National Education Longitudinal Study of 1988 (NELS:88)

Data from the National Education Longitudinal Study of 1988 (NELS:88), a nationally representative longitudinal survey that followed 8th-grade students through high school and beyond. This subsample of 2,000 students contains 299 variables spanning base-year (1988) and first follow-up (1990) waves, including student demographics, family background, school characteristics, attitudes, behaviors, and standardized test scores. Used extensively in regression teaching for model building, handling large variable sets, and longitudinal analysis of educational achievement.

Usage

nels_data

Format

A tibble with 2,000 rows and 299 columns. Key variable groups:

Identifiers and demographics:

id

Student identifier. Type: numeric. Range: (124966, 9199195).

bys2a

Student sex. Type: numeric. Values: 1 = male, 2 = female. Code 8 = missing/refused.

birthmo

Birth month. Type: numeric. Range: (1, 12). Code 98 = missing.

birthyr

Birth year (last two digits). Type: numeric. Range: (72, 76). Code 98 = missing.

Parent occupation and education (base year):

bys4a

Father's education level. Type: numeric. Range: (1, 7). Codes 8, 9 = don't know/missing.

bys4occ

Father's occupation code. Type: numeric. Range: (1, 18). Code 99 = not applicable.

bys5a

Mother's education. Type: numeric. Range: (1, 7). Codes 8, 9 = don't know/missing.

bys7a

Mother's occupation. Type: numeric. Range: (1, 7). Codes 8, 9 = don't know/missing.

bys7occ

Mother's occupation code. Type: numeric. Range: (1, 18). Code 99 = not applicable.

Family structure and home environment:

bys12

Family composition. Type: numeric. Values: 1 = two parents, 2 = other. Code 8 = missing.

bys18

Number of siblings. Type: numeric. Code 99 = missing.

bys20

Times changed schools. Type: numeric. Code 98 = missing.

bys22

Home language. Type: numeric. Code 99 = missing.

Parental involvement items (bys27a through bys27d): Rated on a scale where lower values indicate more involvement. Code 9 = missing.

Student attitudes and self-concept items:

bys29

Educational expectations. Type: numeric. Code 9 = missing.

Television and homework habits (bys31a through bys31d): Hours spent on various activities. Codes 8, 9, 99 = missing.

Extracurricular activities: bys34a, bys34b (participation codes, 98 = missing).

School-related items (bys35a through bys39c): Various school experience and behavior items rated on Likert-type scales. Code 8 = missing.

Time use items: bys41 (hours working, code 8 = missing), bys42a, bys42b (hours on activities, code 98 = missing), bys43 (absences, code 8 = missing).

School behavior items (bys44a through bys44m): Items measuring academic and social behaviors, rated on Likert-type scales. Code 8 = missing.

Additional student items: bys45 through bys46 (educational plans), bys52 through bys54 (peer influences, codes 8/98 = missing), bys55a through bys55f (risk behaviors, code 8 = missing), bys56a through bys56e (parenting practices), bys57a through bys57c (problem behaviors).

Locus of control and self-concept items (bys58a through bys60d): Likert-type attitude items rated 1-4, with code 8 = missing.

Teacher report items (bys66a through bys68b): Teacher assessments of student behavior. Code 8 = missing.

Drug/alcohol items (bys74 through bys77): Risk behavior items. Code 8 = missing.

Behavioral ratings (bys80): Behavioral scale. Range: (0, 5). 90 NA.

Course-taking patterns (bys81a through bys83j): Items on courses taken and academic track. Codes 8, 98 = missing.

Composite and derived variables:

bys82a through bys82u

Self-concept items. Type: numeric. Rated 1-4. Code 8 = missing. 21 items.

bys83a through bys83j

Locus of control items. Type: numeric. Rated 1-4. Code 8 = missing. 10 items.

byscenrl

School enrollment size category. Type: numeric. Range: (1, 7).

g8urban

Urbanicity. Type: numeric. Values: 1 = urban, 2 = suburban, 3 = rural.

g8regon

Census region. Type: numeric. Range: (1, 4). Code 8 = missing.

g8minor

Percent minority enrollment in school. Type: numeric. Code 998 = missing.

g8lunch

Percent eligible for free lunch. Type: numeric. Code 998 = missing.

