| Briones‐Buixassa et al. (2021) | Frustration (range: 0–100, greater values mean greater frustration) | NSSI (binary) | Between‐person, concurrent association | None | OR = 1.05* (adjusting for: other negative affect variables, person‐level affective problems, Borderline Personality symptoms, metacognitive skill). Several significant interactions are reported |
| Dillon et al. (2021) | Anger/hostility (range: 0–4, greater values mean greater anger/hostility) | NSSIT (binary) | Within‐person, lagged association | B = 0.42* | None |
| Dillon et al. (2021) | Anger/hostility (range: 0–4, greater values mean greater anger/hostility) | NSSI (binary) | Within‐person, lagged association | B = 0.32* | None |
| Hepp et al. (2021) | Negative affect (range: 1–5, greater values mean greater negative affect) | NSSIT (binary) | Within‐person, concurrent association | None | OR = 7.50* (adjusting for a large number of covariates relating to the affective, interpersonal and environmental context) |
| Hepp et al. (2021) | Negative affect (range: 1–5, greater values mean greater negative affect) | NSSIT (binary) | Between‐person, concurrent association | None | OR = 5.00* (adjusting for a large number of covariates relating to the affective, interpersonal and environmental context) |
| Hepp et al. (2021) | Negative affect (range: 1–5, greater values mean greater negative affect) | NSSIT (binary) | Within‐person, lagged association | None | OR = 2.38, n.s. (adjusting for a large number of covariates relating to the affective, interpersonal and environmental context) |
| Houben et al. (2017) | Negative affect (range: 0–100, greater values mean greater negative affect) | NSSI (binary) | Mixed‐model, lagged association | B = 0.03* | None |
| Houben et al. (2017) | Positive affect (range: 0–100, greater values mean greater positive affect) | NSSI (binary) | Mixed‐model, lagged association | B = −0.01, n.s. | None |
| Hughes et al. (2019) | Negative affect (range: 0–110, greater value means greater negative affect) | NSSIT (range: 0–10, greater value means more intense thoughts) | Mixed‐model, lagged association | B = 0.01, RR = 1.01 a , * | B = 0.01, RR = 1.01 a , * (adjusting for repetitive negative thinking and its interaction with negative affect) |
| Hughes et al. (2019) | Anxiety (range: 0–10, greater value means greater negative affect) | NSSIT (range: 0–10, greater value means more intense thoughts) | Mixed‐model, lagged association | B = 0.05, RR = 1.05 a , * | B = 0.04, RR = 1.04 a , * (adjusting for repetitive negative thinking and its interaction with negative affect) |
| Hughes et al. (2019) | Feeling overwhelmed (range: 0–10, greater value means greater negative affect) | NSSIT (range: 0–10, greater value means more intense thoughts) | Mixed‐model, lagged association | B = 0.05, RR = 1.05 a , * | B = −0.01, RR = 0.99, a n.s. (adjusting for gender, repetitive negative thinking and its interaction with negative affect) |
| Hughes et al. (2019) | Negative affect (range: 0–110, greater value means greater negative affect) | NSSI frequency | Mixed‐model, lagged association | B = 0.03, RR = 1.03 a , * | B = 0.02, RR = 1.02 a , * (adjusting for repetitive negative thinking and its interaction with negative affect) |
| Hughes et al. (2019) | Anxiety (range: 0–10, greater value means greater negative affect) | NSSI frequency | Mixed‐model, lagged association | B = 0.07, RR = 1.08 a , * | B = 0.02, RR = 1.02, a n.s. (adjusting for repetitive negative thinking and its interaction with negative affect) |
| Hughes et al. (2019) | Feeling overwhelmed (range: 0–10, greater value means greater negative affect) | NSSI frequency | Mixed‐model, lagged association | B = 0.08, RR = 1.09 a , * | B = −0.01, RR = 0.99, a n.s. (adjusting for repetitive negative thinking and its interaction with negative affect) |
| Kiekens et al. (2020) | Negative affect (range: 0–6, greater value means greater negative affect) | NSSIT (range: 0–6, greater value means more intense thoughts) | Mixed‐model, concurrent associations | B = 0.48 (credibility interval: 0.43, 0.53) b | B = 0.28 (credibility interval: 0.22, 0.33) b (adjusting for: positive affect, self‐efficacy to resist NSSI) |
| Kiekens et al. (2020) | Positive affect (range: 0–6, greater value means greater negative affect) | NSSIT (range: 0–6, greater value means more intense thoughts) | Mixed‐model, concurrent associations | B = −0.33 (credibility interval: −0.37, −0.29) b | B = −0.05 (credibility interval: −0.09, −0.00) b (adjusting for: negative affect, self‐efficacy to resist NSSI) |
| Kiekens et al. (2020) | Negative affect (range: 0–6, greater value means greater negative affect) | NSSIT (range: 0–6, greater value means more intense thoughts) | Within‐person, lagged associations | B = 0.17 (credibility interval: 0.11, 0.23) b | B = 0.09 (credibility interval: 0.02, 0.17) b (adjusting for: positive affect, self‐efficacy to resist NSSI) |
| Kiekens et al. (2020) | Positive affect (range: 0–6, greater value means greater negative affect) | NSSIT (range: 0–6, greater value means more intense thoughts) | Within‐person, lagged associations | B = −0.10 (credibility interval: −0.15, −0.05) b | B = 0.00 (credibility interval: −0.05, 0.06) b (adjusting for: negative affect, self‐efficacy to resist NSSI) |
