UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Influence of family and child characteristics on the outcomes of a parent rated developmental screening… Riesch, Lindsay 2007

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2007-0569.pdf [ 3.62MB ]
Metadata
JSON: 831-1.0054354.json
JSON-LD: 831-1.0054354-ld.json
RDF/XML (Pretty): 831-1.0054354-rdf.xml
RDF/JSON: 831-1.0054354-rdf.json
Turtle: 831-1.0054354-turtle.txt
N-Triples: 831-1.0054354-rdf-ntriples.txt
Original Record: 831-1.0054354-source.json
Full Text
831-1.0054354-fulltext.txt
Citation
831-1.0054354.ris

Full Text

I N F L U E N C E O F F A M I L Y A N D C H I L D C H A R A C T E R I S T I C S O N T H E O U T C O M E S O F A P A R E N T R A T E D D E V E L O P M E N T A L S C R E E N I N G I N S T R U M E N T by L I N D S A Y R 1 E S C H B . A . The Univers i ty o f W i s c o n s i n - M a d i s o n , 2001 A T H E S I S S U B M I T T E D I N P A R T I A L F U L F I L L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F M A S T E R O F A R T S in The Facul ty O f Graduate Studies (School Psycho logy) T H E U N I V E R S I T Y O F B R I T I S H C O L U M B I A J u l y 2007 © L indsay R ie sch , 2007 11 A B S T R A C T The development o f ch i ldren is an important issue for school psychologists , researchers, healthcare practitioners, educators, and parents. Methods wi th demonstrated efficacy to screen for developmental delays in chi ldren are necessary because early detection o f delays can begin the process o f des igning interventions, educating parents, and preparing young ch i ld ren for school . Parents p lay a major role in the detection process. M a n y research studies report that parents are accurate at report ing observable ch i ld development; however, sensi t ivi ty on screening tools is s t i l l modest for most instruments. In this study the role o f f ami ly and c h i l d characteristics on agreement between the N i p p i s i n g Dis t r ic t Deve lopmenta l Screen ( N D D S ) and developmental assessment using the Stanford B ine t Intelligence Scales, 5 l h edi t ion ( S B 5 ) , the Peabody Picture V o c a b u l a r y Test, 3 r d edition ( P P V T - I I I ) , and the B a y l e y Scales o f Infant Development , 2 n d edi t ion (BS1D-2) was investigated. Pearson Chi-squared statistics were used to test hypothesized associations between the explanatory variables (ch i ld and f a m i l y characteristics) and results from the parent-completed N D D S . Var iables found to have significant associations were then analyzed using binary logist ic regression analyses to determine w h i c h ch i ld and f ami ly characteristics predicted agreement between the N D D S and the developmental assessment. The hypotheses were part ial ly supported. The results f rom the bivariate analyses support hypotheses for ch i ld gender and age, as w e l l as f ami ly income and maternal age for ch i ldren without delays (True Negat ives) , but not for chi ldren w i t h delays (True Posit ives. These f indings indicate the sensitivity and specif ic i ty rates for the N D D S vary largely by c h i l d gender and age, but that some fami ly characteristics (specif ical ly , f ami ly income and maternal age) are also associated wi th the l i ke l ihood o f ach iev ing True Posi t ive and True Negat ive results Ill T A B L E OF CONTENTS Abstract .' ii Table of Contents... iii List of Tables iv List of Figures v Acknowledgements v l CHAPTER ONE Introduction 1 CHAPTER TWO Review of the Literature 6 CHAPTER THREE Methodology 27 CHAPTER FOUR Results 38 CHAPTER FIVE Discussion 62 References • 79 IV L I S T O F T A B L E S 39 Table 4.1 • 40 Table 4.2 • ...42 Table 4.3 • 43 Table 4.4 45 Table 4.5 : Table 4.6 4 6 49 Table 4.7 • 50 Table 4.8 53 Table 4.9 Table 4.10 5 5 Table 4.11 5 5 Table 4.12 " 5 7 Table 4.13 : 5 7 L I S T O F F I G U R E S Figure 1.1 vi A C K N O W L E D G E M E N T S I am thankful for the support I was g iven by m y advis ing professor and research supervisor D r . Laur ie Fo rd . Laurie encouraged me to enter the Schoo l P s y c h o l o g y program and has been a mentor and professor throughout the three years o f study. L a u r i e ' s tireless commitment to teaching the assessment o f ch i ld ren w i l l no doubt help me be an ethical and informed school psychologist in the future. Laur i e also gave me the opportunity to participate in the research study that led to this thesis. W o r k i n g wi th Dr. Susan Dahinten, whose log ic , analytic sk i l l s , and patience were invaluable to the complet ion o f this thesis has been cha l lenging and rewarding. B o t h Laur ie and Susan have been enthusiastic in encouraging m y research interests. I am also thankful for Dr . C o n n i e C a n a m ' s encouragement and thoughtful feedback throughout this process. Las t ly , Dr . Shel ley H y m e l has been a wonderful professor for several o f my courses in this Master ' s program and 1 am thankful for her wi l l ingness to be a part o f m y committee and m y preparation as a schoo l psychologist . I am appreciative o f the ch i ldren and famil ies who participated in the val idat ion studies that provided the data for this study. The chi ldren and their families were the source o f enjoyable interactions and confi rmed m y interest i n this f ield. M y involvement in the developmental assessments w i l l make the l i n k between research and practice clearer. I am grateful to al l o f the people I have met i n the School P s y c h o l o g y Program at the i Univer s i ty o f B r i t i s h C o l u m b i a . I w i l l have l i fe long friends and colleagues because o f this experience. Las t ly , I am thankful for the never ending support 1 have received from both m y mother and boyfr iend throughout m y graduate school career and the wr i t i ng o f this thesis. 1 Chapter One: Introduction The development o f chi ldren is an important issue for school psychologis ts , researchers, healthcare practitioners, educators, and parents. Methods wi th demonstrated efficacy to screen for developmental delays i n ch i ldren are necessary because early detection o f delays can begin the process o f designing interventions, educating parents, and preparing young chi ldren for school . R i sk factors for developmental delays are environmental and b io log ica l (Glascoe , 2005). Studies demonstrate that detecting developmental delay early is beneficial to not on ly the c h i l d and family , but also to entire communi t ies (Bai ley , Skinner , & Warren, 2005; Glascoe , 2001). Screening for developmental delays has taken many forms. Intricate instruments that require experienced professionals are typ ica l ly reserved for chi ldren who have been identif ied as at r isk for delay, whereas parent-completed developmental screening instruments have gained acceptance, and use, over the last decade to identify chi ldren at r isk for delays. Developmenta l screening by professionals and parents requires accuracy to identify chi ldren who are in need o f services. Thus, the focus o f this study is the accuracy o f parent reporting on developmental screening instruments. Parents play a major role i n the developmental delay detection process. Research provides evidence that parents can be accurate when asked to report observable ch i ld behavior (Glascoe, 2001). Parents spend valuable t ime wi th their chi ldren and they share the c o m m o n goal o f healthy ch i ld development w i t h healthcare practitioners (Commit tee on C h i l d r e n W i t h Disab i l i t i es , 2006). Healthcare practitioners see chi ldren routinely for immuniza t ion administration, w e l l - c h i l d vis i ts , and i l lness care. Therefore, p rov id ing healthcare practitioners w i t h an efficient instrument that can be completed by parents has become a powerful technique for detecting developmental delay early. 2 R e l y i n g on parents as informants makes it possible for healthcare providers to spend time and money investigating the development o f chi ldren who may need addi t ional assessment and fol low-up. Because parents are valuable to the developmental screening process, it is important to understand factors that may influence accuracy o f their reporting. Several ideas have been suggested in the literature. For example, e l ic i t ing parental concerns (Glascoe, 1999) and asking parents specific questions about c h i l d behaviors in contrast to general questions about development may enhance, accuracy as does the use o f instruments wi th demonstrated sensit ivi ty and specifici ty. There is some evidence that parent and ch i ld characteristics may influence parents' assessment o f appropriate developmental progress in their ch i ld . Because parents p lay such a valuable role i n developmental screening, it is important to examine what parental and chi ld characteristics influence accuracy o f their observations as measured by developmental screening instruments. A c c o r d i n g to Glascoe (1999; 1997; 1990), parents are accurate in perce iv ing developmental delays i n their chi ldren regardless o f parent education level or experience wi th mult iple chi ldren. Howeve r , other researchers do not agree. Lagerberg (2005) points out that, "parents are good at assessing c h i l d characteristics - at least some parents, sometimes, and i n some respects" (p. 1007). A l t h o u g h research by Glascoe has been w i d e l y accepted, her studies main ly investigate the accuracy o f parent report on her o w n developmental instrument, the Parent 's Eva lua t ion o f Deve lopmenta l Status ( P E D S ) . In addit ion, caution should be taken when discussing sensit ivi ty and specif ici ty o f an instrument i n the context o f parent reporting on ch i ld development because even a tool w i th good sensit ivi ty and specif ici ty may not accurately predict delays (Lagerberg, 2005). 3 Quest ioning the accuracy o f parent reporting is log ica l when the screening instrument has l imited psychometric properties. However , instruments wi th relat ively strong psychometr ic properties that rely on parent report may y i e l d false negative and false posi t ive screens.-The literature to date ove rwhe lming ly supports parent reports o f development; however it is not clear whether the research really focused on the appropriate aspects o f parent reporting. In this study parent and c h i l d characteristics o f their accuracy on parent reported screening instruments is investigated Purpose of the Study The purpose o f this study is to examine f ami ly and ch i ld characteristics that may influence agreement between parent report o f developmental delay using the N i p p i s i n g Dis t r ic t Developmental Screen ( N D D S ) and in-depth assessments o f ch i ld development by a professional. Data was taken from a normative, community-based sample o f chi ldren and their parents from a val idat ion study o f the N D D S . 1. T o what extent do ch i ld characteristics (gender, age, and health status) predict agreement between parent reports on the N D D S and developmental assessments o f the ch i ld us ing the Stanford-Binet: F i f th E d i t i o n ( S B 5 ) , the Peabody Picture Vocabu la ry Test, T h i r d Ed i t ion (PPVT-I I1) , and the B a y l e y Scales o f Infant Development ; Second E d i t i o n ( B S I D - 2 ) ? 2. T o what extent do fami ly characteristics (maternal educational, f ami ly income, fami ly structure, and maternal age) predict agreements between parent report on the N D D S and developmental assessment o f the c h i l d us ing the S B 5 , the P P V T - I I I and the B S I D - 2 ? 4 3. What set o f ch i ld and fami ly characteristics best accounts for agreement between parent report on the N D D S and developmental assessments o f the c h i l d us ing the S B 5 , the P P V T - J J 1 , and the B S I D - 2 ? Definitions of Key Terms Developmental Screening is defined i n the literature and in the present study as a process o f moni tor ing children to identify those w h o are in need o f further evaluation Glascoe , 1999; 1990; H a m i l t o n , 2006; Lee & Harr i s , 2005). Universal Screening as defined in the literature and in the present study is a form o f developmental screening that measures mul t ip le areas such as cogni t ive abi l i ty , language, motor, and behaviour, and is used for early detection o f developmental delays for populations o f chi ldren (Hami l ton , 2006). Targeted Screening as defined i n the literature and used in the present study, also a form o f developmental screening, aims to identify ch i ldren w h o are high risk for developmental delay, and may target specific groups o f chi ldren who are at higher risk for delays (Lee & Harris , 2005). In addit ion, targeted screening cou ld also be used as a term to investigate one specific domain o f development more in-depth. Developmental Assessment in this study is an in depth evaluation o f chi ldren 's abilities that typ ica l ly fo l lows developmental screening. M a n y times chi ldren are referred for developmental assessment after rece iv ing a posi t ive screen for potential developmental delays. Developmental assessments are usual ly conducted by professionals w i th graduate training in the area (Glascoe, 1999; H a m i l t o n , 2006; Lee & Harr i s , 2005). In this study w i d e l y used and w e l l validated developmental assessment tools inc lud ing the Stanford-Binet Intelligence Scales, Fifth 5 Edi t i on , Peabody Picture Vocabu la ry Test- T h i r d E d i t i o n and the B a y l e y Scales o f Infant Development -Second Ed i t ion were used. Developmental Delay is a term that is subjective according to who is using the term, and what is being measured (Glascoe, 1990; 1999). In addit ion, the k i n d o f delay is significant. Developmenta l delays can be global , meaning an overa l l cogni t ive delay incorporat ing the fo l l owing delays: a language delay, a delay in motor sk i l l s , or a delay in behaviour. Fo r the purpose o f this study a developmental delay is considered a standard score o f one standard deviation below the mean on the cogni t ive measures, and one flag on the screening measures. Parent characteristics are psychosocia l descriptive traits o f the mother that may influence her observations, perceptions, or abili t ies to be accurate at reporting her ch i ld ' s developmental behavior. In the present study, parent characteristics are defined as maternal education, fami ly income, fami ly structure, and maternal age. Child characteristics are descriptive traits that pr ior research has suggested may influence a parent's observations, perceptions, or abili t ies to be accurate in reporting ch i ld developmental behavior. In the present study, parent characteristics are defined as age, gender, and health status. Benefits of the Study The intended benefits o f this study are to better understand the influence o f f ami ly and c h i l d characteristics on a parents' ab i l i ty to report and detect developmental delays in their chi ldren. It is expected this study w i l l inform research w i t h i n the field o f developmental screening to take parent and c h i l d characteristics into account when deve lop ing and using parent report screening instruments in the future. 6 Chapter T w o : R e v i e w o f the Literature The purpose o f this chapter is to rev iew the current research on agreement between parents and professionals on developmental screening parameters. Parent-completed developmental screens w i l l be cri t iqued and parent and ch i ld characteristics that influence agreement w i l l be reviewed. This chapter is organized around the f o l l o w i n g concepts: (a) definitions and categories o f developmental screening, the screening instruments currently in use, and the role o f parents in developmental screening; (b) parent characteristics that may influence parent observations o f their ch i ld ; and (c) the ch i ld characteristics that m a y influence parental observations o f their ch i ld . Developmental Screening Developmental screening plays an important role o f early ch i ldhood professionals inc luding psychologists , early educators, nurses and others. Often on ly those chi ldren who are considered to be at h igh r isk for developing delays are screened on a regular basis. Ep idemio log i ca l evidence, however, shows that what determines developmental delays is complex and diff icul t to detect at the ind iv idua l l eve l (Hami l ton , 2006) and thus there is a need for more widespread screening. Th is need has led to increased interest in universal , fami ly , and community-based developmental screening throughout the early ch i ldhood years and an increased demand for screening instruments that are re la t ively inexpensive and easi ly administered. Screening instruments in current use. T w o instruments are used w i d e l y across Nor th A m e r i c a in a universal approach to developmental screening. These include the A g e s and Stages Questionnaire ( A S Q , Squires & Br icker , 1999), and Parent 's Eva lua t ion o f Developmenta l Status 7 ( P E D S , Glascoe , 1999). A third, N i p p i s i n g Dis t r ic t Developmental Screen ( N D D S , www.ndds .ca retrieved A p r i l 29, 2007) has gained widespread use across Canada and is the focus o f the present study. Un ive r sa l developmental screening instruments, w h i c h are those that measure mul t ip le areas o f concern, typical ly are i nd iv idua l ly administered and norm referenced as are other ind iv idua l ly administered measures o f abil i t ies such as intelligence tests or socia l -emotional-behavioral rating scales. They also incorporate the psychometr ic properties. Examples o f re l iabi l i ty are demonstrated through test-retest, inter-rater, and internal consistency. V a l i d i t y evidence is demonstrated through concurrent, discr iminate, and sometimes predictive studies (Glacoe, 2005; Hami l t on , 2006; Lee & Har r i s 2005). In recent reviews o f current universa l developmental screening instruments, Glascoe (2005), Lagerberg (2005), and The Commi t t ee on Ch i ld r en W i t h Disab i l i t i e s (2001) emphasized that a good screening instrument is one that is accurate, meaning it has sufficient sensit ivity and specif ici ty ( typica l ly 70-80%). Sens i t iv i ty is defined as the probabi l i ty that the screening instrument w i l l correctly identify ch i ld ren w h o show signs o f developmental delays. Speci f ic i ty is defined as the probabi l i ty that the screening instrument w i l l detect ch i ldren who are developing normal ly . Screening instruments are not "error free" (Glascoe, 2005, p. 174), yet h igh levels o f accuracy can be achieved. Lagerberg (2005), however, goes on to say that screening instruments that do not have predict ive va l id i ty do not give accurate information about a ch i ld ' s development. "It can be conc luded that effective tests and screening devices with strong predict ive va l id i ty do not exist at present" (p. 1007). The ASQ was created by Dianne B r i c k e r and Jane Squires f rom the Unive r s i ty o f Oregon in 1980(ht tp: / /www.brookespublishing.com/tools/asq/ index.htm retrieved on A p r i l 3, 2007). The 8 measure was revised in 1994 and currently contains 19 questionnaires for chi ldren ages 4-60 months. Scor ing the A S Q takes between 1-5 minutes and can be scored by paraprofessionals or staff trained by professionals. Studies demonstrate that re l iabi l i ty is consistently high (rel iabi l i ty greater that .80, sensitivity ranging from 5 1 % at the 