1.INTRODUCTION

This analysis explores depressive symptoms in Austria using data from the European Social Survey (ESS Round 11). After extracting data related to the Austrian country, we specifically examined responses to the 8-item version of the CES-D8 scale, which includes both negative and positive affect items.

The CES-D8 items are grouped into three conceptual domains:

2. METHODOLOGY

2.1 - We first visualized the distribution of responses to these items using Likert-scale percentages.

2.1.1 - Define CES-D8 variable names and Reverse Positive Items

vnames = c("fltdpr", "flteeff", "slprl", "fltlnl", "enjlf", "cldgng", "fltsd", "wrhpp")
likert_df <- DataAT[, vnames]
likert_df <- as.data.frame(lapply(likert_df, as.character), stringsAsFactors = FALSE)

2.1.2 - Basic Likert Plot and Table

# Convert items to ordered factors (Likert requires this)
likert_df[] <- lapply(likert_df, function(x) {
  factor(as.character(x),
         levels = c("1", "2", "3", "4"),
         labels = c("Rarely", "Sometimes", "Often", "Always"),
         ordered = TRUE)
})

# Create basic likert object
likert_obj <- likert::likert(likert_df)
plot(likert_obj)

2.1.3 - Append Mean and Count

# Convert to numeric 
likert_numeric_df = as.data.frame(lapply(DataAT[, vnames], as.numeric))

# Calculation of means
likert_means = c()
for (v in vnames) {
  likert_means[v] = mean(likert_numeric_df[[v]], na.rm = TRUE)
}

# Calculation of counts
likert_counts = c()
for (v in vnames) {
  likert_counts[v] = sum(!is.na(likert_numeric_df[[v]]))
}

# Summary table
likert_table = likert_obj$results
likert_table$Mean = round(unlist(likert_means), 3)
likert_table$Count = unlist(likert_counts)

2.1.4 - Rename Items and Round Percentages

str(likert_table)
## 'data.frame':    8 obs. of  7 variables:
##  $ Item     : chr  "fltdpr" "flteeff" "slprl" "fltlnl" ...
##  $ Rarely   : num  0 0 0 0 0 0 0 0
##  $ Sometimes: num  0 0 0 0 0 0 0 0
##  $ Often    : num  0 0 0 0 0 0 0 0
##  $ Always   : num  0 0 0 0 0 0 0 0
##  $ Mean     : num  1.36 1.59 1.63 1.29 2.81 ...
##  $ Count    : int  2350 2348 2346 2350 2337 2347 2345 2338
# Set descriptive labels
likert_table$Item = c(
  "Felt depressed",
  "Everything was an effort",
  "Sleep was restless",
  "Felt lonely",
  "Enjoyed life",
  "Could not get going",
  "Felt sad",
  "Felt happy")

expected_cols = c("Item", names(likert_obj$results)[2:6], "Mean", "Count")
existing_cols = intersect(expected_cols, names(likert_table))

# Keep only those columns that exist
likert_table = likert_table[, existing_cols]

# Round Likert percentage columns if they exist
percent_cols = intersect(names(likert_table), names(likert_obj$results)[2:6])
likert_table[, percent_cols] = round(likert_table[, percent_cols], 1)

2.1.5 - Display Formatted Table

kable_styling(
  kable(likert_table, caption = "Distribution of depression-related responses in Austria (ESS11)"),
  bootstrap_options = "striped"
)
Distribution of depression-related responses in Austria (ESS11)
Item Rarely Sometimes Often Always Mean Count
Felt depressed 0 0 0 0 1.359 2350
Everything was an effort 0 0 0 0 1.591 2348
Sleep was restless 0 0 0 0 1.633 2346
Felt lonely 0 0 0 0 1.286 2350
Enjoyed life 0 0 0 0 2.813 2337
Could not get going 0 0 0 0 1.363 2347
Felt sad 0 0 0 0 1.361 2345
Felt happy 0 0 0 0 2.873 2338

2.1.6 - Replot from Table

# Plot again using only percentage columns
plot(likert(summary = likert_table[, 1:6]))

2.2 - Next, we computed a total CES-D8 score to estimate overall depressive symptom burden, the predictors of clinically significant depression. A threshold was applied to define clinically significant depression.

