SSEI Project SEM Analysis
Manyu Li
Data cleaning and recoding
data_all <- read.csv ("ssei_working_RSQupdate.csv", header=TRUE)
data_all$id<-1:421Here are our notes on the variables treatment: (just for our record). The data were recoded previously using codes in ssei_SEM.R. The data were then saved as "ssei_recodedclean.csv" and used here for further analysis.
Amy: I've done only the CTQ and SSEI. The only differences is that I got more usable scores with my method. I used the (statistical: mean) method, which takes the average of available items. The original scoring did not compute a mean if any component items had missing data. The method I use simply takes the average of what is available.
SESr is the categorical "levels of victimzation variable" where 0 = no victimization, 1 = unwanted contact, 2 = verbal coercion, 3 = attempted rape, and 4 = rape. SESr_di is dichotomized to reflect the presence or absence of any victimization.
I ran correlations between MSQ and SSEI factors and found that all three MSQ factors were pretty strongly correlated (rs between .47 and .67) with two of the SSEI subscales (skills and experience and adaptiveness; these two subscales were also pretty strongly correlated with each other). Which suggests that the MSQ doesn't provide a ton of unique variance beyond that provided by the SSEI, so maybe we don't need to include it.
For SCS, PositiveAttitude and AwareDiscuss are indeed pro-consent variables. LackControl is anti-consent. The other two are less clear. Indirect is about using non-verbal strategies to establish consent. It doesn't strike me as necessarily good or bad. In a psychometric evaluation of the measure (Humphreys and Brousseau, 2010), the Indirect subscale was positively correlated with sexual assertiveness (a positive indicator of sexual health) but also positively correlated with a measure of sexual sensation seeking (which measures tendencies toward taking sexual risks, which seems like a more negative indicator of sexual health). Sexual consent norms is about the perception that establishing consent is more important in new relationships and/or the beginning of a sexual encounter than in later on. Which seems kind of neutral in terms of sexual health. It was unrelated to the two other things they measured. If they aren't fitting well, we could remove Indirect and Norms from the model, as they are less clearly indicative of sexual health? And for the SSEI, yes, higher scores mean a more positive attitude toward one's sexual behavior. For sexual victimization (SES), the SESr is ordinal data based on victimization severity. Not sure about the assumptions of SEM whether ordinal data can be modeled as continuous, or if it is better to use the dichotomous variable. Amy
I think if we can statistically and conceptually justify creating an overall efficacy score, that should be fine. The original measure included more subscales than we used in this study anyway, so we are already deviating from how the measure was originally intended.
Hung-Chu: Please note that MSQ has been recoded to be on the 0-4 scale. I suggested that we do not use the categorical approach to scoring attachment working models but dimensional approach (greater statistical power and much more supported by the literature). Thus, RSQself and RSQother were scored.
But I looked at the correlation of the two RSQ subscales (attachment anxiety and avoidance), the correlation was high (.56, p <.0001). I think we can enter both into the measurement model of cognitive-emotional functioning. This makes sense theoretically because these two are two distinct aspects of attachment insecurity.
Column AL: ACE Sum (the higher the score, the more ACEs) Column OX: derssum (we may want to try this summed score for the measurement model first) (the higher the score, the greater difficulty in emotional regulation) (if you would like to look into the subscales, the higher the scores, the greater difficulty in ER) Column OJ: RSQself (the higher the score, the higher levels of attachment insecurity/the more negative views about the self) Column OK: RSQother(the higher the score, the higher levels of attachment insecurity/the more negative views about others) Thus, for cognitive-emotional functioning, all RSQself, RSQother, and derssum (as well as all the subscales of ders) are reversely related to positive functioning.
Manyu: I further conducted a factor analysis and I found the results most supportive of 1 factor (eigenvalues for 2nd and 3rd factors are 1.4 and 1.0). The three subscales are also highly correlated (rs > .62). But a disadvantage of using the overall score is that it is against the original design of the scale. We can still put the three subscales in the SEM model and then account for the inter-correlations among the three subscales.
Meeting notes: Hung-Chu: For the cognitive emotional functioning latent variable, make sure DERS and internal working model goes in the same direction.
CTQ/ACE issues Amy: Do not use CTQ_min. What can ACEs tell us that CTQ can't? CTQ sum score + ACEs in the model?
data <- dplyr::select(data_all, id, ACE.Sum, CTQ_PA, CTQ_EA, CTQ_SA, CTQ_PN, CTQ_EN, CTQ_min, CTQtotal
, RSQself, RSQother, derssum
, dersnonaccept, dersgoals, dersawareness, dersstrategies, dersclarity
, SCS_LackControl, SCS_PosAttitude, SCS_Indirect, SCS_Norms,
SCS_AwareDiscuss
, SSEI_SnE, SSEI_Att, SSEI_Con, SSEI_Adapt, SSEI_Mor
, ProtectionEfficacy, STIEfficacy, SexCommEfficacy, OverallEfficacy
, SES, dersmean
,gender, year, age, single, sexor, race
,ACE1:ACE10, finance
)
#reverse so that higher score = higher control; recoding doesn't affect model performance or estimates. It just makes interpretation easier.
data$SCS_LackControl_r<- 8 -(data$SCS_LackControl)
#dichotomizing the SES variable into 0=no assault; 1=any assault experiences, i.e., should negatively relate to Sexual WB
data$SES_di[data$SES==0]<-0
data$SES_di[data$SES>0]<-1
data$SES_di<-as.factor(data$SES_di)
data$sex[data$gender==1]<-1
data$sex[data$gender==2]<-0
data$sex[data$gender==6]<-NA
#include only complete cases of the final sem model
finalsemvar<-dplyr::select(data, RSQself,RSQother,dersmean, SCS_LackControl_r, SCS_PosAttitude, SSEI_SnE , SSEI_Att , SSEI_Con,SSEI_Adapt,SSEI_Mor, OverallEfficacy, sex , ACE.Sum)
finalsemvar_screened<-finalsemvar[complete.cases(finalsemvar), ]
data<-data[complete.cases(finalsemvar), ]checking outliers and normality SEM is not very sensitive to normality, so we may get away with the few nomality issues we have. We can note them in the paper if we want.
I only examined multivariate outliers using Mahalanobis Distance and a cutoff score a chi-square with k degrees of freedom (k = number of variables) and .001 alpha probability. A score above the cutoff is identified as potential outliers. 5 outliers were identified, but they didn't really change the model results. We may include this outlier analysis at the end of the result section.
