NLSY79 Appendix 25: Attitudinal Scale Scoring

This appendix contains descriptions of scale score computations and programs that create some of the scale scores for the most recent survey year for the following attitudinal scales:

• Center for Epidemiologic Studies Depression Scale (CES-D)
• Rosenberg Self-Esteem Scale
• Pearlin Mastery Scale
• Rotter Locus of Control Scale
• Satisfaction with Life Scale
• General Anxiety Disorder Scale

Center for Epidemiologic Studies Depression (CES-D) Scale

Scoring computations are described below.

Please note: Item response theory (IRT) parameter estimates, IRT scores, and their standard errors of measurement for the CES-D variables are also available, along with a custom-weighted z-score and percentile rank. See Appendix 27: IRT Item Parameter Estimates, Scores and Standard Errors for a detailed description and discussion of the IRT scoring for these scales.

The CES-D is a self-report scale that measures the current prevalence of depression symptoms. Respondents rate a series of statements regarding how they felt during the week prior to the interview. While the full 20-item battery was included in the 1992 (round 14) survey, a collapsed 7-item battery was administered to respondents in the 1994 (round 16) survey. A slightly expanded 9-item battery was included in the 40+ Health Module, in all years of the 50+ Health Module, in 2018 of the 60+ Health Module, and was asked of all respondents in the 2020 and 2022 interviews. The CES-D scale included in the 40+, 50+ and 60+ (initiated in 2018) Health Modules and administered to all respondents in 2020 and 2022 contained the consistent 7-item scale. Table 1 details the questions asked and the survey years and modules in which they were included.

Possible responses for all 16 negative-symptom questions are as follows:

• 0 point : Rarely or none of the time (< 1 day)
• 1 point : Some or a little of the time (1-2 days)
• 2 points: Occasionally or a moderate amount of the time (3-4 days)
• 3 points: Most or all of the time (5-7 days).

Response categories for 4 positive-symptom questions asked only in 1992 (CESD~000002A, CESD~000005A, CESD~000006A, CESD~000007B) are reversed.

Scale scores for 1992 (20-item and 7-item), 1994, 2020, 2022, and the 40+, 50+, and 60+ Health Modules (7-item) CES-D are all computed by summing the points for each answer across either the 20 items or 7 items in the specific survey year or module. The 20-item scale contains both negative- and positive-symptom questions. No positive-symptom questions are included in the 7-item scales. CES-D 20-item scale score: Points for all 20 items were summed. Possible scores range from 0 to 60, with the higher scores indicating the presence of more depressive symptoms. If more than four items are missing, the scale score is coded as missing. A flag that indicates the number of missing items for CES-D 20-item scale is also present.

CESD 7-item scale score: Points are summed across 7 items (CESD~000001, CESD~000003, CESD~000004, CESD~000005, CESD~000006, CESD~000008, CESD~000009). The possible range of scores is 0 to 21. If one item is missing, the scale score is coded as missing.

Click below to view programming code.

Q11-CESD~000001
Q11-CESD~000002
Q11-CESD~000003
Q11-CESD~000004
Q11-CESD~000005
Q11-CESD~000006
Q11-CESD~000007
Q11-CESD~000008
Q11-CESD~000009
* calculate ces_d_score_7_item;
'CESD_SCORE_7_ITEM'n = -4;
array cesd7 (*) 'Q11-CESD~000001'n 'Q11-CESD~000003'n 'Q11-CESD~000004'n 'Q11-CESD~000005'n
'Q11-CESD~000006'n 'Q11-CESD~000008'n 'Q11-CESD~000009'n;
vskipnum=0;
validnum=0;
missnum=0;
do j=1 to 7;
if cesd7(j)=-4 then vskipnum=vskipnum+1;/*valid skip*/
else if cesd7(j) in (-3,-2,-1) then missnum=missnum+1; /* missing/dk/ref */
else if cesd7(j) in (0,1,2,3) then validnum=validnum+1;/*valid answer*/
end;
if vskipnum=7 then 'CESD_SCORE_7_ITEM'n=-4; /*all 7 items are "valid skip"*/
else if missnum ge 1 then 'CESD_SCORE_7_ITEM'n=-3;
else if validnum=7 then 'CESD_SCORE_7_ITEM'n=sum(of cesd7(1), cesd7(2), 
cesd7(3), cesd7(4), cesd7(5), cesd7(6), cesd7(7));
else 'CESD_SCORE_7_ITEM'n=-3;

References

Radloff, Lenore. S. 1977. "The CES-D Scale: A self-report depression scale for research in the general population." Applied Psychological Measurement, V.1, No. 3 (Summer): 385-401.

