CESD Analysis Report
Melinda Higgins
March 21, 2016
Background
The Center for Epidemiologic Studies Depression Scale (CESD) was created in 1977 by Laurie Radloff (Radloff 1977). It was revised in 2004 by William Eaton and others (Eaton et al. 2004) with additional validation published in 2011 by Nicholas T. Van Dam and Mitch Earleywine (Dam and Earleywine 2011). Later in 2012, Eaton, Ybarra and Schwab (Eaton, Ybarra, and Schwab 2012) published a letter to the editor advertising that the CESD-R was available on the web at http://cesd-r.com/.
More information can also be found at the website of the Americal Psychological Association.
The original and revised versions have 20 items. However, there have also been shortened versions validated as:
- a 10 item version among HIV-Positive People in British Columbia, Canada (Zhang et al. 2012) and
- a 8 item version in Older Adults (Karim et al. 2014)
Methods
The analyses and the report were completed using R (R Core Team 2015b) with the following packages:
foreign(R Core Team 2015a)ggplot2(Wickham 2009)mnormt(Azzalini and Genz 2016)likert(Bryer and Speerschneider 2015)psych(Revelle 2015)knitr(Xie 2016; Xie 2015; Xie 2014)knitcitations(Boettiger 2015)
Data Summary
The following hypothetical dataset has 65 subjects and 30 variables which are the following:
Table 1: Variables in Dataset
| Label | |
|---|---|
| age | Age |
| raceW | Race |
| cesd01 | ces_d1: 1: Bothered by things that usually dont bother me. |
| cesd02 | ces_d2: 2: Did not feel like eating; my appetite was poor. |
| cesd03 | ces_d3: 3: I felt that I could not shake off the blues |
| cesd04 | ces_d4: 4: I felt I was just as good as other people |
| cesd05 | ces_d5: 5: I had trouble keeping mind on what was doing |
| cesd06 | ces_d6: 6: I felt depressed |
| cesd07 | ces_d7: 7: I felt that everything I did wasan effort |
| cesd08 | ces_d8: 8: I felt hopeful about the future |
| cesd09 | ces_d9: 9: I thought my life had been a failure |
| cesd10 | ces_d10: 10: I felt fearful. |
| cesd11 | ces_d11: 11: My sleep was restless. |
| cesd12 | ces_d12: 12: I was happy. |
| cesd13 | ces_d13: 13: I talked less than usual. |
| cesd14 | ces_d14: 14: I felt lonely. |
| cesd15 | ces_d15: 15: People were unfriendly. |
| cesd16 | ces_d16: 16: I enjoyed life. |
| cesd17 | ces_d17: 17: I had crying spells. |
| cesd18 | ces_d18: 18: I felt sad. |
| cesd19 | ces_d19: 19: I felt that people dislike me. |
| cesd20 | ces_d20: 20: I could not get going. |
| cesd04.rev | ces_d4 reversed: 4: I felt I was just as good as other people |
| cesd08.rev | ces_d8 reversed: 8: I felt hopeful about the future |
| cesd12.rev | ces_d12 reversed: 12: I was happy. |
| cesd16.rev | ces_d16 reversed: 16: I enjoyed life. |
| nmiss_cesd | Number of Missing or Skipped Items (out of 20) |
| cesd_total | CESD Sum Total of 20 items (items 4,8,12,15 reversed) |
| cesd_total_meanSub4 | CESD Sum Total w/mean sub up to 4 items |
| cesd_gte16 | CESD Total sub 4 - >= 16 |
Demographics summary
This dataset contains information on 2 demographic variables: age, raceW. The average age of the 65 subjects was 53.97 (SD 12.56) with 38.5% white.
There was a significant difference in Ages between the races (t=-2.265; df=51.8; p=0.02771) given the group means listed below. The 95% confidence interval for these group differences was (-13.21,-0.799)
Table 2: Average Age by Race
| Group Mean | |
|---|---|
| AA non-White | 51.275 |
| White | 58.280 |
CESD Response Completion - Missing/Skipped Items
Of the 65 subjects, 6.154% had one or more items missing. The subjects who had missing or skipped items were subjects 5, 23, 38, 61. The amount of missing or skipped items ranged from 0 to 2.
NOTE: The CESD scoring instructions allow up to 4 items to be missing on the 20-item version (20% of the items). The maximum number of missing items for this dataset was 2 which is is 10% of the total items.
Of the 20 items, 25% had one more subjects missing or skipped responses for that item. The items with missing or skipped responses were: cesd01, cesd07, cesd11, cesd12r, cesd14.
Scoring the CESD
Of the 20 items, 4 need to be reverse coded: items 4,8,12, and 16. Here is an example of the recoding for item 4. The table below shows the frequencies before and after recoding item 4 and item 12.