4-month age to 84% at the 12-month age, and specif ic i ty ranging from 83 .9% at the 4-month to 100% at the 60-month age) (Br icker & Squires, 1998). The A S Q was val idated using the B a y l e y Scales o f Infant Development , the Stanford-Binet Intelligence Scale, and the M c C a r t h y Scales o f C h i l d r e n ' s Ab i l i t i e s (Squires & B r i c k e r , 1999). It measures gross motor sk i l l s , fine motor sk i l l s , communica t ion , cogni t ion, and behavior (Glascoe, 2006; Lee & Harr is 2005). Examples o f items at the 36-month-old level include gross motor items such as hopping and standing on one foot for a length o f time and fine motor items such as drawing and putting puzzle pieces together (Squires & Br icke r , 1999). The P E D S was developed by Francis Page Glascoe and publ ished in 1997. This screen is unique i n that it is based on ly on parent reports o f concerns (Hami l ton , 2006). It contains eight yes/no questions and two open-ended questions. The purpose o f the P E D S is to "identify when a c h i l d should be referred, provided a second screen, counseled, or moni tored for development, behavior , and academic progress" (p 30). It was wri t ten at a fifth grade reading level , takes approximate ly five minutes, and can be completed by parents w h i l e wa i t ing for a healthcare provider appointment (Glascoe, 2005). Scor ing the P E D S is s imple and categorizes risk as low, m e d i u m , or h igh for the development o f a delay. The P E D S psychometr ic properties fall w i th in the no rma l range for current screening instruments. The sensi t ivi ty for a l l ages combined was 7 5 % and the specifici ty for a l l ages combined was 74%. N o predict ive va l id i ty studies have been completed to date. The P E D S concurrent va l id i ty was evaluated through studies compar ing it w i t h the Woodcock-Johnson Psychoeducat ional Bat te ry-Revised : Tests o f Achievement , 9 Stanford-Binet Intelligence Scale, Four th E d i t i o n , and the B a y l e y Scales o f Infant Development-Second E d i t i o n (Hami l ton , 2006). A third c o m m o n l y used instrument in Canada is the N i p p i s i n g Dis t r ic t Developmental Screen ( N D D S ) . The N D D S w i l l be described in more depth in the Chapter Three o f this document. The N D D S is a parent-completed measure, w i th 13 versions, ranging from one month to s ix years (http:/ /www.ndds.ca retrieved A p r i l 3 2007). It assesses seven major developmental domains: v i s i o n , hearing, speech-language, gross motor, fine motor, cogni t ive , and self-help sk i l l s . The N D D S is a yes/no checklis t , the need for referral is indicated when the selection o f one or more ' n o ' responses (i.e., the behavior in question has not been observed) is specified. I f the health professional makes a referral based on one "no" answer, it is k n o w n as the one-flag rule. A more conservative flag rule is ca l led the two-f lag rule, w h i c h requires more than one " n o " response on the N D D S to refer the ch i ld . A l t h o u g h the N D D S has become a popular screen to use in 'many health care facilit ies across Canada, there have not been many evaluation studies on the N D D S . It was or ig ina l ly developed i n Ontar io in 1993. One study that d id attempt to evaluate the N D D S was the Ontar io Hea l thy Babies /Hea l thy C h i l d r e n ( H B H C ) evaluation, w h i c h eventual ly changed the scor ing rules o f the screen (Dahinten & F o r d , 2004). V a l i d a t i o n w o r k by Dah in ten and Ford suggests that the sensi t ivi ty and specif ici ty o f the N D D S varies by the severi ty o f the delay being targeted, the scor ing rule used, the age version, ch i ld ' s gender, and the developmental domain o f interest. F o r example , for infants and toddlers, they found sensit ivity rates ranging from 5 0 % (when using the two-f lag rule and cut-off o f < 1.0 S D below the mean on the cr i ter ion measure) to 100% w h e n the N D D S was used to identify those who were severely delayed (i.e., < 2.0 S D ) (Dahinten, F o r d , Canam, Lapointe , & M e r k e l , 2007). Sens i t iv i ty rates were lower for 36-month 10 olds, and there were gender differences, wi th the instrument showing lower sensit ivi ty for females than for males. The advantage o f the N D D S when compared to the A S Q and the P E D S is that it requires a short administrat ion time, shorter than the A S Q , w h i l e s t i l l capturing a majority o f domains important to early c h i l d development. A s discussed above, there are l imi ted va l ida t ion studies o f the N D D S . H o w e v e r , it was created in Canada to address Canadian health care needs for a developmental screening instrument unlike most o f the other developmental screening instruments that were developed in the Uni ted States and brought to Canada. The N D D S also asks more specific developmental information than the P E D S , thereby attempting to identify areas o f learning and development that may signal potential delays, but that may not be highl ighted in parental reports. In addit ion, the N D D S has an accompanying parent handout that includes age-appropriate activities for parents to engage i n w i t h their chi ldren , as a means o f increasing parent education about ch i ld development and learning (ht tp: / /www.ndds.ca retrieved A p r i l 3 2007). The A S Q , P E D S , and N D D S are three developmental screens that possess the characteristics recommended by Glascoe as useful for ident i fying chi ldren who may be at risk for developmental delay. N e x t , what is known about the parent's role in developmental screening w i l l be examined. The parent's role in developmental screening. A l t h o u g h sensi t ivi ty rates i n the 7 0 % range have been w i d e l y established for several developmental screening instruments, screening tests are s t i l l not w i d e l y used (Ba i ley , Skinner & Warren , 2005; H a m i l t o n , 2006) by practitioners who see and assess chi ldren on a regular basis. In the past, parents were exc luded from the screening process because health care workers considered parents biased. Healthcare workers 11 d id not believe parents could recall and report an accurate representation o f their ch i ld ' s development due to social desirabil i ty or other factors that may bias a parents ' opin ion about development (Henderson & Meise l s , 1994). A c c o r d i n g to H a m i l t o n (2006), 35-45% o f parents po l led by the Nat iona l Survey o f Ea r ly C h i l d h o o d Health d id not recal l ever having their pediatr ician inquire about developmental delays. Research has shown that re ly ing on parental concerns, ma in ly by asking parents to complete a screening measure, may save time and money. Parents tend to be more reliable when responding to specific items than when responding to general inquiries about developmental milestones or c l i n i ca l judgment in a visi t (Glascoe, Foster, & W o l r a i c h , 1997; Hami l t on , 2006). A s expla ined by Glascoe and D w o r k i n (1995), tools that rely on parental descriptions o f current achievements can provide important information wi th moderate re l iab i l i ty and val id i ty . In a study investigating how parents appraise their ch i l d ' s development, Glascoe and M a c L e a n (1990) investigated whether parents could be accurate w i t h a modest amount o f knowledge about c h i l d development. In a sample o f 100 parent-chi ld dyads from predominately urban/suburban neighborhoods, parents were asked to complete the P E D S to assess their ch i ld ' s development, than they were g iven a questionnaire to record demographic data, and asked questions about " information thought to effect their perceptions o f their ch i l d ' s development" (p. 281). Glascoe and M a c L e a n chose this populat ion because the ch i ld ren were considered overrepresented for risk o f developmental delays (e.g., l ow S E S and premature infants). Results indicated that the parents who had posit ive appraisals o f their o w n ch i ld ren (no concerns) compared their chi ldren in a posi t ive w a y to other chi ldren, and based their appraisals on experience, literature about development, or visi ts w i th health care professionals. Parents w i th concerns about their ch i ld ' s development also engaged in compar i son but, interestingly, most compared their c h i l d unfavorably to other chi ldren. . Th i s study investigated parents concerns about ch i ld development, but d id not investigate whether parents were accurate at ident ifying developmental delays, much less whether parent characteristics influenced their abi l i ty to identify delays. The Glascoe and M a c L e a n study provides evidence that parents engage in comparisons to better understand and conf i rm their beliefs about h o w their ch i ld is developing. In support o f the Glascoe and M a c L e a n study, D ichde l t e lmi l l e r et a l . (1992) conducted a study to investigate the relationship between parental experience and knowledge and the development o f l ow birth weight infants. In a study o f 40 preterm infants and their mothers, the infants were assessed for cognit ive, motor, social emot ional , neurologica l , and medical problems. Mothers were asked to complete the K n o w l e d g e o f Infant Deve lopment Inventory ( K I D I ) that measures parent knowledge o f ch i ld rearing practices, developmental processes, and infant norms. T h e y were also asked to complete the Cata log o f Prev ious Exper ience wi th Infants ( C O P E ) that measures a parent's exposure to babies and their overa l l op in ion regarding how much they think they k n o w about babies. The i r findings indicated that there was little difference i n parent's knowledge o f infant development between parents o f preterm infants and parents o f ful l term infants. Furthermore, their results support Glascoe and M a c L e a n ' s (1990) findings that many parents re ly on socio-cultural indicators to tell them i f their c h i l d is not developing normal ly . That is, they compare against other chi ldren i n the commun i ty to assess their ch i ld ' s development. Therefore, it seems parents w i th l imi ted knowledge o f c h i l d development, those who have h igh risk chi ldren, or those come from high r isk populations can recognize atypical development when they compare their chi ldren against others either favorably or unfavorably. Literature supports parents as part o f the developmental screening process, but neither o f these studies addresses characteristics that may influence their accuracy on reporting." Henderson and M e i s e l s (1994) also investigated parental input i n the screening process. They argued that developmental screening tools must address the mul t ip le risk factors that influence development o f a ch i ld , as chi ldren are exposed to mul t ip le risk factors inf luencing their development. They hypothesized that parents ho ld the key to understanding mul t ip le risk factors. T h e y studied the predictive value o f a developmental screening instrument when a parent questionnaire was added to the assessment. They hypothesized that "by combin ing parental input w i t h direct assessment it w o u l d be possible to improve accuracy o f the developmental screening process and to reduce the l i ke l ihood o f classif icat ion errors that are associated wi th single sources o f screening data" (p. 143). The study used the E a r l y Screening Inventory (ESI ; M e i s e l s et a l . , 1992), w h i c h is completed by a professional and not by parents, either alone or a long wi th a parent questionnaire developed by the researchers. T h e y then compared both results w i th the results from the M c C a r t h y Scales o f C h i l d r e n ' s A b i l i t i e s ( M S C A ; M c C a r t h y , 1972). A l t h o u g h this study was conducted solely w i t h parents o f preschool aged chi ldren, their results showed that adding a parent questionnaire d i d help decrease false posi t ive results on the ESI . Results from this study showed that a parent questionnaire helped guide practitioners to more accurate screening wi th the E S I . Howeve r , this study does not provide evidence, that characteristics o f parents have influence over the accuracy, just that taking parent information about their ch i ld ' s development helped researchers wi th accuracy on the screening instrument. Therefore, it m a y be that developmental screening instalments are inherently inaccurate, wh i l e parents m a y be more accurate than once thought. 14 Parent reporting of development. Parents can play an important role i n the early screening and detection o f developmental delays for their chi ldren (Lagerberg, 2005). There is evidence that parents who are assisted in parent-completed screening tools are successful in identifying developmental delays (Glascoe, 1999; Glascoe & D w o r k i n , 1995; M o n t g o m e r y , 1999; Squires, 1996). Investigations examin ing parents' abili t ies to correctly identify developmental problems in their chi ldren have shown that parental concerns regarding fine motor sk i l l s , overa l l functioning, or speech and language development tend to more accurately predict actual delays than when concerns centered around social and self-help sk i l l s or gross motor development (Glascoe & D w o r k i n , 1995). Parent reports can be completed in a number o f different settings (e.g., over the telephone, i n the wa i t i ng room, during an interview wi th the pediatrician, etc.), and have been demonstrated to be h igh ly accurate when parents are asked to report on observable ch i ld behavior in a systematic, structured format (Glascoe & D w o r k i n , 1995; R y d z , Sheve l l , Majnemer , & O s k o u i , 2005; Squires, Potter, B r i c k e r , & Lamorey , 1998). In addi t ion, re l iabi l i ty • and va l id i ty o f parent-screening questionnaires are highest when questions are w e l l structured and w e l l worded, detailed, pertain to specific behaviors, and focus on descriptions o f current ch i ld achievements (Glascoe & D w o r k i n , 1995; R y d z et al . 2005). Glascoe (1997) found that the discrepancy between accurate and inaccurate results on the P E D S was the result o f parent language barriers. In other words , parents for w h o m Eng l i sh was their second language were not as accurate as parents for w h o m Eng l i sh was their first language. In addit ion, a different k i n d o f language barrier has also hindered the screening process. T e r m i n o l o g y differences concerning ch i ld development between parents and health care can be very different (Glascoe , 1999). Summar iz ing studies about parent knowledge o f development 15 and health care provider abi l i ty to elici t this parental knowledge, Glascoe conc luded that the questions were more o f a problem than the parents who were answering them. Parents may not have the language needed to describe their concern to the healthcare provider . The disconnect in language may account for some discrepancies found between healthcare provider results and parent results on developmental screening tools. Parents may need assistance w i t h comple t ing the screening tool because the language used may not be famil iar to them. P r o v i d i n g parents wi th a screening tool that elici ts observable c h i l d behaviors may decrease the l i k e l i h o o d that terminology w i l l hinder the process (1999). Parent Characteristics The B i o - e c o l o g i c a l M o d e l (Bronfenbrenner & M o r r i s , 1998) provides a framework for the selection o f f ami ly and c h i l d characteristics that may potential ly influence parent accuracy and agreement w i t h developmental screening instruments. The B i o - e c o l o g i c a l M o d e l includes two defining proposi t ions relevant to this study. The first is that human development takes place through processes o f progressively more complex reciprocal interaction between an active, evo lv ing b io -psycho log ica l human organism and the persons, objects, and symbols in its immediate external environment. Interaction must occur on a fair ly regular basis over extended periods o f t ime w i t h one or more persons w i t h w h o m the c h i l d develops a strong, mutual , irrational attachment, and who are commit ted to that ch i ld ' s development, preferably for l ife. T o do this, the ava i lab i l i ty and active involvement o f another adult who assists, encourages, spells off, gives status to, and expresses admirat ion and affection for the person car ing for and engaging in jo in t act ivi ty wi th the ch i ld (quali ty o f relationships in the fami ly) is necessary. Endur ing interactions are the pr imary engines o f development. Th is propos i t ion guides the selection o f f ami ly structure. The parent w o u l d provide the active involvement o f another adult. 16 The second proposi t ion dictates that the form, content, and direct ion o f the processes affecting development may vary systematical ly as a joint function o f the characteristics o f the developing person; o f the environment; o f the nature o f the developmental "outcomes under consideration; and o f the social continuities and changes occur r ing over t ime through the life course. Th is second proposi t ion guides the selection o f ch i ld characteristics part icularly health, age, and gender and the fami ly characteristics o f age, education, and income. A l t h o u g h Bronfenbrenner and M o r r i s (1998) articulate s ix b io -psycholog ica l constructs to consider in their mode l , three o f them are directly pertinent to this study: (a) f ami ly resources that constitute b io -psycholog ica l l iabi l i t ies and assets such as maternal education, fami ly income, partner support, and ch i ld health; (b) maternal age; (c) and ch i ld gender (See F igure 1). A g a i n , this model is merely to help decide w h i c h characteristics o f parents and chi ldren are important to investigate when quest ioning the accuracy o f parent reporting o f c h i l d development. Figure 1.1: B i o - E c o l o g i c a l M o d e l G u i d i n g Study Concepts (adapted from Bronfenbrenner & M o r r i s , 1998) 17 Investigators w h o have conducted studies on parents and their communi t ies provide evidence that characteristics o f parents and chi ldren play a role in the development o f a ch i ld . The characteristics that serve as the focus o f the present study are reviewed here and include maternal education, f ami ly income, fami ly structure, maternal age and c h i l d gender, age, and health status. Maternal education Simi l a r to f ami ly income level , maternal education and the contr ibut ion educat ion makes to a parent's abi l i ty to accurately report developmental status is disputed in the literature. Investigators ( B o w m a n , 1992; Dichde l t e lmi l l e r et a l . , 1992; Henderson & M e i s e l s , 1994) provide evidence that chi ldren wi th mothers who have less than a high school d i p l o m a are at greater risk for developmental delays. Researchers surmise that mothers w i th l o w levels o f education are not fami l ia r w i t h c h i l d development or possess few parenting sk i l l s ( B o w m a n , 1992). A c c o r d i n g to Bronfenbrenner (1998), educational attainment o f the parent is one o f the variables that is unaffected by fami ly processes because educational attainment happens before the formation o f the fami ly , yet has significant effects on the fami ly i f attainment is l o w . Parent educational attainment is typ ica l ly l inked to whether or not the fami ly l ives i n poverty. Poverty , in turn, creates an unstable and poor environment for consistent and posit ive parent c h i l d interaction (Bronfenbrenner, 1998). Stanton-Chapman, Chapman , and Scott (2001) measured ind iv idua l and population-based r isk factors for ch i ldren w i t h learning disabil i t ies. Mate rna l education level was one o f the highest risk factors at the ind iv idua l - leve l for a learning disabi l i ty later in life, and was one o f the highest populat ion- level r isk factors for a learning disabi l i ty . Th i s study focused on the behaviors often associated wi th mothers who attained an education level o f less than h igh school . 