Cross-system analysis
Two variables, two different systems, one chart. These are the questions no single system can answer, because each system can describe itself and none can describe the relationship between them.
One question per chart. Two systems per question.
The assessment system holds checkpoint scores. The management information system holds attendance, health and family services. Classroom observations sit in a third place, keyed to the classroom rather than the child. Every finding a program can act on lives in the relationship between them, and the join is the product.
Do children with better attendance have higher literacy results?
Share meeting or exceeding literacy expectations at the spring checkpoint.
Yes, and it reads as a cliff rather than a slope. Children attending 95% or more met literacy expectations at 96%, and the bands down to 85% are barely distinguishable from them. Below 85%, the chronically absent group met at 75%, a spread of +21 pp against the top band. 121 children sit in that bottom group. The shape matters for the response: this is not an argument for pushing attendance from 92% to 95%, it is an argument for finding the children below 85%.
Strong evidenceEhrlich, Gwynne, Stitziel Pareja and Allensworth (2014), Preschool Attendance in Chicago Public Schools, UChicago Consortium on School Research; Ehrlich et al. (2018), Early Childhood Research Quarterly. Children absent 10 percent or more of enrolled pre-K days showed lower kindergarten readiness in mathematics, letter recognition, pre-literacy and behavior, and were more likely to be chronically absent in later grades.
Does a second year in the program show up in the assessment results?
The second year is where the difference shows. Children in a second year met cognitive expectations at 84% against 78% for first-year children, a difference of +7 pp. Third-year children met at 79%, which is lower than the second year rather than higher; there are only 24 of them, so read that figure as a group too small to carry a conclusion. 128 children are in a second year.
Moderate evidenceHead Start Family and Child Experiences Survey, Office of Planning, Research and Evaluation; Wen, Leow, Hahs-Vaughn, Korfmacher and Marcus (2012), Early Childhood Research Quarterly. Two years of Head Start is associated with larger school readiness gains than one, though the results are correlational and confounded by why families enroll early. Treat this as descriptive rather than causal.
Are children with untreated oral health needs missing more school?
Average attendance rate. Dots rather than bars, because the spread is narrow and a bar from zero would flatten it while a truncated bar would exaggerate it.
Yes, and the gap is 15 pp of attendance. Children diagnosed as needing treatment who did not receive it averaged 76.4% attendance, against 91.7% for those whose treatment was completed. 25 children are currently in the untreated group, and they are named on the health services action list.
Moderate evidenceJackson, Vann, Kotch, Pahel and Lee (2011), American Journal of Public Health; Seirawan, Faust and Mulligan (2012), same journal. Children with poor oral health status were significantly more likely to miss school and to have lower academic performance, and dental pain is a documented driver of absence in this age group. The direction is not settled here: untreated need may cause absence, or the same family circumstances may cause both.
Do children whose families engage more deeply score higher on social-emotional development?
A spread of +35 pp, concentrated at the low end. Children whose families completed a partnership agreement and received three or more services met social-emotional expectations at 96%; children whose families have no agreement on file met at 61%. 98 children are in that last group.
Moderate evidenceHead Start Family and Child Experiences Survey; Mendez (2010), Journal of Applied Developmental Psychology; Sheridan et al. (2010). Parent engagement is associated with stronger social-emotional and behavioral outcomes, with the strongest experimental evidence coming from structured parent programs rather than service counts. Selection is a real concern here: families able to engage may differ systematically from families who cannot.
Do classrooms with higher Instructional Support produce larger language gains?
Percentage-point gain in the share meeting language expectations, fall to spring.
Yes, and the gradient runs the whole length of the scale. Classrooms scoring 3.5 and above averaged a +26 pp language gain against +4 pp below 2.5, with the middle band in between. The 2 classrooms in the lowest band hold 25 children. Classroom-level counts are small enough that one room moves a band, so this points at which rooms to observe next rather than settling how much of the gain the classroom caused.
Moderate evidenceBurchinal, Vandergrift, Pianta and Mashburn (2010), Early Childhood Research Quarterly; Mashburn et al. (2008), Child Development; Hamre et al. (2014). Associations between Instructional Support and language and literacy gains are consistently positive but modest, and several studies find the effect concentrated above a quality threshold rather than increasing steadily. The classroom-level count here is 32, small enough that one outlier room moves the average, so treat this as directional.
How do children experiencing homelessness differ on attendance and literacy?
The attendance gap is the larger of the two. Children in families that experienced homelessness averaged 82.4% attendance against 88.8%, a difference of 6.4 pp. On spring literacy the two groups are closer: 88% against 91%. 43 children are in the group, small enough that the literacy figure moves several points on a handful of children.
Moderate evidenceCutuli, Desjardins, Herbers, Long, Heistad, Chan, Hinz and Masten (2013), Child Development; Fantuzzo, LeBoeuf, Chen, Rouse and Culhane (2012), Children and Youth Services Review. Homeless and highly mobile children show lower achievement and higher absenteeism than low-income peers who are stably housed, with residential mobility a consistent independent predictor.
Are dual language learners falling behind, or growing faster?
Read this way, dual language learners trail by 9 pp and the finding reads as an achievement gap.
Read this way, dual language learners gained +43 pp against +30 pp: they started further back and moved faster.
The same data supports opposite conclusions depending on the framing. A program that reports only status will write an intervention plan for a problem it does not have, and will miss the children who genuinely are not progressing. 171 children in this program are dual language learners.
MethodologicalOffice of Head Start guidance on assessing dual language learners; Halle, Hair, Wandner, McNamara and Chien (2012), Early Childhood Research Quarterly. Guidance directs programs to assess these children in the home language where possible and to interpret English-medium results as a measure of English acquisition rather than of underlying development. Growth-based interpretation is the standard for this population, and status-only comparison systematically misattributes language acquisition to developmental delay.
What this page does and does not establish
These are associations, not causes. Every comparison on this page is descriptive. Children with low attendance differ from children with high attendance in many ways beyond attendance, and none of these charts controls for those differences. They are built to tell a program where to look, not to establish what caused what.
Group sizes and suppression. Any group below ten children is suppressed and not charted. Group counts are printed on every axis label. Classroom-level comparisons rest on 32 classrooms, small enough that a single outlier room moves the average.
The direction is often ambiguous. Untreated dental need may cause absence, or the same family circumstances may cause both. Family engagement may improve child outcomes, or families of children who are doing well may find engagement easier. Where the direction matters for a decision it is stated as an open question rather than resolved in the chart.
Why no single system produces this page. Each chart joins two or three systems on child ID. The assessment system does not hold attendance. The management information system does not hold checkpoint scores in an analyzable form. Observations live in a third system keyed to the classroom rather than the child.
Every chart on this page reads the 436 children with a finalized checkpoint at all three points, except the attendance and oral health comparisons, which read every child with attendance recorded.