For missing data, as mentioned above, I just removed them altogether. We may consider substitution with mean/median or other imputation methods.
averaging across items and ignore missing cells
out<-outlier(finalsemvar_screened, print=F)## Warning in plot.window(...): "print" is not a graphical parameter## Warning in plot.xy(xy, type, ...): "print" is not a graphical parameter## Warning in axis(side = side, at = at, labels = labels, ...): "print" is not a
## graphical parameter
## Warning in axis(side = side, at = at, labels = labels, ...): "print" is not a
## graphical parameter## Warning in box(...): "print" is not a graphical parameter## Warning in title(...): "print" is not a graphical parameter## Warning in text.default(Chi2[n.obs:(n.obs - bad + 1)], D2[worst[1:bad]], :
## "print" is not a graphical parameter
#finalsemvarout <- data.frame(finalsemvar,out)
alpha <- .001
cutoff <- (qchisq(p = 1 - alpha, df = ncol(finalsemvar_screened)))
names_outliers_MH <- which(out > cutoff)
excluded_mh <- names_outliers_MH
#finalsemvar_cleaned <- finalsemvar_screened[-excluded_mh, ]
data_clean_mh <- data[-excluded_mh, ]
#we only have 5 multivariate outliers. Removing them or not doesn't really matter to the final results. We may include this in the result report if we want.
#normality is violated in the following scales, but then SEM is quite tolerance of normality so it probably doesn't matter.
#shapiro.test(data_clean_mh$ACE.Sum)
#shapiro.test(data_clean_mh$dersmean)
#shapiro.test(data_clean_mh$SCS_LackControl_r)
#shapiro.test(data_clean_mh$SCS_PosAttitude)
#shapiro.test(data_clean_mh$SSEI_SnE)
#shapiro.test(data_clean_mh$SSEI_Att)
#shapiro.test(data_clean_mh$SSEI_Con)
#shapiro.test(data_clean_mh$SSEI_Adapt)
#shapiro.test(data_clean_mh$SSEI_Mor)
#shapiro.test(data_clean_mh$OverallEfficacy)Descriptive stats and Pearson correlations among the chosen variables
Please do not worry about the format of the tables too much at this point. We can finalize how we present the tables at the end of the manuscript writing problems. Point out, though, if you see any numbers that are off.
Sample Characteristics
samplechar<-dplyr::select(data, gender, year, age, single, sexor, race)
samplechar %>%
mutate(
gender=factor(gender, levels=c(1,2,6), labels=c("Male","Female","Non-binary")),
single=factor(single, levels=c(1,0), labels=c("Single","In a relationship")),
sexor=factor(sexor, levels=c(1,2,3,4,5), labels=c("Heterosexual","Homosexual","Bisexual", "Asexual","No Responses")),
race=factor(race, levels=c(1,2,3,4,5,7), labels=c("Asians","Black","White","Hispanic","Native Americans","Multiracial/ethnic"))
) %>%
tbl_summary(label=list(year~"Year in College", age~"Age", single~"Relationship Status",
sexor~"Sexual Orientation", race~"Race/Ethnicity"),
statistic = list(age ~ "{mean} ({sd})"),
digits=list(age~c(1,2)),
missing_text = "No Responses"
)| Characteristic | N = 3291 |
|---|---|
| gender | |
| Male | 96 (29%) |
| Female | 233 (71%) |
| Non-binary | 0 (0%) |
| Year in College | |
| 1 | 173 (53%) |
| 2 | 89 (27%) |
| 3 | 36 (11%) |
| 4 | 30 (9.1%) |
| 6 | 1 (0.3%) |
| Age | 19.3 (2.41) |
| Relationship Status | |
| Single | 167 (51%) |
| In a relationship | 161 (49%) |
| No Responses | 1 |
| Sexual Orientation | |
| Heterosexual | 276 (84%) |
| Homosexual | 14 (4.3%) |
| Bisexual | 24 (7.3%) |
| Asexual | 7 (2.1%) |
| No Responses | 8 (2.4%) |
| Race/Ethnicity | |
| Asians | 3 (0.9%) |
| Black | 73 (22%) |
| White | 223 (68%) |
| Hispanic | 8 (2.5%) |
| Native Americans | 0 (0%) |
| Multiracial/ethnic | 19 (5.8%) |
| No Responses | 3 |
| 1 n (%); Mean (SD) | |
Descriptive Statistics of Tested Variables The list may be adjusted after we determine what variables we are including in the model.
var<-dplyr::select(data, ACE.Sum, CTQ_PA,CTQ_EA,CTQ_SA,CTQ_PN,CTQ_EN,CTQ_min, CTQtotal, RSQself,RSQother,derssum,dersmean, SCS_LackControl_r, SCS_PosAttitude, SCS_AwareDiscuss
,SSEI_SnE , SSEI_Att , SSEI_Con,SSEI_Adapt,SSEI_Mor
,OverallEfficacy
,SES_di)
tbl_summary(var
,statistic = list(all_continuous()~ "{mean} ({sd}; {min}-{max})")
,missing_text = "Missing"
, digits=list(all_continuous()~c(1,2))
)| Characteristic | N = 3291 |
|---|---|
| ACE.Sum | |
| 0 | 166 (50%) |
| 1 | 88 (27%) |
| 4 | 31 (9.4%) |
| 5 | 16 (4.9%) |
| 6 | 12 (3.6%) |
| 7 | 11 (3.3%) |
| 8 | 5 (1.5%) |
| CTQ_PA | 1.4 (0.60; 1.0-4.80) |
| CTQ_EA | 1.7 (0.96; 1.0-5.00) |
| CTQ_SA | 1.1 (0.48; 1.0-4.40) |
| CTQ_PN | 1.3 (0.50; 1.0-3.80) |
| CTQ_EN | 1.7 (0.81; 1.0-4.40) |
| CTQ_min | 2.8 (1.17; 1.0-5.00) |
| CTQtotal | 1.4 (0.49; 1.0-4.16) |
| Missing | 35 |
| RSQself | 0.1 (2.67; -5.7-6.30) |
| RSQother | -0.5 (1.78; -5.5-5.00) |
| derssum | 98.1 (21.89; 51.0-166.00) |
| Missing | 46 |
| dersmean | 2.5 (0.70; 1.1-4.74) |
| SCS_LackControl_r | 5.9 (1.23; 1.3-7.00) |
| SCS_PosAttitude | 5.5 (1.40; 1.0-7.00) |
| SCS_AwareDiscuss | 4.0 (1.58; 1.0-7.00) |
| Missing | 4 |
| SSEI_SnE | 4.3 (1.09; 1.0-6.00) |
| SSEI_Att | 3.8 (1.37; 1.0-6.00) |
| SSEI_Con | 4.5 (1.07; 1.1-6.00) |
| SSEI_Adapt | 4.0 (0.81; 1.0-5.71) |
| SSEI_Mor | 4.2 (1.12; 1.0-6.00) |
| OverallEfficacy | 4.1 (0.85; 1.0-5.00) |
| SES_di | |
| 0 | 108 (41%) |
| 1 | 153 (59%) |
| Missing | 68 |
| 1 n (%); Mean (SD; Minimum-Maximum) | |
Note: ACE was formed by summing items (i.e., range = 0-10). RSQ self (i.e., Positive Model of Self) was computed by (Secure + Dismissing) - (Preoccupied + Fearful), and RSQ other (i.e., Positive Model of Others) was computed by (Secure + Preoccupied) - (Dismissing + Fearful). The possible range of RSQ self is -30 to 40, and RSQ other is -36 to +36. Difficulties in Emotion Regulation (DERS) was computed by summing scores of 36 items rated on a 5-point Likert scale (i.e., range = 36-180). Sexual Experience Survey (SES) was dichotomized as experience of sexual assault (1) and no experience of sexual assualt (0).