2.2.1 - Reverse-code Positive Items ‘Happy’ and ‘Enjoyed life’ are reversed so that higher values mean more depressive symptoms

# Check and convert if numeric version doesn't exist
if (!"enjlf_num" %in% names(DataAT)) {
  DataAT$enjlf_num = as.numeric(DataAT$enjlf)
}
if (!"wrhpp_num" %in% names(DataAT)) {
  DataAT$wrhpp_num = as.numeric(DataAT$wrhpp)
}
DataAT$enjlf_rev = 5 - DataAT$enjlf_num
DataAT$wrhpp_rev = 5 - DataAT$wrhpp_num

2.2.2 - Compute new total CES-D8 score using consistent directionality

In order to be able to do so we need to also convert the other CES-D items to numeric variables:

DataAT$fltdpr_num  = as.numeric(DataAT$fltdpr)
DataAT$flteeff_num = as.numeric(DataAT$flteeff)
DataAT$slprl_num   = as.numeric(DataAT$slprl)
DataAT$fltlnl_num  = as.numeric(DataAT$fltlnl)
DataAT$cldgng_num  = as.numeric(DataAT$cldgng)
DataAT$fltsd_num   = as.numeric(DataAT$fltsd)
DataAT$CES_D8_new = rowSums(DataAT[, c(
  "fltdpr_num", "flteeff_num", "slprl_num", "fltlnl_num", 
  "cldgng_num", "fltsd_num", "enjlf_rev", "wrhpp_rev"
)], na.rm = TRUE)

2.2.3 - Plot of histogram of depressions scores to visualize distribution to choose clinical cutoff

hist(DataAT$CES_D8_new, 
     breaks = 20, 
     main = "Distribution of CES-D8 Depression Scores", 
     xlab = "Total Score", 
     col = "skyblue", 
     border = "white")

2.2.4 - Create binary variable for clinically significant depression

Each CES-D8 item is scored from 1 to 4, yielding a total range from 8 to 32. Based on the score distribution, a threshold of 16 was chosen, as it marks the right tail where symptom severity appears clinically significant.

DataAT$clin_dep = ifelse(DataAT$CES_D8_new >= 16, 1, 0)

2.2.5 - Frequency table

Clinical Depression Status (0 = not clinically depressed, 1 = depressed)
Var1 Freq
0 1877
1 477
Proportion of Sample with Clinical Depression
Var1 Freq
0 79.74
1 20.26

2.3 - Regression Models

Finally, we examine which psychosocial predictors are associated with clinical depression using logistic regression. Key predictors include life satisfaction, perceived control, trust, job insecurity, and self-rated health.

2.3.1 - Logistic Regression Model (Clinical Depression)