Psychometric composite scores (base year):

bylocus1

Locus of control, IRT-based. Type: numeric. Code 99.98 = missing.

bylocus2

Locus of control, standardized. Type: numeric. Code 99.98 = missing.

bycncpt1

Self-concept, IRT-based. Type: numeric. Code 99.98 = missing.

bycncpt2

Self-concept, standardized. Type: numeric. Code 99.98 = missing.

byses

Socioeconomic status composite. Type: numeric. Range: (-2.89, 1.98). Code 100 = missing. Standardized composite of parental education, occupation, and family income.

bypared

Parental education (highest of either parent). Type: numeric. Range: (1, 7).

byfamsiz

Family size. Type: numeric. Code 98 = missing.

byfcomp

Family composition. Type: numeric. Code 98 = missing.

byparmar

Parental marital status. Type: numeric. Code 98 = missing.

byfaminc

Family income category. Type: numeric. Range: (1, 15). Code 98 = missing.

byhmlang

Home language. Type: numeric. Code 8 = missing.

bypsepln

Post-secondary education plans. Type: numeric. Code 98 = missing.

byhomewk

Hours of homework per week. Type: numeric. Code 98 = missing.

bylep

Limited English proficiency. Type: numeric. Binary (0/1). Code 8 = missing.

bylm

Language minority. Type: numeric. Binary (0/1). Code 8 = missing.

bygrads

GPA composite. Type: numeric. Range: (0.5, 9.8).

byrisk

Number of risk factors. Type: numeric. Range: (0, 5).

Base-year standardized test scores:

by2xrstd

Reading standardized score. Type: numeric. Range: (31.95, 70.55). 77 NA. Mean approximately 51.

by2xmstd

Mathematics standardized score. Type: numeric. Range: (34.10, 77.20). 75 NA. Mean approximately 51.

by2xsstd

Science standardized score. Type: numeric. Code 99.99 = missing.

by2xhstd

History standardized score. Type: numeric. Code 99.99 = missing.

by2xrpro

Reading proficiency level. Type: numeric. Range: (0, 4). Code 9 = missing.

by2xmpro

Math proficiency level. Type: numeric. Range: (0, 5). Code 9 = missing.

by2xspro

Science proficiency level. Type: numeric. Range: (0, 3). Code 9 = missing.

First follow-up (1990) student survey items (f1s prefix): Items f1s8a through f1s99s cover attitudes, school experiences, behavior, and educational plans at the first follow-up wave (10th grade). Most items are rated on Likert-type scales (1-4 or 1-6). Code 8 = missing/not applicable. Approximately 186 NA values per item for students lost to follow-up.

First follow-up derived variables:

f1stat

Follow-up status. Type: numeric. Range: (0, 5).

f1dostat

Dropout status. Type: numeric. Range: (0, 5).

f1ses

SES at follow-up. Type: numeric. Code 100 = missing.

f1locus1

Locus of control, follow-up. Type: numeric. Code 99.98 = missing.

f1locus2

Locus of control standardized, follow-up. Code 99.98 = missing.

f1cncpt1

Self-concept, follow-up. Code 99.98 = missing.

f1cncpt2

Self-concept standardized, follow-up. Code 99.98 = missing.

f1drps89

Dropout between base year and 1989. Type: numeric. Code 8 = missing.

f1drpf89

Dropout by fall 1989. Code 8 = missing.

f1drps90

Dropout by spring 1990. Code 8 = missing.

f1hsprog

High school program. Type: numeric. Code 8 = missing.

f12xrstd

Reading score at follow-up. Type: numeric. Code 99.99 = missing. Mean approximately 57.

year2

Derived time variable 2. Type: numeric.

year1

Derived time variable 1. Type: numeric.

Source

National Center for Education Statistics (NCES). National Education Longitudinal Study of 1988 (NELS:88). U.S. Department of Education. Original data file: nels_data.dta

Details

This dataset is used across multiple chapters for model building exercises, particularly in Chapter 7 (Model Building) for variable selection and in general regression contexts for examining predictors of educational achievement. The large number of variables allows practice in identifying relevant predictors from a rich set of candidates. Note that many variables use special missing value codes (8, 9, 98, 99, 99.98, 99.99, 100, 998) rather than NA, which requires careful data cleaning before analysis. Key analyses include: predicting math achievement from SES and school characteristics, stepwise and purposeful model building, and examining longitudinal change in test scores from base year to follow-up.

Examples

data(nels_data)
# Regression of math score on SES (after cleaning missing codes)
nels_clean <- nels_data[nels_data$byses < 90, ]
lm(by2xmstd ~ byses, data = nels_clean)
#> 
#> Call:
#> lm(formula = by2xmstd ~ byses, data = nels_clean)
#> 
#> Coefficients:
#> (Intercept)        byses  
#>      51.821        6.423  
#>