Early Childhood Research Quarterly 27 (2012) 90 103

Contents lists available at ScienceDirect

Early Childhood Research Quarterly

School indThe me

Claudia Ga Language, Lit emic Ib Center on Sch

a r t i c l

Article history:Received 20 FeReceived in reAccepted 29 M

Keywords:School and hoEarly childhooAchievement g

ost ins papuencstudedentredictmathents rten.

One way to ensure that all students succeed and graduate fromhigh school is to ensure that they get off to a strong start. Achilds experiences in the early years of schooling may set her orhim on a le(Rimm-Kauthat childrelater achievSameroff, &students inacademic ex& McLanahunderstandrience acad

The twolearning aning to Brondevelopmetext, as welparticipatioin each sethome and schildrens ddegree to w

CorresponE-mail add

relationships with one another helps determine childrens aca-demic success. She argues that schools implementing practices thatpromote strong school, family, and community partnerships should

0885-2006/$ doi:10.1016/j.arning trajectory that affects an entire school careerfman & Pianta, 2001). Studies indicate, for example,ns performance in kindergarten is predictive of theirement (Alexander, Entwisle, & Dauber, 1993; Gutman,

Cole, 2003). Also, children construct their identity as the early years of schooling, which impacts their laterperiences (Farkas & Beron, 2004; Rouse, Brooks-Gunn,

an, 2005). It is vital, then, that researchers identify and the inuences and support that help all students expe-emic success at the beginning of their school careers.

most inuential contexts in which young childrensd development occur are home and school. Accord-fenbrenner and Morris (1998), childrens behavior andnt are inuenced by their interactions within each con-l as the connections between settings, including jointn, communication, and the existence of informationting about the other. Epstein (2001) argues that thechool constitute overlapping spheres of inuence onevelopment and academic achievement, and that thehich educators and family members maintain positive

ding author.ress: galindo@umbc.edu (C. Galindo).

be better able to help children succeed academically because theseoutreach activities create greater consistency between childrenshome and school contexts (Epstein, 2001).

Decades of studies, reviews, and syntheses conrm this and haveconcluded that parents and family members are powerful inu-ences on student achievement across grades (Epstein & Sheldon,2006; Henderson & Mapp, 2002; Jeynes, 2005; Snow, Burns, &Grifn, 1998; Wang, Haertel, & Walberg, 1993). Not only does eachcontext uniquely inuence child development and learning, butthe nature and quality of interactions among parents, teachers,and children have consequences for a variety of outcomes (Epstein,2001).

We drew from ecological and sociological theories, as wellas prior research on family involvement, to guide our analysesof the Early Childhood Longitudinal Study-Kindergarten Cohort(ECLS-K) database estimating the extent to which schools out-reach to families was associated with three indicators of familyinvolvement involvement at home, involvement at school, andparents educational expectations. We then estimated whetherfamily involvement was related to students math and reading gainsbetween the fall and spring of kindergarten. Finally, we investigatedwhether school outreach efforts were associated with childrensachievement gains and whether family involvement was a mediat-ing mechanism by which school efforts to engage families translate

see front matter 2011 Elsevier Inc. All rights reserved.ecresq.2011.05.004 and home connections and childrens kdiating role of family involvement

alindoa,, Steven B. Sheldonb

eracy, and Culture, University of Maryland, Baltimore County, 1000 Hilltop Circle, Acadool, Family, and Community Partnerships, Johns Hopkins University, United States

e i n f o

bruary 2010vised form 19 May 2011ay 2011

me connectionsdains

a b s t r a c t

Childrens home and school are the mespecially during early childhood. Thitheory of overlapping spheres of intionships to family involvement and sample of kindergartners (16,425 stumunicate with and engage families pstudent achievement in reading and family involvement at school and paring and math achievement in kindergaresearchers, and practitioners.ergarten achievement gains:

V (A Wing), Room 403, Baltimore, MD 21250, United States

uential contexts in which learning and development occur,er builds on Bronfenbrenners ecological theory and Epsteinse to examine school and family connections and their rela-nts achievement gains. We used a nationally representatives from 864 schools) and found that schools efforts to com-ed greater family involvement in school and higher levels of

at the end of kindergarten. We also found that, on average,educational expectations were associated with gains in read-

We discuss the implications of our ndings for policy makers,

2011 Elsevier Inc. All rights reserved.

C. Galindo, S.B. Sheldon / Early Childhood Research Quarterly 27 (2012) 90 103 91

into childrens academic gains. Together, these analyses tested ourhypotheses that school outreach relates to gains in achievement inkindergarten by getting their parents more engaged.

1. Family iachieveme

This stument, condenitions family and volunteerinwith the coApostolerisof involvemdistinguishfocusing oninteractionsWalker, Hoo2005; McW1998). All ofamily invonot use omnunderstandment.

For this at home, fato test wheto kindergaconceptualidistinguishparenting btinctions arlarge natioshown thating school eare associatior, and att1999; Mulle

Studies ointeractionscognitive d(2004) founchildrens snicant bodwith more ents tend tachievemenments (FairMcWayne, 22002).

Althouginvolvemenor achieveminteractionsstronger nu2009). We time at homless importKnabe, 199& Doucet, 2instruction to be a fam2005). Famiin which ou

performances in school, promoting reading and math achievementdifferently.

Studies have also shown that family involvement at school canhave a positive inuence on young childrens education and cog-

devethat,, famgartece iily intaryce igituded atition001

as as anhool, Cox

of chthat t

pred, Lun, 19s (FaLin, 2

ool oes

earch invo

eduto end, 2005er line wiol annallpect

of m, Coxorteds tre m

yearing tudyes prer sthoolsive ahool

Alsoith aanceent Sheldn import

stud studleagunvolvement theory and childrens academicnt

dy adopted a multidimensional denition of involve-sistent with current research and theory. Theseinclude Epsteins (2001), framework of six types ofcommunity involvement (parenting, communication,g, learning at home, decision making, and collaboratingmmunity) and Grolnicks (Grolnick, Benjet, Kurowski, &, 1997; Grolnick & Slowiaczek, 1994) three dimensionsent (cognitive, affective, and school-based). Others

between family involvement at home and at school, the location in which parentteacher or parentchild

occur (e.g., Fantuzzo, Tighe, & Childs, 2000; Green,ver-Dempsey, & Sandler, 2007; Hoover-Dempsey et al.,ayne, Campos, & Owsianik, 2008; Zellman & Waterman,f these conceptualizations share the understanding thatlvement is multidimensional, and that researchers can-ibus measures or any single type of involvement to fully

how families shape childrens education and develop-

study, we distinguished between family involvementmily involvement at school, and parental expectationsther each type of engagement was differentially relatedrteners gains in reading and math achievement. Thiszation is most closely aligned with studies that haveed forms of involvement based on the locale of theehaviors (e.g., Hoover-Dempsey et al., 2005). The dis-e also consistent with other studies that have usednal databases focused on older students which have

family practices such as PTA/PTO membership, attend-vents, monitoring behavior, and homework completioned with students higher academic achievement, behav-endance (Domina, 2005; Keith et al., 1998; McNeal,r, 1993; Simon, 2001).f young children have demonstrated that parentchild

at the home are a strong inuence on childrensevelopment. Fantuzzo, McWayne, Perry, and Childsd that family involvement at home was associated withchool readiness competencies. On reading, there is a sig-y of literature demonstrating that children from homesbooks and who experience more reading with par-

o perform higher on literacy assessments and readingt tests than do children from less reading-rich environ-

es, Nichols, & Rickelman, 2000; Ginsburg-Block, Manz, &009; Scarborough & Dobrich, 1994; Snchal & LeFevre,

h less is known about whether and how familyt affects young childrens math skill developmentent, studies are emerging that suggest parentchild

with board and counting games are associated withmeracy skills (Ginsburg-Block et al., 2009; LeFevre et al.,also know that young children generally spend lesse in math activities, and that parents view math as

ant, compared to reading (Musun-Miller & Blevins-8; Sonnenschein, Baker, Moyer, & LeFevre, 2005; Tudge004). Additionally, parents may conceptualize math

as teachers responsibility, whereas reading is perceivedilyteacher shared responsibility (Sonnenschein et al.,ly involvement at home, then, may be an important waytside-of-school experiences affect childrens academic

nitive found gartenkinderformanin famelemenformanthe loninvolv

AddChen, 2tationsserve awith scPiantaaspectfound futuresEnglun& OlsogroundYan &

2. Schoutcom

Resfamilychildseffort Bertranet al., 2strongnicativat schoAdditiotant assurveyPiantaity repfamiliethat thschoolas sendTheir spractic

Othand sca positand sc2010).cate wattendof stud2007; strate aand im

Theily andhis collopment. Dearing, Kreider, Simpkins, and Weiss (2006) after controlling for students achievement in kinder-ilies more involved at their childrens school fromn to fth grade had children with higher literacy per-n fth grade. Moreover, they found that an increasevolvement at school from kindergarten to the upper

grades was associated with an increase in literacy per-n fth grade. Their study provides strong evidence ofinal benets for young children when their parents are

the school.ally, consistent with prior research on this topic (Fan &; Suizzo & Stapleton, 2007), we included parental expec-n aspect of family involvement. Parental expectations

indicator of the family norms and values associateding that pervade childrens family life (Rimm-Kaufman,, & Bradley, 2003), making expectations a salientildrens home environments. Studies have consistentlyhese parental beliefs about their childrens educationaldict student achievement (Davis-Kean & Sexton, 2009;ckner, Whaley, & Egeland, 2004; Entwisle, Alexander,97), across families with diverse racial and ethnic back-n & Chen, 2001; Goyette & Xie, 1999; Hong & Ho, 2005;005).