Other variables were simple average of all items in the subscales/scale. The ranges are 1-5 for CTQ, 1-5 for DERS, 1-7 for SCS, 1-6 for SSEI, and 1-5 for Sexuxl self-efficacy.
Correlations among tested variables
rcorr(as.matrix(var))## ACE.Sum CTQ_PA CTQ_EA CTQ_SA CTQ_PN CTQ_EN CTQ_min CTQtotal
## ACE.Sum 1.00 0.52 0.65 0.40 0.60 0.62 0.57 0.72
## CTQ_PA 0.52 1.00 0.60 0.32 0.45 0.51 0.45 0.75
## CTQ_EA 0.65 0.60 1.00 0.25 0.56 0.74 0.66 0.88
## CTQ_SA 0.40 0.32 0.25 1.00 0.27 0.23 0.24 0.40
## CTQ_PN 0.60 0.45 0.56 0.27 1.00 0.65 0.48 0.75
## CTQ_EN 0.62 0.51 0.74 0.23 0.65 1.00 0.72 0.86
## CTQ_min 0.57 0.45 0.66 0.24 0.48 0.72 1.00 0.70
## CTQtotal 0.72 0.75 0.88 0.40 0.75 0.86 0.70 1.00
## RSQself -0.25 -0.20 -0.38 -0.09 -0.26 -0.31 -0.33 -0.36
## RSQother -0.23 -0.13 -0.34 -0.18 -0.18 -0.31 -0.35 -0.32
## derssum 0.21 0.08 0.36 0.06 0.22 0.23 0.24 0.31
## dersmean 0.22 0.11 0.35 0.11 0.25 0.28 0.26 0.31
## SCS_LackControl_r 0.06 0.05 0.04 0.01 -0.08 0.00 -0.01 0.03
## SCS_PosAttitude 0.09 -0.02 0.08 0.02 -0.07 0.00 0.03 0.04
## SCS_AwareDiscuss 0.16 0.02 0.13 -0.03 0.13 0.07 0.15 0.09
## SSEI_SnE -0.01 0.07 -0.06 0.11 -0.12 -0.10 -0.10 -0.08
## SSEI_Att -0.18 -0.11 -0.31 -0.08 -0.20 -0.30 -0.30 -0.29
## SSEI_Con -0.07 -0.07 -0.20 -0.02 -0.20 -0.18 -0.17 -0.22
## SSEI_Adapt -0.01 0.01 -0.07 0.06 -0.14 -0.14 -0.09 -0.08
## SSEI_Mor -0.01 -0.05 -0.03 0.03 -0.09 -0.04 -0.03 -0.06
## OverallEfficacy 0.08 0.05 0.00 0.11 -0.03 -0.05 -0.04 0.02
## SES_di 0.15 0.21 0.14 0.11 0.26 0.17 0.09 0.23
## RSQself RSQother derssum dersmean SCS_LackControl_r
## ACE.Sum -0.25 -0.23 0.21 0.22 0.06
## CTQ_PA -0.20 -0.13 0.08 0.11 0.05
## CTQ_EA -0.38 -0.34 0.36 0.35 0.04
## CTQ_SA -0.09 -0.18 0.06 0.11 0.01
## CTQ_PN -0.26 -0.18 0.22 0.25 -0.08
## CTQ_EN -0.31 -0.31 0.23 0.28 0.00
## CTQ_min -0.33 -0.35 0.24 0.26 -0.01
## CTQtotal -0.36 -0.32 0.31 0.31 0.03
## RSQself 1.00 0.54 -0.63 -0.62 0.11
## RSQother 0.54 1.00 -0.38 -0.36 0.08
## derssum -0.63 -0.38 1.00 0.77 -0.06
## dersmean -0.62 -0.36 0.77 1.00 -0.15
## SCS_LackControl_r 0.11 0.08 -0.06 -0.15 1.00
## SCS_PosAttitude -0.08 -0.10 0.07 -0.03 0.52
## SCS_AwareDiscuss -0.12 -0.08 0.20 0.11 0.23
## SSEI_SnE 0.28 0.21 -0.15 -0.25 0.25
## SSEI_Att 0.48 0.30 -0.39 -0.47 0.07
## SSEI_Con 0.40 0.30 -0.32 -0.37 0.26
## SSEI_Adapt 0.27 0.13 -0.18 -0.31 0.21
## SSEI_Mor 0.25 0.08 -0.25 -0.32 0.18
## OverallEfficacy 0.07 -0.02 -0.03 -0.17 0.36
## SES_di -0.16 -0.10 0.16 0.16 -0.11
## SCS_PosAttitude SCS_AwareDiscuss SSEI_SnE SSEI_Att SSEI_Con
## ACE.Sum 0.09 0.16 -0.01 -0.18 -0.07
## CTQ_PA -0.02 0.02 0.07 -0.11 -0.07
## CTQ_EA 0.08 0.13 -0.06 -0.31 -0.20
## CTQ_SA 0.02 -0.03 0.11 -0.08 -0.02
## CTQ_PN -0.07 0.13 -0.12 -0.20 -0.20
## CTQ_EN 0.00 0.07 -0.10 -0.30 -0.18
## CTQ_min 0.03 0.15 -0.10 -0.30 -0.17
## CTQtotal 0.04 0.09 -0.08 -0.29 -0.22
## RSQself -0.08 -0.12 0.28 0.48 0.40
## RSQother -0.10 -0.08 0.21 0.30 0.30
## derssum 0.07 0.20 -0.15 -0.39 -0.32
## dersmean -0.03 0.11 -0.25 -0.47 -0.37
## SCS_LackControl_r 0.52 0.23 0.25 0.07 0.26
## SCS_PosAttitude 1.00 0.35 0.08 -0.01 0.04
## SCS_AwareDiscuss 0.35 1.00 0.00 -0.11 -0.12
## SSEI_SnE 0.08 0.00 1.00 0.41 0.54
## SSEI_Att -0.01 -0.11 0.41 1.00 0.45
## SSEI_Con 0.04 -0.12 0.54 0.45 1.00
## SSEI_Adapt 0.10 0.08 0.66 0.30 0.41
## SSEI_Mor 0.12 0.17 0.33 0.26 0.38
## OverallEfficacy 0.42 0.22 0.27 0.16 0.21
## SES_di 0.00 0.11 0.02 -0.15 -0.23
## SSEI_Adapt SSEI_Mor OverallEfficacy SES_di
## ACE.Sum -0.01 -0.01 0.08 0.15
## CTQ_PA 0.01 -0.05 0.05 0.21
## CTQ_EA -0.07 -0.03 0.00 0.14
## CTQ_SA 0.06 0.03 0.11 0.11
## CTQ_PN -0.14 -0.09 -0.03 0.26
## CTQ_EN -0.14 -0.04 -0.05 0.17
## CTQ_min -0.09 -0.03 -0.04 0.09
## CTQtotal -0.08 -0.06 0.02 0.23
## RSQself 0.27 0.25 0.07 -0.16
## RSQother 0.13 0.08 -0.02 -0.10
## derssum -0.18 -0.25 -0.03 0.16
## dersmean -0.31 -0.32 -0.17 0.16
## SCS_LackControl_r 0.21 0.18 0.36 -0.11
## SCS_PosAttitude 0.10 0.12 0.42 0.00
## SCS_AwareDiscuss 0.08 0.17 0.22 0.11
## SSEI_SnE 0.66 0.33 0.27 0.02
## SSEI_Att 0.30 0.26 0.16 -0.15
## SSEI_Con 0.41 0.38 0.21 -0.23
## SSEI_Adapt 1.00 0.52 0.29 0.03
## SSEI_Mor 0.52 1.00 0.35 -0.07
## OverallEfficacy 