CHAPTER II. METHODS AND DESIGN 61
I. METHODS FOR THE QUANTITATIVE STUDY (STUDY 1) 67
I.3 Instruments used for the quantitative study 71
The data for the quantitative study comes from two sources: 1. UBC´s data on children´s language and literacy skills, which were measured with the Spanish Woodcock Muñoz Language Survey-‐Revised battery (WMLS-‐R), and 2. UBC´s data on these children´s families, demographics and HLLE, acquired through the application of a parent questionnaire developed by the UBC Project.
1.3.1 Children’s language and literacy skills’ data measured with the Spanish Woodcock
Muñoz Language Survey-Revised battery (WMLS-R)
Children´s emergent language and literacy skills were measured with the Woodcock Muñoz Language Survey Revised Spanish Form (WMLS-‐R; Woodcock, Muñoz, Sandoval, Ruef, & Alvarado, 2005). The WMLS-‐R is a major revision of the Woodcock-‐Muñoz Language Survey (WMLS) (Woodcock & Muñoz-‐Sandoval, 1993, 2001) and the tests from the Spanish version were adapted from the parallel tests in English and reviewed by professionals from different regions of the Spanish-‐speaking world. The WMLS-‐R was normed on 8,800 individuals between two and 90 years of age, belonging to more than 100 geographically diverse communities from Argentina, Costa Rica, Mexico, Peru, Puerto Rico, Spain, and the United States. For each test, the calibration-‐equating data was obtained from approximately 2,000 native Spanish-‐speaking subjects from the US (reviewed by Barata, 2011). Past research has demonstrated high levels of internal reliability and predictive validity for these three subtests and also that the reliability and the validity of the Spanish Baterıa tests are comparable to the English WJ III versions (Schrank, McGrew, Ruef, Alvarado, Muñoz-‐Sandoval & Woodcock, 2005). The WMLS-‐R battery is composed of seven tests: Test 1: Picture Vocabulary, Test 2: Verbal Analogies, Test 3: Letter-‐Word Identification, Test 4: Dictation, Test 5: Understanding Directions, Test 6: Story Recall, Test 7: Passage Comprehension. The UBC project only used four tests, however, which measured the following areas:
Test 1. Vocabulary: this test measured receptive and expressive vocabulary by asking children to name the objects represented in certain pictures. This test had a median reliability of .91 in the five to 19 age range (Woodcock, Muñoz-‐Sandoval, Ruef, & Alvarado, 2005).
Test 3. Letter-‐Word Identification: this test evaluated letter and word reading identification skills. The first items measured children´s abilities to identify some
letters and the later items measured the children´s abilities to read some isolated words. This test had a median reliability of .97 in the five to 19 age range (Woodcock et al., 2005).
Test 4. Dictation: this test assessed prewriting skills (for example tracing or copying letters), as well as letterform, spelling, punctuation, capitalization and word usage. The administrator dictated the tasks orally to the child. This test had a median reliability of .94 in the five to 19 age (Woodcock et al., 2005).
Test 7. Passage Comprehension: this test assessed children´s understanding of orally-‐ read written discourse. The tasks included matching a rebus with a picture of the object, pointing to the picture represented by a phrase, and also inferring the missing word or words from a provided passage. This test had a median reliability of .82 in the five to 19 age range (Woodcock et al., 2005).
Appendix D shows the test summary and reliability statistics reported by Woodcock et al., (2005) for the aforementioned four tests, for children aged three to five years, which is the range of ages of the sample in the current study.
All 1,132 children in the current study´s sample were assessed by UBC through these four WMLS-‐R subtests at three different times during the two years of the UBC intervention. Research assistants, trained by UBC, administered the subtests individually in the children’s respective schools.
Since the current research aims at exploring and understanding the existing HLLE of low SES children as they enter preschool, the emergent literacy skills data used was the data collected by UBC during the first of the three measurement periods. More specifically, the data collected at the beginning of (pre-‐K), which corresponded to April 2008 for cohort 1 students and April 2009 for Cohort 2 students.
The scores from the WMLS-‐R tests were interpreted using a SPSS syntax created by the UBC team for this purpose (this syntax can be seen in Appendix D).
1.3.2 Children’s family and home data measured using a parent questionnaire
Even though the importance of the HLLE is widely recognized, there are few instruments for measuring the HLLE, which are in Spanish and also have reported reliability scores. This is partly due to the fact that there are few quantitative studies of the HLLE of Latino populations and most of these are of Latino immigrants to the US. Hence, they tend to use HLLE-‐measuring instruments in English.
reliability exist, include: 1) Romero-‐Contreras Family Environment Survey (Encuesta sobre Ambiente Familiar, EAF) (Romero-‐Contreras, 2006), 2) the Spanish version of the Familia Inventory of family literacy practices (Taylor, 1996), and 3) a version of the Home Observation for Measurement of the Environment (HOME) by Caldwell & Bradley which was translated in Chile in 1979 by a team of researchers led by M.I. Lira from CEDEP (Centro de Estudios de Desarrollo y Estimulación Psicosocial).