utreach to families and students and families

ers have shown that schools are critical inuences onlvement. Parents are more likely to be involved in theircation when the school or teachers make a strongergage them in their childrens learning (Deslandes &

005; Epstein, 2001; Green et al., 2007; Hoover-Dempsey; Sheldon & Van Voorhis, 2004). When educators fosterks with families by being responsive to and commu-th parents, families are expected to feel more welcomed become more involved in their childrens education.y, outreach to, and engagement of, families are impor-s of early childhood educators professional roles. In aore than 3000 kindergarten teachers across the nation,, Taylor, and Early (1999) found that the vast major-

using at least some practices to facilitate children andansition into school. The researchers noted, however,ost frequent practices tended to take place after the

began and were low-intensity generic contacts, suchhome brochures, iers, or invitations to school events., however, did not examine the degree to which theseedicted parental behaviors or student outcomes.udies show that the close relationships between parents

are important for student success. Research indicatesssociation between school outreach to involve familieslevels of math prociency (Sheldon, Epstein, & Galindo,, studies have shown that school efforts to communi-nd engage families is related to higher levels of student, lower levels of chronic absenteeism, and lower levelsbehavior problems (Epstein & Sheldon, 2002; Sheldon,on & Epstein, 2002). Together, these studies demon-portant connection between school outreach to familiesant student outcomes.ies that examine school outreach as a predictor of fam-ent outcomes have important limitations. Sheldon andes (Sheldon, 2007; Sheldon & Epstein, 2005; Sheldon

92 C. Galindo, S.B. Sheldon / Early Childhood Research Quarterly 27 (2012) 90 103

et al., 2010), for example, relied on measures of school practicesas predictors of student outcomes (i.e., attendance, achievement,disciplinary incidents), but aggregated to the school level. As aresult, the associations found cannot be taken as evidence thatschool outrvidual famiinvestigatinhave reliedtors of famDeslandes &limited becily involvemfamily memfamily invoprovide impily involvemand parenta

In particfamilies, anDodge (200Study Kinschools thasuccessful thigher on aover, they dparents invfamilies wit

The studin several imof transitionthe effects oilies in theiactivities, thnication anthroughoutonly on theschool and ates the effaccounting the effects have been a

The preresearch in to which scdictive of frelationshipthe school ygarten mayteachers to families. Wfall and sprderived fromThird, we ily involvemschool outrestudent achlimitations conducted practices toto student a

3. Researc

This studpractices de

on parental behaviors and students academic gains. Three researchquestions and hypotheses guided this study:

1) To what extent is school outreach to families associated withily inentsothe

witpsteen th

reso encvitieool oontrool oal exm-Ke les(Golectedhat

h anarchily inl., 20t at ngerlvemitioneens. M

schoeen

read7).hat

drenshipiatechoopsteiitive s, inage fand mnd

studinvolectateen

thod

datK) Kal Cens c

fromistagresechoomplinion Sindeeach to involve families is related to the actions of indi-lies or outcomes of individual students. Other studiesg the effect of school outreach on students and families

on parent perceptions of school outreach as predic-ily involvement behaviors (Anderson & Minke, 2007;

Bertrand, 2005; Green et al., 2007). These studies areause connections between school outreach and fam-ent may be over-estimated as a result of the fact thatbers are reporting on school outreach and their ownlvement at the same time. Although the prior studiesortant clues that school outreach is predictive of fam-ent, additional research that measures school outreachl involvement independently is needed.ular, research is needed that collects data about schools,d children. One such study by Schulting, Malone, and5), used the national Early Childhood Longitudinaldergarten Cohort Database (ECLS-K), and showed thatt implemented more activities to help families make aransition into kindergarten had students who scoredchievement tests by the spring of kindergarten. More-emonstrated that this effect was partially mediated byolvement at school, and that the effect was stronger forh lower-incomes.y by Schulting and her colleagues, however, was limitedportant ways. First, by focusing on the implementation

activities for families, their study could not account forf school and teachers on-going efforts to engage fam-r childrens education. Effects attributed to transitionerefore, may be an artifact of the home-school commu-

d family involvement practices that schools implement the school year. Also, Schulting et al. (2005) focused

effects of school transition practices on involvement inthe degree to which this one type of involvement medi-ect of school outreach on student achievement. By notfor the multidimensional nature of family involvement,of other types of family involvement on students mayttributed to involvement at school.sent study builds upon existing family involvementa number of ways. First, this study focused on the extenthool outreach to families during the school year was pre-amily involvement and student outcomes. A familys

with the school is dynamic and evolves throughoutear. Childrens adjustment and performance in kinder-

be inuenced by the on-going efforts of schools andcommunicate and maintain positive relationships withe also focused on students academic gains between theing of kindergarten to improve upon previous results

cross-sectional data (See Henderson & Mapp, 2002).relied on independently collected measures of fam-

ent and school outreach, using principal reports ofach and parent reports of family involvement to predictievement through kindergarten. By addressing thesein prior research on family involvement, the analyseslled important gaps in our understanding of how school

engage families in their childrens education are relatedchievement through kindergarten.

h questions and hypotheses

y built on previous research to analyze the inuences ofsigned to engage families in their childrens education

famparhypated& EGivandmayactischIn cschtion(Rimto bing exp

2) To wmatresefamet amenstroinvoAddbetwgainandbetwfor 200

3) To wchiltionmedof s& Eposgainenging to andily expbetw

4. Me

The(ECLS-Nationchildremationa multally rep1000 sthe saEducatfrom kvolvement (family involvement at home, in school, and educational expectations)? Based on prior research, wesized that school outreach would be positively associ-h family involvement at home and at school (Dauberin, 1993; Green et al., 2007; Schulting et al., 2005).at school outreach may provide parents informationurces about educational activities to use at home orourage parents to participate more actively in schools, we expected to observe a positive association betweenutreach and family involvement at school and at home.ast, we did not expect to see an association betweenutreach and parents educational expectations. Educa-pectations are indicative of family norms and valuesaufman et al., 2003; Yan & Lin, 2005), have been showns malleable in the rst few years of childrens school-denberg, Gallimore, Reese, & Garnier, 2001), and are

to be stable through childrens year of kindergarten. extent is family involvement associated with studentsd reading gains in kindergarten? Based on previous, we expected to see a signicant association betweenvolvement and childrens achievement gains (Dearing06; Schulting et al., 2005). Specically, family involve-home and parental expectations were expected to have

relationships with childrens gains in achievement thanent at school (Fan & Chen, 2001; Jeynes, 2003, 2005).ally, we expected to observe different associations

family involvement indicators and math and readingath achievement tends to be more sensitive to teacherol effects, so we expected to nd stronger associations

family involvement and childrens achievement gainsing than for math (Rimm-Kaufman, Fan, Chiu, & You,

extent is school outreach to families associated withs math and reading gains in kindergarten? Is the rela-

between outreach to families and achievement gainsd by family involvement? Consistent with prior studiesl outreach to families (Schulting et al., 2005; Sheldonn, 2005), we expected to nd a statistically signicantrelationship between school outreach and achievementdicating that schools implementing more practices toamilies had students that made greater gains in read-ath through kindergarten. However, we also expected

that that the relationship between school outreachents achievement gains would be mediated by fam-vement at home and at school. Parents educationalions were not hypothesized to mediate the relationship

school outreach and student achievement.

a came from the Early Childhood Longitudinal Studyindergarten Class of 19981999, sponsored by thenter of Education Statistics. ECLS-K focused on youngognitive and non-cognitive growth and collected infor-

students, parents, teachers, and administrators. Usinge probability sampling design, ECLS-K included a nation-ntative sample of about 21,000 kindergarteners in overls. For more details of the ECLS-K study, includingg frame and data structure, see National Center fortatistics (2001). In this article, we analyzed informationrgarten (fall 1998 and spring 1999).

C. Galindo, S.B. Sheldon / Early Childhood Research Quarterly 27 (2012) 90 103 93

4.1. Sample and missing cases

The study sample for the math analysis included 16,430 studentsfrom 870 schools and the reading analysis included 15,960 studentsfrom 860 scthe nearestsample was(Imputationmultiple imsample. ICEchain equatness, that irst and soand Broh (2tion, separaprocedure tdeveloped. sets that wepercentagedeviations procedures

After constudents fro(1610 studof kindergabetween fanot have anschools didkindergarte16,430 stud15,960 stud

The samWhites, 145% Asians, the samplelanguage spin homes wsiblings at h

4.2. Key va

4.2.1. StudeThese va

two-stage adomains ba(NAEP) framThe same tevery childthe same rthey answemedium, antion of the tand operatianalysis, stfunctions. Tiarity, lettesounds, woprehensionpersonal rebility estimof kindergaon the ECLECLS-K asseStatistics (2

In this sscores obta

C2RSCALE). These scale scores are criterion-reference measures ofachievement that place students performance within a commonand continuous scale (64-point scale for math and 92-point scale forreading). IRT procedures estimate patterns of responses for ques-

ased item-abilo ademen

twot sce. W

icatod mle recoeffromeadl

Famis var0 = noerga

ing a, PTOrent teaventittee;