Model Summary

## 
## Call:
## glm(formula = clin_dep ~ ctrlife + stflife + ppltrst + testji9 + 
##     stfhlth, family = binomial(link = "logit"), data = DataAT)
## 
## Coefficients:
##                                     Estimate Std. Error z value Pr(>|z|)  
## (Intercept)                         33.33678 2058.24320   0.016   0.9871  
## ctrlife1                           -17.48729 2520.82311  -0.007   0.9945  
## ctrlife2                           -29.68550 2058.24344  -0.014   0.9885  
## ctrlife3                           -16.84719 1455.39839  -0.012   0.9908  
## ctrlife4                           -17.01211 1455.39819  -0.012   0.9907  
## ctrlife5                           -18.18562 1455.39799  -0.012   0.9900  
## ctrlife6                           -17.81210 1455.39797  -0.012   0.9902  
## ctrlife7                           -17.78121 1455.39793  -0.012   0.9903  
## ctrlife8                           -17.87502 1455.39793  -0.012   0.9902  
## ctrlife9                           -18.62558 1455.39793  -0.013   0.9898  
## ctrlifeComplete control            -18.76733 1455.39791  -0.013   0.9897  
## stflife1                           -16.28935 1455.39860  -0.011   0.9911  
## stflife2                           -14.95665 1455.39852  -0.010   0.9918  
## stflife3                           -13.34326 1455.39833  -0.009   0.9927  
## stflife4                           -14.21871 1455.39808  -0.010   0.9922  
## stflife5                           -14.91058 1455.39789  -0.010   0.9918  
## stflife6                           -16.07806 1455.39787  -0.011   0.9912  
## stflife7                           -16.81729 1455.39786  -0.012   0.9908  
## stflife8                           -17.02964 1455.39787  -0.012   0.9907  
## stflife9                           -17.50833 1455.39786  -0.012   0.9904  
## stflifeExtremely satisfied         -17.99883 1455.39789  -0.012   0.9901  
## ppltrst1                             0.95725    0.92210   1.038   0.2992  
## ppltrst2                             0.09383    0.91727   0.102   0.9185  
## ppltrst3                             0.27871    0.81769   0.341   0.7332  
## ppltrst4                            -0.36069    0.85916  -0.420   0.6746  
## ppltrst5                             0.38083    0.80184   0.475   0.6348  
## ppltrst6                             0.23792    0.81516   0.292   0.7704  
## ppltrst7                            -0.33981    0.82249  -0.413   0.6795  
## ppltrst8                            -0.65668    0.83952  -0.782   0.4341  
## ppltrst9                            -0.03446    0.92692  -0.037   0.9703  
## ppltrstMost people can be trusted   -0.39374    1.35445  -0.291   0.7713  
## testji91                            -0.01021    0.45679  -0.022   0.9822  
## testji92                             0.20249    0.41329   0.490   0.6242  
## testji93                             0.19086    0.42733   0.447   0.6551  
## testji94                            -0.26258    0.49132  -0.534   0.5930  
## testji95                             0.22812    0.44125   0.517   0.6052  
## testji96                             0.41988    0.57584   0.729   0.4659  
## testji97                            -1.39141    0.75642  -1.839   0.0658 .
## testji98                             0.67279    0.55147   1.220   0.2225  
## testji99                             0.14389    0.69377   0.207   0.8357  
## testji9Extremely likely             -0.68168    0.77383  -0.881   0.3784  
## stfhlth1                            -1.64731    1.25784  -1.310   0.1903  
## stfhlth2                            -0.13447    1.00966  -0.133   0.8940  
## stfhlth3                             0.73434    0.87273   0.841   0.4001  
## stfhlth4                             0.02530    0.87715   0.029   0.9770  
## stfhlth5                             0.58407    0.85671   0.682   0.4954  
## stfhlth6                            -0.08458    0.89358  -0.095   0.9246  
## stfhlth7                             0.27161    0.83947   0.324   0.7463  
## stfhlth8                            -0.22994    0.84620  -0.272   0.7858  
## stfhlth9                             0.12697    0.86760   0.146   0.8837  
## stfhlthExtremely good                0.80575    0.93533   0.861   0.3890  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for binomial family taken to be 1)
## 
##     Null deviance: 727.46  on 748  degrees of freedom
## Residual deviance: 562.44  on 698  degrees of freedom
##   (1605 observations deleted due to missingness)
## AIC: 664.44
## 
## Number of Fisher Scoring iterations: 14