Ross, Catherine E., and John Mirowsky. 1989. "Explaining the Social Patterns of Depression: Control and Problem Solving--or Support and Talking?" Journal of Health and Social Behavior, V.30, No. 2 (June): 206-219.

Rosenberg Self-Esteem Scale

Scoring computations are described below.

Please note: Item response theory (IRT) parameter estimates, IRT scores, and their standard errors of measurement for the Rosenberg Self-Esteem variables are also available, along with a custom-weighted z-score and percentile rank. See Appendix 27 for a detailed description and discussion of the IRT scoring for these scales.

The Rosenberg Self-Esteem Scale is designed to measure the self-evaluation that an individual makes and customarily maintains. It has been administered to NLSY79 respondents in survey years 1980, 1987 and 2006. Table 2 lists the questions by survey year.

Possible response categories for items 3, 5, 8, 9, 10 are:

• 0 points: strongly agree
• 1 point: agree
• 2 points: disagree
• 3 points: strongly disagree

Response categories for items 1, 2, 4, 6, 7 are reversed so that a higher score indicates higher self-esteem.

Points for each of the 10 items were summed to create the scale scores. Scores range from 0 to 30 points. If one item is missing, the scale score is coded as missing.

Click below to view programming code.

ROSENBERG_ESTEEM~000001
ROSENBERG_ESTEEM~000002
ROSENBERG_ESTEEM~000003
ROSENBERG_ESTEEM~000004
ROSENBERG_ESTEEM~000005
ROSENBERG_ESTEEM~000006
ROSENBERG_ESTEEM~000007
ROSENBERG_ESTEEM~000008
ROSENBERG_ESTEEM~000009
ROSENBERG_ESTEEM~000010
* calculate rosenberg_esteem_score;
'rosenberg_esteem_score'n = -4;
array rosenberg_rev (*) 'rosenberg_esteem~000001rev'n 'rosenberg_esteem~000002rev'n
‘rosenberg_esteem~000004rev'n 'rosenberg_esteem~00000rev6'n 'rosenberg_esteem~000007rev'n;
do I = 1 to 5;
if (rosenberg_rev >= 0) then rosenberg_rev = (3 – rosenberg_rev);
end;
array rosenberg (*) 'rosenberg_esteem~000001rev'n 'rosenberg_esteem~000002rev'n
‘rosenberg_esteem~000003'n 'rosenberg_esteem~000004rev'n ‘rosenberg_esteem~000005'n
'rosenberg_esteem~000006rev'n 'rosenberg_esteem~000007rev'n 'rosenberg_esteem~000008'n
'rosenberg_esteem~000009'n 'rosenberg_esteem~000010'n;
vskipnum=0;
validnum=0;
missnum=0;
do j=1 to 10;
if rosenberg(j)=-4 then vskipnum=vskipnum+1;/*valid skip*/
else if rosenberg(j) in (-3,-2,-1) then missnum=missnum+1; /* missing/dk/ref */
else if rosenberg(j) in (0,1,2,3) then validnum=validnum+1;/*valid answer*/
end;
if vskipnum=10 then 'ROSENBERG_ESTEEM_SCORE'n=-4; /*all 10 items are "valid skip"*/
else if missnum ge 1 then 'ROSENBERG_ESTEEM_SCORE'n=-3;
else if validnum=10 then 'ROSENBERG_ESTEEM_SCORE'n=sum(of rosenberg(1), rosenberg(2), rosenberg(3), rosenberg(4), rosenberg(5), rosenberg(6), rosenberg(7), rosenberg(8), rosenberg(9), rosenberg(10);
else 'ROSENBERG_ESTEEM_SCORE'n=-3;

References

Rosenberg, Morris. 1965. Society and the Adolescent Self-Image. Princeton: Princeton University Press.

Pearlin Mastery Scale

Scoring computations are described below.

Please note: Item response theory (IRT) parameter estimates, IRT scores, and their standard errors of measurement for the Pearlin Mastery variables are also available, along with a custom-weighted z-score and percentile rank. See Appendix 27 for a detailed description and discussion of the IRT scoring for this scale.

The Pearlin Mastery scale is designed to measure self-concept and references the extent to which individuals perceive themselves in control of forces that significantly impact their lives. It was administered in NLSY79 survey year 1992 (round 14). Table 3 lists the Pearlin Mastery questions.

Possible response categories for items 4, 6 are:

• 4 points: strongly agree
• 3 points: agree
• 2 points: disagree
• 1 point: strongly disagree

Response categories for items 1, 2, 3, 5, 7 are reversed so that higher scores represent greater sense of mastery.