Famis varyda

of pl of kafts, sciene of ser or

Parenanothl variouldMD, oemene ans edt schey wily in

Schooitionefforring

= 0.6 schogs; ome;

parited

mat schemenhools. The unweighted sample sizes were rounded to 10 because of restricted license requirements. Each

dened by a two-step process. We rst applied the ICE by Chained Equations) algorithm in STATA to deriveputation of missing values using the entire kindergarten

handles complex data structures by tting a sequence ofions to impute variables in order of increasing missing-s, the variable with the least missing values is imputed

on (Royston, 2005). Following Downey, Von Hippel,004), we imputed student and school level informa-tely, for missing data. To the best of our knowledge, ao impute missing values with nested data has yet to beThe imputation procedure resulted in ve plausible datare analyzed with HLM 6.08 software. Table 1 shows the

of missing cases, means or percentages, and standardfor all variables before applying multiple imputation.ducting multiple imputation procedures, we droppedm the original ECLS-K sample who did not have math

ents) or reading test scores (2320 students) at springrten (time 2 or outcome variable), changed schoolsll and spring of kindergarten (510 students), or didy school-level information (2720 students). About 150

not answer the principals survey in the spring ofn. The study sample for the math analysis includedents from 870 schools and the reading analysis includedents from 860 schools.ple for this study included 60% native non-Latino

% native non-Latino Blacks, 18% Latinos of any race;and 3% other race/ethnicities. Most of the students in

were born in the U.S. (82%) and English was the mainoken at home (89%). About 66% of the students livedith two biological parents and the average number ofome was 1.43.

riables in the study

nts math and reading achievementriables were measured using individually administereddaptive math and reading tests, with content areas andsed on the National Assessment of Educational Progressework (National Center for Education Statistics, 2001).ests were administered across schools, although not

received the same items. All students responded toouting questions (rst set of common questions) andred questions at one of three levels of difculty (high,d low) based on their performance in the routing por-est. The math tests measured number sense, propertiesons; measurement; geometry and spatial sense; dataatistics, and probabilities; and patterns, algebra, andhe reading tests measured basic skills (print famil-

r recognition, beginning and ending sounds, rhymingrd recognition); receptive vocabulary; and reading com-

skills (initial understanding, developing interpretation,ection, and demonstrating critical stance). The relia-ates for the reading and math test scores in the springrten were 0.95 and 0.94, respectively. For more detailsS-K assessments and the psychometric properties ofssment instruments, see National Center for Education001) and Rock and Pollack (2002).tudy, we used the item response theory (IRT) scaleined from the Base Year Code Book (C2MSCALE and

tions band onguess

Twinvolvcreatedment aat homan indimputeAll scaalpha scales (see Ch

4.2.2. Thi

scale (of kindincludof PTAthe paparentclass ecomm

4.2.3. Thi

4 = eveipationthe faland cror do outsidtogeth

4.2.4. As

ordinachild wa PhD, involvat homparentment athat thof fam

4.2.5. Add

reach the spyear, by themeetinsent hon-oneare invfamilyfor theinvolv on patterns of right, wrong, and omitted responses parameters of difculty, discriminating ability, and

ity (Rock & Pollack, 2002).ditional variables were the foci of analyses: Familyt and school outreach efforts to involve families. We

ordinal scales of family involvement: family involve-hool and family involvement in educational activitiese also included parents educational expectations as

r of family involvement. To create the scales, we rstissing data for each item and then calculated the scales.liabilities are estimated on the study sample and thecients are based on the non-imputed values. Similar

the ECLS-K have been used in many published articlese, 2008; Crosnoe & Cooper, 2010; Schulting et al., 2005).

ly involvement at schooliable was measured using the average of a 7-item

and 1 = yes, = 0.58) on parents reports in the springrten of their participation in school-related activitiesttending open house or back-to-school nights; meetings, or parentteacherstudent organization; meetings ofadvisory group or policy council; regularly-scheduledcher conferences or meeting with teachers; school ors; acting as a volunteer at the school or serving on a

and fundraising for school.

ly involvement in educational activities at homeriable was the average response (1 = never toy, = 0.75) to 11 items on the frequency of partic-arent and child on the following activities collected inindergarten: read books, tell stories, sing songs, do artdo chores, play games or do puzzles, talk about naturece projects, play sports, child looked at picture bookschool, child read or pretended to read, and build things

play with construction toys.

ts educational expectations for their childrener indicator of family involvement, we included thisable indicating how far in school parents believed their

go (1 = receive less than a high school diploma to 6 = getr other higher degree). The correlations between familyt indicators were small (r = 0.24 between involvementd school; r = 0.14 between involvement at home anducational expectations; and r = 0.12 between involve-ool and parents educational expectations), suggestingere modestly related, but largely independent aspectsvolvement.

l outreach effortsally, we used an 8-item scale to measure school out-ts to involve families based on principals reports inof kindergarten (1 = never to 5 = seven or more times a4) of how often the following activities were conductedol: PTA, PTO, or parentteacherstudent organizationwritten reports (report cards) of childs performance

teacherparent conferences; home visits to do one-ent education; school performances to which parents; classroom programs like class plays, book nights, orh nights; fairs or social events planned to raise fundsool; and workshops for teachers that focus on parentt.

94 C. Galindo, S.B. Sheldon / Early Childhood Research Quarterly 27 (2012) 90 103

Table 1Weighted descriptive statistics for the ECLS-K sample.

Mean or % SD % of missing

Key analytical variablesMath IRT w 19.48Math IRT w 27.55 Reading IR 22.28 Reading IR 32.07 Family inv 0.54 Family inv 2.84 Parents ed 4.05 School out 3.07

Level-1 contRace/ethn

White 56.99 Black 16.71 Latino 18.91 Asian 2.99 Other 4.40

Gender: fe 48.27 Generatio

First gen 3.00 Second g 17.09 Third-pl 79.91

Non-Engli 13.03 Kindergar 95.24 Family typ

Two bio 64.05Two par 8.86 Single-pOther

Age at kinParents edFamily incNumber o

Level-2 contKindergarSchool yeaSector (%)

PublicCatholicOther reOther pr

CompositiMean WMean BlMean LaMean As

Compositi

Note: N = 16,44All descriptive

4.3. Backgr

We inclumodels to sFamily type group; twoent at homeparents. Nuents highesfrom 1 = somwas a conted as WhiLatino of aHawaiian, Oand more thas rst genents; secondand third-p(the referenentry (contigartener (dave 1ave 2

T wave 1 T wave 2olvement in school olvement at home ucational expectations reach effortsrol variables (students and families)icity (%)

male (%) nal status (%)eration enerationus generation sh speaking homes (%) ten status: rst time (%) e (%) logical parentsents, one biological

arent 23.09

4.00 dergarten entry (months) 68.62 ucational level (2 = high school diploma and 3 = some college) 2.98ome 51,891f siblings 1.43rol variables (schools)ten enrollment 55.04 r length 177.30

65.16 9.62 ligious 14.23 ivate 10.99 on-race hite non-Latino 0.64 ack non-Latino 0.12 tino 0.13ian 0.05 on-parents education 3.15

0 students and 870 schools. % = percentages and SD = standard deviation. Percentages of m statistics were computed using ECLS-K cross-sectional weights: c2cw0 for student leve

ound and control variables

ded several student, family, and school variables in ourtatistically control for important background factors.was measured as two biological parents the reference

parents with one biological; only one biological par-; all other structures, including guardian and adoptivember of siblings at home was a continuous variable. Par-t educational level was an ordinal variable with values

e high school to 5 = graduate studies. Family incomeinuous variable (in 1000s). Students race was identi-te non-Latino the reference group, Black non-Latino,ny race, Asian, and Other. Other race includes nativether Pacic Islander, American Indian, Alaska native,an one race. Students generational status was measurederation, non-U.S. born students to non-U.S born par-

generation, U.S born students to non-U.S born parents;lus generation, U.S. born students to U.S.-born parentsce group). Additional controls were age at kindergartennuous), whether the student was a second-time kinder-ummy); whether student was female (dummy); and

whether a nvariables exin our regre

At the scposition, avlength of thcategories: gious, and of Hispanicwas measuat the schoby aggregalevel. Lengtment were the ECLS-K(2001).

4.4. Data a

We useddents repre7.26 9.338.76 0.078.31 13.1410.22 2.890.23 6.720.48 12.701.11 13.090.57 1.27

0.330.0619.180.0112.703.43

8.26

1.16 3.43.40 55,348.37 3.43

1.12 3.43

46.52 2.77 24.54 4.86

00

0.33 0.23 0.20 0.11 0.74 0.23

issing data were calculated based on 16,440 students or 870 schools.l variables and s2saqw0 for school level variables.

on-English language was spoken at home (dummy). Allcept for dummies were treated as continuous variablesssion models.hool level, we controlled for sector, racial/ethnic com-erage educational level of parents, enrollment size, ande school year. School sector was represented by fourpublic the reference category, Catholic, other reli-other private. School racial composition (i.e., percent

students and percent of African American students)red by aggregating student racial/ethnic informationol level. The school educational average was measuredting parents highest education attained at the schoolh of school year (number of days) and school enroll-continuous variables. For a detailed description of all

variables see National Center for Education Statistics

nalyses

two-level hierarchical linear modeling (HLM) with stu-senting the level-1 units and schools representing the

C. Galindo, S.B. Sheldon / Early Childhood Research Quarterly 27 (2012) 90 103 95

level-2 units. HLM gives valid and accurate estimates when dealingwith nested data (as in this case where students are nested withinschools) because it takes into account the complex structure of theerror terms (Raudenbush & Bryk, 2002). With nested data, OrdinaryLeast Squarcance givenviolated. Intend to be mselected.

We usedon student-least two unwere severateacher. It students neschools) wi

To analywith familels 1, 2, ameasures (ment in edexpectationcontrolled including aof kindergament in thinvolvementheir childrwere treatmodels.

To analystudents awe estimat(Models 4, The rst thmeasures sfamily invoachievemenscores fromscale scoresThese modecontrols.