0.29 0.35 1.00 -0.01
## SES_di 0.03 -0.07 -0.01 1.00
##
## n
## ACE.Sum CTQ_PA CTQ_EA CTQ_SA CTQ_PN CTQ_EN CTQ_min CTQtotal
## ACE.Sum 329 329 329 329 329 329 329 294
## CTQ_PA 329 329 329 329 329 329 329 294
## CTQ_EA 329 329 329 329 329 329 329 294
## CTQ_SA 329 329 329 329 329 329 329 294
## CTQ_PN 329 329 329 329 329 329 329 294
## CTQ_EN 329 329 329 329 329 329 329 294
## CTQ_min 329 329 329 329 329 329 329 294
## CTQtotal 294 294 294 294 294 294 294 294
## RSQself 329 329 329 329 329 329 329 294
## RSQother 329 329 329 329 329 329 329 294
## derssum 283 283 283 283 283 283 283 260
## dersmean 329 329 329 329 329 329 329 294
## SCS_LackControl_r 329 329 329 329 329 329 329 294
## SCS_PosAttitude 329 329 329 329 329 329 329 294
## SCS_AwareDiscuss 325 325 325 325 325 325 325 291
## SSEI_SnE 329 329 329 329 329 329 329 294
## SSEI_Att 329 329 329 329 329 329 329 294
## SSEI_Con 329 329 329 329 329 329 329 294
## SSEI_Adapt 329 329 329 329 329 329 329 294
## SSEI_Mor 329 329 329 329 329 329 329 294
## OverallEfficacy 329 329 329 329 329 329 329 294
## SES_di 261 261 261 261 261 261 261 232
## RSQself RSQother derssum dersmean SCS_LackControl_r
## ACE.Sum 329 329 283 329 329
## CTQ_PA 329 329 283 329 329
## CTQ_EA 329 329 283 329 329
## CTQ_SA 329 329 283 329 329
## CTQ_PN 329 329 283 329 329
## CTQ_EN 329 329 283 329 329
## CTQ_min 329 329 283 329 329
## CTQtotal 294 294 260 294 294
## RSQself 329 329 283 329 329
## RSQother 329 329 283 329 329
## derssum 283 283 283 283 283
## dersmean 329 329 283 329 329
## SCS_LackControl_r 329 329 283 329 329
## SCS_PosAttitude 329 329 283 329 329
## SCS_AwareDiscuss 325 325 281 325 325
## SSEI_SnE 329 329 283 329 329
## SSEI_Att 329 329 283 329 329
## SSEI_Con 329 329 283 329 329
## SSEI_Adapt 329 329 283 329 329
## SSEI_Mor 329 329 283 329 329
## OverallEfficacy 329 329 283 329 329
## SES_di 261 261 221 261 261
## SCS_PosAttitude SCS_AwareDiscuss SSEI_SnE SSEI_Att SSEI_Con
## ACE.Sum 329 325 329 329 329
## CTQ_PA 329 325 329 329 329
## CTQ_EA 329 325 329 329 329
## CTQ_SA 329 325 329 329 329
## CTQ_PN 329 325 329 329 329
## CTQ_EN 329 325 329 329 329
## CTQ_min 329 325 329 329 329
## CTQtotal 294 291 294 294 294
## RSQself 329 325 329 329 329
## RSQother 329 325 329 329 329
## derssum 283 281 283 283 283
## dersmean 329 325 329 329 329
## SCS_LackControl_r 329 325 329 329 329
## SCS_PosAttitude 329 325 329 329 329
## SCS_AwareDiscuss 325 325 325 325 325
## SSEI_SnE 329 325 329 329 329
## SSEI_Att 329 325 329 329 329
## SSEI_Con 329 325 329 329 329
## SSEI_Adapt 329 325 329 329 329
## SSEI_Mor 329 325 329 329 329
## OverallEfficacy 329 325 329 329 329
## SES_di 261 257 261 261 261
## SSEI_Adapt SSEI_Mor OverallEfficacy SES_di
## ACE.Sum 329 329 329 261
## CTQ_PA 329 329 329 261
## CTQ_EA 329 329 329 261
## CTQ_SA 329 329 329 261
## CTQ_PN 329 329 329 261
## CTQ_EN 329 329 329 261
## CTQ_min 329 329 329 261
## CTQtotal 294 294 294 232
## RSQself 329 329 329 261
## RSQother 329 329 329 261
## derssum 283 283 283 221
## dersmean 329 329 329 261
## SCS_LackControl_r 329 329 329 261
## SCS_PosAttitude 329 329 329 261
## SCS_AwareDiscuss 325 325 325 257
## SSEI_SnE 329 329 329 261
## SSEI_Att 329 329 329 261
## SSEI_Con 329 329 329 261
## SSEI_Adapt 329 329 329 261
## SSEI_Mor 329 329 329 261
## OverallEfficacy 329 329 329 261
## SES_di 261 261 261 261
##
## P
## ACE.Sum CTQ_PA CTQ_EA CTQ_SA CTQ_PN CTQ_EN CTQ_min CTQtotal
## ACE.Sum 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
## CTQ_PA 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
## CTQ_EA 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
## CTQ_SA 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
## CTQ_PN 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
## CTQ_EN 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
## CTQ_min 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
## CTQtotal 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
## RSQself 0.0000 0.0004 0.0000 0.1089 0.0000 0.0000 0.0000 0.0000
## RSQother 0.0000 0.0195 0.0000 0.0014 0.0010 0.0000 0.0000 0.0000
## derssum 0.0004 0.1776 0.0000 0.3248 0.0002 