- Neighborhood Clusters
- Longitudinal Cohort Study
- Infant Assessment Unit
- Community Survey
- Systematic Social Observation
The PHDCN Scientific Directors defined neighborhoods spatially, as a collection of people and institutions occupying a subsection of a larger community. The project collapsed 847 census tracts in the city of Chicago to form 343 neighborhood clusters (NCs). The predominant guideline in formation of the NCs was that they should be as ecologically meaningful as possible, composed of geographically contiguous census tracts, and internally homogenous on key census indicators. The project settled on an ecological unit of about 8,000 people, which is smaller than the 77 established community areas in Chicago (of which the average size is almost 40,000 people), but large enough to approximate local neighborhoods. Geographic boundaries (e.g., railroad tracks, parks, and freeways) and knowledge of Chicago's neighborhoods guided this process.
Composition of Neighborhood Clusters
|Racial/ethnic stratum||Socioeconomic Status|
|≥20% Latino and ≥ 20% White||6||40||12|
|≥20% Latino and ≥ 20% Black||9||4||0|
|≥20% Black and ≥ 20% White||2||4||11|
|NCs not classified above||8||15||12|
Longitudinal Cohort Study
As part of the longitudinal cohort study, 800-900 participants in each of seven age groups were sampled from households in 80 of the 343 NCs.
The 80 NCs were sampled from the 21 strata defined in the table above with the aim of representing these 21 cells as nearly equally as possible to eliminate the confounding between ethnic mix and SES.
PHDCN used a three-stage sampling design. At the first stage, 343 neighborhoods, containing all residents of Chicago, were cross-classified by two census-derived stratification variables--racial-ethnic mix (seven levels) and socioeconomic status (three levels). A stratified probability sample of 80 neighborhoods was selected for inclusion in the study. Next, block groups were selected at random within each of the sample neighborhoods and a complete listing of dwelling units collected for sampled block groups. Finally, residents were contacted and the household composition enumerated. Children within 6 months of birth, ages 3, 6, 9, 12, 15, and 18 were selected for longitudinal study.
|Racial/ethnic stratum||Socioeconomic Status|
|> 75% Black||9||4||4|
|> 75% White||0||4||8|
|> 75% Latino||4||4||0|
|≥ 20% Latino and ≥ 20% White||4||5||4|
|≥ 20% Latino and ≥ 20% Black||4||4||0|
|≥ 20% Black and ≥ 20% White||2||4||4|
|NCs not classified above||4||4||4|
Infant Assessment Unit
The participants in the Infant Assessment Unit were drawn from the youngest project cohort included in the Longitudinal Cohort Study. The youngest cohort of study participants numbered 1,261 children who were sampled at some point between the time their mothers were pregnant with them and their first birthday. Approximately 412 infants and their primary caregivers from the 1,261 cohort 0 participants were assessed for the Infant Assessment Unit when the infants were between 5-7 months of age.
The 343 NCs contain all of the dwelling units in Chicago and every NC was represented in the sample. Thus, the NCs are "strata" rather than sampling units for purposes of the Community Survey. The basic design for the Community Survey had three stages:
- City blocks were sampled within each NC
- Dwelling units were sampled within blocks
- One adult resident was sampled within each selected dwelling unit.
Although this three-stage design within each stratum is straightforward, the sample size and method of sampling differed depending upon whether an NC had been selected for the project's Longitudinal Cohort Study (LCS): 80 NCs were selected for the LCS and the remaining 263 were not selected for the LCS.
NCs Not Selected for the Longitudinal Cohort Study
The target sample size for each NC that was not selected for the LCS was 20. The sampling plan called for selection of nine blocks within an NC, three dwelling units within a block, and one resident within a dwelling unit, potentially producing 9 x 3 = 27 interviews. Given a dwelling unit occupancy rate of .90 and a response rate of .85, one would expect about 20 completed interviews for each of the NCs not selected for the LCS.
Blocks were selected with probability proportional to size; dwelling units within blocks were selected at random; and persons within dwelling units were also selected at random. The tasks at each stage are described below.
Selecting blocks: The method of selection was a systematic random sample with probability proportional to the number of dwelling units in the block. To achieve this, blocks were listed in block number order. Associated with each block was the number of dwelling units in that block as given in the 1990 Census, along with the cumulative total number of dwelling units. Nine equal intervals between a and N + a were computed, where N is the number of dwelling units in the NC and a is a random number between 0 and N19. Associated with the endpoint of each interval was a particular block, and that block was selected into the sample. (If a block contained more than N/9 dwelling units, that block was selected with certainty and removed from the list. A new interval was then constructed having width N18 and a systematic random sample of size 8 was drawn. Also, blocks with fewer than three dwelling units were treated as if they had three dwelling units.)
Selecting dwelling blocks: All of the dwelling units within a selected block were listed and, from the list, three were selected at random.
Selecting respondents: Within each selected dwelling unit, a list of persons 18 years old and older was obtained, and one person from that list was selected at random for the interview.
NCs Selected for the Longitudinal Cohort Study
For the 80 NCs selected for the LCS, the target sample size was 50. The Community Survey sampling plan for these NCs built upon work already done in selecting blocks for the LCS. Within these blocks, a systematic random sample of, on average, 65.4 dwelling units, was selected with one respondent sampled per dwelling unit. Again, given a .90 dwelling-unit occupancy rate and a response rate of .85, this sample was expected to yield about 50 completed interviews per sample NC.
Selecting and listing of blocks: These blocks constituted a simple random sample of all blocks in the NC.
Selecting dwelling units: A systematic random sample of dwelling units was drawn within the set of listed blocks in each sample NC. First, dwelling units were sorted by census tract, by block group within census tract, by block within block group, and by address within block. Equal intervals of length K165.4 between β and K + β were computed, where K is the number of addresses in the NC and β is a random number between 0 and K/65.4. Associated with the endpoint of each interval was a particular address, and that address was selected into the sample.
Selecting respondents: Just as in the case of nonsample NCs, a list of persons 18 years old and older was obtained within each selected dwelling unit, and one person from that list selected at random for the interview.
Systematic Social Observation
After the 80 NCs were selected for the Longitudinal Cohort Study, the National Opinion Research Center (NORC) prioritized block faces by selecting random neighborhoods by stratum. Three neighborhoods were regarded as a unit of work. The priority list was followed essentially in order, although some areas were switched for logistic reasons or because some neighborhoods, deemed more dangerous or traffic-dense, were moved to Sunday morning taping slots.
For the 27,734 block faces identified, the following information was gathered:
- 22,418 block faces had complete observation logs and videotapes
- 1,395 block faces had incomplete observation logs and complete videotapes
- 2 block faces had incomplete observation logs and no videotapes
- 1 block face had a complete observation log, but no videotape
- 3,918 block faces did not exist. (These block faces were determined not to be block faces once out in the neighborhood. For example, new construction eliminated some block faces.)
A random subsample of 64% of the videotaped block faces was chosen to be coded. In those NCs consisting of 150 or fewer block faces, all block faces were coded. In the remaining block faces, sample sizes were calculated to approximate a balanced design as closely possible in order to maximize statistical power for comparisons of NCs. A total of 15,141 block faces were actually coded for an average of 189 block faces per NC. Thirty of the coded block faces had coding records but no observation logs and were thus excluded when the coding records and observation logs records were merged.