18 Mothers w i t h a less than high school education were found to be more l i k e l y to smoke and engage in more risk behaviors pre and post-natal than their better-educated counterparts. A l t h o u g h Stanton-Chapman et a l . (2001) do not provide evidence that mothers w i th a less than high school educational attainment are inaccurate in reporting their ch i l d ' s developmental progress, they do document that w o m e n wi th a less than high school education may not understand the impl ica t ions their o w n behaviors have for their ch i ldren ' s health and development. Th is lack o f understanding may l imi t their abi l i ty to observe signs their chi ldren may have developmental delays. It should be noted that this study d id not ment ion whether the mothers themselves had learning disabil i t ies , w h i c h w o u l d also influence their abi l i ty to understand development and ul t imate ly report on their ch i ld ' s development. It is important to consider maternal education as a characteristic that m a y influence a parent's abi l i ty to accurately report developmental delay as it is an important characteristic for parents i n predict ing other developmental accomplishments in chi ldren. Glascoe and D w o r k i n (1995) summarized studies from the 1960s and 1970s regarding parental education attainment. They found that there are discrepancies w i th in the literature. It was reported that early studies found that parents w i th higher levels o f education made more accurate reports o f their ch i ld ' s development than d id parents w i t h less than high school education. In the Glascoe and M a c L e a n (1990) study, they found that maternal education d id not affect the abi l i ty to make accurate appraisals, however , they ca l l for more research to investigate maternal education as influential in accuracy for reporting on development o f a ch i ld . Regardless, none o f these studies actually investigated the accuracy o f parents report according to the developmental status o f the c h i l d (i.e., i f the c h i l d was reported to be delayed or not delayed both on a parent completed 19 developmental screening instrument and on a developmental assessment), a focus o f the present study. Family income In the literature, it has been wide ly disputed as to whether or not fami ly soc ioeconomic status plays a role in accurate reporting o f a ch i ld ' s development (Glascoe 1995; Squires, Potter, Br i cke r , & Lamorey , 1998). M u c h o f the dispute is a result o f new advancements in the screening instruments; as the screening instruments have become stronger psychometr ica l ly , researchers have found that parental characteristics l ike soc ioeconomic status have less influence on the accuracy o f the results (Glascoe, 1999). A l t h o u g h screening instruments have become increasingly v a l i d and rel iable, part ial ly because o f parent report o f developmental milestones, socioeconomic status may continue to play a role in parent accuracy in reporting. F a m i l y income/soc io-economic status ( S E S ) is w ide ly researched and connected to many outcomes for c h i l d development. It is clear that demographic and environmental characteristics contribute to ch i ld development. Fo r instance, chi ldren who come from disadvantaged backgrounds have an increased chance o f being developmental ly delayed. N o t on ly is there evidence that ch i ldren f rom l o w income or disadvantaged famil ies are at r isk, but also that neighborhoods, such as h i g h l y stressful and dangerous areas impact a ch i ld ' s development ( B o w m a n , 1992). A l t h o u g h some investigators have demonstrated certain characteristics put ch i ldren at r isk for developmental delays, others' f indings have been contradictory regarding these same characteristics o f parents and the influence they have on the parent's abi l i ty to detect and report a developmental delay. B o w m a n (1992) and H a m i l t o n (2006) reported that parents o f h igh socioeconomic status provide a better and more developmental ly s t imulat ing environment for 20 their young chi ldren. However , understanding the influence o f soc ioeconomic status on a parent's abi l i ty to report on their ch i ld ' s development is less clear. A c c o r d i n g to a study by Squires, Potter, B r i cke r , and Lamorey (1998), parents, us ing the A S Q , were able to accurately report their ch i ld ' s development regardless o f parent income status. These investigators focused on the abi l i ty o f parents from low and midd le - income levels to accurately understand and report their ch i ld ' s development. The researchers were a iming to validate the use o f the A S Q wi th low- income parents by reporting that there was no statistically significant difference in abi l i ty o f parents from either income category to accurately report on their ch i ld ' s development compared to results on the professionally administered B a y l e y Scales o f Infant Deve lopment (Squires, Potter, B r i cke r , & Lamorey , 1998). These results are often interpreted to mean that parents o f a l l income levels have the abi l i ty to read and understand the questions on the A S Q and therefore are accurately able report on their ch i ld ' s development. A l t h o u g h s imi lar findings are established i n the literature (Glascoe 2005, 1999, 1997; Squires & B r i c k e r , 1991), there is an inherent bias in the Squire et al (1998) study. The authors' m a y have over interpreted the statistical results. For example , the study did not f ind enough chi ldren wi th delays to calculate sensit ivity percentages, m u c h less conclude that low- income parents were as able as middle- income parents to accurately complete the A S Q . Furthermore, in taking a closer look at the statistical analysis, the creation o f a "total agreement" category (encompassing a l l o f the specif ici ty and sensi t ivi ty numbers) creates a false interpretation o f the results. The number o f parents that accurately identif ied a delay when there was in fact a delay was on ly seven (sensitivity). A s the authors stated, this number is too smal l to have any sort o f power in a statistical analysis. However , the number o f parents that were able to accurately identify chi ldren who d id not have delays was large ( N = 260), and therefore creates 21 the best scenario for statistical analysis. The number o f false positive, screens i n this study was twenty-nine and false negative screens were four. It is diff icult to report that a screen is sensitive in detecting developmental delays when there were not many chi ldren wi th delays in that particular populat ion. In addi t ion, for the smal l number o f chi ldren who d id in fact have a developmental delay, the results show that only ha l f were detected in one age group and o n l y one third were detected in another age group, a l l w i t h i n the low- income parent c h i l d dyad. Therefore, it cannot be concluded that parents o f l o w and high soc ioeconomic status have the same abi l i ty to detect developmental delays i n their chi ldren. In a literature review by M c L o y d (1990), parents o f low income and l o w soc ioeconomic status were demonstrated to have a decreased capacity to cue into their chi ldren 's soc ia l -emotional needs, and less l i ke ly to provide a supportive, car ing and consistent environment for their chi ldren. M o r e evidence is needed to understand the role o f soc ioeconomic status in inf luencing accurate parent reporting. Family structure Jackson, B r o o k s - G u n n , Huang , and Glassman (2000) explored single motherhood and its impl icat ions on chi ldren 's preschool outcomes. Results provide evidence to support the b io-eco logica l model . They found that maternal educational attainment, together w i t h the f inancia l stress o f being a single mother w i th on ly one income, was related to maternal depression. Financia l stress coupled wi th l o w educational attainment and maternal depression lead to . decreased parenting quali ty and poor c h i l d outcomes in behavior and cogni t ive development (Jackson et al , 2000). W h i l e these studies are important in exp la in ing reasons for invest igat ing these characteristics i n the literature, none offer any evidence relevant to the question o f whether single parenthood influence parents' abil i t ies to accurately report on their ch i l d ' s development. 22 Va l ida t ion studies o f developmental screening tools include demographic data that include characteristics such as single motherhood. None has found a significant effect on the accuracy o f parent reporting (Glascoe, 2000; 1999; 1995; Squires, Potter, B r i c k e r & Lamorey , 1998). A s stated earlier, high risk populations are not always w e l l represented in these studies, and therefore, investigating single motherhood as a characteristic that cou ld influence accuracy on developmental screening tools is wor thwhi l e (Dichde l te lmi l le r et a l , 1992; Henderson & Meise l s , 1994; Tervo , 2005). Maternal age Chi ld ren o f teen mothers are at r isk for developmental delays. Furstenberg, B r o o k s -G u n n , and M o r g a n (1987) summarized the results o f the famous Bal t imore study, a longi tudinal study f o l l o w i n g over 300 teenage mothers i n the U n i t e d States. Adolescent motherhood, l ike educational attainment, co-varies w i t h other characteristics for its negative influence on chi ldren. Mothers who bare chi ldren as teenagers have a l ike l ihood o f addit ional risk factors such as dropping out o f school , l ow income levels, and less marital stability (Bronfenbrenner, 1998; Furstenberg, B r o o k s - G u n n , & M o r g a n , 1987). In addit ion, as study by T a m i s - L e M o n d a et al . (2002) studied maternal knowledge o f development when mothers were less than 18-years o f age. S ix ty adolescent mothers (ranging i n age from 13-18-years-old) were asked to estimate ages infants met developmental milestones. These mothers were able to name the order o f milestones, but were much less l i ke ly to k n o w when (i.e., t iming) the milestones should take place for the infant.' O v e r a l l , the mothers were observed as expecting developmental milestones to happen much earlier and in less t ime that developmental milestones occur in general. A l though this study provided evidence that young mothers were inaccurate at estimating the onset o f developmental milestones, the connect ion was not made to accuracy o f report ing on 23 observable developmental milestones. Therefore, the literature is l ack ing evidence that young mothers can be accurate i n reporting development, even i f they were shown to be inaccurate at estimating t iming o f developmental milestones. M a n y studies investigating the accuracy o f parents as reporters include adolescent mothers in the demographic descript ion. H o w e v e r few, i f any, studies actually investigate whether this particular characteristic influences accuracy o f reporting on developmental screening tools. M a n y val idat ion studies completed on developmental screening tools fai l to include large numbers o f teenage mothers, and therefore it is un l ike ly that current research is sufficient in this area (Dichde l t e lmi l l e r et a l , 1992; Henderson & M e i s e l s , 1994). Child Characteristics 1 C h i l d characteristics cannot be ignored when considering parent accuracy i n reporting on their ch i ld ' s development. Research on the va l id i ty o f developmental screening tools provides some o f the best evidence o f the influence o f c h i l d characteristics on the abi l i ty o f parents to ( accurately report on the ch i l d ' s development. C h i l d age, gender, and health status have been examined to better understand their influence on parent report accuracy. Child age A s chi ldren increase i n age they progress through developmental milestones. A greater number o f older chi ldren have developmental delays because they have gone through more development (Glascoe, 1990). Therefore, it is log ica l that parents w o u l d have greater accuracy when reporting on older chi ldren. D i a m o n d and Squires (1993) summarize studies that are contradictory in expla in ing the equ ivoca l relationship between c h i l d age and maternal report. They reported that some studies p rov ided evidence o f greater accuracy o f maternal report w i th preschool aged chi ldren than wi th infants. Some studies suggested that there was greater 24 accuracy in maternal reports when the developmental level o f the ch i ld was lower, wh i l e others found no difference in accuracy and the age and the developmental level o f chi ldren. In the study by Glascoe and M a c L e a n (1990) investigating parent's appraisals o f c h i l d development, they found that parents w i t h concerns about their ch i ld ' s development had chi ldren that were s ignif icant ly older than parents who d i d not have concerns. H o w e v e r , i n this study there was not a cr i ter ion measure for accuracy. The study, as described previous ly was conducted to understand how parents appraise their ch i ld ' s development. Glascoe (2003) conducted another study where parent's concerns were used to help predict ch i ldren ' s performance on the Eyeberg C h i l d B e h a v i o r Inventory ( E C B I ; Eyeberg, 2000), and the Poss ible Problems Check l i s t , w h i c h is part o f the C h i l d Development Inventory ( C D 1 ; Ireton, 1994). Parents who reported concerns about their ch i l d ' s behavior successfully predicted mental health problems. Results were significant for chi ldren under the age o f 4 years, 5 months, indicat ing that parents' concerns can be accurate for ch i ldren less than five years (Glascoe, 2003). Te rvo ' s (2006) research suggested that parents o f younger chi ldren are less l i k e l y to have concerns as parents o f older chi ldren. H i s research also suggested that age o f the c h i l d has not been w i d e l y investigated in terms o f predic t ing ch i ld outcomes from parental concerns. Understanding how age relates to parent abi l i ty to accurately report development w i l l certainly better in fonn research i n developmental screening tools. Child gender Gender is important to investigate, especial ly wi th young chi ldren because rates o f development vary w i d e l y by gender (Dahinten & F o r d , 2004; Frankenburg, 1994). F e w studies focused upon invest igat ing c h i l d gender effects on parent accuracy in reporting developmental delays. Howeve r , there are numerous studies p rov id ing evidence o f discrepancy i n referral rates 25 o f boys and girls because parents rate boys higher in behavioral problems than they do gir ls . A study by Tervo (2006) found parents o f boys reported more problems wi th attention and behavior than parents o f girls . In a study invest igat ing the uti l i ty o f C H I L D S E R v a communi ty-based developmental screening service, more boys than girls were referred for services ( M c K a y , Shannon, Vater & D w o k i n , 2006). The question must be raised as to whether the parent's expectations for the c h i l d is inf luenced by their gender. It is also important to consider that parent accuracy is influenced by inherent bias in the creation o f developmental screening instruments. Therefore, investigating the effect o f the ch i ld ' s gender on a parent's abi l i ty to accurately report on their ch i ld ' s development is important. Child health status Parents who rate their ch i ld ren as having poor health or chronic health problems m a y have a more unique perspective o n their ch i ld ' s development. In a study by Tervo (2006), some parents w h o in i t ia l ly had health concerns about their ch i ld , in fact had confused developmental concerns wi th health concerns. In a study examin ing parents' consideration o f their ch i ld ' s development when parents provided concerns about their c h i l d ' s development, almost one th i rd o f parents wi th inaccurate reports based their rating on pr ior health concerns (Glascoe, 1999; 1990). Therefore, health o f the c h i l d seems to hold a unique status i n parents' op in ion about development. Some hypothesized that lower income parents actually don' t consider developmental problems to be developmental in nature; rather they v i e w the problems they observe as health problems. T h i s may be w h y some chi ldren 's problems go unnoticed or are inaccurately recognized and diagnosed. 26 Summary This chapter summarized what is k n o w n through publ ished studies about the parent's role in developmental screening and the contr ibut ion o f selected parent and ch i ld characteristics i n the accuracy o f their assessment. The B i o - E c o l o g i c a l M o d e l was proposed as a framework to organize the fami ly and c h i l d characteristics o f interest. It can be concluded, from the literature, that the parent's accuracy in comple t ing developmental screens can be influenced by (a) the screens themselves; (b) by education, income, f ami ly structure, and age; and (c) by their ch i l d ' s age, gender, and health status. The exact nature o f parent and c h i l d characteristics lacks clar i ty because the literature is l imi ted by a chronic under representation o f parents o f low educational attainment, low income, young age, and single parent homes. This study contributes some clari ty to the field because o f the divers i ty o f its participants. The literature is somewhat more congruent regarding c h i l d characteristics in that the older the ch i ld , the more l ike ly the parent is to be accurate on the screen. It remains unclear whether c h i l d gender influences accuracy. Developmental delay, when observed by parents, may be considered a health condi t ion. Taken together, the literature indicates there is a gap i n the f ie ld 's knowledge about parental accuracy in reporting c h i l d developmental delays. Th i s study has potential to reduce the gap i n this knowledge. 27 Chapter Three: M e t h o d o l o g y This study's purpose was to examine c h i l d and parent characteristics that may influence agreement between parent report o f developmental delay us ing the N i p p i s i n g Dis t r ic t Developmenta l Screen ( N D D S ) and in-depth assessments o f c h i l d development by a professional. In this chapter, the methods for the study are explained. The participants, measures, and procedures for data col lec t ion and analysis are described. Research Questions The fo l lowing research questions are proposed: 1. T o what extent do c h i l d characteristics (gender, age, and health status) predict agreement between parent reports on the N D D S and developmental assessment o f the c h i l d us ing the Stanford-Binet: F i f th E d i t i o n (SB5) , the Peabody Picture Vocabu la ry Test: T h i r d Ed i t ion ( P P V T - I I I ) , and the B a y l e y Scales o f Infant Development : Second Ed i t i on (BSID-2)? Hypothesis. It was hypothesized that higher rates o f True Posi t ive results and True negative results w o u l d be obtained for older ch i ld ren (preschool age), females, and children wi th no health concerns. 