Odds Ratios with Confidence Intervals

Odds Ratios and 95% Confidence Intervals
OddsRatio CI_lower CI_upper
(Intercept) 3.005939e+14 0.000 NA
ctrlife1 0.000000e+00 0.000 1.546760e+02
ctrlife2 0.000000e+00 NA 2.261190e+126
ctrlife3 0.000000e+00 NA 1.713987e+120
ctrlife4 0.000000e+00 NA 3.120425e+120
ctrlife5 0.000000e+00 NA 2.009736e+120
ctrlife6 0.000000e+00 NA 3.115596e+120
ctrlife7 0.000000e+00 NA 3.700404e+120
ctrlife8 0.000000e+00 NA 3.287505e+120
ctrlife9 0.000000e+00 NA 1.566809e+120
ctrlifeComplete control 0.000000e+00 NA 1.452455e+120
stflife1 0.000000e+00 0.000 0.000000e+00
stflife2 0.000000e+00 0.000 0.000000e+00
stflife3 0.000000e+00 NA 1.427472e+94
stflife4 0.000000e+00 NA 6.905451e+121
stflife5 0.000000e+00 NA 7.236426e+121
stflife6 0.000000e+00 NA 2.452844e+121
stflife7 0.000000e+00 NA 1.221110e+121
stflife8 0.000000e+00 NA 9.668870e+120
stflife9 0.000000e+00 NA 6.097507e+120
stflifeExtremely satisfied 0.000000e+00 NA 3.557520e+120
ppltrst1 2.605000e+00 0.455 1.117100e+01
ppltrst2 1.098000e+00 0.191 7.432000e+00
ppltrst3 1.321000e+00 0.289 4.899000e+00
ppltrst4 6.970000e-01 0.139 4.313000e+00
ppltrst5 1.464000e+00 0.333 8.306000e+00
ppltrst6 1.269000e+00 0.281 7.367000e+00
ppltrst7 7.120000e-01 0.155 4.179000e+00
ppltrst8 5.190000e-01 0.108 3.118000e+00
ppltrst9 9.660000e-01 0.163 6.628000e+00
ppltrstMost people can be trusted 6.750000e-01 0.078 8.281000e+00
testji91 9.900000e-01 0.484 2.015000e+00
testji92 1.224000e+00 0.547 2.784000e+00
testji93 1.210000e+00 0.524 2.368000e+00
testji94 7.690000e-01 0.356 1.989000e+00
testji95 1.256000e+00 0.528 3.004000e+00
testji96 1.522000e+00 0.473 4.603000e+00
testji97 2.490000e-01 0.073 9.600000e-01
testji98 1.960000e+00 0.650 5.732000e+00
testji99 1.155000e+00 0.383 4.195000e+00
testji9Extremely likely 5.060000e-01 0.146 2.067000e+00
stfhlth1 1.930000e-01 0.014 2.107000e+00
stfhlth2 8.740000e-01 0.118 4.273000e+00
stfhlth3 2.084000e+00 0.400 1.269500e+01
stfhlth4 1.026000e+00 0.192 6.211000e+00
stfhlth5 1.793000e+00 0.355 6.834000e+00
stfhlth6 9.190000e-01 0.167 3.797000e+00
stfhlth7 1.312000e+00 0.270 4.935000e+00
stfhlth8 7.950000e-01 0.161 4.616000e+00
stfhlth9 1.135000e+00 0.657 4.510000e+00
stfhlthExtremely good 2.238000e+00 0.372 1.518300e+01

2.3.2 - Linear Regression (Continuous CES-D8 Score)

In addition to the logistic model, we fit a linear regression using the continuous CES-D8 score (CES_D8_new) as the dependent variable. This allows us to capture variation across the full spectrum of depressive symptoms.