0.0000 0.0000 0.0000
## dersmean 0.0000 0.0435 0.0000 0.0520 0.0000 0.0000 0.0000 0.0000
## SCS_LackControl_r 0.3035 0.3245 0.4428 0.9133 0.1717 0.9779 0.9266 0.6438
## SCS_PosAttitude 0.0991 0.7655 0.1593 0.6959 0.2133 0.9305 0.5380 0.5073
## SCS_AwareDiscuss 0.0034 0.6540 0.0217 0.5424 0.0179 0.2118 0.0066 0.1317
## SSEI_SnE 0.8559 0.1796 0.2935 0.0428 0.0283 0.0706 0.0785 0.1565
## SSEI_Att 0.0014 0.0428 0.0000 0.1489 0.0004 0.0000 0.0000 0.0000
## SSEI_Con 0.1765 0.2099 0.0003 0.7197 0.0003 0.0008 0.0017 0.0002
## SSEI_Adapt 0.8485 0.9013 0.2187 0.2803 0.0110 0.0133 0.1039 0.1737
## SSEI_Mor 0.8919 0.4086 0.5873 0.6497 0.0936 0.4439 0.6251 0.3302
## OverallEfficacy 0.1260 0.4074 0.9954 0.0410 0.5591 0.3281 0.4307 0.7253
## SES_di 0.0132 0.0007 0.0197 0.0823 0.0000 0.0064 0.1613 0.0004
## RSQself RSQother derssum dersmean SCS_LackControl_r
## ACE.Sum 0.0000 0.0000 0.0004 0.0000 0.3035
## CTQ_PA 0.0004 0.0195 0.1776 0.0435 0.3245
## CTQ_EA 0.0000 0.0000 0.0000 0.0000 0.4428
## CTQ_SA 0.1089 0.0014 0.3248 0.0520 0.9133
## CTQ_PN 0.0000 0.0010 0.0002 0.0000 0.1717
## CTQ_EN 0.0000 0.0000 0.0000 0.0000 0.9779
## CTQ_min 0.0000 0.0000 0.0000 0.0000 0.9266
## CTQtotal 0.0000 0.0000 0.0000 0.0000 0.6438
## RSQself 0.0000 0.0000 0.0000 0.0413
## RSQother 0.0000 0.0000 0.0000 0.1703
## derssum 0.0000 0.0000 0.0000 0.3010
## dersmean 0.0000 0.0000 0.0000 0.0078
## SCS_LackControl_r 0.0413 0.1703 0.3010 0.0078
## SCS_PosAttitude 0.1442 0.0695 0.2429 0.6083 0.0000
## SCS_AwareDiscuss 0.0250 0.1532 0.0008 0.0392 0.0000
## SSEI_SnE 0.0000 0.0000 0.0136 0.0000 0.0000
## SSEI_Att 0.0000 0.0000 0.0000 0.0000 0.2389
## SSEI_Con 0.0000 0.0000 0.0000 0.0000 0.0000
## SSEI_Adapt 0.0000 0.0225 0.0018 0.0000 0.0001
## SSEI_Mor 0.0000 0.1332 0.0000 0.0000 0.0012
## OverallEfficacy 0.1883 0.7144 0.6231 0.0016 0.0000
## SES_di 0.0096 0.1054 0.0180 0.0092 0.0647
## SCS_PosAttitude SCS_AwareDiscuss SSEI_SnE SSEI_Att SSEI_Con
## ACE.Sum 0.0991 0.0034 0.8559 0.0014 0.1765
## CTQ_PA 0.7655 0.6540 0.1796 0.0428 0.2099
## CTQ_EA 0.1593 0.0217 0.2935 0.0000 0.0003
## CTQ_SA 0.6959 0.5424 0.0428 0.1489 0.7197
## CTQ_PN 0.2133 0.0179 0.0283 0.0004 0.0003
## CTQ_EN 0.9305 0.2118 0.0706 0.0000 0.0008
## CTQ_min 0.5380 0.0066 0.0785 0.0000 0.0017
## CTQtotal 0.5073 0.1317 0.1565 0.0000 0.0002
## RSQself 0.1442 0.0250 0.0000 0.0000 0.0000
## RSQother 0.0695 0.1532 0.0000 0.0000 0.0000
## derssum 0.2429 0.0008 0.0136 0.0000 0.0000
## dersmean 0.6083 0.0392 0.0000 0.0000 0.0000
## SCS_LackControl_r 0.0000 0.0000 0.0000 0.2389 0.0000
## SCS_PosAttitude 0.0000 0.1443 0.8132 0.4690
## SCS_AwareDiscuss 0.0000 0.9939 0.0438 0.0284
## SSEI_SnE 0.1443 0.9939 0.0000 0.0000
## SSEI_Att 0.8132 0.0438 0.0000 0.0000
## SSEI_Con 0.4690 0.0284 0.0000 0.0000
## SSEI_Adapt 0.0853 0.1280 0.0000 0.0000 0.0000
## SSEI_Mor 0.0332 0.0019 0.0000 0.0000 0.0000
## OverallEfficacy 0.0000 0.0000 0.0000 0.0032 0.0002
## SES_di 0.9958 0.0682 0.7995 0.0138 0.0001
## SSEI_Adapt SSEI_Mor OverallEfficacy SES_di
## ACE.Sum 0.8485 0.8919 0.1260 0.0132
## CTQ_PA 0.9013 0.4086 0.4074 0.0007
## CTQ_EA 0.2187 0.5873 0.9954 0.0197
## CTQ_SA 0.2803 0.6497 0.0410 0.0823
## CTQ_PN 0.0110 0.0936 0.5591 0.0000
## CTQ_EN 0.0133 0.4439 0.3281 0.0064
## CTQ_min 0.1039 0.6251 0.4307 0.1613
## CTQtotal 0.1737 0.3302 0.7253 0.0004
## RSQself 0.0000 0.0000 0.1883 0.0096
## RSQother 0.0225 0.1332 0.7144 0.1054
## derssum 0.0018 0.0000 0.6231 0.0180
## dersmean 0.0000 0.0000 0.0016 0.0092
## SCS_LackControl_r 0.0001 0.0012 0.0000 0.0647
## SCS_PosAttitude 0.0853 0.0332 0.0000 0.9958
## SCS_AwareDiscuss 0.1280 0.0019 0.0000 0.0682
## SSEI_SnE 0.0000 0.0000 0.0000 0.7995
## SSEI_Att 0.0000 0.0000 0.0032 0.0138
## SSEI_Con 0.0000 0.0000 0.0002 0.0001
## SSEI_Adapt 0.0000 0.0000 0.6582
## SSEI_Mor 0.0000 0.0000 0.2282
## OverallEfficacy 0.0000 0.0000 0.8233
## SES_di 0.6582 0.2282 0.8233Checking measurement model
data$SES_di<-as.numeric(data$SES_di)
#model 1 doesn't work with SEM even though it shows good measurement fit. This is because, for some unknown reasons (likely power issues), the SSEI subscales have negative covariances in the SEM model.