Table 3: Summary Frequencies for Item 4 Before and After Reverse Coding
| cesd04 | cesd04.rev | |
|---|---|---|
| Rarely or none | 12 | 36 |
| Some or a little | 5 | 12 |
| Ocassionally or moderate | 12 | 5 |
| Most or all | 36 | 12 |
| NA | 0 | 0 |
Table 4: Summary Frequencies for Item 12 Before and After Reverse Coding
| cesd12 | cesd12.rev | |
|---|---|---|
| Rarely or none | 6 | 18 |
| Some or a little | 22 | 17 |
| Ocassionally or moderate | 17 | 22 |
| Most or all | 18 | 6 |
| NA | 2 | 2 |
After reverse coding items 4, 8, 12 and 16, the items are summed together across all 20 items. Since each item is scored from 0 to 3, the expected CESD total scores should range from 0 to 60.
The data included here were scored twice.
- First, using complete case scoring for subjects who completed all items and
- Secons, using within subject mean substitution on 4 or fewer missing items (complete summary already provided above).
Table 5: Summary Statistics for CESD: Complete Cases v Mean Substitution
| Complete Cases | Mean Substition | |
|---|---|---|
| N | 61.00 | 65.00 |
| Mean | 15.84 | 15.97 |
| SD | 9.90 | 10.24 |
| Median | 15.00 | 15.00 |
| Minimum | 0.00 | 0.00 |
| Maximum | 46.00 | 46.00 |
Visual Summary of the CESD (the likert Package)
The likert package by (Bryer and Speerschneider 2015) is very helpful for reviewing the response patterns across all items and by group.
Here is a figure showing the response patterns for the 1st 10 items. For this plot the items are shown in the order as presented in the CESD instrument. Also the responses are centered about the midpoint in the 4 point reponse scale.
Figure 1: Likert Plot of CESD Items 1-10
Here is a variation on this plot - this time with the categories NOT centered about the midpoint of the 4 point response scale.
Figure 2: Likert Plot of CESD Items 1-10 [not centered]
Here is one more variation showing these responses by the 2 Race groups. To make the plot less cluttered we’ll just focus on the 1st 5 items.
Figure 3: Likert Plot of CESD Items 1-5 By Race
Another option is to view the responses across the 4 ordinal categories using density curves. Here are the desnity curves by race.
Figure 4: Likert Density Curves CESD Responses of Items 1-5 By Race
Finally, another top level visualization to see response category concentrations across the items is to use a heatmap to visualize the overall response categories by item.
Figure 5: Likert Heatmap of 20 CESD Items
Psychometric Summary of the CESD
Given this dataset with 65 subjects, the reliability Cronbach’s alpha = 0.8826 (std = 0.8873) with average inter-item correlations = 0.2824. The leave one out alpha evaluation table is also provided below.
Table 5: Summary Statistics for CESD: Complete Cases v Mean Substitution
| raw_alpha | std.alpha | G6(smc) | average_r | S/N | alpha se | |
|---|---|---|---|---|---|---|
| cesd01 | 0.8779 | 0.8831 | 0.9285 | 0.2845 | 7.5549 | 0.0313 |
| cesd02 | 0.8766 | 0.8810 | 0.9271 | 0.2803 | 7.4012 | 0.0315 |
| cesd03 | 0.8733 | 0.8782 | 0.9244 | 0.2751 | 7.2107 | 0.0321 |
| cesd04r | 0.8796 | 0.8838 | 0.9282 | 0.2858 | 7.6027 | 0.0310 |
| cesd05 | 0.8769 | 0.8818 | 0.9273 | 0.2820 | 7.4632 | 0.0315 |
| cesd06 | 0.8675 | 0.8724 | 0.9209 | 0.2646 | 6.8351 | 0.0331 |
| cesd07 | 0.8818 | 0.8857 | 0.9301 | 0.2897 | 7.7504 | 0.0306 |
| cesd08r | 0.8836 | 0.8865 | 0.9310 | 0.2914 | 7.8125 | 0.0303 |
| cesd09 | 0.8748 | 0.8795 | 0.9260 | 0.2776 | 7.3016 | 0.0318 |
| cesd10 | 0.8747 | 0.8799 | 0.9261 | 0.2782 | 7.3238 | 0.0318 |
| cesd11 | 0.8818 | 0.8857 | 0.9306 | 0.2897 | 7.7487 | 0.0306 |
| cesd12r | 0.8737 | 0.8795 | 0.9233 | 0.2775 | 7.2977 | 0.0320 |
| cesd13 | 0.8813 | 0.8863 | 0.9338 | 0.2909 | 7.7947 | 0.0307 |
| cesd14 | 0.8715 | 0.8768 | 0.9203 | 0.2724 | 7.1142 | 0.0324 |
| cesd15 | 0.8787 | 0.8835 | 0.9300 | 0.2853 | 7.5855 | 0.0311 |
| cesd16r | 0.8734 | 0.8795 | 0.9233 | 0.2776 | 7.3003 | 0.0321 |
| cesd17 | 0.8814 | 0.8873 | 0.9294 | 0.2930 | 7.8753 | 0.0307 |
| cesd18 | 0.8707 | 0.8752 | 0.9239 | 0.2696 | 7.0114 | 0.0326 |
| cesd19 | 0.8827 | 0.8887 | 0.9297 | 0.2958 | 7.9819 | 0.0304 |
| cesd20 | 0.8792 | 0.8845 | 0.9311 | 0.2872 | 7.6568 | 0.0310 |
References
Azzalini, Adelchi, and Alan Genz. 2016. The R Package mnormt: The Multivariate Normal and \(t\) Distributions (Version 1.5-4). http://azzalini.stat.unipd.it/SW/Pkg-mnormt.