We alsodifference ment at tim1 as a variab2 and time robust. Regterns of assostatisticallyvariables.

Then, tociated withmediated bachievemenmeasure ondent achievof family inrespectivelyment (Modstatisticallyvariables. Wupper level able and levused the So1982).

The most exhaustive HLM model is specied as follows.Between-student equation

1 Model : (Test score wave 2)ijj +

j (famj

(

ween

Mo

10

20

the of ki) is

ent tors, -leve

averects

parspt w

the crosnd d(End

to ere cdentregres in sociaas wed aly sicted t acto eemening tmeasrrentns fo

erro wher st.

o, to varf thetcom

assomilyhool

the imat

variolveodel stimaes analyses tend to overestimate the levels of signi- that the assumption of independence of variables isdividuals within social contexts such as in schools,ore alike than if the sample of students was randomly

two-level models because the focus of this study was and school-level effects. Additionally, HLM requires atits per cluster (i.e. two teachers per school), and therel schools in the ECLS-K data with only one kindergartenwas not possible, then to use three-level models (i.e.,sted within classrooms and classrooms nested withinth these data.ze whether school outreach to parents was associatedy involvement, we estimated three models (Mod-nd 3), to examine each of the family involvementi.e., family involvement at school, family involve-ucational activities at home, and parents educationals) as dependent variables. These models statisticallystudent, family, and school background variables,

measure of prior math achievement from the fallrten. We included a measure of prior math achieve-ese models to control for the possibility that familyt could be a reaction to parents perceptions ofen academic skills. All family involvement variablesed as continuous variables in the HLM regression

ze whether family involvement was associated withchievement gains from fall to spring of kindergarten,ed four models for each achievement outcome math5, 6, and 7) and reading (Models 8, 9, 10, and 11).ree models included each of the family involvementeparately whereas the fourth model included all threelvement measures simultaneously. To capture gains int, we included in our regression models the IRT scale

the fall assessment as a control variable and the IRT from the spring assessment as the dependent variable.ls included student and family background variables as

re-estimated all the cognitive outcomes models usingscores. In these models, rather than having achieve-e 2 as the dependent variable and achievement at timele predictor, we specied the difference between time1 as the dependent variable. Overall, our results wereardless of model specication, we observed similar pat-ciation between different specications. All of the tests

controlled for student, family, and school background

analyze whether schools outreach efforts were asso- students gains and whether these relationships werey family involvement, we estimated two models for eacht outcome. One model included the school outreachly to estimate the association of outreach and stu-ement gains prior to adjusting for potential inuencesvolvement (Models 12 and 14 for math and reading,). Then, we added all three measures of family involve-els 13 and 15 for math and reading). All of the tests

controlled for student, family, and school backgrounde followed Kenny, Kashy, and Bolger (1998) to study anmediation (when the association between level-2 vari-el-1 outcome is mediated by level-1 variables) and webel test to formally test the mediation model (Sobel,

Level-= 0+2+3

Bet

Level-2

1j = 2j = wherespringwave 2represindicatschoolon thethe eff

Forinterceso thatsame auous amean appliedyses wfor stu

All cientthe asment, we ustisticalcondustudenment achievanalyzment the cunectioType Ichancethat ousample

Alsfor thesion othe ouof theand fathe sction ofTo estlevel-1ily invthe mized e2010).1j (test score wave 1)ilyinvolvement indicators)individual and family background) + rij

(1)

-school equation

del : 0j = 00 + 01 (school outreach)+ 02 (school background) + 0 j (2a)

(2b)

(2c)

overall average achievement score (by subject) in thendergarten for a student i in school j (Yij test scorea function of student-level variables (1j, 2j, and 3jhe effect of previous achievement, family involvementand individual and family controls, respectively) andl variables (01 represents the effect of school outreachage achievement level in school j, and 02 representsof school-level controls).imony, only the level-1 model and the level-2 modelere set as random, whereas all level-2 slopes were xedeffects of all level-1variables were constrained to be thes schools. In every model, the level-1 and level-2 contin-ichotomous variables were centered around the granders & Toghi, 2007). Thus, the estimated parametersthe average student in the sample. All statistical anal-omputed using ECLS-K cross-sectional weights: c2cw0

level variables and s2saqw0 for school level variables.ssion estimates are reported as unstandardized coef-the tables. Given the extensive research base ontion between school outreach and family involve-ell as family involvement and student achievement,

more conservative alpha of 0.01 for discussing sta-gnicant relationships. Far fewer studies have beenanalyzing the association between school outreach andhievement, or the mediating role of family involve-xplain why school outreach might predict studentt. Thus, we chose to use an alpha of 0.05 whenhe association between school outreach and achieve-ures. As a result, our analyses are able to extend

body of research on the role of school-family con-r student achievement, while also guarding againstrs (where we nd associations between variables byn there no relationship exists) that stem from the fact

udy contains a large number of analyses with a large

measure the magnitude of importance or effect sizesiables of interest, we divided the standardized ver-

coefcient by level-specic standard deviation ofe variable. For example, to estimate the effect sizeciation between school outreach (a level-2 variable)

involvement, we divided the standardized version ofoutreach coefcient by the level-2 standard devia-unconditional model (i.e., models without predictors).e the effect size for family involvement at school (aable), we divided the standardized version of the fam-ment measure by the level-1 standard deviation ofwith only math test score at time. These standard-tes are equivalent to effect sizes (Galindo & Fuller,

96 C. Galindo, S.B. Sheldon / Early Childhood Research Quarterly 27 (2012) 90 103

Table 2HLM regression xed and random estimates of family involvement from school outreach efforts.

PI school PI home Education expectationsModel 1 Model 2 Model 3

Fixed effect coefcientsIntercept 0.55** 2.84** 4.09**

(0.00) (0.01) (0.01)Level-2 key variable (schools)

School outreach efforts 0.02** 0.01 0.04(0.01) (0.01) (0.02)

Level-1 control variables (students and families)Math score at wave 1 0.00** 0.00 0.02**

(0.00) (0.00) (0.00)Age at kindergarten entry 0.00 0.01** 0.01**

(0.00) (0.00) (0.00)Black 0.04** 0.00 0.30**

(0.01) (0.02) (0.05)Latino 0.01 0.01 0.23**

(0.01) (0.02) (0.04)Asian 0.07** 0.05 0.17**

(0.01) (0.03) (0.06)Other 0.03** 0.03 0.09

(0.01) (0.02) (0.05)Female 0.01 0.04** 0.08**

(0.00) (0.01) (0.02)First generation 0.08** 0.06 0.28**

(0.01) (0.04) (0.07)Second generation 0.03** 0.04 0.27**

(0.01) (0.02) (0.04)Non-English speaking homes 0.04** 0.16** 0.32**

(0.01) (0.02) (0.05)First time kindergartener 0.04** 0.10 0.29**

(0.01) (0.03) (0.06)Two parents, one biological 0.05** 0.02 0.00

(0.01) (0.02) (0.04)One single parent 0.06** 0.01 0.08**

(0.01) (0.01) (0.03)Other type of family 0.04** 0.03 0.13

(0.01) (0.03) (0.06)Number of siblings 0.00 0.01** 0.05**

(0.00) (0.00) (0.01)Educational level 0.04** 0.05** 0.20**

(0.00) (0.01) (0.01)Income 0.00** 0.00 0.00**

(0.00) (0.00) (0.00)Level-2 control variables (schools)

Kindergarten enrollment 0.00 0.00 0.00(0.00) (0.00) (0.00)

School year length 0.00 0.00 0.00(0.00) (0.00) (0.00)

Catholic 0.07** 0.03 0.05(0.01) (0.02) (0.04)

Other religious 0.00 0.07** 0.08(0.02) (0.02) (0.05)

Other private 0.09** 0.03 0.16**(0.02) (0.03) (0.06)

Mean black 0.03 0.03 0.23**(0.02) (0.03) (0.07)

Mean Latino 0.07** 0.01 0.32**

(0.02) (0.03) (0.07)Mean Asian 0.00 0.03 0.25

(0.03) (0.05) (0.12)Mean education 0.03** 0.00 0.10**

(0.01) (0.01) (0.02)Random effect coefcients

Level-1 variance (between students) 0.038 0.207 1.046Level-2 variance (between schools) 0.004** 0.005** 0.022**

Note: Robust standard errors in parentheses. p-values are based on estimations with robust standard errors. For parsimony only the level-1 equation and the level-2 interceptwere included as random (all other level-2 slopes were xed). Patterns and trends did not change when using reading test scores as control variable instead of math testscores for these models. We estimated all statistics models using appropriate ECLS-K cross-sectional weights.

** p 0.01.

C. Galindo, S.B. Sheldon / Early Childhood Research Quarterly 27 (2012) 90 103 97

5. Results

5.1. Association between school outreach to parents andinvolvement

Table 2 iindicators otional activactivities, apendent varof school ouunstandardily variablesin all modelditional moof the varities was bein parents educationarespectively

In Table itive associaschool efforties paid ofschool. All twas associain school aoutreach efinvolvemen(Models 2 a

In standof the associnvolvemention for par0.12 for invto school-le1 (family areect the tations, invoa school by ocase on thetion of the otake into aachievemen

Table 2 tics that wbeing equallevels had all three fachanged wistudents living homes higher educnon-immigParents of Black parenWhite studAlso, parenof siblings increased a

At the sccational levof schools)at school aeffects weregested that

family involvement measures. Such differential patterns of associ-ation should be analyzed in future work.