model1<-'
childhood=~ ACE.Sum+CTQtotal
cogemo=~RSQself+RSQother+dersmean
sexwb=~OverallEfficacy+
SCS_LackControl_r
+ SCS_PosAttitude
+SSEI_SnE + SSEI_Att + SSEI_Con+SSEI_Adapt+SSEI_Mor
#+SES_di
SSEI_SnE ~~SSEI_Att + SSEI_Con+SSEI_Adapt+SSEI_Mor+SCS_LackControl_r
SSEI_Att ~~SSEI_Con+SSEI_Adapt+SSEI_Mor
SSEI_Con~~SSEI_Adapt+SSEI_Mor+SCS_LackControl_r
SSEI_Adapt~~SSEI_Mor
SCS_LackControl_r~~SCS_PosAttitude
OverallEfficacy~~ SCS_PosAttitude
'
#Model 2 has acceptable measurement properties and fit in the SEM
model2<-'
childhood=~ACE.Sum
cogemo=~RSQself+RSQother+dersmean
sexwb=~OverallEfficacy+
SCS_LackControl_r
+ SCS_PosAttitude
+SSEI_SnE + SSEI_Att + SSEI_Con+SSEI_Adapt+SSEI_Mor
+SES_di
SSEI_SnE ~~SSEI_Att + SSEI_Con+SSEI_Adapt+SCS_LackControl_r
SSEI_Adapt~~SSEI_Mor
SCS_LackControl_r~~SCS_PosAttitude
OverallEfficacy~~ SCS_PosAttitude+SCS_LackControl_r
'
fit <- cfa(model2, data=data
# , se="bootstrap"
# , bootstrap=1000, check.gradient = FALSE
)
summary(fit, standardized=T, fit.measures=T)## lavaan 0.6-11 ended normally after 77 iterations
##
## Estimator ML
## Optimization method NLMINB
## Number of model parameters 36
##
## Used Total
## Number of observations 261 329
##
## Model Test User Model:
##
## Test statistic 125.290
## Degrees of freedom 55
## P-value (Chi-square) 0.000
##
## Model Test Baseline Model:
##
## Test statistic 1027.353
## Degrees of freedom 78
## P-value 0.000
##
## User Model versus Baseline Model:
##
## Comparative Fit Index (CFI) 0.926
## Tucker-Lewis Index (TLI) 0.895
##
## Loglikelihood and Information Criteria:
##
## Loglikelihood user model (H0) -4896.546
## Loglikelihood unrestricted model (H1) -4833.901
##
## Akaike (AIC) 9865.092
## Bayesian (BIC) 9993.415
## Sample-size adjusted Bayesian (BIC) 9879.279
##
## Root Mean Square Error of Approximation:
##
## RMSEA 0.070
## 90 Percent confidence interval - lower 0.054
## 90 Percent confidence interval - upper 0.086
## P-value RMSEA <= 0.05 0.023
##
## Standardized Root Mean Square Residual:
##
## SRMR 0.069
##
## Parameter Estimates:
##
## Standard errors Standard
## Information Expected
## Information saturated (h1) model Structured
##
## Latent Variables:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## childhood =~
## ACE.Sum 1.000 2.325 1.000
## cogemo =~
## RSQself 1.000 2.204 0.864
## RSQother 0.464 0.054 8.662 0.000 1.022 0.569
## dersmean -0.235 0.021 -10.953 0.000 -0.517 -0.736
## sexwb =~
## OverallEfficcy 1.000 0.282 0.340
## SCS_LckCntrl_r 1.266 0.330 3.841 0.000 0.357 0.287
## SCS_PosAttitud -0.229 0.382 -0.601 0.548 -0.065 -0.046
## SSEI_SnE 2.019 0.482 4.190 0.000 0.569 0.537
## SSEI_Att 2.911 0.656 4.439 0.000 0.820 0.607
## SSEI_Con 2.609 0.574 4.547 0.000 0.735 0.680
## SSEI_Adapt 1.447 0.344 4.212 0.000 0.408 0.509
## SSEI_Mor 2.066 0.492 4.195 0.000 0.582 0.505
## SES_di -0.398 0.146 -2.723 0.006 -0.112 -0.227
##
## Covariances:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## .SSEI_SnE ~~
## .SSEI_Att 0.120 0.061 1.971 0.049 0.120 0.125
## .SSEI_Con 0.169 0.053 3.219 0.001 0.169 0.239
## .SSEI_Adapt 0.309 0.043 7.228 0.000 0.309 0.501
## .SCS_LackControl_r ~~
## .SSEI_SnE 0.000 0.047 0.003 0.997 0.000 0.000
## .SSEI_Adapt ~~
## .SSEI_Mor 0.228 0.045 5.016 0.000 0.228 0.332
## .SCS_LackControl_r ~~
## .SCS_PosAttitud 0.904 0.119 7.625 0.000 0.904 0.543
## .OverallEfficacy ~~
## .SCS_PosAttitud 0.483 0.075 6.431 0.000 0.483 0.443
## .SCS_LckCntrl_r 0.295 0.063 4.661 0.000 0.295 0.317
## childhood ~~
## cogemo -1.355 0.362 -3.741 0.000 -0.264 -0.264
## sexwb -0.038 0.049 -0.782 0.434 -0.059 -0.059
## cogemo ~~
## sexwb 0.435 0.104 4.192 0.000 0.700 0.700
##
## Variances:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## .ACE.Sum 0.000 0.000 0.000
## .RSQself 1.647 0.360 4.574 0.000 1.647 0.253
## .RSQother 2.185 0.212 10.303 0.000 2.185 0.677
## .dersmean 0.226 0.027 8.234 0.000 0.226 0.458
## .OverallEfficcy 0.608 0.056 10.899 0.000 0.608 0.885
## .SCS_LckCntrl_r 1.419 0.128 11.058 0.000 1.419 0.918
## .SCS_PosAttitud 1.956 0.171 11.415 0.000 1.956 0.998
## .SSEI_SnE 0.800 0.084 9.572 0.000 0.800 0.712
## .SSEI_Att 1.152 0.126 9.148 0.000 1.152 0.631
## .SSEI_Con 0.628 0.078 8.058 0.000 0.628 0.538
## .SSEI_Adapt 0.475 0.045 10.487 0.000 0.475 0.741
## .SSEI_Mor 0.989 0.098 10.069 0.000 0.989 0.745
## .SES_di 0.230 0.021 11.218 0.000 0.230 0.948
## childhood 5.407 0.473 11.424 0.000 1.000 1.000
## cogemo 4.858 0.642 7.565 0.000 1.000 1.000
## sexwb 0.079 0.033 2.430 0.015 1.000 1.000Structural equation model
Here is the model. As we can see above, most correlations were significant. However, it is not good to include all correlations in our model. We should only add correlation for residuals if we believe that theoretically they are related or if we believe that the latent model cannot account for those correlations. In this analysis, I only included subscales as residual correlates and significant, large correlations. (And actually, technically speaking, including residual correlations of subscales are not ideal... because we assume they are accounted for in the measurement model, but things are usually not that perfect)