Points for each of the 10 items were summed to create the scale scores. Scores can range from 7 to 28 points. If one item is missing, the scale score is coded as missing.

Click below to view programming code.

PEARLIN_1
PEARLIN_2
PEARLIN_3
PEARLIN_4
PEARLIN_5
PEARLIN_6
PEARLIN_7
* calculate pearlin_score;
pearlin_score = -4;
array pearlin_rev (*) pearlin_1rev pearlin_2rev pearlin_3rev pearlin_5rev pearlin_7rev;
do I = 1 to 5;
if (pearlin_rev >= 0) then pearlin_rev = (5 – pearlin_rev);
end;
array pearlin (*) pearlin_1rev pearlin_2rev pearlin_3rev pearlin_4 pearlin_5rev pearlin_6 pearlin_7rev;
vskipnum=0;
validnum=0;
missnum=0;
do j=1 to 7;
if pearlin(j)=-4 then vskipnum=vskipnum+1;/*valid skip*/
else if pearlin(j) in (-3,-2,-1) then missnum=missnum+1; /* missing/dk/ref */
else if pearlin(j) in (1,2,3,4) then validnum=validnum+1;/*valid answer*/
end;
if vskipnum=7 then PEARLIN_SCORE=-4; /*all 7 items are "valid skip"*/
else if missnum ge 1 then PEARLIN_SCORE=-3;
else if validnum=7 then PEARLIN_SCORE=sum(of pearlin(1), pearlin(2), pearlin(3), pearlin(4), pearlin(5), pearlin(6), pearlin(7);
else 'PEARLIN_SCORE'n=-3;

References

Pearlin, Leonard I.; Lieberman, Morton A.; Menaghan, Elizabeth G.; and Joseph T. Mullan. 1981. “The Stress Process.” Journal of Health and Social Behavior, V.22, No. 4 (December):337-356.

Rotter Locus of Control Scale

Scoring computations are described below.

The Rotter Locus of Control scale is designed to measure the extent to which individuals believe they have control over their lives through self-motivation or self-determination (internal control) as opposed to the extent that the environment (that is, chance, fate, luck) controls their lives (external control). It was included in the 1979, 2014, 2016 and 2018 (for those not interviewed in 2014 and 2016 respectively) NLSY79 surveys. Table 4 lists the question names and reference numbers in each year.

Points for each pair of items are as follows:

Internal control statement:

• 1 point: Much closer
• 2 points: Slightly closer

External control statement:

• 3 points: Much closer
• 4 points: Slightly closer

Points for each of the 4 pairs were summed to create scale scores. Scores range from 4 to 16 points. If one item is missing, the scale score is coded as missing.

Click below to view programming code.