Boettiger, Carl. 2015. Knitcitations: Citations for ’Knitr’ Markdown Files. https://CRAN.R-project.org/package=knitcitations.
Bryer, Jason, and Kimberly Speerschneider. 2015. Likert: Functions to Analyze and Visualize Likert Type Items. https://CRAN.R-project.org/package=likert.
Dam, Nicholas T. Van, and Mitch Earleywine. 2011. “Validation of the Center for Epidemiologic Studies Depression ScaleRevised (CESD-R): Pragmatic Depression Assessment in the General Population.” Psychiatry Research 186 (1). Elsevier BV: 128–32. doi:10.1016/j.psychres.2010.08.018.
Eaton, William W., Charles Muntaner, Corey Smith, Allen Tien, and Michelle Ybarra. 2004. “Center for Epidemiologic Studies Depression Scale: Review and Revision (CESD and CESR-R).” In The Use of Psychological Testing for Treatment Planning and Outcomes Assessment: Colume 3: Instruments for Adults (3rd Ed), edited by Mark E. Maruish, 363–77. Mahwah, New Jersey: Lawrence Erlbaum Associates Publishers.
Eaton, William W., Michele Ybarra, and Joseph Schwab. 2012. “The CESD-R Is Available on the Web.” Psychiatry Research 196 (1). Elsevier BV: 161. doi:10.1016/j.psychres.2011.08.018.
Karim, Jahanvash, Robert Weisz, Zainab Bibi, and Shafiq ur Rehman. 2014. “Validation of the Eight-Item Center for Epidemiologic Studies Depression Scale (CES-d) Among Older Adults.” Current Psychology 34 (4). Springer Science \(\mathplus\) Business Media: 681–92. doi:10.1007/s12144-014-9281-y.
R Core Team. 2015a. Foreign: Read Data Stored by Minitab, S, SAS, SPSS, Stata, Systat, Weka, DBase, . https://CRAN.R-project.org/package=foreign.
———. 2015b. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. https://www.R-project.org/.
Radloff, L. S. 1977. “The CES-d Scale: A Self-Report Depression Scale for Research in the General Population.” Applied Psychological Measurement 1 (3). SAGE Publications: 385–401. doi:10.1177/014662167700100306.
Revelle, William. 2015. Psych: Procedures for Psychological, Psychometric, and Personality Research. Evanston, Illinois: Northwestern University. http://CRAN.R-project.org/package=psych.
Wickham, Hadley. 2009. Ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York. http://ggplot2.org.
Xie, Yihui. 2014. “Knitr: A Comprehensive Tool for Reproducible Research in R.” In Implementing Reproducible Computational Research, edited by Victoria Stodden, Friedrich Leisch, and Roger D. Peng. Chapman; Hall/CRC. http://www.crcpress.com/product/isbn/9781466561595.
———. 2015. Dynamic Documents with R and Knitr. 2nd ed. Boca Raton, Florida: Chapman; Hall/CRC. http://yihui.name/knitr/.
———. 2016. Knitr: A General-Purpose Package for Dynamic Report Generation in R. http://yihui.name/knitr/.
Zhang, Wendy, Nadia O’Brien, Jamie I. Forrest, Kate A. Salters, Thomas L. Patterson, Julio S. G. Montaner, Robert S. Hogg, and Viviane D. Lima. 2012. “Validating a Shortened Depression Scale (10 Item CES-d) Among HIV-Positive People in British Columbia, Canada.” Edited by Shilpa J. Buch. PLoS ONE 7 (7). Public Library of Science (PLoS): e40793. doi:10.1371/journal.pone.0040793.