5.2. Family involvement and math and reading achievement

le 3 g andemenee mndenily in

wittes (nvolstratergaies wnd re in and 0st, faith rtanden faf fam.05 aentsns) aadint all evel en aode

ation invoath tionrentsall.haveolvemsize g andducaemen

(Daes, &f parSo, aeas

les arainsdditaractant eh acrage,ationmathgarteen Lang h-spe

resurth nncludes the three models that give attention to differentf family involvement: family involvement in educa-ities at home, family involvement in school-relatednd parents educational expectations. The main inde-iable of interest in all these models is: principals reporttreach to parents. Regression estimates are reported asized coefcients in the table. Level-1 (student and fam-) and level-2 (school variables) controls were includeds. Analyses of the intracluster correlations of the uncon-dels (i.e., models without predictors) revealed that 21%ance in family involvement in school-related activi-tween schools. A smaller proportion of the varianceseducational expectations and family involvement in

l activities at home were between schools, 9% and 4%,.2, Model 1 shows that school outreach efforts had a pos-tion with family involvement at school, suggesting thatts to encourage families to participate in school activi-f in terms of getting parents to interact at and with thehings being equal, each unit increase on school outreachted with a 0.02 point increase of family involvementctivities. In contrast, the associations between schoolforts and parents educational expectations and familyt at home were positive but not statistically signicantnd 3).ard deviation units, we observed that the magnitudeiations between school outreach and the three familyt indicators were modest (0.07 of a standard devia-ents expectations; 0.08 for involvement at home; andolvement at school). These effect sizes correspondedvel effect sizes, which are not comparable to level-

nd students) effect sizes. The school-level effect sizeseffect on the outcome (i.e., parents educational expec-lvement in school and involvement at home) of movingne standard deviation on a school-level variable (in this

school outreach efforts) relative to the standard devia-utcome variable between schools. These net effect sizesccount all of the covariates, including students matht.indicates additional student and family characteris-ere associated with family involvement. Other things, students math achievement and parents educationalpositive and statistically signicant associations withmily involvement indicators. Patterns of associationth different family involvement measures. For instance,ing with immigrant parents and in non-English speak-had lower levels of family involvement at school, butational expectations than their counterparts living inrant families and English speaking homes, respectively.racial/ethnic minority students, specically Asian andts, reported higher educational expectations than didents, but lower levels of family involvement at school.ts educational expectations decreased as the numberat home increased, but family involvement at homes the number of sibling increased.hool level, some school characteristics (e.g., mean edu-el of students parents and racial/ethnic composition

further explained differences in family involvementnd parents educational expectations, although fewer

observed for parents involvement at home. This sug-school and family mechanisms related differently to the

gains

Tabreadininvolvall thrindepeof famtationscovariamore idemonof kindactivitmath aincreasa 0.17 contraated w

In sbetwesizes owere 0for paring gaiand reaccounment l

Whmath mexpectfamilywith massociaand payet sm

To ily inveffect readinents eachievtationsRhoadsizes ogains. ment mvariabment g

In aual chimportin matdents expectlower kinderbetwespeakiEnglish

Theare woreports four models for each achievement outcome: math. The rst three models tested each of the familyt measures separately, whereas the last model includedeasures of involvement to check their simultaneous,t effects. The results show the signicant associationsvolvement at school and parents educational expec-h math and reading gains, even after controlling allModels 4, 6, 8, and 10). Students whose parents wereved at school or had higher educational expectationsed greater gains in reading and math skills at the endrten. Each unit increase in family involvement in schoolas associated with a 1.10 and 0.97 points increase ineading achievement, respectively. Similarly, each unitparents educational expectations was associated with.18 points increase in these achievement measures. In

mily involvement at home was not signicantly associ-eading or math gains (Models 5 and 9).ard deviation units, we observed modest associationsmily involvement and cognitive gains. Overall, effectily involvement at school for reading and math gains

nd 0.04, respectively. Smaller effect sizes were observed educational expectations (0.04 for both math and read-nd family involvement at home (0.01 and 0.02 for mathg gains, respectively). These net effect sizes take intoof the covariates, including students previous achieve-in reading or in math.ll family involvement measures were included in thel (Model 7 and 11), school involvement and parentss coefcients remained statistically signicant, andlvement at home remained non-signicantly associatedand reading achievement gains. This suggests that thes to achievement gains of parent involvement in school educational expectations are independent and robust,

a better sense of the relative importance of the fam-ent indicators, we compared their effect sizes to the

of parents own educational attainment for students math gains. Numerous studies have shown that par-tional attainment is a consistent predictor of childrenst, as well as parental behaviors and educational expec-

vis-Kean, 2005; Hill et al., 2004; Lareau, 2003; Luster, Haas, 1989). In standard deviation units, the effectents education were 0.09 for both reading and mathlthough the absolute effect size of our family involve-ures seemed small (0.040.05), in relative terms thesee moderately important predictors of students achieve-

in kindergarten.ion, Table 3 indicates that several family and individ-eristics included as covariates in these analyses hadffects on reading and math gains in kindergarten. Gainsoss kindergarten were positively associated with stu-

parents educational levels, and parents educationals. Compared with White students, Black students had

gains, but Asian students had higher math gains inn. There was no statistical difference in math gainstino and White students. Children living in non-Englishomes had lower math gains than children living inaking homes.lts were similar for reading, although some differencesoting. For instance, students gender and the number of

98 C. Galindo, S.B. Sheldon / Early Childhood Research Quarterly 27 (2012) 90 103

Table 3HLM regression xed and random estimates of students achievement from family involvement.

Math Reading

Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 Model 10 Model 11

Fixed effect Intercept 94****

7) Level-1 ke

Involvem 3**

5) Involvem 1

2) Educatio 5**

5)Level-1 co

Test sco 1**

1) Age at k 6**

1) Black .37**

7) Latino .42

Asian

Other

Female

First gen

Second g

Non-Eng

First tim

Two par

One sing

Other ty

Number

Educatio

Income

Random effeLevel -1 vaLevel -2 va

Note: Robust swere includedvariances of thbase model fo

** p 0.01.

siblings at hmath gains.students, busiblings at hgarten had Additionallbetween chEnglish spedid not takenon-Englishstudents w2009).

Overall, involvemencontrolling dents whoscoefcients27.94** 27.93** 27.93** 27.(0.07) (0.07) (0.07) (0.0

y variables (students and families)ent at school 1.10** 1.0

(0.24) (0.2ent at home 0.13 0.0

(0.11) (0.1nal expectations 0.17** 0.1

(0.05) (0.0ntrol variables (students and families)re at wave 1 0.92** 0.92** 0.92** 0.9

(0.01) (0.01) (0.01) (0.0indergarten entry 0.06** 0.06** 0.06** 0.0

(0.01) (0.01) (0.01) (0.01.32** 1.37** 1.42** 1(0.17) (0.17) (0.17) (0.10.38 0.39 0.44 0

(0.18) (0.18) (0.18) (0.18)0.68** 0.60 0.57 0.65**

(0.25) (0.25) (0.25) (0.25) 0.49 0.52 0.53 0.50 (0.24) (0.24) (0.24) (0.24) 0.03 0.03 0.04 0.04 (0.09) (0.09) (0.09) (0.09)

eration 0.54 0.46 0.41 0.49 (0.35) (0.36) (0.35) (0.35)

eneration 0.11 0.08 0.03 0.07 (0.16) (0.16) (0.16) (0.16)

lish speaking home 0.69** 0.71** 0.78** 0.74**(0.20) (0.20) (0.20) (0.20)

e kindergartener 0.58 0.61 0.57 0.54 (0.25) (0.25) (0.25) (0.25)

ents, one biological 0.36 0.42** 0.42** 0.36 (0.16) (0.16) (0.16) (0.16)

le parent 0.08 0.15 0.16 0.10 (0.14) (0.14) (0.14) (0.14)

pe of family 0.73** 0.78** 0.76** 0.72**(0.24) (0.24) (0.24) (0.24)

of siblings 0.03 0.03 0.02 0.02 (0.04) (0.04) (0.04) (0.04)

nal level 0.31** 0.35** 0.32** 0.28**

(0.05) (0.05) (0.05) (0.05) 0.00 0.00 0.00 0.00 (0.00) (0.00) (0.00) (0.00)

ct coefcientsriance (between students) 22.502 22.538 22.514 22.480 riance (between schools) 2.478** 2.503** 2.498** 2.477**

tandard errors in parentheses. p-values are based on estimations with robust standard err as random (all other level-2 slopes were xed). We estimated all statistical models usinge base math model (i.e., with only math in the fall of kindergarten as a predictor) were 2.r reading were 6.108 and 31.735.

ome were related to students reading gains, but not to Female students had higher reading gains than did malet students reading gains decreased as the number ofome increased. Children who were retained in kinder-lower reading gains than did rst time kindergarteners.y, there was no statistical difference in reading gainsildren living in non-English speaking homes and inaking homes. Students who were not English procient

the reading test, thus the estimated reading scores for speaking students in kindergarten were based only on

ho were procient in oral English (Reardon & Galindo,

results in Table 3 show positive effects of familyt on achievement gains in kindergarten, even afterother inuential background measures. On average, stu-e parents were more involved at school and who had

higher educduring kind

5.3. School achievemen

Whereastudent achabout involschool outrassociated. explorationmodels is ption to studincluded paand involve32.50** 32.50** 32.50** 32.50**

(0.11) (0.11) (0.11) (0.11)

0.97** 0.85**

(0.29) (0.29)0.26 0.14(0.12) (0.12)

0.20** 0.18**

(0.05) (0.06)