2. T o what extent do fami ly demographic or soc io -economic characteristics (maternal educational attainment, f ami ly income, f ami ly structure, and maternal age) predict agreement between parent report on the N D D S and developmental assessment o f the. c h i l d using the S B 5 , the P P V T - I I I and the B S I D - 2 ? Hypothesis. It was hypothesized that older maternal age (age 23 or older at the birth o f the first chi ld) , higher maternal education (greater than h igh school), mothers i n two-parent famil ies, and those wi th a f ami ly income greater than L I C O w o u l d be associated wi th higher levels o f agreement. 3. Wha t set o f ch i ld and fami ly characteristics best accounts for agreement between parent report on the N D D S and developmental assessment o f the ch i ld using the S B 5 , the P P V T - l l I , and the BS1D-2? Hypothesis. It was hypothesized that both c h i l d and fami ly characteristics associated wi th lower risk factors w o u l d best account for agreement between the developmental assessments and the developmental screening instrument. Those c h i l d and fami ly characteristics were; females, preschool age chi ldren , chi ldren wi th no health concerns, mothers w i th greater than high school education, famil ies above L I C O , two-parent famil ies, and mothers whose first c h i l d was born after age 22. Participants A secondary data source was analyzed to answer the research questions. The participants in this study were part o f a larger va l id i ty study conducted on the N i p i s s i n g Dis t r ic t Deve lopmenta l Screen. The sample was compr ised o f 395 chi ldren who were 4, 18, 24, or 36 months o ld . Fami l i e s were recruited from a two smaller communi t ies outside a large urban centre ( in southwestern B r i t i s h C o l u m b i a through newspaper advert is ing and by direct advert ising at the publ ic health immuniza t ion c l in i c , a communi ty development centre, and c h i l d health fairs. Recrui tment posters were also distributed to phys ic ian ' s offices, day care centres, preschools, supermarkets, recreation centres, and the l ibrary. E l i g i b i l i t y cri teria required on ly that the responding parent and c h i l d speak Eng l i sh (Dahinten & F o r d , 2004). The sample was approximate ly evenly d iv ided by gender w i t h 5 3 % female chi ldren. The sample was also evenly d iv ided by age group, w i th 5 1 % o f the chi ldren in the ' p reschool ' group (i.e., the 36-month olds), 29 and 4 9 % in the infant/toddler group (i.e., 4, 18 or 24 months olds , wi th 81, 73, and 39 chi ldren, respectively). F i v e percent o f the chi ldren l ived in one-parent famil ies and 2 7 % had been b o m to mothers w h o began chi ldbear ing prior to 23 years o f age. Twenty-one percent o f the mothers reported having high school education or less, and 17% reported hav ing a fami ly income be low the low- income cut-off point. Data Collection Trained graduate student research assistants from the U n i v e r s i t y o f Br i t i sh C o l u m b i a completed the ind iv idua l ized developmental assessment battery. The research assistants conducted individual developmental assessments wi th the c h i l d and assisted the parents in comple t ing the parent questionnaire and the N D D S . C h i l d r e n were assessed at the communi ty health unit or in their homes. The N D D S was completed by the parent, either over the phone, pr ior to the visi t , or at the time o f the assessment. Th is study draws on data from the f o l l o w i n g developmental assessments: the B a y l e y Infant Developmenta l Instrument: Second Ed i t ion , the Stanford Bine t Intelligence Scale: Fi f th Ed i t i on , and the Peabody Picture Vocabu la ry Test: T h i r d E d i t i o n . Measures The Nippising District Developmental Screen (NDDS). The N D D S is a parent-completed screening instrument that measures chi ldren 's development i n the fo l l owing areas: v i s i o n , hearing, speech-language, gross motor, fine motor, cogni t ive , social-emotional , and self-help sk i l l s . Parents answer "yes or n o " to questions about their ch i ld ' s development across a l l seven domains. The N D D S can be used to screen chi ldren f rom 1 month to six years o f age and has 13 different versions according to the age o f the c h i l d . A n answer o f "yes" on the screen indicates that the parent has observed the ch i ld do ing the behavior in question. A n answer o f 30 " n o " (referred to as a ' f lagged ' item) on the screen indicates that the parent has not yet observed the behavior in question. The score on the N D D S consists o f the total number o f " n o " answers or flagged items. The developers created "f lag rules" that indicate when the c h i l d should be referred for further assessment. The one flag rule means that the ch i ld w i l l be referred i f the parent selects one or more "no" answers. The two flag rule means the ch i ld w i l l be referred i f the parent selects two or more "no" answers (Dahinten & F o r d , 2004). Fo r the purpose o f this study, one or more flags on the N i p i s s i n g Dis t r ic t Deve lopmen ta l Screen were considered to be a posi t ive screening result. The one flag rule was chosen because it encompassed the chi ldren wi th m i l d to moderate delays, wh i l e the two-f lag rule is more w i d e l y accepted as ident ifying the more severe delays wh i l e under-identifying chi ldren w i t h m i l d to moderate delays (Dahinten and Ford , 2004), and therefore the sample size was greater when using the one-flag rule. Few studies have been completed to assess the re l iab i l i ty and va l id i ty o f the N D D S . In 2001, an evaluation o f the N D D S was conducted as part o f the Ontar io Hea l thy Babies / Heal thy Ch i ld r en study ( H B H C , N a g y , R y a n , & Rob inson , 2002). The study provided evidence o f inter-rater re l iabi l i ty (71%) between the parent and caregiver other than the parent (for example a daycare worker) , and overal l agreement rates for 12 month-olds on the N D D S when compared to the 12 month-old A S Q (78% for 1 flag mie , 9 3 % for 2 flag rule). The H B H C val idat ion study on ly looked at the 12 month N D D S and relied on one cr i ter ion measure ( A S Q ) . Therefore, Dahin ten and Ford (2004) conducted a val idat ion study o f the N D D S for ch i ldren ages 4, 18 and 24 months. In this study, concurrent va l id i ty was assessed b y compar ing results from the N D D S to results o f the M e n t a l Development Index on the B S I D - 2 . T h e y found that sensi t ivi ty rates ranged from 4 4 % to 100% depending on w h i c h flag-rule was used and whether they used a 2 standard deviat ion cut-off or 2.5 standard devia t ion cut -off on the B S I D - 2 . Spec i f ic i ty rates 31 ranged from 6 8 % to 86%. Current ly there are no other studies investigating the N D D S using more advanced statistical analyses (i.e., factor analysis) or addit ional studies investigating concurrent val id i ty . The Stanford-Binet Intelligence Scales- Fifth Edition (SB5). The S B 5 is an i nd iv idua l ly administered test o f intell igence designed for ages 2.5 years to 85 years, measuring cogni t ive abi l i ty using five major factor indices: F l u i d Reasoning , V i sua l -Spa t i a l Processing, K n o w l e d g e , Quantitative Reasoning, and W o r k i n g M e m o r y . The measurement o f these specific cogni t ive domains is supported by Ca t t e l l -Horn -Car ro l l ( C H C ) theory ( R o i d , 2003). The S B 5 provides an overal l composite Intelligence Quotient (IQ) wi th the F u l l Scale 1Q ( F S I Q ) , as we l l as a Nonve rba l IQ ( N V 1 Q ) and a V e r b a l IQ ( V I Q ) . The five factor indices also generate factor scores specific to each ind iv idua l domain (i.e. V i sua l -Spa t i a l Process ing Factor) . The S B 5 yields standard scores wi th a mean o f 100 and standard deviat ion o f 15 ( R o i d , 2003). The S B 5 is a well-respected, reliable and v a l i d measurement o f cognit ive abi l i ty at the 36 month-old age range. For chi ldren aged 2 to 7 years, the re l iabi l i ty coefficients ranged from mid-.70s to mid-.90s for the five Nonve rba l and five V e r b a l subtests. These coefficients for the broader IQ indexes ( F S I Q , N V I Q , and V I Q ) us ing the norms for chi ldren ages 2 to 7 years consistently ranged from the middle to h igh .90 range. Spec i f ica l ly , scores have y ie lded median re l iabi l i ty coefficients o f .98, .96, and .96 respectively for the 3-year-old age range. R e l i a b i l i t y coefficients for the factor index scores were also acceptable for preschool-aged chi ldren, ranging from .84 to .95 range ( R o i d , 2003). Criterion-related va l id i ty was demonstrated by the moderate to h igh correlations (ranging from .40 to .90) found between examinees' scores on the S B 5 and their scores on the S B 1 V and other measures o f cogni t ive abi l i ty (e.g., Wechs le r Preschool and Pr imary Scale o f Intelligence-32 Rev i sed , Wechsler Intelligence Scale for C h i l d r e n - T h i r d E d i t i o n , and the Woodcock- Johnson III Tests o f Cogni t ive Abi l i t i e s ) . Predict ive va l id i ty was demonstrated by correlating S B 5 scores wi th measures o f academic achievement (e.g., Woodcock - Johnson III Tests o f Achievement , and the Wechs le r Individual Achievement Test-II), y i e ld ing coefficients in the moderate to h igh range (ranging from .33 to .84) ( R o i d , 2003). The Bayley Scales of Infant Development-Second Edition (BSID-2). The B S I D - 2 is one o f the most wide ly used measures o f development o f infants and toddlers in both research and c l in i ca l practice. Va l ida t ion studies o f the M e n t a l Deve lopment Index ( M D I ) have yie lded median rel iabi l i ty coefficients o f .88, .92, and .92 for chi ldren ages 4, 18, and 24 months o f age respectively, wi th a median re l iabi l i ty o f .88 among al l ages and test retest re l iabi l i ty o f .87. The stabil i ty coefficient for the M D I has been found to be .83 for the 1-to 12-month age range and .91 for the 24 to 36-month range (Bay ley , 1993.) The construct va l id i ty evidence consists o f intercorrelation data for the M D I , and intercorrelations o f M D I scores wi th other measures o f preschool cognit ive functioning. The M D I tends to correlate h igh ly w i t h other measures o f preschool cogni t ive functioning (Wescher Preschool and P r i m a r y Scales o f Inte l l igence-Revised ( W P P S I - R ) , M c C a r t h y Scales o f Ch i ld r en ' s A b i l i t i e s ( M S C A ) , and the Different ial Ab i l i t i e s Scale ( D A S ) ) . The moderate correlations between the M D I , W P P S I - R (.73), and M S C A (.79) provide evidence o f concurrent va l id i ty for ch i ldren at the older age range (36 to 42 months) o f the B S I D - 2 . The correlations for the preschool age range is not relevant to this study as the B S I D - 2 was on ly used wi th 4, 18, and 24 month-olds in this study, but does provide evidence o f its strong re l iabi l i ty as a criterion measure. The BS1D-I1 was chosen because o f the psychometr ic properties and history, as w e l l as its appropriateness for chi ldren 4 to 24 months o f age. L i k e the S B 5 , the B S I D - 2 y ie lds standard scores wi th a mean o f 100 and a standard 33 deviat ion o f 15 (Bay ley , 1993). The M D I is currently the o n l y outcome measure for intell igence on the B a y l e y - 2 . Th is is different from the S B 5 , w h i c h y ie lds scores for verbal and nonverbal intell igence as w e l l as scores for the factor indices. The Peabody Picture Vocabulary Test, Third Edition (PPVT-III). The P P V T - I I I is an i nd iv idua l l y administered test o f receptive vocabulary designed for ages 2.6 to 90 years. The P P V T - I I I was chosen because o f its strong psychometr ic properties, and ut i l i ty wi th young chi ldren. A receptive vocabulary test l ike the P P V T - I I I is also useful because it does not require any verbal answers from the ch i ld ; he or she may just point to the answers. The rel iabi l i ty coefficients for the P P V T - I I I range from .88 to .96 for the preschool age. The criterion-related va l id i ty evidence is also good. Correlat ions wi th the Wechs le r Intel l igence Scale for Ch i ld ren , T h i r d Ed i t ion (WISC-I I I ) are .91 and .92. L o w e r correlations were found wi th the nonverbal subtests, w h i c h is to be expected as they require less language and measure different abilities than the P P V T - I I I . Concurrent va l id i t y again, was correlated higher w i t h the verbal subtests on the W I S C - I I I than the performance subtests, h igh ly correlated w i t h the crys ta l l ized intelligence on the K a u f m a n Adolescent and A d u l t Intelligence Test, and w i t h the vocabulary subtest more than wi th matrices subtest on the K a u f m a n B r i e f Intelligence Test (Dunn , D u n n , W i l l i a m s , & W a n g , 1997). Explanatory Variables Child characteristics. Three c h i l d characteristics o f interest are gender, age group, and health status. A g e group was defined as infant/toddler (i.e., 4, 18, or 24 months old) or preschool (i.e.,36 months old): Heal th status was measured by a single i tem o n the parent questionnaire asking the parent to describe their ch i l d ' s health as poor, fair, good or excellent. Th is variable was later recoded as 'excel lent ' and 'less than excellent ' . A l t h o u g h this cutoff is at 'excel lent ' 34 and not ' good , ' g lobal self-evaluations o f health are an often-used measure o f health status that has been shown, w i t h i n the adult population, to be reflective o f phys ica l and mental health, physical functioning, and health behaviours (KLrause & Jay, 1994; V i n g i l i s , Wade , & A d l a f , 1998). Therefore, there is reason to believe that parental self-reports o f ch i ldren ' s health (using a single item or global evaluation o f health) may also be predict ive o f ch i ldren ' s developmental outcomes. The responses o f poor, fair, good, and excellent were d iv ided into excellent vs. less than excellent for two reasons, one pragmatic and more theoretical: 1) because 7 1 % o f chi ldren were identified as be ing i n excellent health vs. 2 7 % i n good health and 2 % in poor health, and 2) because it was hypothesized that excellent meant no concerns at a l l , whereas ' g o o d ' a l lows for some concerns about the ch i ld ' s health. Family characteristics. Four fami ly demographic and soc io-economic characteristics are included in the analysis: maternal educational attainment, f ami ly income, f ami ly structure, and maternal age at first chi ldbear ing. Each o f these were measured on the parent questionnaire and later recoded as fo l lows : Materna l educational attainment was recoded as "greater than high school educat ion" or "less than or equal to a high school educat ion." F a m i l y income was later recoded as "above LICO," or "be low LICO." F a m i l y structure was recoded as " two parent f a m i l y " or "one parent f ami ly . " A l though the fami ly structure variable d id not a l l ow for any other f ami ly structure (e.g., extended family , same-sex couples, or addi t ional help for the mother l ike l i v i n g wi th her parents) the question on the parent questionnaire l imi ted the mother to answer i f she was marr ied or not. Therefore, it was log ica l to recode the variables as one-parent or two-parent fami ly . Las t ly , maternal age at first ch i ldbear ing was later recoded as "greater than 23 years" or "less than or equal to 22 years." Research invest igat ing the effects o f young maternal age on chi ldren ' s developmental outcomes has been inconsistent across studies w i th 35 respect to whether maternal age at first birth or maternal age at the target ch i l d ' s birth is used as a predictor, and m i x e d results, even for the few studies that have used both ages (Moore et a l . , 1997 and Tur ley , 2003). Thus, given that there is some evidence o f res idual effects o f age at first chi ldbear ing for subsequent chi ldren bora when the mother is older, maternal age at first chi ldbi r th was chosen as the. variable o f interest for this study. A n a l y s i s o f Data The p r imary aim o f this study is to identify the set o f ch i ld and f ami ly characteristics that best accounts for agreement between parent report on the N D D S and cogni t ive assessment o f the ch i ld . Correspondence between the screening results and developmental assessments also provides the basis for assessing the sensi t ivi ty and specif ici ty rates for the N D D S . Agreement Status and the Calculation of Sensitivity and Specificity Rates. Sensi t iv i ty and specif ici ty rates are calculated based on the agreement between the parent report on the N D D S and the results o f the professional ly administered direct c h i l d assessment. A true posi t ive result refers to a positive screen for a ch i ld w h o was assessed as hav ing a developmental delay (i.e., the parent-reported screening results are consistent w i t h the results obtained on the professional ly administered direct assessment). F o r the purpose o f this study, a developmental delay is considered a standard score o f one standard deviat ion be low the mean on one o f the developmental assessments (Mean=100, SD=15 ; therefore, delay = a score o f < 85). A false posi t ive refers to a posit ive screen for a delay for a c h i l d who was assessed as not having a delay. A false negative refers to a negative screen for a delay for a ch i ld w h o has a delay. A true negative refers to a negative screen for a c h i l d w h o does not have a delay. Thus , the true posit ives and true negatives represent cases where there is agreement between the parent-reports and the direct assessments, whereas false posit ives and false negatives represent disagreement. 