Language Development and Enjoyment of Reading: Impacts of Early Parent-Child Activities in Two Growing up in Scotland Cohorts

Footnotes

1. http://growingupinscotland.org.uk/

2. At the time of writing, the Primary 6 fieldwork is ongoing.

3. At the time of writing, the age 5 fieldwork is ongoing.

4. During 2012 and 2013, the PlayTalkRead buses visited every local authority in Scotland. However, there were differences in the number of visits made to each local authority, ranging from just one visit in West Dunbartonshire and West Lothian to 71 in Glasgow.

5. Dialogic Book Reading is a technique used to improve children's oral language and preliteracy skills. It is a specific way of sharing a book with the child where instead of just reading a book to the child, the adult actively engages the child in a conversation by talking about what they see in the pictures. A variety of techniques can be used to do this, the most common being asking simple "what" questions, e.g. "What do you call this?" and then the parent repeats the child's answer back to them. The adult also expands on what the child says "What is the dog doing?", "What's your doggie called?" and then, if the child can manage these questions, asking open ended questions, e.g "What do you think he's trying to do?".

6. The GUS sample is generated in two stages. The first stage randomly selects geographic areas or clusters, the second stage selects individuals within those clusters. The standard errors are adjusted to take account of the geographic clustering of the sample at the first stage.

7. Cases with missing values on any of the variables were excluded from the models.

8. A detailed descriptive comparison of cognitive ability scores across the two cohorts at age 3 is provided in Bradshaw, Knudsen and Mabelis (2015).

9. A summary of the socio-economic and other characteristics of families in each cohort when the child was aged 3 is provided in the Appendix B. For a more detailed consideration of these differences please see Bradshaw, Knudsen and Mabelis (2015).

10. Full results are provided in Table A.1.1 in the Technical Annex.

11. Full results are provided in Table A.1.2 in the Technical Annex.

12. The answer categories for this question differed slightly between the cohorts. For BC2, the list of answer categories consisted of four items: By him/herself, with his/her mother, with his/her father, or with someone else. For BC1, the list of answer categories was more extensive, also including, for example, grandparents and nursery teachers. For the purposes of analysis in this report, the BC1 categories have been combined to match the BC2 categories.

13. Factors controlled for: Parental level of education, family type, number of children in the household, whether child was first born, the child's sex, languages spoken in the household, and employment status of the child's main carer. Outcomes are presented in Tables A.2.1 and A.3.1 in the Technical Annex.

14. The figures presented here differ slightly to those reported by Bradshaw et al. (2013) though the pattern of change is the same ( i.e. an increase in reading in BC2). This is due to a difference in the bases used for the analysis. The analysis conducted by Bradshaw et al. (2013) was restricted to singleton births and cases where interviews were conducted with the child's natural mother, while the analysis presented in this report includes all cases where valid data was available ( i.e. the analysis presented here includes cases which were not singleton births and cases where the interview was conducted with someone other than the child's natural mother). The prior analysis was part of a broader section on parenting where it was deemed appropriate to minimise variation on these factors.

15. Multivariable regression analysis was undertaken. Outputs are provided in Tables A.4.1 and A.4.2 in the

16. Doing frequent activities is defined here as having a score of at least 16 out of a possible total of 28 across the four types of activities. This is equivalent to having done, on average, each of the four activities on at least four days in the last week.

17. Frequency of playing at recognising letters, words, shapes or numbers at age 3 did not vary significantly by area deprivation.

18. For both age 10 months and 3 years, figures showing frequency of activities by household income and level of area deprivation are provided in appendix C.

19. Here defined as those with no formal qualifications, and those with lower Standard Grades or equivalent.

20. Multivariable logistic regression analysis was undertaken. Factors controlled for: Parental level of education, number of children in household, whether child was first born, family type, languages spoken in the household, child's sex, and employment status of child's main carer. Note that this analysis was undertaken for all cases across the two cohorts, i.e. across all levels of parental education. Analysis outcomes are presented in Table A.5 in the Technical Annex.

21. Multivariable logistic regression analysis was undertaken to explore factors associated with a child having done all four activities with his or her father. Note that the analysis included only cases where 1) a child had undertaken all four measured activities in the last week and 2) where a partner was present in the household (family type was therefore not included as an independent variable). Analysis outputs are provided in Table A.6 in the Technical Annex.

