Table 4. Regression Adjusted Means for Average Sample Child, by Participation in P-3 Program Components.

	Reading	Math	Teacher Report Reading	Teacher Report Math	Teacher Report Approaches to Learning	Held Back	Special Education
				Full Sample
Group 1 (No P-3 Components)	49.08	48.88	−.92	−.98	−1.56	.15	.08
Group 2 (Pre + FullK + School Stability K-3)	49.90	50.59	.85	.59	−.95	.07	.07
Group 3 (Group 2 + High Involvement + High Instruction + Teacher Certification)	52.12	52.53	1.28	1.33	.63	.04	.05
				Disadvantaged Sample
Group 1 (No P-3 Components)	43.88	44.20	−4.73	−4.32	−3.68	.22	.11
Group 2 (Pre + FullK + School Stability K-3)	44.76	46.03	−2.19	−1.51	−3.02	.11	.07
Group 3 (Group 2 + High Involvement + High Instruction + Teacher Certification)	47.55	48.38	−1.74	−1.08	−2.01	.09	.07

Notes:

¹Sample includes only first-time kindergarten public school children. Full Sample size for this study is 6,761. Disadvantage is defined as having a mother or father with less than a high school diploma or living in poverty in kindergarten fall. Disadvantaged sample size is 2,013. The means represent the score for the average child in the described group holding constant: household income, race, parental education, family structure, family size, non-English language spoken in the home, child’s height and weight, region of the country, urbanicity, and child’s race. Reading and Math test scores have means of 50 and standard deviation of 10. Teacher reports of reading, math and approaches to learning have full sample means of 0 and standard deviations of 10. Held Back and Special education are dichotomous measures (1=yes, 0=no).

²Preschool in these estimates does not include Head Start participation, as it is likely that our selected set of covariates does not adequately adjust for the very disadvantaged characteristics of Head Start attendees. We do not include experiencing a small class size because of concerns that children that attended small public school classes differed from those who experienced larger classes in ways we were not able to adjust for with the included covariates. We use median splits to define both high levels of high levels of reading instruction. Teachers’ reports of time spent on reading and language work and projects were averaged over the three waves of data (kindergarten, first grade, and third grade). More than an average of 60-90 minutes per day of reading and language activities is considered high instruction. Having parents who indicated they were involved in their child’s school along 3 or more of 6 dimensions (attending a parent-teacher conference, open house, parent advisory group, Parent-Teacher association meeting, participating in a school fundraiser or volunteering at the school) is considered a high level of parent involvement.