36 F i n a l l y , sensi t ivi ty refers to proport ion o f the true posit ives among chi ldren wi th delays and specif ic i ty refers to the proport ion o f true negatives among chi ldren without delays. Bivariate Analysis. Research questions one and two were analyzed using a bivariate analysis. Chi-squared statistics were completed to test the f o l l o w i n g hypotheses generated based on the literature review provided in chapter two. ( T w o - w a y tables w i th chi-square statistics were used because each o f the variables were measured at the n o m i n a l leve l , each wi th two categories.) It was hypothesized that certain parent and c h i l d characteristics w o u l d predict higher levels o f agreement between the results o f the N D D S and the c h i l d assessments. Separate analyses were completed for chi ldren with delays and for ch i ld ren without delays in both the bivariate and multivariate analyses. Multivariate Analysis. The third research question was analyzed us ing a binary logis t ic regression to investigate the c h i l d and parent characteristics that best predict agreement between parent report on the N D D S and developmental assessment o f the c h i l d us ing the Stanford-Binet: Fi f th Ed i t ion ( S B 5 ) and the B a y l e y Scales o f Infant Development : Second Edi t ion ( B S I D - 2 ) . The dependent variable was measured as 1 = True Posi t ive for ch i ldren w h o were assessed as delayed, and 0 = False Negat ive . F o r chi ldren who were assessed as not delayed, 1 = True Nega t ive , and 0 = False Pos i t ive . The predictor variables were the parent and c h i l d characteristics supported by the bivariate analysis. Assumpt ions i n v o l v e d in logist ic regression are s imi la r to that o f linear regression. Specif icat ion o f Er ro r and Co l l inea r i ty are discussed. Specif ica t ion o f Error refers to the adequacy o f the model . Does the mode l incorporate a l l the relevant independent variables and exclude irrelevant independent variables? This study was l imi ted because it was conducted on a secondary data set. In logis t ic regression tests for "goodness o f f i t" are based on the log l ike l ihood . "The concern is less wi th the overal l fit o f the 37 mode l and more w i t h the accuracy wi th w h i c h the mode l predicts actual category membership on the dependent var iable" (Menard , 2002, p. 41). The H o s m e r - L e m e s h o w Test is the test for the goodness o f fit statistic. " A non-significant result indicates that the mode l fits; a significant result indicates that it does not fit" (Munro , 2005, p. 311). The specific tests used to examine the accuracy o f category membership were the C o x and Sne l l R Square and the Nege lkerke R Square tests. T h e y assessed the percentage o f variance explained by each o f the variables in the mode l ( M u n r o , 2005). ' Co l l i nea r i t y is the correlation o f independent variables. In this study it was expected the fami ly soc io-economic status variables w o u l d be correlated. Thus after cont ro l l ing for the one fami ly characteristic, a significant result may not be found. Therefore sub-analyses us ing a series o f logis t ic regression were run where relevant fami ly characteristics were entered one at a t ime. 38 Chapter Four : Results Chapter Four presents the analysis and results o f the study. Firs t , sample characteristics are described. Then , the overal l results for sensit ivity and specif ic i ty analysis are described. N e x t the hypothesized associations between screening results on the N D D S and c h i l d characteristics (gender, age, and health status) and fami ly characteristics (maternal education, fami ly income, f ami ly structure, and maternal age) were tested using two-way tables and Pearson chi-square statistics. F i n a l l y , binary logist ic regression analyses were conducted to identify fami ly characteristics and ch i ld characteristics that predict agreement between parent-completed screening result and the developmental assessments. Discuss ion and impl ica t ions o f the findings are presented in Chapter F i v e . Descr ip t ion o f the Sample Init ial descriptive analyses were used to describe the sample used i n this study. O f the total sample ( N = 395), the majority were chi ldren wi th no delays (n = 343, 86.8%,.per the results o f the developmental assessment) whereas chi ldren wi th delays (n = 52) accounted for 13.2% o f the sample. F o r both groups (chi ldren wi th delays and ch i ld ren wi thout delays), the sample represents males and females, and the infant/toddler and preschool age groups evenly. C h i l d health status is less evenly distributed i n both groups. O f the f a m i l y characteristics, fami ly income and maternal age at first chi ldbear ing are not w e l l distributed across the sample. Notab ly , characteristics that are generally considered to put chi ldren at r isk for developmental delays, young maternal age and l o w fami ly income, were more c o m m o n among the chi ldren who were assessed as hav ing delays. See Table 4.1. Table 4.1 Demographic Characterist ics o f the Sample Children Without Tota l Sample - ( N = 395) Ch i ld ren W i t h Delays (n = 52) Delays (ri-= 343) C h i l d Characteristics n % n % ft % Gender M a l e Female 184 211 46.6 53.4 27 25 51.9 48.1 157 186 .45.8 54.2 A g e Infant/Toddlers Preschoolers 193 ' 202 48.9 51.1 23 29 44.2 55.8 170 173 49.6 50.4 Health Status Excel lent Less than Exce l len t 278 . 115 70.7 29.1 . 34 18 65.4 34.6 244 97 71.1 28.4 F a m i l y Characteristics Maternal Educa t i on ' < H i g h Schoo l > H i g h Schoo l 81 308 20.8 79.2 11 41 21.2 78.8 70 267 20.8 79.2 F a m i l y Income B e l o w L I C O A b o v e L I C O 65 320 16.9 83.1 15 35 30.0 70.0 50 285 14.9 85.1 F a m i l y Structure One-parent Two-parent 20 374 5.1 94.7 3 49 5.8 94.2 17 325 5.0 95.0 Mate rna l A g e < 22 years > 22 years 106 283 27.2 72.8 22 30 42.3 57.7 84 253 24.9 75.1 40 Overa l l Sensi t ivi ty and Spec i f ic i ty Rates Table 4.2 shows the correspondence between the parent-reported screening results and the developmental assessments. A s described in Chapter 2, the true posit ives and true negatives represent cases where there is agreement between the parent-reports and the developmental assessments, whereas false posit ives and false negatives represent disagreement. Sensi t iv i ty refers to proportion o f the true posit ives among chi ldren wi th delays and specif ic i ty refers to the proport ion o f true negatives among children without delays. Therefore, in this sample, the sensit ivity rate for the N D D S was 3 7 % (19/52) and the specif ici ty rate was 8 2 % (282/343). Table 4.2 A n a l y s i s o f Sensi t iv i ty and Speci f ic i ty Assessment Results N D D S Results De l ayed N o t Delayed Tota l Posi t ive Negat ive 19 True Posi t ives 33 False Negat ives 61 False Posi t ives 282 T a i e Negat ives 80 H 5 Total 52 343 395 Separate analyses were conducted for chi ldren wi th and without delays in both the bivariate and multivariate analyses. F o r chi ldren wi th delays, agreement is represented by a True Posi t ive screening result; among chi ldren wi th delays, agreement is represented by a True Negat ive screening result. Research Quest ion One 41 Research question one. T o what extent do ch i ld characteristics (gender, age, and health status) predict agreement between parent report on the N D D S and developmental assessment o f the ch i ld us ing the S B 5 , the P P V T - I I I , and the B S I D - 2 ? Bivar ia te analyses, using two-way tables w i th chi-square statistics, were conducted to determine the relat ionship between ch i ld characteristics and agreement between parent report on the N D D S and professional assessment o f the c h i l d for chi ldren wi th and wi thout delays. First, bivariate results are described for chi ldren wi th delays. Second, bivariate results are described for chi ldren without delays. Children with delays. F o r chi ldren wi th delays, there were significant associations between each o f the three c h i l d characteristics under investigation and the agreement status between parent reports and direct assessment (i.e., True Pos i t ive vs False Nega t ive results). Parents o f males were s ignif icant ly more l i ke ly than parents o f females to report delays (x = 5.68, df= l,p = .017). A p p r o x i m a t e l y 5 0 % o f the male chi ldren wi th delays were identif ied by their parents as delayed, compared wi th on ly 2 0 % o f the female chi ldren w i t h delays. Parents o f infants and toddlers were s ignif icant ly more l i ke ly than parents o f preschoolers to report true positives (y2 = 10.53, df= \,p= .001). Less than one-fifth o f the preschoolers w i t h delays were identified by their parents as delayed. There was also a significant associat ion between c h i l d health status and parent report o f delay. Parents who reported their ch i ldren ' s health as less than excellent were s ignif icant ly more l i ke ly to report a delay than parents o f ch i ldren w i t h excellent health (y2 = 4.29, df= l,p = .038). O n l y one-quarter o f the chi ldren wi th excellent health were identif ied by their parents as delayed, compared wi th approximately h a l f o f the ch i ldren w i t h less than excellent health. Thus , among the sub-sample o f chi ldren who were assessed as delayed, true posit ive screening results were more l i ke ly to occur for male chi ldren , younger ch i ldren (4, 42 18, and 24 months), and those perceived by the parents to have less than excellent health. See Table 4.3. Table 4.3 Assoc ia t ion Be tween C h i l d Characteristics and Parent-Reported Screening Resul ts for C h i l d r e n wi th Delays (n = 52) C h i l d Characterist ic M a r g i n a l Probabi l i t ies True Posit ives z2 P n % n % Gender M a l e 27 51.9 14 51.9 5.679 .017 Female 25 48.1 5 20.0 A g e Infant/Toddlers 23 . 44.2 14 60.9 10.529 .001 Preschoolers 29 55.8 5 17.2 Hea l th Status Excel lent 34 65.4 9 26.5 4.294 .038 Less than excellent 18 34.6 10 55.6 Children without delays. F o r chi ldren without delays, there were significant associations between two o f the three ch i ld characteristics and parent reports o f delay (i.e., True Negat ive vs False Posi t ive results). There was a significant association between c h i l d gender and parent report o f delay. Parents o f females were more l ike ly to correctly report their ch i ld as non-delayed than parents o f males (jr2 = 5.243, df= 1, p = .022). There was also a significant association between c h i l d age and agreement status (/}f2= 37.79, df= 1, p = .000). Parents o f preschool chi ldren were more l i k e l y to report true negative results compared w i t h parents o f infants and toddlers. There was not a significant association between parent-reported c h i l d health status and parent report o f delay (%l= 3.129, df= 1, p = .077). 43 Table 4.4 Associa t ions Between C h i l d Characterist ics and Parent-Reported Screening Results for C h i l d r e n Wi thou t Delays (n = 343) C h i l d Characteristic M a r g i n a l Probabil i t ies True Negatives z2 P n % n % Gender M a l e 157 45.8 121 77.1 5.243 .022 Female 186 54.2 161 86.6 A g e Infant/Toddlers 170 49.6 118 69.4 37.792 .000 Preschoolers 173 50.4 164 94.8 Heal th Status Exce l len t 244 71.1 206 84.4 3.129 .077 Less than 97 28.4 74 76.3 Exce l l en t T o summarize the analyses o f the c h i l d characteristics, bivariate analyses provided evidence o f significant associations between ch i ld characteristics and parent report o f delay both when chi ldren had delays and when they d id not have delays. Parents were more l i ke ly to identify a true posit ive for males i n the delayed group, and more l i k e l y to identify true negatives for females in the group without delays. In addit ion, parents were more l i ke ly to identify true posi t ives among younger ch i ldren i n the delayed group and more l i k e l y to identify true negatives among older preschool ch i ldren i n the group without delays. C h i l d health status had a significant association wi th parent report on ly i n the delayed group. Parents i n the sub-sample o f delayed chi ldren, w i t h less than excellent health, were more l i ke ly to report true posit ives than parents o f ch i ldren w i t h excellent health. Research Quest ion T w o Research question two. T o what extent do parent demographic characteristics (maternal education, fami ly income, f ami ly structure, and maternal age) predict agreement between parent 44 reports on the N D D S and developmental assessment o f the ch i ld us ing the S B 5 , the P P V T - I I I , and the B S I D - 2 ? Bivariate analyses, us ing two-way tables w i th chi-square statistics, were conducted to determine the associations between fami ly characteristics wi th agreement between parent report on the N D D S and developmental assessment o f the c h i l d for chi ldren wi th and without delays. The analyses were conducted for chi ldren wi th delays first, fo l lowed by analyses for chi ldren without delays. Children with delays. F o r chi ldren wi th delays, the findings are mixed in that on ly two o f the four parent characteristics were associated wi th agreement status between parent-report screening results and developmental assessments (i.e., True Pos i t ive vs. False Negative results). N o significant association was found for maternal age at first chi ldbear ing or fami ly structure and parent report o f delay, but maternal education and fami ly income showed significant associations with parent report o f delay. Mothers w i th a h igh school education or less were almost three times more l i ke ly than mothers w i th education beyond h igh school to report True Pos i t ive screening results 7.88, df= 1, p - .005). Fami l i es w i t h household income be low the L o w Income Cu to f f ( L I C O ) were also much more l i ke ly to report delays (i.e., True Posi t ives) than parents wi th incomes above L I C O (jf= 5.357, df= 1, p = .021). 45 Table 4.5 Associa t ions Between F a m i l y Characteristics and Parent-Reported Screening Results for C h i l d r e n wi th Delays (n = 52) F a m i l y Characteristic M a r g i n a l Probabi l i t ies True Posi t ives P n % n % Materna l Education < H i g h School 11 21.2 8 72.7 7.88 .005 > H i g h School 41 78.8 11 28.6 F a m i l y Income B e l o w L I C O 15 30.0 9 60.0 5.357 .021 A b o v e L I C O 35 70.0 9 25.7 F a m i l y Structure One-parent ~> 5.8 2 66.7 1.246 .264 Two-parent 49 94.2 17 34.7 Mate rna l A g e < 22 years 22 42.3 9 40.9 .314 .575 > 22 years 30 57.7 10 33.3 Children without delays. F o r chi ldren without delays, two o f the four parent characteristics were associated w i t h agreement status between parent-reported screening results and developmental assessments (i.e., True Negat ives vs False Posi t ives) , although these are not the same as for chi ldren w i t h delays. F a m i l y income, and maternal age at first chi ldbear ing were found to be signif icantly associated w i t h agreement status, but maternal education and fami ly structure were not. Mothers w i t h fami ly income above L I C O were more l i ke ly to identify true negatives than mothers w i t h f ami ly income be low L I C O (JC2 = 4.069, df= 1, p = .044). Mothers w h o were age 23 or older when their first ch i ld was born were more l i ke ly to correctly report True Negatives than mothers who were age 22 or less (j?= 7.012, df= 1,/? = .008). 46 Table 4.6 Associa t ions Be tween F a m i l y Characteristics and Parent-Reported Screening Results for Ch i ld ren without Delays (n = 343) F a m i l y Characterist ic M a r g i n a l Probabili t ies True Nej; natives 7 r P ii % . n % Materna l Educat ion < H i g h Schoo l 70 20.8 52 74.3 3.778 .052 > H i g h School 267 79.2 225 84.3 F a m i l y Income B e l o w L I C O 50 14.9 36 72.0 4.069 .044 A b o v e L I C O 285 85.1 239 83.9 F a m i l y Structure One-parent 17 5.0 11 64.7 3.720 .054 Two-parent 325 95.0 .270 83.1 Materna l A g e < 22 years 84 24.9 61 72.6 7.012 .008 . > 22 years 253 75.1 216 85.4 In summary, fami ly characteristics had several significant associations for parent report o f delays for ch i ldren wi th delays and chi ldren wi thout delays. In the sub-sample o f chi ldren wi th delays, mothers w i t h lower educational attainment were more l i ke ly to identify true posit ives, w h i l e in the sub-sample o f chi ldren without delays, maternal education was not significant. In both groups fami ly income was significant. In the delayed group, famil ies be low L I C O were l i ke ly to identify true posit ives, and in the group o f ch i ldren without delays famil ies above L I C O were l i ke ly to identify true negatives. F o r both ch i ldren wi th delays and chi ldren without delays, fami ly structure had no significant association w i t h parent report. Las t ly , the analysis o f the association between maternal age and parent report o f delay resulted in m i x e d findings. In the group o f chi ldren w i t h delays, maternal age was not significant, whereas in the group o f chi ldren 47 without delays, mothers who began chi ldbear ing after age 22 were more l i ke ly to identify true negatives than younger mothers. Research Question Three Research question three. W h a t set o f ch i ld and f a m i l y characteristics best accounts for agreement between parent report on the N D D S and cogni t ive assessment o f the ch i ld us ing the S B 5 , the P P V T - I I 1 , and the B S I D - 2 ? Mul t ivar ia te analyses us ing logist ic regression were conducted to determine the set o f c h i l d and fami ly characteristics that best accounts.for agreement between parent report on the N D D S and professional assessment o f the c h i l d us ing the S B 5 , the PPVT-1I1 , and the B S I D - 2 for chi ldren wi th and without de lays . .As w i t h the bivariate analyses, the analyses and findings are described for children wi th delays fo l lowed by the analyses and findings for chi ldren without delays. Results for sub-analyses b y c h i l d gender and ch i ld age group are also presented. Children with delays. The first binary logist ic regression analysis was conducted wi th ch i ld ren identified as having a delay by developmental assessment. A l l three o f the ch i ld variables were included in the model , but on ly the two fami ly characteristics (maternal education and fami ly income) that were found to be statistically significant i n the bivariate analyses were inc luded in the model . T w o o f the three ch i ld characteristics were found to be statistically significant in this multivariate analysis, but neither o f the f ami ly characteristics were found to be significant. B o t h ch i ld gender and c h i l d age were statistically significant i n association wi th agreement status. C h i l d gender for the group o f delayed chi ldren was shown to be associated w i t h agreement status (i.e., True Pos i t ive vs False Nega t ive results) ( O R = .118, df= I, p = .027). Table 4.7 indicates that females were on ly about l / 8 l h as l i ke ly as males to be reported as delayed 48 when in fact they were delayed. C h i l d age was also predict ive o f agreement ( O R = .129, df= 1, p = .011). A g a i n , consistent wi th the bivariate analysis, chi ldren in the preschool group were only about l / 8 l h as l i ke ly as infants/toddlers to be accurately identified by their parents as delayed. C h i l d health status was not found to be a statistically significant predictor o f agreement in the sample o f chi ldren wi th delays. A c c o r d i n g to the C o x and Sne l l and Nage lke rke estimates, between 3 7 % and 5 0 % o f the variance in agreement is accounted for by this mode l . The Hosmer and Lemeshow Test also supports the fit o f the mode l (p = .887). H o w e v e r , g iven the strong association found in the bivariate analysis for maternal education and f ami ly income wi th agreement status, but the non-significant results in the logist ic regression above (coupled wi th the strong association between maternal education and fami ly income, ( / 2 = 21.429, df= \ ,p = .000), two other models were run for the chi ldren wi th delays, each inc lud ing o n l y one o f the two fami ly variables. Inc luding maternal education w i t h the three ch i ld variables d i d not y ie ld a significant result for maternal education, but when fami ly income was inc luded in the model , it was found to be significant. Thus, chi ldren from families wi th low incomes were seven times more l i ke ly to be identif ied as delayed (true positives) than chi ldren from famil ies w i t