| Kiekens et al. (2020) | Negative affect (range: 0–6, greater value means greater negative affect) | NSSI (binary) | Within‐person, lagged associations | B = 0.26 (credibility interval: 0.12, 0.41) b | None |
| Kiekens et al. (2020) | Positive affect (range: 0–6, greater value means greater negative affect) | NSSI (binary) | Within‐person, lagged associations | B = −0.19 (credibility interval: −0.32, −0.09) b | None |
| Kranzler (2016) | Negative affect (range: 0–130, greater value means greater negative affect) | NSSIT (range: 0–10, greater value means more intense thoughts) | Mixed‐model, lagged association | B = 0.02, RR = 1.02* | B = 0.02, RR = 1.02* (adjusting for positive affect, age, gender, person‐level positive and negative affect, lifetime NSSI frequency) |
| Kranzler (2016) | Positive affect (range: 0–80, greater value means greater positive affect) | NSSIT (range: 0–10, greater value means more intense thoughts) | Mixed‐model, lagged association | B = −0.02, RR = 0.98* | B = −0.01, RR = 0.99* (adjusting for negative affect, age, gender, person‐level positive and negative affect, lifetime NSSI frequency) |
| Kranzler (2016) | Negative affect (range: 0–130, greater value means greater negative affect) | NSSI frequency | Mixed‐model, lagged association | B = 0.01, RR = 1.01* | B = 0.04, RR = 1.04* (adjusting for positive affect, age, gender, person‐level positive and negative affect, lifetime NSSI frequency) |
| Kranzler (2016) | Positive affect (range: 0–80, greater value means greater negative affect) | NSSI frequency | Mixed‐model, lagged association | B = 0.00, RR = 1.00, n.s. | B = 0.01, RR = 1.01, n.s. (adjusting for negative affect, age, gender, person‐level positive and negative affect, lifetime NSSI frequency) |
| Kranzler et al. (2018) | Negative affect (range: 0–130, greater value means greater negative affect) | NSSIT (range: 0–10, greater value means more intense thoughts) | Mixed‐model, lagged association | None | B = 0.02, RR = 1.02* (adjusting for: momentary positive affect, trait positive and negative affect, lifetime NSSI, depression and BPD symptoms) |
| Kranzler et al. (2018) | Positive affect (range: 0–80, greater value means greater positive affect) | NSSIT (range: 0–10, greater value means more intense thoughts) | Mixed‐model, lagged association | None | B = −0.01, RR = 0.99* (adjusting for: momentary negative affect, trait positive and negative affect, lifetime NSSI, depression and BPD symptoms) |
| Kranzler et al. (2018) | Negative affect (range: 0–130, greater value means greater negative affect) | NSSI frequency | Mixed‐model, lagged association | None | B = 0.03, RR = 1.03* (adjusting for: momentary positive affect, trait positive and negative affect, lifetime NSSI, depression and BPD symptoms) |
| Kranzler et al. (2018) | Negative affect (range: 0–130, greater value means greater negative affect) | NSSI frequency | Mixed‐model, lagged association | None | B = −0.00, n.s. (adjusting for: momentary positive affect, trait positive and negative affect, lifetime NSSI, depression and BPD symptoms) |
| Selby et al. (2013) | Affective instability (across ESM period) | NSSI events | Between‐person, concurrent association | None | B = 0.91* (adjusting for rumination instability; also a significant interaction between rumination instability and affective instability) |
| Victor et al. (2019) | Internalising negative affect (range: 1–5, greater values means greater negative affect) | NSSI urges (binary) | Within‐person, lagged | None | β = .24 (credibility interval: .15, .33) b (adjusting for drug use, alcohol use, externalising negative affect, past‐month NSSI urges, depression and BPD symptoms) |
| Victor et al. (2019) | Average internalising negative affect (across ESM period) | NSSI urges (binary) | Between‐person, lagged | None | β = .49 (credibility interval: −.05, .92) b (adjusting for drug use, alcohol use, externalising negative affect, past‐month NSSI urges, depression and BPD symptoms) |
| Victor et al. (2019) | Externalising negative affect (range: 1–5, greater values means greater externalising negative affect) | NSSI urges (binary) | Within‐person, lagged | None | β = .01 (credibility interval: −.11, .12) b (adjusting for drug use, alcohol use, internalising negative affect, past‐month NSSI urges, depression and BPD symptoms) |
| Victor et al. (2019) | Average externalising negative affect (across ESM period) | NSSI urges (binary) | Between‐person, lagged | None | β = −.42 (credibility interval: −.90, .14) b (adjusting for drug use, alcohol use, internalising negative affect, past‐month NSSI urges, depression and BPD symptoms) |
| Zaki et al. (2013) | Negative emotion differentiation (aggregate from across ESM period) | Average NSSI acts and urges (across ESM period) | Between‐person, concurrent | None | B = −0.41, β = −.16 (adjusting for rumination, number of diary entries; also a significant interaction between rumination and negative emotion differentiation) |
Note: β = standardised coefficient; B non‐standardised coefficient. RR = Relative risk ratio.
a
It is not clear from the paper whether these regression coefficients are standardised or unstandardised in nature.
b
This analysis used Bayesian inference with credibility intervals. Credibility intervals that do not cross zero are considered to indicate meaningful effects.
*
p < .05; n.s. = p < .05.