See more https://pdfs.semanticscholar.org/df3d/7d8f2a3177b1146ee155ccbb936330c68fe0.pdf
Model without SES as an indicator of sexual well-being
model<-'
#measurement model
childhood=~ ACE.Sum+CTQtotal
# +CTQ_PA+CTQ_EA
# +CTQ_SA
# +CTQ_PN+CTQ_EN
# +CTQ_min
#childhood=~ACE1+ACE2+ACE3+ACE4+ACE5+ACE6+ACE7+ACE8+ACE9+ACE10
cogemo=~RSQself+RSQother+dersmean
sexwb=~OverallEfficacy+
SCS_LackControl_r
+ SCS_PosAttitude
+SSEI_SnE + SSEI_Att + SSEI_Con+SSEI_Adapt+SSEI_Mor
#+SES_di
#regression
sexwb~c1*childhood+b1*cogemo+sex
#+b2*SES_di
cogemo~a*childhood+sex
#SES_di~a2*childhood+sex
#residual correlations
SSEI_SnE ~~SSEI_Att + SSEI_Con+SSEI_Adapt+SSEI_Mor+SCS_LackControl_r
SSEI_Att ~~SSEI_Con+SSEI_Adapt+SSEI_Mor
SSEI_Con~~SSEI_Adapt+SSEI_Mor+SCS_LackControl_r
SSEI_Adapt~~SSEI_Mor
SCS_LackControl_r~~SCS_PosAttitude
OverallEfficacy~~ SCS_PosAttitude
#mediation
indirect:=a*b1
total:=c1+a*b1
'
fit <- sem(model, data=data
# , se="bootstrap"
# , bootstrap=1000, check.gradient = FALSE
)## Warning in lav_object_post_check(object): lavaan WARNING: some estimated ov
## variances are negativesummary(fit, standardized=T, fit.measures=T) ## lavaan 0.6-11 ended normally after 105 iterations
##
## Estimator ML
## Optimization method NLMINB
## Number of model parameters 45
##
## Used Total
## Number of observations 294 329
##
## Model Test User Model:
##
## Test statistic 155.423
## Degrees of freedom 59
## P-value (Chi-square) 0.000
##
## Model Test Baseline Model:
##
## Test statistic 1433.823
## Degrees of freedom 91
## P-value 0.000
##
## User Model versus Baseline Model:
##
## Comparative Fit Index (CFI) 0.928
## Tucker-Lewis Index (TLI) 0.889
##
## Loglikelihood and Information Criteria:
##
## Loglikelihood user model (H0) -5379.739
## Loglikelihood unrestricted model (H1) -5302.028
##
## Akaike (AIC) 10849.478
## Bayesian (BIC) 11015.239
## Sample-size adjusted Bayesian (BIC) 10872.531
##
## Root Mean Square Error of Approximation:
##
## RMSEA 0.075
## 90 Percent confidence interval - lower 0.060
## 90 Percent confidence interval - upper 0.089
## P-value RMSEA <= 0.05 0.003
##
## Standardized Root Mean Square Residual:
##
## SRMR 0.072
##
## Parameter Estimates:
##
## Standard errors Standard
## Information Expected
## Information saturated (h1) model Structured
##
## Latent Variables:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## childhood =~
## ACE.Sum 1.000 1.454 0.697
## CTQtotal 0.347 0.054 6.433 0.000 0.504 1.030
## cogemo =~
## RSQself 1.000 2.278 0.848
## RSQother 0.447 0.048 9.394 0.000 1.018 0.574
## dersmean -0.229 0.019 -12.036 0.000 -0.521 -0.740
## sexwb =~
## OverallEfficcy 1.000 0.254 0.293
## SCS_LckCntrl_r 0.947 0.366 2.591 0.010 0.241 0.195
## SCS_PosAttitud 0.042 0.337 0.124 0.901 0.011 0.008
## SSEI_SnE 2.294 0.653 3.514 0.000 0.583 0.540
## SSEI_Att 4.072 1.082 3.765 0.000 1.035 0.766
## SSEI_Con 2.902 0.786 3.694 0.000 0.738 0.689
## SSEI_Adapt 1.836 0.518 3.545 0.000 0.467 0.581
## SSEI_Mor 2.414 0.693 3.483 0.000 0.613 0.546
##
## Regressions:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## sexwb ~
## childhood (c1) 0.002 0.010 0.188 0.851 0.011 0.011
## cogemo (b1) 0.077 0.021 3.632 0.000 0.694 0.694
## sex 0.055 0.033 1.671 0.095 0.216 0.098
## cogemo ~
## childhood (a) -0.642 0.100 -6.424 0.000 -0.409 -0.409
## sex 0.810 0.298 2.723 0.006 0.356 0.162
##
## Covariances:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## .SSEI_SnE ~~
## .SSEI_Att 0.043 0.125 0.346 0.729 0.043 0.055
## .SSEI_Con 0.232 0.099 2.350 0.019 0.232 0.330
## .SSEI_Adapt 0.279 0.071 3.939 0.000 0.279 0.470
## .SSEI_Mor 0.033 0.089 0.375 0.708 0.033 0.039
## .SCS_LackControl_r ~~
## .SSEI_SnE 0.145 0.052 2.781 0.005 0.145 0.132
## .SSEI_Att ~~
## .SSEI_Con -0.134 0.143 -0.934 0.350 -0.134 -0.199
## .SSEI_Adapt -0.153 0.098 -1.569 0.117 -0.153 -0.270
## .SSEI_Mor -0.240 0.132 -1.825 0.068 -0.240 -0.294
## .SSEI_Con ~~
## .SSEI_Adapt 0.012 0.074 0.168 0.867 0.012 0.024
## .SSEI_Mor -0.003 0.099 -0.034 0.973 -0.003 -0.005
## .SCS_LackControl_r ~~
## .SSEI_Con 0.119 0.061 1.949 0.051 0.119 0.127
## .SSEI_Adapt ~~
## .SSEI_Mor 0.187 0.074 2.534 0.011 0.187 0.304
## .SCS_LackControl_r ~~
## .SCS_PosAttitud 0.699 0.099 7.054 0.000 0.699 0.422
## .OverallEfficacy ~~
## .SCS_PosAttitud 0.369 0.064 5.748 0.000 0.369 0.325
##
## Variances:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## .ACE.Sum 2.243 0.355 6.319 0.000 2.243 0.515
## .CTQtotal -0.014 0.036 -0.396 0.692 -0.014 -0.060
## .RSQself 2.034 0.352 5.776 0.000 2.034 0.281
## .RSQother 2.102 0.192 10.925 0.000 2.102 0.670
## .dersmean 0.225 0.025 8.917 0.000 0.225 0.453
## .OverallEfficcy 0.686 0.058 11.803 0.000 0.686 0.914
## .SCS_LckCntrl_r 1.462 0.121 12.076 0.000 1.462 0.962
## .SCS_PosAttitud 1.879 0.152 12.356 0.000 1.879 1.000
## .SSEI_SnE 0.827 0.107 7.732 0.000 0.827 0.709
## .SSEI_Att 0.754 0.220 3.427 0.001 0.754 0.413
## .SSEI_Con 0.601 0.126 4.772 0.000 0.601 0.525
## .SSEI_Adapt 0.427 0.063 6.765 0.000 0.427 0.662
## .SSEI_Mor 0.888 0.120 7.396 0.000 0.888 0.702
## childhood 2.114 0.432 4.892 0.000 1.000 1.000
## .cogemo 4.185 0.563 7.430 0.000 0.806 0.806
## .sexwb 0.032 0.014 2.317 0.020 0.493 0.493
##
## Defined Parameters:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## indirect -0.050 0.015 -3.204 0.001 -0.284 -0.284
## total -0.048 0.016 -2.969 0.003 -0.273 -0.273Model with SES as an indicator of sexual well-being
model3<-'