ROTTER_LINT
ROTTER_1A
ROTTER_1B
ROTTER_2A
ROTTER_2B
ROTTER_3A
ROTTER_3B
ROTTER_4A
ROTTER_4B 
 
/* recode items, compute total score */
data two;
set one;
/* recode two-part Rotter items into a single item */
score1=-4;
if rotter_1a=1 and rotter_1b=1 then score1=1;
else if rotter_1a=1 and rotter_1b=2 then score1=2;
else if rotter_1a=2 and rotter_1b=2 then score1=3;
else if rotter_1a=2 and rotter_1b=1 then score1=4;
else if rotter_1a=-3 or rotter_1b=-3 then score1=-3;
else if rotter_1a=-2 or rotter_1b=-2 then score1=-2;
else if rotter_1a=-1 or rotter_1b=-1 then score1=-1;
score2=-4;
if rotter_2a=1 and rotter_2b=1 then score2=1;
else if rotter_2a=1 and rotter_2b=2 then score2=2;
else if rotter_2a=2 and rotter_2b=2 then score2=3;
else if rotter_2a=2 and rotter_2b=1 then score2=4;
else if rotter_2a=-3 or rotter_2b=-3 then score2=-3;
else if rotter_2a=-2 or rotter_2b=-2 then score2=-2;
else if rotter_2a=-1 or rotter_2b=-1 then score2=-1;
score3=-4;
if rotter_3a=1 and rotter_3b=1 then score3=1;
else if rotter_3a=1 and rotter_3b=2 then score3=2;
else if rotter_3a=2 and rotter_3b=2 then score3=3;
else if rotter_3a=2 and rotter_3b=1 then score3=4;
else if rotter_3a=-3 or rotter_3b=-3 then score3=-3;
else if rotter_3a=-2 or rotter_3b=-2 then score3=-2;
else if rotter_3a=-1 or rotter_3b=-1 then score3=-1;
score4=-4;
if rotter_4a=1 and rotter_4b=1 then score4=4;
else if rotter_4a=1 and rotter_4b=2 then score4=3;
else if rotter_4a=2 and rotter_4b=2 then score4=2;
else if rotter_4a=2 and rotter_4b=1 then score4=1;
else if rotter_4a=-3 or rotter_4b=-3 then score4=-3;
else if rotter_4a=-2 or rotter_4b=-2 then score4=-2;
else if rotter_4a=-1 or rotter_4b=-1 then score4=-1;
array score (*) score1 score2 score3 score4;
vskipnum=0;
ivskipnum=0;
dknum=0;
refnum=0;
validnum=0;
do k=1 to 4;
if score(k)=-4 then vskipnum=vskipnum+1; /*valid-skip*/
else if score(k)=-3 then ivskipnum=ivskipnum+1; /*invalid-skip*/
else if score(k)=-2 then dknum=dknum+1; /*don't know*/
else if score(k)=-1 then refnum=refnum+1; /*refused*/
else if score(k)>=1 then validnum=validnum+1; /*valid answer*/
end;
/* Per Rosella Gardecki, we should no longer code CV missing values as anything but -3 */
/* for the created variable, ROTTER_SCORE */
/* Regardless of the type of missingness, ROTTER_SCORE should be set to -3 */
if vskipnum=4 then rotter_score=-4; /*all 4 pairs are "valid skip"*/
else if ivskipnum=4 then rotter_score=-3; /*all 4 pairs are "invalid skip"*/
else if dknum=4 then rotter_score=-3; /*all 4 pairs are "don't know"*/
else if refnum=4 then rotter_score=-3; /*all 4 pairs are "refused"*/
else if validnum>0 and validnum<4 then rotter_score=-3;/*if number of valid responses < 4 then code it as -3*/
else if validnum=4 then rotter_score=sum(of score1-score4);
else rotter_score=-3;

Q11-H60DIENER~000005
Q11-H60DIENER~000004
Q11-H60DIENER~000003
Q11-H60DIENER~000002
Q11-H60DIENER~000001
array swls 'Q11-H60DIENER~000001'n 'Q11-H60DIENER~000002'n 'Q11-H60DIENER~000003'n
'Q11-H60DIENER~000004'n 'Q11-H60DIENER~000005'n;
misscount=0; * initialize count of missing values;
do i=1 to 5;
if swls[i] in (-4,-3,-2,-1) then do;
swls[i]=.; * recode missing items to system missing;
misscount=misscount+1; * count number of missing items;
end;
end;
/* only compute summary scale for those with no missing data on any items */
if misscount=0 then do;
'H60-DIENER_SWLS_SCORE'n=sum('Q11-H60DIENER~000001'n,'Q11-H60DIENER~000002'n,
'Q11-H60DIENER~000003'n, ‘'Q11-H60DIENER~000004'n,'Q11-H60DIENER~000005'n);
end;
else 'H60-DIENER_SWLS_SCORE'n=-3;

Q11-H60GAD-7~000001
Q11-H60GAD-7~000002
Q11-H60GAD-7~000003
Q11-H60GAD-7~000004
Q11-H60GAD-7~000005
Q11-H60GAD-7~000006
Q11-H60GAD-7~000007
array gad 'Q11-H60GAD-7~000001'n 'Q11-H60GAD-7~000002'n 'Q11-H60GAD-7~000003'n
'Q11-H60GAD-7~000004'n 'Q11-H60GAD-7~000005'n 'Q11-H60GAD-7~000006'n 'Q11-H60GAD-7~000007'n;
vskipnum=0;
missnum=0;
validnum=0;
do j=1 to 7;
if gad(j)=-4 then vskipnum=vskipnum+1; /*valid skip*/
else if gad(j) in (-3,-2,-1) then missnum=missnum+1;/*invalid-skip*/
else if gad(j) in (0,1,2,3) then validnum=validnum+1;/*valid answer*/
end;
/* calculate sum of 7 items just for those with no missing items */
if validnum=7 then 'H60-GAD-7_SCORE'N=sum('Q11-H60GAD-7~000001'n, 'Q11-H60GAD-7~000002'n,
'Q11-H60GAD-7~000003'n, 'Q11-H60GAD-7~000004'n, 'Q11-H60GAD-7~000005'n,
'Q11-H60GAD-7~000006'n, 'Q11-H60GAD-7~000007'n);
else if missnum ge 1 then 'H60-GAD-7_SCORE'N=-3;
else 'H60-GAD-7_SCORE'N=-3;