Model Summary

Linear Regression Coefficients for Predicting CES-D8 Depression Score
Estimate Std. Error t value Pr(>|t|)
(Intercept) 32.4765542 4.1912147 7.7487212 0.0000000
ctrlife1 -7.0700746 5.3150226 -1.3302059 0.1838849
ctrlife2 -5.3355298 4.2921019 -1.2431042 0.2142468
ctrlife3 -3.8451822 3.4026731 -1.1300475 0.2588444
ctrlife4 -1.9652152 3.2317575 -0.6080949 0.5433224
ctrlife5 -3.6955703 3.1113120 -1.1877852 0.2353221
ctrlife6 -4.5400130 3.0991372 -1.4649280 0.1433908
ctrlife7 -4.1137048 3.0743125 -1.3380894 0.1813031
ctrlife8 -4.6629521 3.0745014 -1.5166531 0.1298070
ctrlife9 -5.0054597 3.0713659 -1.6297178 0.1036123
ctrlifeComplete control -5.2088821 3.0614248 -1.7014568 0.0893026
stflife1 -10.3613303 3.6658183 -2.8264714 0.0048410
stflife2 -7.7595656 3.5080841 -2.2119098 0.0272960
stflife3 -8.5169838 3.2153308 -2.6488671 0.0082588
stflife4 -10.2000938 3.1603684 -3.2275015 0.0013071
stflife5 -10.9464352 3.0633933 -3.5733039 0.0003768
stflife6 -12.9688024 3.0490121 -4.2534440 0.0000239
stflife7 -14.1004303 3.0339154 -4.6476017 0.0000040
stflife8 -14.6637558 3.0347168 -4.8320014 0.0000017
stflife9 -15.8625276 3.0321323 -5.2314760 0.0000002
stflifeExtremely satisfied -16.5528260 3.0306820 -5.4617496 0.0000001
ppltrst1 0.5471289 1.0169432 0.5380132 0.5907394
ppltrst2 -1.0834372 0.9596093 -1.1290399 0.2592689
ppltrst3 -0.3282984 0.8663231 -0.3789561 0.7048357
ppltrst4 -1.3079594 0.8891537 -1.4710161 0.1417374
ppltrst5 -0.8150263 0.8454315 -0.9640358 0.3353618
ppltrst6 -0.8068982 0.8582320 -0.9401866 0.3474471
ppltrst7 -1.1991328 0.8477109 -1.4145540 0.1576452
ppltrst8 -1.1780053 0.8545334 -1.3785362 0.1684795
ppltrst9 -1.8121930 0.9297781 -1.9490597 0.0516885
ppltrstMost people can be trusted -1.5493409 1.2015523 -1.2894494 0.1976690
testji91 -0.0187425 0.4603224 -0.0407159 0.9675340
testji92 0.1405106 0.4132024 0.3400528 0.7339191
testji93 -0.1633152 0.4365794 -0.3740791 0.7084591
testji94 -0.3483347 0.4620653 -0.7538646 0.4511848
testji95 0.0762718 0.4364748 0.1747450 0.8613306
testji96 0.3067147 0.5821774 0.5268405 0.5984717
testji97 -0.3903762 0.5433030 -0.7185239 0.4726748
testji98 0.5925059 0.5543458 1.0688381 0.2855121
testji99 0.4064911 0.5955749 0.6825189 0.4951373
testji9Extremely likely -1.3971119 0.6795785 -2.0558506 0.0401683
stfhlth1 -0.6179849 1.1351344 -0.5444157 0.5863293
stfhlth2 -0.0440316 0.9958708 -0.0442142 0.9647463
stfhlth3 1.3227040 0.8844553 1.4955013 0.1352356
stfhlth4 0.3757689 0.8847030 0.4247402 0.6711570
stfhlth5 0.7879890 0.8640406 0.9119814 0.3620934
stfhlth6 0.3644092 0.8730122 0.4174159 0.6765026
stfhlth7 0.6073071 0.8396200 0.7233119 0.4697306
stfhlth8 0.5988654 0.8331515 0.7187953 0.4725076
stfhlth9 0.2208950 0.8495072 0.2600273 0.7949194
stfhlthExtremely good 0.8956431 0.9089649 0.9853440 0.3247965

This model helps interpret how psychosocial variables relate to depression symptom severity across the whole distribution, rather than just above a clinical threshold.

3 - FINAL VISUAL REPRESENTATIONS

3.1 - Histogram of CES-D8 Depression Scores

This shows the distribution of depression scores to define a meaningful cutoff for “clinically depressed” individuals (suggested: CES-D8 ≥ 16).

3.2 - Bar Chart: Frequency of Clinical Depression

This visualizes prevalence of clinically significant depression in the Austrian sample using a binarized version of CES-D8.

3.3 - Odds Ratios for Predictors of Clinical Depression

This highlights which factors (life control, satisfaction, trust) are protective or risky for clinical depression.

4 - CONCLUSION

This study explored the prevalence and predictors of depression in Austria using data from the European Social Survey (ESS Round 11). We applied both a continuous scoring approach and a clinical classification based on the CES-D8 threshold. Results revealed that depressive symptoms are widespread, with approximately one-fifth of the Austrian sample meeting criteria for clinically significant depression.

By reverse-coding positive affect items and summing the eight indicators, we created a total depression score. A cutoff of ≥16 was used to classify respondents as clinically depressed, following CES-D literature and the distribution of scores in our sample.

By combining descriptive statistics and Likert-scale visualization this report provides a well-rounded picture of how depression manifests and varies within the Austrian population. The findings highlight the importance of subjective wellbeing and perceived autonomy in mental health with potential relevance for targeted prevention, early detection, and public health interventions.