0.93** 0.93** 0.93** 0.92**

(0.01) (0.01) (0.01) (0.01)0.05** 0.05** 0.05** 0.05**

(0.01) (0.01) (0.01) (0.01)0.80** 0.84** 0.90** 0.85**(0.21) (0.21) (0.21) (0.21)0.03 0.02 0.03 0.02

(0.21) (0.21) (0.21) (0.21)1.28** 1.21** 1.16** 1.24**

(0.33) (0.33) (0.33) (0.33)0.36 0.40 0.40 0.38(0.26) (0.26) (0.26) (0.26)0.72** 0.71** 0.71** 0.70**

(0.10) (0.10) (0.10) (0.10)0.27 0.34 0.41 0.33(0.47) (0.47) (0.47) (0.47)0.10 0.08 0.02 0.06(0.21) (0.21) (0.20) (0.20)0.06 0.05 0.16 0.10(0.28) (0.28) (0.28) (0.29)1.47** 1.48** 1.45** 1.41**

(0.32) (0.32) (0.32) (0.32)0.31 0.36 0.36 0.31(0.18) (0.18) (0.18) (0.18)0.37 0.42** 0.44** 0.39(0.16) (0.15) (0.15) (0.16)0.95** 0.10** 0.97** 0.94**(0.28) (0.28) (0.28) (0.28)0.15** 0.15** 0.14** 0.14**(0.05) (0.05) (0.05) (0.05)0.38** 0.41** 0.38** 0.34**

(0.05) (0.05) (0.05) (0.05)0.00 0.00 0.00 0.00(0.00) (0.00) (0.00) (0.00)

30.997 31.009 30.987 30.9635.890** 5.942** 5.918** 5.884**

ors. For parsimony only the level-1 equation and the level-2 intercept appropriate ECLS-K cross-sectional weights. The level-1 and level-2753 and 22.981, respectively. The level-1 and level-2 variances of the

ational expectations had higher math and reading gainsergarten, regardless of their initial skills in the fall.

outreach to parents and math and readingt gains

s this study conrms that involvement is important forievement gains in math and reading, it takes questionsvement further than prior studies by exploring whethereach and students achievement in kindergarten areTable 4 presents four models that give attention to this. The main independent variable of interest in theserincipals report of school outreach to parents. In addi-ent, family, and school covariates, Models 13 and 15rents educational expectations, involvement in school,ment at home to analyze the mediating power of the

C. Galindo, S.B. Sheldon / Early Childhood Research Quarterly 27 (2012) 90 103 99

Table 4HLM regression xed and random estimates of students achievement from school outreach efforts.

Math Reading

Model 12 Model 13 Model 14 Model 15

Fixed effect Level-2 ke

School o .37*

.15)Level-1 ke

Involvem .03**

.25) Involvem .01

.12)Educatio .15**

.05)Random effe

Level-1 va 2.476Level-2 va .472**

Note: Robust s rd errwere included s usingvariances of th ere 2.base model fo (scho

* p 0.05.** p 0.01.

family invoand school vshown in Tintra-clustewithout premath and re

Table 4 school outrmath (Modment and areport of scincrease in less of theirmath and rvided moreand interaceffects of prachievemenand 0.09 foreffect sizes

In modetors to analbetween scparing diffefamily invoin the modand cognitisignicant, cient decrmeasures, tachievemen

To formasteps of assment outcobetween thcomes (seerespectivelyassociated wresults presmediator vawith the ououtreach mTable 4, res

r anting mredicl expy useemench anor reemen

outr test 6, p

en outed bndenn kin

cussi

earchdispahe qtribu

outrcoefcientsy variables (schools)utreach efforts 0.40** 0

(0.15) (0y predictors (students and families)ent at school 1

(0ent at home 0

(0nal expectations 0

(0ct coefcientsriance (between students) 22.537 2riance (between schools) 2.496** 2

tandard errors in parentheses. p-values are based on estimations with robust standa as random (all other level-2 slopes were xed). We estimated all statistical modele base math model (i.e., with only math in the fall of kindergarten as a predictor) wr reading were 6.108 and 31.735. Level-1 (student and family variables) and level-2

lvement variables. All models included student, family,ariables as controls. These background controls are not

able 4 but were included in all models. Analyses of ther correlations of the unconditional models (i.e., modelsdictors) revealed that 22% and 23% of the variance inading achievement were between schools.shows positive and signicant associations betweeneach and students achievement gains in reading andels 12 and 14). After controlling for previous achieve-ll other covariates, each unit increase on principals

hool outreach was associated with a 0.40 and 0.41 pointreading and math achievement, respectively. Regard-

starting skills in the fall, students had greater gains ineading, on average, if they attended schools that pro-

opportunities for family and school communicationstions. In standard deviation units, we observed modestincipals report of school outreach on math and readingt in kindergarten (0.12 of a standard deviation for math

reading). These effect sizes correspond to school-levelon the school-level variables.ls 13 and 15, we added the family involvement indica-yze whether these variables mediated the relationshiphool outreach to parents and cognitive gains, by com-rences in coefcients before and after accounting for

Aftefor tesonly pcationawe onlinvolvoutreatests finvolvschool(Sobeltic = 1.9betwemediaindepegains i

6. Dis

Restional or to tto conschoollvement. After including family involvement indicatorsels, the association between school outreach activitiesve outcomes in reading and math remained statisticallyalthough the magnitude of the school outreach coef-eased slightly. After including all family involvementhe associations between outreach and math and readingts decreased by 10% and 8%, respectively.lly test for mediation, we followed Kenny et al. (1998)ociations and estimated a Sobel test for each achieve-me (Sobel, 1982). We rst analyzed the relationshipe level-2 variable (school outreach) and the level-1 out-

Models 12 and 14 for math and reading in Table 4,). We then analyzed whether the level-2 variable wasith the mediator variables of family involvement (seeented in Table 2). Finally, we analyzed whether theriables (family involvement measures) were associatedtcome variable in a model already including the schooleasure (see Models 13 and 15 for math and reading inpectively).

with readinthe extent between scment. The study that contexts hemath throuin part by thigher in sclies in theirdiscussed wthis study.

6.1. School

The resuesis 1. The student bament as a 0.41* 0.38*

(0.19) (0.18)

0.74**

(0.27)0.06(0.11)0.19**

(0.05)

31.573 31.5185.699** 5.677**

ors. For parsimony only the level-1 equation and the level-2 intercept appropriate ECLS-K cross-sectional weights. The level-1 and level-2753 and 22.981, respectively. The level-1 and level-2 variances of theol variables) controls are not shown but were included in all models.

alyzing the required models to fulll the steps necessaryediation, we found that principals report of outreach

ted family involvement at school and not parents edu-ectations and family involvement at home. As a result,d the Sobel test to determine the extent to which familyt at school mediated the relationship between schoold childrens achievement gains. After running separateading and math, the Sobel test indicated that familyt at school was a signicant mediator of the inuence ofeach efforts on achievement gains for reading and mathstatistic = 2.03, p = 0.04 for math and Sobel test statis-

= 0.05 for reading). We concluded that the associationtreach and achievement in kindergarten was partiallyy family involvement at school, and that outreach had ant and positive relationship with childrens achievementdergarten above and beyond family involvement.

on

ers and policy makers have debated whether educa-rities are intrinsically related to the home environmentuality of the school a child attends. This study aimedte to this discussion by examining the extent to whicheach to parents and family involvement are associated

g and math achievement gains in kindergarten, and

to which family involvement explains the relationshiphool outreach to involve families and student achieve-ndings support the theoretical assumptions guiding thethe interactions of people across the home and schoollp explain childrens achievement gains in reading andgh kindergarten. The relationship can be explained onlyhe fact that family involvement at school tends to behools that implement more practices to engage fami-

childrens learning. The main ndings of the paper areithin the context of the research questions that guided

outreach to parents and involvement

lts of this study demonstrate partial support for hypoth-analyses showed that, after controlling for family andckground variables including students prior achieve-measure of student ability, schools that conducted

100 C. Galindo, S.B. Sheldon / Early Childhood Research Quarterly 27 (2012) 90 103

more activities to engage families tended to have parents whoreported higher levels of involvement at school but not higherinvolvement at home nor higher educational expectations. Thisnding is consistent with established theory and research aboutthe importaily involvemHoover-Dem& Hoover-Dthat increations with and school occur.

The fact school outreducationalrelated to tSpecicallyreach is keyeffect on pof school oaimed to inat school. Ittion was fouat home orthat involvewere less seat school, gance for schindicators.