22. In addition to partner's employment status, the following factors were also significantly associated with children having done all four activities with their father: cohort, parental level of education, whether child was first born, number of children living in the household, languages spoken in the household, child's sex, and maternal employment status.

23. The vast majority of those who were read to frequently at 10 months were also read to frequently at age 3 (95% in both cohorts). Thus including both measures risks a high degree of overlap in the cases they identify which can affect the results of the analysis. By not including reading activities in the age 3 activity variable it is possible to test if there are any independent associations between non-reading activities and expressive vocabulary while still being able to test associations with frequency of reading through the inclusion of the measure of frequency of reading at 10 months.

24. See Table A.7.1 in the Technical Annex.

25. An interaction effect between parental education and each of the two activity measures (frequency of being read to at 10 months and frequency of doing non-reading activities at age 3) was added to the multivariable models to test whether the association between doing frequent activities and better vocabulary differed according to parental level of education. This tested, for example, whether the association was stronger or weaker amongst children whose parents had lower levels of education, compared with those whose parents had higher levels of education. This analysis was carried out separately for each of the two cohorts. Outputs are provided in Table 7.2 in the Technical Annex.

26. Those who remembered receiving the Bookbug pack but who didn't remember if they had used any of the resources have been classified as 'Received but did not use'.

27. Charts outlining receipt and use of Bookbug by household income and level of area deprivation are provided in Appendix C.

28. Charts outlining the % of parents who had accessed PlayTalkRead by household income and level of area deprivation are provided in appendix C.

29. See outcomes in Tables A.8 and A.9 in the Technical Annex.

30. This was done by fitting an interaction effect between receipt and use of Bookbug and parental level of education to the multivariable regression model predicting frequency of reading. Outputs are provided in Table A.8 in the Technical Annex.

31. Multivariable logistic regression analysis was undertaken. Further factors controlled for: Number of children in household, whether child was first born, family type, languages spoken in the household, child's sex, and employment status of child's main carer. Analysis outputs are provided in Table A.10.1 in the Technical Annex.

32. Multivariablee linear regression analysis was undertaken. The factors included are those outlined previously. Outputs are provided in Table A.11 in the Technical Annex. The activity variable used as the dependent variable at age 3 included all four activities measured: Reading or looking at books; singing songs or reciting nursery rhymes; drawing or painting; and playing at recognising letters, words, numbers or shapes.

33. This relationship was also significant when using an activity variable consisting only of non-reading activities, i.e. singing songs or reciting nursery rhymes; drawing or painting; and playing at recognising letters, words, numbers or shapes. Outputs are provided in Table A.12 in the Technical Annex.

34. An interaction effect was fitted between accessing PlayTalkRead website and parental level of education in the multivariable regression models predicting frequent activities at 10 months and age 3. Outcomes are provided in Tables A.10.1 and A.11 in the Technical Annex. The activity variable used as the dependent variable at age 3 included all four activities measured: reading or looking at books; singing songs or reciting nursery rhymes; drawing or painting; and playing at recognising letters, words, numbers or shapes.

35. Separate multivariable regression models were run for sub-groups of parents with different levels of education: 1) Lower levels of parental education, incl. No qualifications, Lower Standard Grades ( SGs) or Vocational Qualifications ( VQs) or 'Other' qualifications, and Upper level SGs or Intmed VQs; 2) Higher levels of parental education, incl. Higher Grades or Upper level VQs and Degree level or VQs). Analysis outputs are provided in Table A.10.2 in the Technical Annex.

36. First, an interaction effect between receipt and use of Bookbug and parental level of education was fitted to the multivariable regression model - results are provided in Table A13.1 in the Technical Annex. Second, separate multivariable regression models were run for sub-groups of parents with different levels of education: 1) Lower levels of parental education, incl. No qualifications, Lower SGs or VQs or 'Other' quals, and Upper level SGs or Intmed VQs; 2) Higher levels of parental education, incl. Higher Grades or Upper level VQs and Degree level or VQs). Results from this analysis are provided in Table A.13.2 in the Technical Annex.

37. Results of this multivariable analysis are shown in Tables A.14.1 and A.14.2 in the Technical Annex.

38. See Tackling Inequalities in the Early Years: Key Messages from 10 years of the Growing Up in Scotland Study (Scottish Government, 2015) for a summary of evidence on grandparental involvement from GUS.