In Chile, Bustos, Herrera & Mathiesen (2001) applied the HOME to 60 four-‐ to five-‐ year old preschoolers from low SES and middle SES backgrounds and concluded that the instrument was reliable and feasible in the Chilean context. However, there is evidence to the contrary from other Chilean-‐based studies. For example, in 1979, the CEDEP team applied their translated version of the HOME to a sample of 60 low SES families and then, in 1981, reapplied it to a sample of 60 medium to high SES families. They reported that its validity and reliability was lower than it was reported for the same instrument by US studies.
Similarly, Gonzalez, Taylor, McCormick, Villareal, Kim, Perez, Darensbourg & Haynes (2011) examined the Spanish version of the Familia Inventory of family literacy practices (Taylor, 1996) and provided evidence that the Inventory varied between the Spanish and English versions. Moreover, they also said that there was a poor fit between the model suggested by the structure of the inventory and the results (the Spanish data) they analysed.
Romero-‐Contreras mentioned two other challenges that are also related to the
selection of instruments for measuring the HLE. The first is that some instruments are not appropriate for a study with a large sample because their administration is time-‐ consuming, involving home visits and application on a one-‐to-‐one basis (Romero-‐ Contreras, 2006). This is the case for the HOME instrument, for example (Caldwell & Bradley, 1984), which involves an interview with the caregiver and a semi-‐structured home observation of the parent-‐child dyad. This is also the case for the Ecological Inventory Form (Baker, Sonneschein, Serpell, Fernandez-‐Fein, & Sher, 1994), which gathers data from interviews with parents and through analysis of parental diaries. A second challenge is that some of the instruments require the student to respond; this, of course, makes them unsuitable for studies of younger children.
UBC´s parent questionnaire was based on Silvia Romero-‐Contreras´s Family
Environment Survey (Encuesta sobre Ambiente Familiar, EAF). The EAF is a survey that has eight components. Its author tested this instrument with a sample of 247 Costa Rican children. Most of its items were relevant in characterizing language and literacy
practices; in fact Romero-‐Contreras’s analysis indicated that all components had moderate reliability as assessed through the Cronbach´s Alpha. Moreover, the EAF also had predictive validity when tested by its author (Romero-‐Contreras, 2006).
The UBC questionnaire was 18-‐pages long and had 97 questions, most of which were in multiple-‐choice format. It considered HLLE and family variables. It gathered
information about several aspects of the family, including the socio-‐economic status (SES) (income, working status, educational level of the parents, home equipment), language and literacy resources in the home (number of books, magazines), parental educational aspirations and expectations and developmental and literacy-‐related beliefs. Furthermore, it asked about home language and literacy practices, such as the frequency of shared reading or conversations with the child. Finally, it also gathered data on aspects that were not considered for this study such as the child´s general health, socioemotional development, special needs and it also enquired about the presence of depressed adults in the child´s home.
It is important to note that UBC made some changes to the parent questionnaire forms between cohort 1’s time 1 (beginning of pre-‐K) and time 2 (end of pre-‐K academic year). For example, the questions relating to socioeconomic status (SES) of the family (income, parent education, etc.) were modified and asked in much more detail at time 2 (see below). Since the variables incorporated were important for this study, it has considered the SES data gathered at time 2. As such, this study is based on an assumption that the UBC intervention had no effect on families SES between times 1 and 2 (the beginning and end of the first year of transition).
The variables relating to frequency of watching TV and playing video games and time spent playing outside the home were considered to be important for the present study and UBC only included them in the parent questionnaire from Cohort 1-‐time 2.
Allegedly these practices could have been affected by UBC´s first year of intervention, which, in part, emphasized the importance of reducing hours of watching TV and improving language and literacy interactions with the children at home. Therefore, the data relating to the cohort 1 children who had participated in the UBC intervention was not considered for these variables. This had the effect that, for these variables, data for 154 of the 1,132 children was excluded.
The UBC research assistants (‘RAs) were responsible for administering the parent questionnaire. Mostly, they handed the questionnaire to the parents during meetings at the children´s schools, and ensured it was filled in. However, in those cases when the parents did not attend the meeting at school, some of the RAs went to the children´s
homes to gather the information. In certain cases, the RAs filled in the information while the parent answered the questionnaire verbally.
Most of the relatives who answered the parent questionnaire (´PQ´) were the mother or the father of the child; however, in certain cases, some other relative or caregiver, who attended the meeting at school or was at home when the RAs visited, answered it. Table 2.5 shows the participants’ relationship to the child(ren) they reported on.
Several researchers have pointed out the limitations of questionnaires and self-‐reports for studying the HLLE. Issues such as social desirability biases or caregivers’ difficulty to fully understand the questions and estimating the frequencies of their practices have been underlined. However, the use of survey reports in HLLE studies has been partly validated by the fact that the data they provide has proven to correlate significantly with observational and diary assessments (Burgess, 2002). Also, as pointed out by Sénéchal et al. (1998), parental self reports have the advantage that they allow the researcher to gather information on a wider range of language and literacy beliefs, expectations and practices that otherwise would have to be collected through the application of several single instruments.