6.2. Familygains

Consistedents in olof young cated with cDavis-Kean1993). On aschool activoutperformagement freven after cincluding sChen (2001ment had st

On avercational exreading gaito achieveminvolvemeninvolvemenexperiencesour study dinvolvemenment gainswith familyto which alachievemen

6.3. School achievemen

We examto childrenates and fou

families in their childrens education was associated with greaterstudents reading and math gains over the course of kindergarten(hypothesis 3). As Bronfenbrenner and Morris (1998) and others(Epstein, 2001; Rimm-Kaufman & Pianta, 2001) have emphasized,

ual re thross amilir thts edeve

nshipadingpartiievemdingn avg bec

pareuppoouldr acadhe retionmecht leatherwiths wit

con, 200r conr cone abd to lize venademhoolrthernismge fadenird

outreasure thportsonales ims, forch wgatef hamensultsch tos stumen

sho func

to hs suing,

engunitince of school invitations and encouragement of fam-ent as predictors of parental behaviors (Epstein, 2001;psey et al., 2005; Walker, Wilkins, Dallaire, Sandler,empsey, 2005). When schools use planned activitiesse school and teacher communications and connec-their students families, greater overlap of the homecontexts can be facilitated and family involvement can

that we did not nd a signicant relationship betweeneach and family involvement at home or parents

expectations, however, suggests important caveatshe way school outreach predicts family involvement., the results suggest that the nature of school out-

in determining whether or not these efforts have anarticular kinds of parental involvement. The measureutreach in this dataset largely included activities thatcrease parents participation at meetings and events

is not surprising, therefore, that no signicant associa-nd between school outreach and parental involvement

parents educational expectations. It is also plausiblement at home and parents educational expectationsnsitive to school interventions than family involvementiven that we observed a higher between-school vari-ool involvement that for the other family involvement

involvement and math and reading achievement

nt with previous research using national datasets of stu-der grades, as well as studies using smaller sampleshildren, we found that family involvement is associ-hildrens early achievement gains (Catsambis, 2001;, 2005; Dearing et al., 2006; Ho & Willms, 1996; Muller,verage, children whose parents were more involved inities and had higher educational expectations tended to

their peers who did not have this support and encour-om family members. These results were signicantontrolling for inuential student and family covariates,tudents previous achievement. In contrast to Fan and) and Jeynes (2005), we did not nd that home involve-ronger effects than school involvement on learning.age, family involvement at school and parents edu-pectations were associated with students math andns. Involvement at home, however, was not relatedent gains. It is important to note that the home

t measure largely focuses on quantitative indicators oft as a set of basic activities, rather than on the quality of

and interactions that a student has at home. So whileoes suggest that children who experience more familyt at home are not more likely to experience achieve-

in kindergarten than students who interact less often members, our analyses could not determine the extentl family involvement at home is related to childrenst gains in kindergarten.

outreach to parents and students math and readingt gains

ined the extent to which school outreach was relateds achievement gains after controlling for critical covari-nd that principals reports of school outreach to involve

individment aand actheir flies, nostuden

Howrelatioand rewas a on achthis nhave, oreadinwheremore sthen, cgreatetigate tinstructional studen

Anociated schoolschoolTaylorgreategreateat homexposeinternatasks, eand acwhy scthat fumechato engaand stu

A thschoolthe mBecauspals reor perpracticfamilieoutreainvestitages oinvolvethat reoutrea

Thiinvolvetion bypart, aeffortsndingschooltices toopportbehaviors, academic achievement, and child develop-e consequence of the interactions among people withindifferent settings, in this case between students andes and schools. Neither the students, nor their fami-eir schools are exclusively responsible for explainingucational outcomes.r, family involvement did not completely mediate the

between school outreach efforts and childrens math achievement gains. Only family involvement at schoolal mediator of the inuence of school outreach effortsent gains for reading and math. One explanation for

is that schools with stronger reported outreach effortserage, students that make stronger gains in math andause the schools create a more positive school climatents feel more welcomed and students are immersed in artive learning environment. Students in these schools,

be exposed to a more positive instructional climate andemic support for learning. Future studies should inves-lationship between school outreach and classroom and

al climate within classrooms to better understand addi-anisms by which school outreach may indirectly affectrning.

possible explanation for why outreach is directly asso- student achievement in these analyses may be thath more outreach to families have greater family andsensus about appropriate academic behaviors (Hill &4; McNeal, 1999). With more shared information andsensus about students work in school, there should besistency in the messages students hear at school andout the importance of learning and behavior. Studentsmore consistent messages about math and reading maythese attitudes and work harder in school and on school

in kindergarten. This suggests that student motivationic engagement may be mediating factors that explain

outreach translates into student achievement. It is clear research is needed to investigate and validate variouss underlying the relationship between school outreachmilies, families responses that inuence their children,t achievement.possible explanation is that the relationship betweeneach and student achievement is an artifact of howe of school outreach was collected from principals.is measure of school outreach was based on princi-s, they may reect principals perceptions of the school

values about family involvement, rather than actualplemented. Principals in schools with more supportive

instance, may perceive the school as conducting morehen this is not the case. Future studies are needed to

this possibility. We believe, however, that the advan-ving independent measures of school outreach, familyt, and student achievement outweigh the potential bias

from using principal perceptions to measure school families.dy addressed several gaps in the literature on parentt and early education, making an important contribu-wing that childrens early academic achievement is, intion of schools efforts to engage families and parentselp their children achieve and succeed in school. Theseggest that, at least in the beginning years of childrensteachers and administrators should implement prac-age, involve, and inform students families and createes for family involvement in reading-related and math-

C. Galindo, S.B. Sheldon / Early Childhood Research Quarterly 27 (2012) 90 103 101

related activities at school and at home. By establishing strongerrelationships and connections with families, parents are more likelyto get involved at the school and students are more likely to makegreater gains in reading and math skills.

The prestional pracinvolvementhe beginnischools andment at scwith familiment contrachievemen

Also, thitive of cognof involvemwide range the ndingreading andcators needraise or mational attainto college amuch in theneed to go band begin toand how th

6.4. Limitat

Becauseized experimbetween scachievemenbetween keferential seschool outrables that athe nature aously, it is pother schooabout familily involvemminority stof school quin achievemshould strividentify thestudents ac

Future snature of thextent scholevels durinstudy the echildrens conly two wture true g(Rogosa, 19achievemenalso estimathe dependment scorepatterns ofAs a result,

Future sschool outr

family involvement at school and school outreach scales (0.58 and0.64, respectively) were somewhat low suggesting that the internalconsistency of the items was weak. It is possible that the reliabil-ity of the scales would have been stronger with targeted measures

ily in invomainrs wool

pring or ay w

emenntifyeci

ally, adingsore

on aeracgrouemen/racia003

the ilies.relatildrecono

ndch isat themenect siere

izes varia

relticeag and

wled

s resal In

on Sh suograiate tsearc

not

nces

er, K.vior: lopmen, K. Jerstan

3113brenn. Denels of y.is, S. ndaryal Psycent study has implications for educators and educa-tice. The ndings conrm the importance of familyt on students reading and math achievement fromng of childrens school careers. They also suggest that

educators can increase the frequency of family involve-hool by extending invitations to and communicatinges more frequently, and that this form of involve-ibutes to students growth in reading and mathematicst during kindergarten.s study shows that parental expectations are predic-itive gains and achievement in kindergarten. This forment, however, did not appear to be inuenced by theof efforts schools made to engage families. Nevertheless,s that parental expectations for their children predict

math achievement in kindergarten suggests that edu- to maintain a strong relationship with parents and helpintain expectations regarding their childrens educa-ment. Parents who did not envision their children goingppear less likely to have kindergarteners who learn asir rst year of schooling. To make this happen, educatorseyond the low-intensity contacts (Pianta et al., 1999)

have more rich conversations with families about whateir children are doing in school.

ions and future directions

we used survey data and did not conduct a random-ent, we were not able to estimate a causal relationship

hool outreach and family involvement or studentst gains. It is possible that the observed associationsy variables and achievement gains are driven by dif-lection of students into schools with particular levels ofeach. It also may be that this study omitted key vari-re also associated with the outcomes of interest or withnd extent of school outreach. Also, as we stated previ-ossible that the school outreach variable representedl variables such as principals leadership or attitudesy involvement that affect student outcomes and fam-ent. Although we included schools concentration of

udents and aggregate levels of education as indicatorsality and access to resources, and we modeled changesent from fall to spring of the school year, future studiese to directly measure additional variables in order to

independent, unbiased impact of school outreach onhievement and family involvement.tudies should also take advantage of the longitudinale ECLS-K data set by analyzing whether and to whatol outreach and family involvement vary across gradeg the elementary grades. The data also can be used toffects of time-varying family involvement models onognitive growth over time. Because this study analyzedaves of achievement data, we were not able to cap-rowth in student achievement in reading and math95). To conrm that our estimates using models witht in the fall of kindergarten as control were robust, weted analytical models using difference scores, whereent variable was the difference between spring achieve-s and fall scores. Our analyses found similar levels and

statistical signicance to the ndings presented here.our estimates do appear robust and reliable.tudies are needed that include stronger measures ofeach and family involvement. Cronbachs alphas of the

of famschoolpaper teachethe schnot askportinsame winvolvwill ideence sp

Finthe nilies, mbased the intferent involvethnicCraft, 2ings ofof famin the and chsocioe

Theoutreaand thachievthe effables weffect sential that, inhad noreadin

Ackno

ThiNationCenterresearcship prappreclent reand do

Refere

AlexandbehaDeve

Andersound100,

BronfenIn WmodWile

CatsambsecoSocivolvement at school that separated different kinds oflvement. Also, the school outreach measure used in thisly focused on the extent to which principals believedere working to get parents to events or volunteer atand promote teacherparent communication, but didcipals the extent to which school personnel were sup-

encouraging parentchild interactions at home. In thee need research that examines how different forms oft affect different outcomes, we also need studies that

how different forms of school outreach activities inu-c family attitudes and actions.lthough the results are based on a national sample and

are representative of all kindergarteners and their fam-work is needed on this topic. The ndings here areverage effects and should not be taken to mean thattions and inuence processes are equally valid for dif-ps. Several research studies have shown that familyt levels and processes differ across socioeconomic andl sub-groups (Davis-Kean, 2005; Hill et al., 2004; Hill &; McWayne et al., 2008), making it vital that the nd-present study be examined across different sub-groups

Continued analyses are needed to examine variationsionships among family involvement, school outreach,ns achievement, for families from different racial andmic groups, and at different grade levels.ings from this study provide new evidence that school

effective at getting more families involved at schoolese activities are related to students reading and matht gains during kindergarten. Analyses suggested thatzes of the school outreach and family involvement vari-

modest, yet encouraging. By comparing this studysfor key variables to the effect sizes of other highly inu-bles (i.e., parents educational attainment), we foundative terms, family involvement and school outreachble associations with students achievement gains in

math in kindergarten.

gements

earch was partially supported by a grant from thestitute for Child Health and Human Development to thechool, Family, and Community Partnerships. Additionalpport was provided by the Summer Faculty Fellow-m at the University of Maryland, Baltimore County. Wehe thoughtful comments of Joyce Epstein and the excel-h assistance of Amy Pucino. Opinions are the authorsnecessarily reect those of the funding agency.