h incomes above the L I C O , even after accounting for ch i ld characteristics. See Table 4.8. Table 4.7 Results from the Logis t i c Regression Analyses : C h i l d and Fami ly Characteristics that Predict Agreement between the N D D S and the S B 5 , B S I D - 2 , and the P P V T - I I I in Ch i ld ren wi th Delays (n = 52) B S E Wald d f O R p C h i l d Characteristics Gender (1 = Female, 0 = Male ) -2.136 .968 4.868 1 .118 .027 A g e (1 = Preschool , 0 =Infants/Toddlers) -2.044 .800 6.522 1 .129 .011 Health Status (1 = less than excellent, 0 = excellent) -22.652 40192.991 .000 1 .000 1.000 F a m i l y Characteristics Maternal Educat ion (1 = < H S , 0 = H S + ) .058 1.313 .002 1 1.059 .965 Fami ly Income (1 = < L I C O , 0 = > L I C O ) 1.954 1.291 2.292 1 7.079 .130 50 Table 4.8. Results from the Logis t ic Regression Analyses : C h i l d and F a m i l y Characteristics that Predict Agreement between the N D D S and the S B 5 , B S I D - 2 , and the P P V T - I 1 I in C h i l d r e n wi th Delays (n = 52) B S E W a l d d f O R p C h i l d Characteristics Gender (1 = Female, 0 = M a l e ) -2.147 .937 5.253 1 .117 .022 A g e (1 = Preschool , 0 -2.043 .800 6.529 1 .130 .011 =lnfants/Toddlers) Heal th Status (1 = less than -22.707 40192.991 .000 1 .000 1.00 excellent, 0 = excellent) F a m i l y Characteristics F a m i l y Income (1 = < L I C O , 0 = 1.994 .919 4.709 1 7.346 .030 > L I C O ) 51 Children without delays. The second b inary logist ic regression analysis was conducted wi th the group o f chi ldren without delays. In this analysis on ly two o f the three c h i l d characteristics were included (gender and age) because c h i l d health status d id not have a significant association w i t h agreement in the bivariate analysis. In addit ion, only two o f the four fami ly characteristics that'were found statistically significant in the bivariate analyses were included in the model ( family income and maternal age). B o t h ch i ld characteristics were significant. C h i l d gender for the group o f chi ldren without delays was shown to be associated wi th agreement status ( O R = 2.277, clf~ 1, p = .009). Table 4.9 indicates that females were twice as l i ke ly to be correctly identified as non-delayed (True Negat ives) than males. L i k e w i s e , ch i ld age was significant in predict ing agreement between parent rating and developmental assessment ( O R = .154, df= 1, p. = .000). C h i l d r e n i n the preschool group were seven times more l ike ly to be accurately reported on by their parents than the infant toddler chi ldren. A c c o r d i n g to C o x and Sne l l and Nage lkerke estimates, between 12.6% and 20 .8% o f the variance in agreement is accounted for by this mode l , however the Hosmer and L e m e s h o w Test supports the fit o f the mode l (p = .237). A s wi th ch i ld ren w i t h delays, two further models were run, inc luding on ly one o f the fami ly variables i n each, but the fami ly variables were not found to be significant after account ing for ,chi ld characteristics. Results from the logist ic regression analyses indicate that, for chi ldren without delays, agreement status (i.e., the rate o f True Negat ives , or specif ici ty) was largely influenced by ch i ld gender and age group; none o f the four fami ly variables were found to be significant after accounting for c h i l d characteristics although there was an indicat ion o f some association from the bivariate analyses. Therefore, further sub-analyses were conducted on gender-specific and age-specific sub-samples o f the chi ldren without delays, i n order to further explore possible 52 relationships between agreement status and fami ly characteristics. S imi l a r sub-analyses cou ld not be completed wi th the group o f chi ldren w i t h delays because o f the smaller number o f chi ldren wi th delays (52 vs. 343). Table 4.9 Results from the Log is t i c Regress ion Ana lyses : C h i l d and F a m i l y Characteristics that Predict Agreement between the N D D S and the S B 5 , B S I D - 2 , and the P P V T - I I I in Ch i ld ren without Delays (n = 343) B S E Wald df O R P C h i l d Characteristics Gender (1 = Female , 0 = M a l e ) A g e (1 = Preschool , 0 =Infants/Toddlers) F a m i l y Characterist ics L I C O (1 = < L I C O , 0 = > L I C O ) .823 1.954 -.165 .313 .397 .410 6.900 24.236 .161 2.277 7.059 .848 .009 .000 .688 Mate rna l A g e (1 = < 22 years, 0 = > 22 years) -.210 .354 .352 .810 .553 54 Sub-analysis by Gender for Children without delays. Ma te rna l age predicted agreement when analyz ing the sub-sample o f females wi thout delays ( O R = .293, df= 1, p = .022). For the sub-sample o f females without delays, mothers w h o began chi ldbear ing pr ior to 23 years were only 29.3 % as l i ke ly to correctly identify their chi ldren as not delayed (True Negat ives) as older mothers. However , for the sub-sample o f males without delays, none o f the f ami ly characteristics were s ignif icant ly associated wi th agreement status (i.e., True Negat ives) . Refer to tables 4.10 and 4.11 for details. Table 4.10 F a m i l y Characterist ics that Predict Agreement between the N D D S and the S B 5 , B S I D - 2 and the P P V T - I I I i n Female C h i l d r e n who are not D e l a y e d (n = 179) B S E W a l d d f O R P F a m i l y Characterist ics Materna l Educa t ion (1 = < H S , 0 = H S +) -.125 .597 .044 - 1 .883 .834 F a m i l y Income (1 = < L I C O , 0 = > L I C O ) -.558 .635 .773 1 .572 .379 F a m i l y Structure (1 = one-parent, 0 = two-parent) -.848 1.00 .719 1 .428 .396 Mate rna l A g e (1 = < 22 years, 0 = > 22 years) -1.227 .537 5.219 1 .293 .022 Table 4.11 F a m i l y Characterist ics that Predict Agreement between the N D D S and the S B 5 , B S I D - 2 and the P P V T - I I I who are not D e l a y e d (n = 148) i n M a l e C h i l d r e n B S E W a l d ' d f O R P F a m i l y Characterist ics Mate rna l Educa t i on (1 = < H S , 0 = H S +) -.351 .530 .439 1 .704 .508 F a m i l y Income (1 = < L I C O , 0 = > L I C O ) -.346 .662 .273 1 .707 .601 F a m i l y Structure (1 = one-parent, 0 = two-parent) ' .033 .947 .001 1 1.033 .972 Maternal A g e (1 = < 22 years, 0 = > 22 years) .028 .490 .003 1 1.029 .954 56 Sub-analysis by Age Group for Children without delays. W h e n the sub-sample o f children in the preschool age group was analyzed, maternal age was the on ly f a m i l y characteristic-found to be statistically significant ( O R - . 164, df= 1, p = .020). Y o u n g e r mothers are only 16.4% as l i k e l y as older mothers to accurately report True Negat ives (non-delays) for 36-month o ld chi ldren. Therefore, younger mothers were less l i ke ly to identify True Negat ives than older mothers for preschool children. None o f the fami ly characteristics were found to be statistically significant for the infant/toddler group (ages 4, 18 or 24 months old) . Refer to tables 4.12 and 4.13 for detail . Table 4.12 Family Characteristics that Predict Agreement between the N D D S and the SB5, BSID-2 and the PPVT-III in Preschool Children who are not Delayed (n = 161) . B SE Wald df OR Family Characteristics Maternal Education (1 =< HS, 0 = HS +) 19.000 9229.035 .000 Family Income (1 = < L I C O , 0 = > LICO ) 18.427 15896.799 .000 Family Structure (1 = one-parent, 0 = two-parent) -1.659 25342.850 .000 Maternal Age (1 =< 22 years, 0 = > 22 years) -1.808 .780 5.376 1.8E +008 .0E+008 .190 .164 .998 .999 1.000 .020 Table 4.13 F a m i l y Characteristics that Predict Agreement between the N D D S and the S B 5 , B S I D - 2 and the P P V T - I I I in Infant/Toddler C h i l d r e n who are not D e l a y e d (n - 166) B S E W a l d d f O R p F a m i l y Characteristics Maternal Educa t ion (1 = < H S , 0 = H S + ) -.235 .412 .324 1 .791 .569 . F a m i l y Income (1 = < L I C O , 0 = > L I C O ) -.124 .463 .072 1 .883 .788 F a m i l y Structure (1 = one-parent, 0 = two-parent) -.252 .690 .133 1 .778 .715 Materna l A g e (1 = < 22 years, 0 = > 22 years) -.039 .378 .010 1 .961 .919 58 To summarize the analyses for research question three, statistically significant predictors were present for child characteristics in both sub-samples of children with delays and those without delays. In the sub-sample of children with delays, male gender and younger child age lead more often to the identification of True Positives, but child health was not predictive of agreement between the N D D S and the developmental assessment. For the sub-sample of children without delays, child gender and age were also predictive of agreement. Female gender lead more often to true positives while the preschool age group lead more often to True Negatives. When only a single family characteristic was included in the model, low family income was found to be associated with higher rates of agreement for children with delays, but no family characteristics were found to be statistically significant for the non-delayed group after accounting for child characteristics. The sub-analyses by child gender and child age in children without delays, provided further information about the influence of family characteristics on agreement status. When all four family characteristics were included in the logistic regression model, young maternal age at first childbearing was found to be associated with agreement status for female children who are not delayed and preschool children who are not delayed. That is, younger mothers were less likely than older mothers to correctly identify their female children as not delayed and less likely to identify their preschool children as not delayed (i.e., lower rates of True Negative results, or lower specificity). Summary In Chapter Four, the statistical analyses and findings for three research questions were reported. Bivariate analyses, using the Chi-Square statistic to examine associations among child and family characteristics and parent report of delays, were conducted on data from children with 59 delays and then on data from children without delays. C h i l d gender and c h i l d age were associated w i t h agreement status for both chi ldren wi th delays and chi ldren without delays; but c h i l d health was associated wi th agreement status on ly for ch i ldren w i t h delays. Regard ing fami ly characteristics, maternal education and fami ly income were found to be associated wi th agreement status for chi ldren wi th delays, whereas fami ly income and maternal age were associated w i t h agreement status for chi ldren without delays. F a m i l y Structure was not related to agreement status for either group o f chi ldren. In summary, True Pos i t ive results were more l i k e l y to be found for male chi ldren and infants and toddlers, chi ldren in less than excellent health, ch i ldren whose mothers have not completed any education beyond h igh school , and those who l ive in famil ies w i th incomes be low L I C O . True Nega t ive results were more l i ke ly to be found for female ch i ldren and preschoolers, chi ldren whose mothers began chi ldbear ing at after 22 years o f age, and those who l ive in families w i t h incomes above L I C O . A mult ivariate analyses, using binary logist ic regression, was conducted to examine the best predictors o f accurate parent report were conducted on sub-samples o f chi ldren w i t h delays, and chi ldren without delays. For both groups o f chi ldren, c h i l d gender and age were s ignif icantly associated w i t h agreement status, and in the same direct ion as w i t h the bivariate analysis. N o fami ly characteristics were found to be statistically significant for the sub-sample o f chi ldren without delays after accounting for ch i ld characteristics. H o w e v e r , l o w f a m i l y income was found to be associated w i t h a greater l i ke l i hood o f agreement (True Posi t ives) for the sub-sample o f chi ldren w i t h delays, even after accounting for c h i l d characteristics, but on ly when it was the on ly fami ly variable entered into the model . C h i l d r e n o f famil ies w i t h incomes lower than L I C O were more l i k e l y to be identified as delayed than chi ldren o f famil ies w i th incomes above L I C O . 60 Las t ly , separate sub-analyses were run, us ing on ly the sub-sample o f ch i ld ren without delays, grouped by gender and ch i ld age, to further investigate the potential influence o f family ' characteristics. Ma te rna l age was the only fami ly characteristic found to be s ignif icant ly associated w i t h agreement status, but on ly for female chi ldren and preschool chi ldren. Mothers who began ch i ldbear ing at an earlier age were less l i k e l y than older mothers to correct ly identify their female ch i ld ren as not delayed and less l i k e l y to identify their preschool ch i ld ren as not delayed (i.e., l ower rates o f True Negat ive results, or lower specif ici ty) . Thus , i n contrast to the results found through bivariate analysis,"after account ing for ch i ld characteristics, fami ly income was not found to predict agreement status for the chi ldren without delays. In conc lus ion , hypotheses were part ial ly supported. It was hypothesized that for the ch i ld characteristics higher rates o f True Posi t ive and True Nega t ive results w o u l d be obtained for the older ch i ldren (preschool age), females, and chi ldren w i t h no health concerns. The results from the bivariate analyses support hypotheses for gender and age for ch i ldren wi thout delays (True Negatives) . H o w e v e r , for the chi ldren wi th delays (True Posi t ives) , results were opposite to the hypotheses for gender and age. F o r the fami ly characteristics it was hypothes ized that older maternal age (age 23 or older at the birth o f the first ch i ld) ; higher maternal educat ion (greater than high school) , mothers in two-parent famil ies, and those wi th a f ami ly income greater than L I C O w o u l d be associated w i t h higher levels o f agreement. A g a i n , the hypothes ized True Negatives were supported by the results for f ami ly income and maternal age, but not the hypotheses (True Posi t ives) for the group w i t h delays. These f indings indicate the sensit ivity and specif ic i ty rates for the N D D S vary largely by ch i ld gender and age, but that some fami ly characteristics (specif ical ly , f ami ly income and maternal age) also were associated wi th the l i ke l ihood o f achieving True Pos i t ive and True 61 Negat ive results. Implicat ions o f these results are investigated i n Chapter F ive , where further expansion o f these results to the f ie ld o f research are discussed. In addi t ion, l imitat ions to the study w i l l also be discussed. 62 Chapter F i v e : D i scus s ion Th i s study provided information about f ami ly and c h i l d characteristics w h i c h may influence the outcomes o f parent reported developmental screening instruments. In this chapter, an expansion o f the findings in Chapter Four is discussed according to each research question. Implicat ions o f these conclusions as they relate to c l i n i ca l and research issues w i l l be explored, and l imitat ions and strengths o f this study w i l l be discussed. Direc t ions for future research w i l l be outl ined and important contributions to the f ie ld w i l l also be discussed. O v e r a l l , hypotheses were partially supported. F ind ings from bivariate analyses indicated that the hypotheses d i d not ho ld for the sub-groups o f chi ldren wi th delays for either ch i ld or fami ly characteristics; indeed, significant but contrary relationships were found for gender, ch i ld health status, maternal education, and fami ly income. Howeve r , for the sub-group o f chi ldren without delays, the hypotheses were supported for c h i l d gender, age, f ami ly income, and maternal age. C h i l d gender and ch i ld age were also found to be significant in the binary logist ic regression analyses both for children with delays and chi ldren without delays. H o w e v e r , after accounting for c h i l d gender and age, only fami ly income was significant in the group o f chi ldren wi th delays. F i n a l l y , i n the sub-analysis for ch i ldren without delays, maternal age was significant but for o n l y females and preschool chi ldren. Research Ques t ion One The first research question explored the associations between parent report on the N D D S and ch i ld characteristics us ing chi-squared statistics. Results indicated several significant relationships both i n the sub-sample o f chi ldren w i t h delays and the sub-sample o f chi ldren wi th no delays. 63 Child gender. Results from the bivariate analysis also p rov ided significant associations for c h i l d gender and accuracy o f parent report on the N D D S . In the sub-sample o f chi ldren wi th delays, parents reported more True Posi t ives for males than for females, w h i l e in the sub-sample o f ch i ldren without delays parents reported more True Negat ives for females than for males. W h i l e there are few studies on the influence o f c h i l d gender on parent report o f development, some researchers have found that males are referred more than females i n a variety o f settings ( M c K a y , Shannon, Vater , & D w o k i n , 2006; Tervo , 2006). Th i s is consistent w i th both the higher rates o f True Posi t ives and the lower rate o f True Negatives found i n this study. O n the other hand, it is unclear w h y there is such a l o w rate o f True Posi t ive results for female chi ldren. It is possible that questions on the N D D S are more suitable to recogniz ing male delays because they are based on observable behavior that is more typica l among male ch i ldren . Child age. Parents were more l i k e l y to report True Posi t ive screens for infants and toddlers than for preschool chi ldren and more l i ke ly to report True Nega t ive screens for preschoolers. Thus , both True Posi t ive and False Posi t ive screening results are more c o m m o n among younger chi ldren. D i a m o n d and Squires (1993) conducted a rev iew o f studies focusing on maternal report and the equivocal relationship wi th c h i l d age. The i r f indings were mixed ; p rov id ing evidence that some mothers had greater accuracy w i t h older ch i ldren , some wi th younger chi ldren , and that some studies provided evidence o f no l i n k to c h i l d age. However , in their va l ida t ion study for the Ages and Stages Questionnaire ( A S Q ) , Squires et al . (1999) p rov ided evidence o f lower sensit ivi ty for younger chi ldren. Ear l i e r research by Glascoe and M a c L e a n (1990) also contradicts f indings from the present study. F ind ings from their study invest igat ing parent appraisals showed that parent concerns were more frequently found for older rather than younger chi ldren. Tervo (2006) also found that parents o f older chi ldren had more 64 concerns than parents o f younger chi ldren. A l though i n the current study parents o f preschool chi ldren without delays were accurate at reporting their ch i l d ' s development, parents o f preschool ch i ldren w i t h delays were less accurate in their reports than parents o f infants and toddlers. It seems surprising that parents o f preschool ch i ld ren w o u l d be less accurate i n ident ifying delays. A s Glascoe (1990) pointed out, chi ldren are more l i k e l y to be identif ied when they are older because there are more developmental milestones to be assessed than for younger chi ldren. L imi ta t ions o f the N D D S as a screening instrument at the preschool age level