#measurement model
childhood=~ ACE.Sum+CTQtotal
# +CTQ_PA+CTQ_EA
# +CTQ_SA
# +CTQ_PN+CTQ_EN
# +CTQ_min
#childhood=~ACE1+ACE2+ACE3+ACE4+ACE5+ACE6+ACE7+ACE8+ACE9+ACE10
cogemo=~RSQself+RSQother+dersmean
sexwb=~
SCS_LackControl_r
+ SCS_PosAttitude
+SSEI_SnE + SSEI_Att + SSEI_Con+SSEI_Adapt+SSEI_Mor
+OverallEfficacy
+SES_di
#regression
sexwb~c1*childhood+b1*cogemo+sex
cogemo~a*childhood+sex
#residual correlations
SSEI_SnE ~~SSEI_Att + SSEI_Con+SSEI_Adapt+SCS_LackControl_r
SSEI_Adapt~~SSEI_Mor
SCS_LackControl_r~~SCS_PosAttitude
OverallEfficacy~~ SCS_PosAttitude+SCS_LackControl_r
ACE.Sum~~CTQtotal
#mediation
indirect_consent:=a*b1
total:=c1+a*b1
'
fit2 <- sem(model3, data=data
# , se="bootstrap", bootstrap=1000, check.gradient = FALSE
)## Warning in lav_model_vcov(lavmodel = lavmodel, lavsamplestats = lavsamplestats, : lavaan WARNING:
## Could not compute standard errors! The information matrix could
## not be inverted. This may be a symptom that the model is not
## identified.summary(fit2, standardized=T, fit.measures=T) ## lavaan 0.6-11 ended normally after 91 iterations
##
## Estimator ML
## Optimization method NLMINB
## Number of model parameters 42
##
## Used Total
## Number of observations 232 329
##
## Model Test User Model:
##
## Test statistic 171.806
## Degrees of freedom 77
## P-value (Chi-square) 0.000
##
## Model Test Baseline Model:
##
## Test statistic 1185.911
## Degrees of freedom 105
## P-value 0.000
##
## User Model versus Baseline Model:
##
## Comparative Fit Index (CFI) 0.912
## Tucker-Lewis Index (TLI) 0.880
##
## Loglikelihood and Information Criteria:
##
## Loglikelihood user model (H0) -4432.453
## Loglikelihood unrestricted model (H1) -4346.551
##
## Akaike (AIC) 8948.907
## Bayesian (BIC) 9093.670
## Sample-size adjusted Bayesian (BIC) 8960.552
##
## Root Mean Square Error of Approximation:
##
## RMSEA 0.073
## 90 Percent confidence interval - lower 0.058
## 90 Percent confidence interval - upper 0.087
## P-value RMSEA <= 0.05 0.006
##
## Standardized Root Mean Square Residual:
##
## SRMR 0.078
##
## Parameter Estimates:
##
## Standard errors Standard
## Information Expected
## Information saturated (h1) model Structured
##
## Latent Variables:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## childhood =~
## ACE.Sum 1.000 0.911 0.409
## CTQtotal 0.347 NA 0.317 0.606
## cogemo =~
## RSQself 1.000 2.237 0.859
## RSQother 0.469 NA 1.049 0.582
## dersmean -0.243 NA -0.544 -0.752
## sexwb =~
## SCS_LckCntrl_r 1.000 0.313 0.251
## SCS_PosAttitud -0.464 NA -0.145 -0.102
## SSEI_SnE 1.753 NA 0.549 0.521
## SSEI_Att 2.579 NA 0.808 0.604
## SSEI_Con 2.221 NA 0.695 0.641
## SSEI_Adapt 1.343 NA 0.421 0.529
## SSEI_Mor 1.851 NA 0.580 0.499
## OverallEfficcy 0.809 NA 0.253 0.301
## SES_di -0.414 NA -0.130 -0.263
##
## Regressions:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## sexwb ~
## childhood (c1) 0.041 NA 0.120 0.120
## cogemo (b1) 0.118 NA 0.840 0.840
## sex -0.074 NA -0.235 -0.075
## cogemo ~
## childhood (a) -1.736 NA -0.707 -0.707
## sex 0.625 NA 0.279 0.090
##
## Covariances:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## .SSEI_SnE ~~
## .SSEI_Att 0.165 NA 0.165 0.172
## .SSEI_Con 0.239 NA 0.239 0.320
## .SSEI_Adapt 0.282 NA 0.282 0.465
## .SCS_LackControl_r ~~
## .SSEI_SnE 0.012 NA 0.012 0.011
## .SSEI_Adapt ~~
## .SSEI_Mor 0.235 NA 0.235 0.346
## .SCS_LackControl_r ~~
## .SCS_PosAttitud 0.891 NA 0.891 0.521
## .SCS_PosAttitude ~~
## .OverallEfficcy 0.510 NA 0.510 0.448
## .SCS_LackControl_r ~~
## .OverallEfficcy 0.294 NA 0.294 0.303
## .ACE.Sum ~~
## .CTQtotal 0.567 NA 0.567 0.671
##
## Variances:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## .ACE.Sum 4.138 NA 4.138 0.833
## .CTQtotal 0.172 NA 0.172 0.632
## .RSQself 1.778 NA 1.778 0.262
## .RSQother 2.147 NA 2.147 0.661
## .dersmean 0.227 NA 0.227 0.434
## .SCS_LckCntrl_r 1.457 NA 1.457 0.937
## .SCS_PosAttitud 2.013 NA 2.013 0.990
## .SSEI_SnE 0.808 NA 0.808 0.728
## .SSEI_Att 1.136 NA 1.136 0.635
## .SSEI_Con 0.692 NA 0.692 0.589
## .SSEI_Adapt 0.455 NA 0.455 0.720
## .SSEI_Mor 1.015 NA 1.015 0.751
## .OverallEfficcy 0.643 NA 0.643 0.909
## .SES_di 0.226 NA 0.226 0.931
## childhood 0.830 NA 1.000 1.000
## .cogemo 2.461 NA 0.492 0.492
## .sexwb 0.042 NA 0.429 0.429
##
## Defined Parameters:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## indirect_cnsnt -0.204 -0.594 -0.594
## total -0.163 -0.473 -0.473