L., Entwisle, D. R., & Dauber, S. L. (1993). First-grade classroomIts short- and long-term consequences for school performance. Childnt, 64, 801814.., & Minke, K. M. (2007). Parent involvement in education: Toward anding of parents decision making. The Journal of Educational Research,23.

er, U., & Morris, P. A. (1998). The ecology of developmental processes.ton, & R. M. Lerner (Eds.), Handbook of child psychology: Theoreticalhuman development (5th ed., Vol. 1, pp. 9931028). New York, NY:

(2001). Expanding knowledge of parental involvement in childrens education: Connections with high school seniors academic success.hology of Education, 5, 149177.

102 C. Galindo, S.B. Sheldon / Early Childhood Research Quarterly 27 (2012) 90 103

Cheadle, J. E. (2008). Educational investment, family context, and childrens mathand reading growth from kindergarten through third grade. Sociology of Educa-tion, 81, 131.

Crosnoe, R., & Cooper, C. E. (2010). Economically disadvantaged childrenstransitions into elementary school: Linking family processes, school con-texts, and258291.

Dauber, S. L., &in inner-cischools in a

Davis-Kean, Pon child aenvironme

Davis-Kean, P.child achie285318.

Dearing, E., Kment in stions betw653664.

Deslandes, R., &level schoo

Domina, T. (2parental in

Downey, D., VCognitive Sociologica

Enders, C. K., &multilevel121138.

Englund, M. drens achinvolvemechology, 96

Entwisle, D. RBoulder, C

Epstein, J. L. (2schools. Bo

Epstein, J. L., &dent attenEducationa

Epstein, J. L., &family, anSAGE hand(pp. 1171

Faires, J., Nichodeveloping

Fan, X., & Chenment: A m

Fantuzzo, J., Mfamily invofor urban,

Fantuzzo, J. WmultivariaJournal of E

Farkas, G., & Bedge: Diffe

Galindo, C., & Fgrowth in

Ginsburg-Blocfoster earltenson (Edcompetenc

Goldenberg, Cgitudinal stheir child547582.

Goyette, K., & Determina

Green, C. L., Wents motia theoretic99, 53254

Grolnick, W. Sof parent i89, 53854

Grolnick, W. schooling:Developme

Gutman, L. M.twelfth graDevelopme

Henderson, Aschool, famSouthwestsedl.org/co

Hill, N. E., Castellino, D. R., Lansford, J. E., Nowlin, P., Dodge, K. A., Bates, J. E.,et al. (2004). Parent academic involvement as related to school behavior,achievement, and aspirations: Demographic variations across adolescence. ChildDevelopment, 75, 14911509.

Hill, N. E., & Craft, S. A. (2003). Parent-school involvement and school performance:iated Euro-., & Taeveme., & Weveme, & Ho,t on sic groDempins, AimplicW. H.ority c218.. H.

n elem269.

. Z., Keitudindiffer3536. A., K

ilbert,Vol. 1,. (200

ity of J., Sk9). Ho

scho., Rhonting ily, 51

R. B.ess o144.e, C., inati

t famiC. (19ly resnts, thMillerhematent & l Centeergartol lese of E. C., Coed to ol Jouush, Sysis m, S. F.,readin891.aufmae reslts fro421.aufman to knal of aufmaly inven. Ea., & en clagrade

for Edinfo.aD. (19lemewah, ., Brooiness:, P. (20536.ugh, Hlopme educational policy. American Educational Research Journal, 47,

Epstein, J. L. (1993). Parents attitudes and practices of involvementty elementary and middle schools. In N. Chavkin (Ed.), Families and

pluralistic society (pp. 5371). Albany, NY: SUNY Press.. E. (2005). The inuence of parental education and family incomechievement: The direct role of parental expectations and the homent. Journal of Family Psychology, 19, 294304.

E., & Sexton, H. R. (2009). Race difference in parental inuences onvement: Multiple pathways to success. Merrill-Palmer Quarterly, 55,

reider, H., Simpkins, S., & Weiss, H. B. (2006). Family involve-chool and low-income childrens literacy: Longitudinal associa-een and within families. Journal of Educational Psychology, 98,

Bertrand, R. (2005). Motivation of parent involvement in secondary-ling. The Journal of Educational Research, 98, 164175.

005). Leveling the home advantage: Assessing the effectiveness ofvolvement in elementary school. Sociology of Education, 78, 233249.on Hippel, P., & Broh, B. (2004). Are schools the great equalizer?inequality during the summer months and the school year. Americanl Review, 69, 613635.

Toghi, D. (2007). Centering predictor variables in cross-sectional models: A new look at an old issue. Psychological Methods, 12(2),

M., Luckner, A. E., Whaley, G. J. L., & Egeland, B. (2004). Chil-ievement in early elementary school: Longitudinal effects of parentalnt, expectations, and quality of assistance. Journal of Educational Psy-, 723730., Alexander, K. L., & Olson, L. (1997). Children, schools, & inequality.O: Westview Press.001). School and family partnerships: Preparing educators and improvingulder, CO: Westview Press.

Sheldon, S. B. (2002). Present and accounted for: Improving stu-dance through family and community involvement. The Journal ofl Research, 95, 308318.

Sheldon, S. B. (2006). Moving forward: Ideas for research on school,d community partnerships. In C. F. Conrad, & R. C. Serlin (Eds.), Thebook for research in education: Engaging ideas and enriching inquiry38). Thousand Oaks, CA: Sage Publications, Inc.ls, W. D., & Rickelman, R. J. (2000). Effects of parental involvement in

competent readers in rst grade. Reading Psychology, 21(3), 195215., M. (2001). Parental involvement and students academic achieve-eta-analysis. Educational Psychology Review, 13, 122.cWayne, C., Perry, M. A., & Childs, S. (2004). Multiple dimensions oflvement and their relations to behavioral and learning competencieslow-income children. School Psychology Review, 33, 467480.., Tighe, E., & Childs, S. (2000). Family involvement questionnaire: Ate assessment of family participation in early childhood education.ducational Psychology, 92, 367376.

eron, K. (2004). The detailed age trajectory of oral vocabulary knowl-rences by class and race. Social Science Research, 33, 464497.uller, B. (2010). The social competence of Latino kindergartners and

mathematical understanding. Developmental Psychology, 46, 579592.k, M., Manz, P. H., & McWayne, C. (2009). Partnering with families toy achievement in reading and mathematics. In A. L. Reschly, & S. Chris-s.), The handbook on school family partnerships for promoting studente (pp. 176203). Oxford, UK: Routledge/Taylor and Francis Group.., Gallimore, R., Reese, L., & Garnier, H. (2001). Cause or effect? A lon-tudy of immigrant Latino parents aspirations and expectations, andrens school performance. American Educational Research Journal, 38,

Xie, Y. (1999). Educational expectations of Asian American youths:nts and ethnic differences. Sociology of Education, 71, 2438.alker, J. M. T., Hoover-Dempsey, K. V., & Sandler, H. M. (2007). Par-vation for involvement in childrens education: An empirical test ofal model of parental involvement. Journal of Educational Psychology,4.., Benjet, C., Kurowski, C. O., & Apostoleris, N. H. (1997). Predictorsnvolvement in childrens schooling. Journal of Educational Psychology,8.

S., & Slowiaczek, M. L. (1994). Parents involvement in childrens A multidimensional conceptualization and motivational model. Childnt, 65, 237252., Sameroff, A. J., & Cole, R. (2003). Academic trajectories from rst todes: Growth curves according to multiple risk and early child factors.ntal Psychology, 39, 777790.. T., & Mapp, K. L. (2002). A new wave of evidence: The impact ofily, and community connections on student achievement. Austin, TX:

Educational Development Laboratory. Retrieved from http://www.nnections/resources/evidence.pdf

Medand

Hill, N. Eachi

Ho, E. Sachi

Hong, S.menethn

Hoover-Wilkand

Jeynes, min202

Jeynes, Wurba237

Keith, TLongand 36, 3

Kenny, DD. Ged.,

Lareau, Avers

LeFevre,(200early

Luster, TpareFam

McNeal,tiven117

McWaynexamStar

Muller, famiPare

Musun-matopm

NationakindschoOfc

Pianta, RrelatScho

Raudenbanal

Reardonand 853

Rimm-Kof thResu401

Rimm-KsitioJour

Rimm-Kfamigart

Rock, Dgartrst ter pubs

Rogosa, suppMah

Rouse, Cread

Royston527

ScarboroDevepathways among socioeconomically comparable African AmericanAmerican families. Journal of Educational Psychology, 95, 7483.ylor, L. C. (2004). Parental school involvement and childrens academicnt. American Psychology Society, 13(4), 161164.illms, J. D. (1996). Effects of parental involvement on eighth-gradent. Sociology of Education, 69, 126141.

H. (2005). Direct and indirect longitudinal effects of parental involve-tudent achievement: Second-order latent growth modeling acrossups. Journal of Educational Psychology, 97, 3242.sey, K. V., Walker, J. M. T., Sandler, H. M., Whetsel, D., Green, C. L.,. S., et al. (2005). Why do parents become involved? Research ndingsations. The Elementary School Jou