can not be ignored. The number o f preschool chi ldren accurately identif ied by their parents is surpris ingly l o w w h i c h suggests concerns regarding the screening instrument for use in that age group. In the literature, lower sensit ivi ty rates are usual ly found for younger chi ldren. F o r example, the Denver Developmenta l Screening Test, 2 n d edi t ion, and the Ages and Stages Quest ionnaire have l imi ted items for the infant and toddler age groups, therefore m a k i n g the domains o f development less clear and more compl ica ted to interpret at this age (Dahinten & F o r d , 2004; Frankenburg, 1994; Squires et a l . , 1999). Child health status. Results for ch i ld health found that ch i ldren w i t h delays, whose parents reported their health as 'less than excellent ' were more l i k e l y to be reported as true posit ives than chi ldren w h o were considered to be in excellent health. C h i l d r e n w i t h excellent health were s ignif icant ly under-identified as True Posi t ives . O n c e more, there is l imi t ed research into the influence o f c h i l d health status and its relationship to parent accuracy i n reporting on development. However , Glascoe (1990) found that many parents w h o were inaccurate in their report on development tended to have chi ldren wi th prior health concerns. Therefore, the results have led some to bel ieve that parents who have health concerns for their c h i l d may be less 65 accurate in reporting on development because they mistake developmental concerns for health concerns , w h i c h leads to lower rates o f True Negatives. Te rvo ' s f indings (2006) also suggested that parents tend to mistake developmental problems for health problems leading to under-identification (False Negat ives vs True Posit ives). Thus, the l imi ted research findings i n previous literature are inconsistent w i t h the findings o f the present study. In the current study, parents w h o had health concerns for their chi ldren were actually more accurate in reporting on delays. However , in the sub-sample o f chi ldren without delays there was no significant associat ion between health status and parent accuracy. It seems that i n this sample, health concerns m a y have aided parents i n report ing accurately when chi ldren were developmental ly delayed. Research Quest ion T w o The second research question explored the associations between parent report on the N D D S and fami ly characteristics us ing chi-squared statistics. Results showed several significant relationships both i n the sub-sample o f chi ldren wi th delays and the sub-sample without delays. Maternal education. Resul ts from the bivariate analyses provide evidence o f a significant association between maternal education attainment and parent accuracy on the N D D S in the sub-group o f ch i ldren w i t h delays only. Mothers w h o had a high school education or less were more l i ke ly to report True Posi t ives than mothers w i t h higher levels o f education, al though they were less l i ke ly to report True Nega t ive results. Th i s result for the sub-sample o f ch i ld ren wi th delays was surprising. A l t h o u g h there has been little invest igation o f maternal educational attainment having influence o n the accuracy o f parent reporting, early studies p rov ided by Glascoe and D w o r k i n (2005) showed that mothers w i t h higher levels o f education were more accurate in their reports o f c h i l d development. 66 Resul ts f rom the current study for the sub-sample o f delayed ch i ld ren contradict Glascoe and D w o r k i n ' s f indings and the hypothesis for this study. It is u n k n o w n w h y mothers w i th lower levels o f education were l i ke ly to report True Posi t ives . Research provides evidence that h igh ly educated parents read more about development and seek advice from pediatricians more regularly than lower educated parents, w h o rely more on social comparisons w i t h other chi ldren to assess their ch i ld ' s development (Glascoe & M a c L e a n 1990). One must ask i f these mothers were being soc ia l ly desirable by not reporting delays they may have not iced. Parenting is a diff icult j o b and no parent wants to suspect or be told that their c h i l d m a y have a developmental delay. Therefore, it is reasonable that these mothers may feel ashamed, scared, or even in denial that their c h i l d cou ld have a developmental delay. Weiner (J 974) conducted studies i n v o l v i n g attribution theory. Mothers who may suspect delays may attribute the p rob lem to them or feel responsible due to any number o f reasons (self-serving bias); therefore, it is reasonable to consider that the mothers in this study knew there was something w r o n g but for any number o f reasons d id not report their concerns on the screening instalment. F o r the sub-group o f ch i ldren without delays (True Negatives) , maternal education neared s ignif icance, a trend that w o u l d be expected from the literature. A g a i n , studies supporting a connect ion between l o w levels o f education and accuracy on report ing development are scarce and contradictory depending on the decade the studies were comple ted (Glascoe & D w o r k i n , 1995). The research cited b y Glascoe and D w o r k i n reported that mothers o f l o w educational attainment tended to miss developmental delays when they are present, not over report them as the mothers d id in the current study. Howeve r , research supports the idea that mothers w i t h less education may have less knowledge o f development (Stanton-Chapman, Chapman , & Scott, 2001), and may use socia l comparisons more than higher education mothers 67 (Glascoe and M a c L e a n 1990). Therefore, it is also possible that the mothers f rom the sub-sample o f ch i ldren w i t h no delays s imply were compar ing their chi ldren to others unfavorably. Furthermore, these mothers, as stated by B o w m a n (1992), may not understand the processes o f development and i f compar ing to other chi ldren, may not be l ook ing for the correct developmental milestones. Family income. Resul ts for fami ly income were s imi la r to that for maternal education. Parents wi th fami ly incomes be low the L o w Income C u t o f f ( L I C O ) were more l i k e l y to report True Posi t ive results than parents above L I C O , but less l i k e l y to report True Nega t ive results. The former results were surpris ing. The parents w i th higher soc io-economic status were more l i ke ly to be inaccurate at ident i fying delays than lower S E S parents. Research has prov ided m i x e d results for this f a m i l y characteristic as w e l l . M c L o y d (1990), provided evidence that parents o f l ow income and l o w socioeconomic status demonstrated a lower capacity to cue into their chi ldren 's soc ia l -emot iona l needs, and were less l i k e l y to provide a supportive, car ing and consistent environment for their chi ldren. M a n y more recent studies suggest that that S E S is no longer as influential on accuracy o f parent reported developmental screening instruments because the psychometr ic characteristics o f screeners have improved (Glascoe, 1999). H o w e v e r , as discussed in Chapter 2, there are concerns wi th the statistical analysis in many o f the studies that examine S E S . Therefore, d rawing conclusions such as, S E S has no influence on accuracy, may be incorrect. A l t h o u g h a statistical association is present, it is diff icul t to speculate w h y higher income families w o u l d not identify developmental delays for the same reasons as out l ined for maternal education: higher S E S famil ies have more resources, more s t imulat ing environments (Hami l ton , 2006), and are general ly better informed about the developmental process than lower S E S 68 families ( B o w m a n , 1992). Therefore, it is puzz l ing w h y the higher S E S parents i n the present study were inaccurate in identifying delays, but accurate in ident ifying no rma l development (as the literature predicts). A g a i n , social desirabi l i ty may be an issue. Interview methods to engage parents w i t h inaccuracies and learn more about w h y they reported the w a y they d i d m a y provide insight. Family structure. U n l i k e maternal education and fami ly income, partner status was not found to have a statistically significant association wi th parent accuracy o n the N D D S . Al though the literature supports the idea that single mothers might have a more di f f icul t t ime accurately reporting o n c h i l d development than partnered mothers, the association was not found in this study among either sub-group. This is not surprising given the smal l number o f single mothers (n = 20) that were i n the study. The sub-sample size for single mothers as compared to mothers who were marr ied or had a partner was m u c h smaller and was not large enough for meaningful statistical analysis. H i g h - r i s k populations are not always w e l l represented i n these studies, and therefore, invest igat ing single motherhood as a characteristic that cou ld influence accuracy on developmental screens is st i l l wor thwhi le (Dichde l te lmi l le r et al , 1992; Henderson & M e i s e l s , 1994; Te rvo , 2005). In the future, it w o u l d be beneficial to have a more balanced populat ion o f single and marr ied mothers to better understand the influence o f f ami ly structure on accuracy o f parent reporting. Maternal age. The last f ami ly characteristic analyzed in the bivariate analyses was maternal age. Resul ts showed for the sub-sample o f chi ldren wi th delays there was no significant association between maternal age at first birth and parent accuracy on the N D D S . However , a significant association was found for maternal age in the sub-sample o f chi ldren without delays. Mothers older than age 23 were more l i ke ly to report True Negat ives than 69 younger mothers. Th i s was an interesting result, consistent w i t h the literature, and the research hypothesis. A l t h o u g h the literature is l imi ted in research relating to maternal age and accuracy o f parent reporting on development, findings by T a m i s - L e m o n d a , Shannon, and Spe l lman (2002) indicated that adolescent mothers were more l ike ly to believe that developmental milestones happened earlier then they actual ly do. This w o u l d lead us to expect lower levels o f True Negat ives among younger mothers, as was found in the present study. Research Quest ion Three In the third research quest ion the l ike l ihood that c h i l d and fami ly characteristics cou ld predict agreement between the results o f the developmental screening instrument ( N D D S ) and the results from the professional assessment (the S B 5 , the P P V T - I I I , and the B S I D - 2 ) was examined. A binary logis t ic regression was used to analyze the sub-sample o f ch i ldren wi th delays and the sub-sample o f chi ldren wi th no delays. A d d i t i o n a l sub-analyses were run to investigate the s ignif icance o f c h i l d characteristics for the sub-sample o f chi ldren wi thout delays (age and gender) focus ing on fami ly variables. F o r both groups o f chi ldren , ch i ld gender and age were s ignif icant ly associated wi th agreement status and in the same direction as w i th the bivariate analysis. In future studies it w o u l d be beneficial to study the response patterns for parents o f boys. F o r example, in terms o f the N D D S , d id the parent tend to answer "no" to the questions focusing on social emot ional development more than questions related to cogni t ion , language, or motor? I f so, do these males real ly have developmental delays or should they be referred for p sycho log ica l evaluat ion for mental health problems? F o r c h i l d age, this information w i l l be useful for developers o f the N D D S and for the developers o f other developmental screening instruments. M o r e response 7 0 choices than "yes" or " n o " may be useful for this age group as developmental milestones for the very young chi ldren seem more difficult to differentiate between accurately. N o fami ly characteristics were found to be statistically significant for the sub-sample o f chi ldren without delays after accounting for ch i ld characteristics. H o w e v e r , l o w fami ly income was found to be associated wi th greater l ike l ihood o f agreement (True Posi t ives) for the sub-sample o f chi ldren w i t h delays, even after accounting for c h i l d characteristics, but on ly when it was the on ly f ami ly variable entered into the model . Ch i ld r en and famil ies w i t h incomes lower than L I C O were more l i k e l y to be identified as delayed than chi ldren o f famil ies w i t h incomes above L I C O . A g a i n , there is a lack o f evidence i n the literature to support these findings, especial ly g iven that many studies for other developmental screening instruments do not d ivide the groups into ch i ldren wi th delays and chi ldren without delays, rather they compare families o f low income and midd le to h igh income wi th in an overa l l group o f chi ldren. The results from this study are unique because even wi th the smal l number o f chi ldren w i t h delays ( N = 52), findings provided evidence that l ow- income families were more accurate at ident i fying delays. Recent research supports the f inding that l o w income parents are accurate at report ing delays (Glascoe, 1999; Squires et a l . , 1999). Ove ra l l , these findings provide evidence that f a m i l y income is worth further invest igat ion w i t h other developmental screening instruments. Sub-analysis for children without delays. The sub-analyses were conducted solely on the sub-sample o f ch i ld ren without delays. Further sub-analyses were conducted on gender-specific and age-specific sub-samples o f the ch i ldren without delays, in order to further explore possible relationships between agreement status and fami ly characteristics. S i m i l a r sub-analyses cou ld not be comple ted w i t h the group o f ch i ldren wi th delays because o f the smaller number o f chi ldren wi th delays (52 vs. 343). Mate rna l age was the on ly fami ly characteristic found to 71 signif icant ly associated wi th agreement status, but on ly for female ch i ldren and preschool chi ldren. Thus, i n contrast to the results found through bivariate analysis, after account ing for ch i ld characteristics, f ami ly income was not found to predict agreement status for chi ldren without delays. Wha t is exc i t ing about these f indings is that w i th greater numbers o f chi ldren , it may be possible to f ind significant results among the fami ly characteristics after account ing for c h i l d age and gender, w h i c h influence sensi t ivi ty and specif ici ty. Materna l age is w i d e l y researched in the literature. Bronfenbrenner 's (1994) w o r k provides evidence that teenage mothers are at risk for l o w education levels , single parenthood, and poverty, but this does not provide evidence o f inabi l i ty to accurately report developmental delays. The results from the sub-analyses provide the evidence needed to make the l ink between maternal age and accuracy o n parent completed developmental screening instruments. Limi ta t ions o f the Study The current study had three l imitat ions. First , the sizes o f the sub-samples were less than opt imal . Second, there are l imitat ions to the N D D S . T h i r d , the extent to w h i c h the results can be generalized to other populat ions is restricted. These l imitat ions are discussed be low. Sample Size O v e r a l l , the sample size for the current study was acceptable. However , for the analyses sub-samples were created to investigate the differences between chi ldren w i t h delays and those without. Therefore, the sub-sample o f chi ldren w i t h delays was smal l (n = 52). The smal l sample size resulted in l imi ted abi l i ty to detect statistically significant relationships between fami ly characteristics and agreement wi th the N D D S and professional assessment even when c l i n i c a l l y meaningful associations existed. Th i s was not just the case for the fami ly characteristics, but also 72 for ch i ld health status. A l t h o u g h c h i l d health status proved valuable i n the bivariate analysis, the numbers d id not support significant statistical analysis in the multivariate analysis. L imi ta t ions o f the N D D S The current study used the N D D S for the developmental screening instrument and the B S I D - 2 , or the S B 5 and the P P V T - I I I as criterion measures. T o date, there has been little val idat ion o f the psychometr ic properties o f the N D D S . The study completed in 2002 (Nagy, R y a n , & R o b i n s o n , 2002) used the A S Q as the cri terion measure and focused on the concurrent va l id i ty o f on ly the 12-month vers ion. The study by Dahinten and F o r d found that the N D D S performed w e l l for 18-month olds, but that sensit ivity rates for other age groups tended to be lower than the 7 0 % used as an acceptable cutoff for sensit ivity in the f ie ld . Dahinten and F o r d (2004) suggested that the N D D S is "effective at ident ifying chi ldren w i t h severe delays, but is less effective at ident i fying chi ldren w i t h mi lder delays" (pg. 12). Therefore, one might question the role ch i ld age actually plays on parent accuracy and what is potential ly a more inherent problem w i t h i n the developmental screening instrument at the infant and toddler age group and at the preschool age group from the findings o f the current study. Las t ly , the results o f this study may not generalize to another developmental screening instrument. A b i l i t y to Genera l ize to a Larger Popula t ion The current study was conducted using secondary data from three separate va l ida t ion studies o f the N D D S at the 4, 18, 24 and 36 month ages. A l l the parents and chi ldren invo lved were recruited from two smal l suburban middle-class communit ies i n Southeastern B r i t i s h C o l u m b i a , Canada. The results f rom the current study may not generalize to large urban environments. In addit ion, as w i t h m u c h research, the families were self-selected. A l t h o u g h the response rates were satisfactory, it is very possible that the parents who took part in the studies 73 were different f rom the parents w h o chose not to respond and take part in the studies. It is even difficult to generalize results to parents outside the study who have taken part in developmental screening using the N D D S . Some o f the families had already completed the part o f w e l l ch i ld visits . M a n y communi t ies across southern Br i t i sh C o l u m b i a use the N D D S at earlier age groups through local health centres where they were g iven the N D D S as part o f their communi ty-based health. Therefore, potential bias cou ld be impl icated because parents may have already been given information about their ch i l d ' s development. Strengths o f the Study A l t h o u g h three l imitat ions o f the study were out l ined above, the current study had several strengths. First, the need for this k i n d o f research in the developmental screening w i l l be discussed, the need for further research on the N D D S , and finally, the approach to data analysis. A s stated in the literature review, little is k n o w n about how fami ly characteristics influence parent accuracy on developmental screening instruments in general. M a n y studies use high risk populations i n the standardization samples, but do not analyze i f these fami ly characteristics actually influence parent accuracy on reporting c h i l d development. T h i s study has taken the data from two val ida t ion studies on the N D D S and provided evidence that f ami ly and ch i ld characteristics do influence accuracy on the N D D S . The N D D S is a re la t ively new developmental screening instrument; therefore this study has provided more needed research to validate i t suse w i t h parents. In addit ion, developers c l a i m that it is parent friendly. Th i s study has provided evidence useful to the developers about h igh risk populations o f parents, and w h i c h age groups are l i k e l y el ici t the most accuracy from parents. 