2.4. Statistical procedures
SPSS software version 22.0 [25] was used for demography data
(frequencies, means, range, SD) and multinominal logistic regression.
In the mutinominal logistic regression-analysis Suicidal
ideation and Suicidal plans were the dependent variables while
Depression, Hopelessness, Lack of Insight, Suspiciousness/Perse-
cution, Hallucinations, Gender, Age, Anxiety, Agitation/Excitement,
Negative Symptoms and Drug Use were explanatory
variables. Suicide attempters were excluded from the analysis,
as they were too few to be included. Since the outcome variable
was at a nominal level with more than 2 values a multinominal
logistic regression analysis was performed. The sample size gives
statistical power for the analyses of suicidal ideation and plans
regarding the number of variables and model parameters [3]. We
aimed to determine the associations between Hallucinations,
Suspiciousness/Persecution and other explanatory variables and
Suicidal ideation and Suicidal plans, respectively. Significance level
was set at 0.05, two-sided.
Secondly, a structural equation model (SEM) was estimated in
MPlus version 7.11 with direct and indirect relations between the
explanatory variables [40]. SEM is a methodology for representing,
estimating, and testing a network of relationships between
observed and latent variables [16]. The SEM-model explores
multiple relationships simultaneously, presenting a more complete
picture of associations which makes it clinically more
applicable.
The regression used Maximum Likelihood and SEM used Robust
Maximum Likelihood as the model estimator. The Suicide variable
was split in Suicidal ideation and Suicide plans as in the
Multinominal logistic regression analysis.
2.4 2.4. Statistical procedures
SPSS software version 22.0 [25] was used for demography data
(frequencies, means, range, SD) and multinominal logistic regression.
In the mutinominal logistic regression-analysis Suicidal
ideation and Suicidal plans were the dependent variables while
Depression, Hopelessness, Lack of Insight, Suspiciousness/Perse-
cution, Hallucinations, Gender, Age, Anxiety, Agitation/Excitement,
Negative Symptoms and Drug Use were explanatory
variables. Suicide attempters were excluded from the analysis,
as they were too few to be included. Since the outcome variable
was at a nominal level with more than 2 values a multinominal
logistic regression analysis was performed. The sample size gives
statistical power for the analyses of suicidal ideation and plans
regarding the number of variables and model parameters [3]. We
aimed to determine the associations between Hallucinations,
Suspiciousness/Persecution and other explanatory variables and
Suicidal ideation and Suicidal plans, respectively. Significance level
was set at 0.05, two-sided.
Secondly, a structural equation model (SEM) was estimated in
MPlus version 7.11 with direct and indirect relations between the
explanatory variables [40]. SEM is a methodology for representing,
estimating, and testing a network of relationships between
observed and latent variables [16]. The SEM-model explores
multiple relationships simultaneously, presenting a more complete
picture of associations which makes it clinically more
applicable.
The regression used Maximum Likelihood and SEM used Robust
Maximum Likelihood as the model estimator. The Suicide variable
was split in Suicidal ideation and Suicide plans as in the
Multinominal logistic regression analysis.
การแปล กรุณารอสักครู่..