74 The current study is novel in that the investigation o f parent accuracy was completed without c o m b i n i n g both the delayed sub-sample and the non-delayed sub-sample as most other studies have done when investigating accuracy o f developmental screening instruments. Often the number o f ch i ldren wi th delays is smal l and therefore researchers combine the groups into one "total agreement" category to run analyses. Another approach that has been used in other literature is to focus on whether or not parents have concerns, but without ident i fy ing whether or not the chi ldren actually have delays. In this study two distinct groups; ch i ld ren w i t h delays, and chi ldren without delays were kept separate for a l l analyses p rov id ing new evidence o f how fami ly characteristics influence accuracy on the N D D S . Implicat ions for Further Research Future research in this area should include the continued investigation o f f a m i l y and ch i ld characteristics that m a y influence parent accuracy on parent reported developmental screening instruments, especia l ly us ing larger sample sizes to look at these variables. A l t h o u g h the ch i ld characteristics (ch i ld gender and age) revealed significant results consistent w i t h the literature in the field, c h i l d health and the majority o f f ami ly characteristics were not found to p lay a significant role i n parent accuracy after accounting for variance in sensi t ivi ty and speci f ic i ty by c h i l d gender and age. Th i s impl ies further need for investigation into the advancement o f developmental screening instruments, as w e l l as f ami ly characteristics in general. It w o u l d be advantageous to examine the theoretical underpinnings that drive the study o f certain fami ly characteristics such as h igh-r isk parents; for it may not necessarily be the parent that is the problem but rather an inherent problem w i t h the screening instalment i tself (Dahin ten & Ford , 2004; Dich te lmeie r et a l . , 1992; Frankenburg, 1994). Other f ami ly characteristics may be valuable to investigate as w e l l . In this study, maternal depression was on ly measured in the 36-75 month-old group and therefore was not ult imately used in the analysis. H o w e v e r , maternal depression has been found to have effects on the development o f chi ldren at a l l ages (Johnson & Flake , 2007). Because little research exists about a depressed mother 's abi l i ty to report on her ch i l d ' s development it w o u l d be an addit ional important characteristic to investigate in future research. In addit ion, as stated i n the l imitat ions o f the study section, the N D D S as w e l l as other developmental screening instruments have low sensit ivity numbers for infants and toddlers; therefore, it w o u l d be valuable to continue to investigate ch i ld age as a factor in a l l developmental screening instruments. In this study, the preschool age chi ldren had very l o w sensit ivi ty and further invest igat ion o f these results w o u l d be wor thwhi le for researchers interested in use o f the N D D S w i t h this age group. A s stated previously , it c o u l d be beneficial to take a quali tat ive look at the N D D S data. F o r example, the chi-squared hypothesis testing provided a significant association between c h i l d gender and parent accuracy o n the N D D S . Speci f ica l ly , males were more l i k e l y to be reported as delayed (True Posit ives) and females were more l ike ly to be identif ied as True Negat ives . It w o u l d be beneficial to investigate i tem response by the parent for these chi ldren. Recent research provides evidence that males are often referred more for behavior and attention problems more than females are (Tervo, 2006). A study focusing o n the social and emotional ratings o f parents and gender o f the c h i l d may help shed l ight on the gender influence found i n the current study. B u i l d i n g on a quali tat ive study that investigated item response, it w o u l d be interesting to understand w h y the mothers w i t h higher education and higher income were more l i k e l y to be inaccurate for chi ldren w i t h delays. For this question, a social p sycho logy theoretical 76 perspective m a y be benefic ia l . F o r example, as related to Glascoe and M a c L e a n (1990), is social comparison d r iv ing these ratings? One might hypothesize that this sub-group o f mothers have higher expectations than are reasonable for these chi ldren and that is w h y they are not rating their chi ldren as delayed as often as mothers w i th lower education and lower incomes. Contr ibut ions to the F i e l d The present study provides several important contributions to the fields o f developmental screening, school psycho logy , and health care. First , this study is unique i n that accuracy was investigated accord ing to whether the ch i ld was delayed or not delayed. In previous research to date most studies look at overal l agreement and do not look at accuracy in sub-groups. Second, results from the current study were that ch i ld characteristics (chi ld gender and age) inf luenced accuracy o f parent report on c h i l d development. These results w i l l add to the g r o w i n g research in the development o f more accurate, and parent fr iendly developmental screening instruments. Speci f ica l ly , researchers n o w have further evidence that not on ly are males referred more often for a variety o f services ( M c K a y , Shannon, Vate r & D w o k i n , 2006; Tervo , 2006) , but that male gender is l i k e l y to influence parental report, especia l ly increasing the chances o f True Pos i t ive screening results. In addi t ion, the current study lends support to those who c l a i m parents o f older chi ldren are more accurate i n identifying normal development (Glascoe & M a c L e a n , 1990; Tervo, 2006). Furthermore, the current study lends support to the notion that maternal age is influential in parent accuracy o f parent report. G i v e n that many val ida t ion studies for developmenta l screening instruments have data on young mothers, the findings from the current study may incite further invest igat ion into the influence o f maternal age on accuracy o f parent reporting. 77 Contr ibut ions to other fields are evident as w e l l . For example, the field o f school psychology relies heavi ly on parent report, especia l ly for social and emot ional , and behavior rating scales. Therefore, understanding influences on parent accuracy are important for school psychologists to be aware o f when conduct ing psycho-educational assessments. Furthermore, many school psychologists are becoming more invo lved in early intervention services and therefore w i l l be exposed to the preschool popula t ion. School psychologists h o l d a unique posit ion when it comes to evaluating young chi ldren. U n l i k e developmental screening in general, school psychologists have more o f an opportunity to interview the parents and form a more long tenn; professional relationship w i t h them to facilitate accuracy o n parent completed rating scales. E v e n so, the increasing number o f chi ldren being referred for early intervention services w i l l increase the case loads for school psychologists . Therefore, understanding what characteristics m a y influence parent accuracy w i l l be important for the school psychologis t . Las t ly , these findings are valuable for health care providers in general. The results o f the current study provide strong evidence for more community-based services to target parents to educate them about ch i ld development. P u b l i c health nurses and physicians can be prepared to provide more support for parents o f young chi ldren and young mothers. Conc lus ions In conc lus ion , parent reported developmental screening instruments are ga in ing respect and usage across N o r t h A m e r i c a , Western Europe , and Aust ra l ia . Just as professional tests for ch i ld development have a marg in o f error; parents are not always accurate in their report o f c h i l d development on these developmental screening instruments and developmental screening is expected to be less accurate than diagnostic tests. Th is study investigated the c h i l d and fami ly characteristics that may influence a parent's abi l i ty to accurately report c h i l d development on the 78 N i p p i s i n g Dis t r ic t Developmenta l Screen, a new screening instrument developed in Canada n o w used throughout Nor th A m e r i c a . A l t h o u g h findings were mixed , many o f the results from this study provide evidence that c h i l d and f a m i l y characteristics p lay a role in the accuracy o f a parent to report development. A s research continues on the advancement o f developmental screening and the role o f c h i l d and f ami ly characteristics'it is hoped that results o f the current study are used to better inform researchers in the field. 79 References B a i l e y D . B . , Skinner , D . , & Warren , S.F. , (2005). N e w b o r n screening for developmental disabil i t ies: Ref ra in ing presumptive benefit. American Journal of Public Health, 95, 1889-1893. Bay ley , N . (1993). Bayley Scales of Development-Second Edition (BSID-I1). San A n t o n i o , T X : The Psycho log ica l Corpora t ion . B o w m a n , B . T . (1992). W h o is at r isk for what and why : A keynote address del ivered at the D i v i s i o n for E a r l y C h i l d h o o d A n n u a l Conference in St. L o u i s , M i s s o u r i on N o v e m b e r 14, 1991. Journal of Early Intervention, 16, 101-108. Br i t i sh C o l u m b i a Statistics (2001). 2001 C h i l l i w a c k Dis t r ic t Prof i le . /www:bcs ta ts .gov.ca /da ta /cen01/prof i les /PED_l l . p d f [retrieved September 23 , 2006] . Bronfenbrenner, U . , & M o r r i s , P . A . (1998). The ecology o f developmental processes. Ha ndbook o f C h i l d P s y c h o l o g y (5th Ed . ) , 1, 9.93-1028. Commit tee on C h i l d r e n W i t h Disab i l i t i es , (2001). Developmenta l survei l lance and screening o f infants and young chi ldren, Pediatrics, 108, 192-195. Dahinten, V . S . , & F o r d , L . , (2004). V a l i d a t i o n o f the N i p i s s i n g Dis t r ic t Deve lopmenta l Screen for use w i t h infants and toddlers -working paper. Consor t ium F o r Hea l th , Intervention, Lea rn ing and Deve lopment ( C H I L D ) . Retr ieved: N o v e m b e r 8, 2004. Dahinten, V . S . , Ford , L . , C a n a m , C , Lapointe , V . R . & M e r k e l , C . (2007). The Chilliwack developmental screening study. In J . Hoube (Chair ) , The ecology o f early screening and fo l low-up programs. S y m p o s i u m paper presented at the 2007 B i e n n i a l M e e t i n g o f the Society for Research in C h i l d Development , Bos ton . Dichdel te lmi l le r , M . , M e i s e l s , S.J., Plunket t , J . W . , B o z y n s k i , M . E . A . , C l a f f l i n , C . & Mangelsdorf , S .C . (1992). The relationship o f parental knowledge to the development o f extremely l o w birth weight infants. Journal of Early Intervention, 16, 210-220. D iamond , K . E . , & Squires, .1. (1993). The role o f parental report in the screening and assessment o f young chi ldren. Journal of Early Intervention, 17, 107-115. Dunn , L . M . , D u n n , L . M . , W i l l i a m s , K . T . , W a n g , J . J. (1997). Peabody Picture Vocabulary Test, Third Edition (PPVT-lll). C i r c l e Pines, M N : A m e r i c a n Guidance Service , Inc. Eyberg, S. (2000). Eyberg Child Behavior Inventory. Odessa, F L : P sycho log i ca l Assessment Resources. Frankenburg, W . K . (1994). Prevent ing developmental delays: Is developmental screening sufficient? Pediatrics, 93, 586-593. Furstenburg, F . F . , B r o o k s - G u n n , J . , & M o r g a n , P . S. (1987). Adolescent mothers in later life. N e w Y o r k , N Y : Cambr idge U n i v e r s i t y Press. Glascoe , F . P . , (1999). The value o f parents' concerns to detect and address developmental and behavioral problems. Journal of Paediactric Child Health, 35, 1-8. Glascoe , F . P . , (1997). Parents' concerns about chi ldren 's development: Prescreening technique or screening test? Pediatrics, 99, 522-528. Glascoe , F . P . , (2000). Evidence-based approach to developmental and behavioural survei l lance us ing parents' concerns. Child: Care, Health, and Development, 26, 137-149. Glascoe , F . P . , Foster, E . M . , & W o l r a i c h , M . L . , (1997). A n economic analysis o f developmental detection methods. Pediatrics, 99, 830-837. Glascoe , F . P . , & D w o r k i n , P . H . , (1995). The role o f parents in the detection o f developmental and behavioral problems. Pediatrics, 95, 829-836. 81 Glascoe, F . P . , (1999). U s i n g parents' concerns to detect and address developmental and behavioral problems. Journal for Specialists in Pediatric Nursing, 4, 24-35. Glascoe, F . P . , (2001). A r e overreferrals on developmental screening tests real ly a problem? Archives of Pediatric and Adolescent Medicine, 155, 54-59. Glascoe, F . P . , (2005). Screening for developmental and behavioral problems. Mental Retardation and Developmental Disabilities Research Reviews, 11, 173-179. Glascoe, F . P , M a r t i n , E . D . , & Humphrey , S., (1990). A comparative review o f developmental screening tests. Pediatrics, 86, 547-554. Glascoe, F . P . , & M a c L e a n , W . E . , (1990). H o w parents appraise their ch i ld ' s development. Family Relations, 39 ,280 -283 . Glascoe , F . P . (1999). The value o f parents' concerns to detect and address developmental and behavioural problems. Journal of Paediatric Child Health, 35, 1-8. Glascoe, F . P . (1997). Parents' concerns about chi ldren 's development: Prescreening technique or screening test? Pediatrics, 95, 522-528. Glascoe, F . P . (2003). Parents' evaluat ion o f developmental status: H o w w e l l do parents' concerns identify chi ldren w i t h behavioral and emotional problems? Clinical Pediatrics, 42, 133-138. Hami l ton , S., (2006). Screening for developmental delay: Re l iab le , easy-to-use tools. Journal of Practice, 55, 415-422. Ireton, H . (1994). Child Development Inventories Behavior Science Systems. M i n n e a p o l i s , M N : Behav io r Science Systems. 82 Jackson, A . P . , B r o o k s - G u n n , J . , Huang , C . C . , & Glassman , M . (2000). Single mothers in l o w -wage jobs: F inancia l strain, parenting and preschooler 's outcomes. Child Development, 71, 1409-1423. Krause , N . M . , & Jay, G . M . (1994). W h a t do g lobal self-rated health items measure? Medical Care, 32, 930-942. Lagerberg, D . , (2005). Parental assessment o f developmental delay in chi ldren: Some l imitat ions and hazards. Acta Paediatrica, 94, 1006-1011. L e B u f f e , P . A . & N a g l i e r i , J . A . (1999). Devereaux Early Childhood Assessment, Technical Manual. L e w i s v i l l e , N C : K a p l a n E a r l y Lea rn ing Company . L e e , L . L . S . , & Harr is , S.R., (2005). Psychomet r ic properties and standardization samples o f four screening tests for infants and young chi ldren: A review. Pediatric Physical Therapy, 17, 140-147. M c C a r t h y , D . (1972). The McCarthy Scales of Children's Abilities. N e w Y o r k : Psycho log ica l Corporat ion. M c K a y , K , Shin , A . , Vater, S., & D w o r k i n , P . H . (2006). C H I L D S E R V , Lessons learned from the design and implementat ion o f a communi ty-based developmental survei l lance program. Infants and Young Children, 19, 371-377. M c L o y d , V . C . (1990). The impact o f economic hardship on black famil ies and chi ldren: Psycho log ica l distress, Parenting and socioemot ional development. Child Development, 60, 311-348. M e i s e l s , S.J., W i s k e , M . S . , Henderson, L . W . , M a r s d e n , D . B . , & B r o w n i n g , K . (1992). The Early Screening Inventory (Revised). A n n A r b o r , M I : The Unive r s i ty o f M i c h i g a n . 83 M e n a r d , S. (2002). Applied logistic regression analysis, Second Edition. Thousand Oaks, C A : Sage Publ icat ions . M o n t g o m e r y , M . L . (1999). Use o f the C h i l d Deve lopment Inventory to screen high-r isk populations. Clinical Pediatrics, 38, 535-539. M o o r e , K . A . , M o r r i s o n , D . R. & Greene, A . D . (1997). Effects on the chi ldren born to adolescent mothers. In R. A . M a y n a r d (Ed.) , Kids having kids: Economic costs and social consequences of teen pregnancy (pp. 145-180). Washington , D C : U r b a n Institute Press. M u n r o , B . H . (2005). Logis t ic regression. In M u n r o , B . H . (Ed.) . Statistical methods for health care research, fifth edition (pp. 301-311). Phi ladelphia , P A : L ipp inco t t W i l l i a m s & W i l k i n s . P i n t o - M a r t i n , J . A . , Dunk le , M . , Ear ls , M . , Fl iedner , D . , & Landes , C , (2005). Developmenta l stages o f developmental screening: Steps to implementat ion o f a successful program. American Journal of Public Health, 95, 1928-1932. Reyno lds , C R . & Kamphaus , R . W . (1998). Behavior Assessment System for Children. C i r c l e Pines , M N : A m e r i c a n Guidance Service , Inc. R o i d , G . H . (2003). Stanford-Binet Intelligence Scales. Fifth Edition, Examiner's Manual. Itasca, I L : R ivers ide Publ i sh ing . Squires, J . (1996). Parent-completed developmental questionnaires: A low-cost strategy for ch i ld - f ind and screening. Infants and Young Children, 9, 16-28. Squires, J . , Potter, L . , & Br icker , D . (1999). The ASQ user's guide: Second edition. Bal i t imore: P a u l H . Brookes . 84 Squires, J . K . , Potter, L . , B r i c k e r , D . D . , & Lamorey , S. (1998). Parent-completed developmental questionnaires: Effectiveness wi th l ow and midd le income parents. Early Childhood Research Quaterly, 13, 345-354. Stanton-Chapman, T . L . , Chapman , D . A . & Scott, K . G . (2001). Identification for early r isk factors for learning disabil i t ies . Journal of Early Intervention, 24, 193-206. T a m i s - L e m o n d a , C . K . , Shannon, J . , & Spel lman, M . (2002). L o w - i n c o m e adolescent mothers ' knowledge about domains o f ch i ld development. Infant Mental Health Journal, 23(1-2), 88-103. Tervo , R . C . (2006). Parent 's reports predict their ch i l d ' s developmental problems. Clinical Pediatrics , 601 —611. Tur ley , R . N . L . (2003). A r e chi ldren o f young mothers disadvantaged because o f their mother 's age or fami ly background? Child Development, 74, 465-474. V i n g i l i s , E . Wade, T . J . , & A d l a f , E . (1998). What factors predict student self-rated phys ica l health? Journal of Adolescence, 21, 83-97. ^ 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0054354/manifest

Comment

Related Items