1st Contact data collection
The 1st Contact study was (in all but a few cases) the first point of contact between TEDS and the 16810 families in the ONS sample. As explained above, the 1st Contact study started and continued while families were being recruited by ONS.
The 16810 families in the ONS sample (with a few exceptions, see below) were invited to take part in the 1st Contact study; hence the study included all families in the TEDS sample, with twin births between January 1994 and December 1996. This initial 1st Contact data collection took place between 1995 and 1998. Data collection was not precisely timed to coincide with a particular twin age, but most twins were aged between 1 and 2 years when the booklets were returned, with a median age of around 1.6 years. Since 1998 there have been several further attempts to collect the data from families that did not respond in the initial phase.
The data for this study was collected in a single booklet (pdf), to be completed by the parent or guardian of the twins. The booklet included a consent form on the first page. The booklet was sent along with a covering letter and an information leaflet (pdfs). The pack was addressed to the contact parent identified by ONS, but it was not necessary for the same parent to complete the booklet.
Regular reminders were sent to families who did not return the booklet promptly. Up to 5 reminders were sent, over a period up to 11 months after the original booklet was sent to each family. In some cases, a second copy of the booklet was sent or a phone call was made.
TEDS sent the 1st Contact booklet to all except roughly 500 of the 16810 families in the ONS sample. The 500 or so exceptions can be explained in various ways. In some cases, the postcards returned to ONS contained ambiguous responses, or were spoiled in some way. When TEDS contacted these families for clarification, before sending the 1st Contact booklet, a few of them changed their minds and withdrew from TEDS. Other families returned their postcards late, and were sent the 2 Year booklets before the 1st Contact booklet (because the twins had reached their second birthdays); some of these families withdrew from TEDS, or became address problems, before the 1st Contact booklet could be sent. In all, around 450 families withdrew from TEDS, and around 50 became address problems, before the 1st Contact booklet could be sent.
There are further pages describing the 1st Contact data returns, and comparisons of samples and returns for different TEDS studies.
Because of the essential nature of the TEDS 1st Contact data, there have been several further attempts to collect the data from families who did not return it when first asked (especially if they returned valuable data in a later study). In the "You and Your Family" study, carried out in 2000 and 2001, a shortened form (pdf) of the 1st Contact booklet was mailed to families that had returned data at ages 2, 3 or 4, but had not returned the original 1st Contact booklet (it was renamed to "You and Your Family" because it was no longer the first contact with the family).
Between 2004 and 2006, telephone interviews were used to gather the data, using the same shortened form of the booklet; especially from families that had returned data at age 7 or later.
In 2013, another attempt was made to collect 1st contact data (by post and by telephone), using an even shorter version of the booklet; targeted families were those that had participated at 12, 14 or 16.
Inactive families and the TEDS sample
The ONS recruitment process identified 16810 families who had expressed an interest in taking part in TEDS. However, it was in the 1st Contact booklet that a parent from each family gave consent to take part in the TEDS study. The first page of the 1st Contact booklet was a consent form, explicitly for this purpose.
The 1st Contact data collection eventually provided TEDS with a sample of 13722 families who had given explicit consent to take part in the study. Although the "ONS sample" included 16810 families, the effective initial "TEDS sample" of consenting and participating families included roughly 13700 families.
Hence, there were roughly 3000 families who had expressed an interest in TEDS to ONS but did not complete the 1st Contact booklet and therefore did not give explicit consent for ongoing participation. As described above, further attempts were made to involve these families in TEDS, both in follow-up waves of 1st Contact data collection and in subsequent data collections at ages 2, 3, 4 and so on.
Eventually, in 2004, following the 9 Year study and before the 10 Year study, a decision was made in TEDS not to contact such families again; namely, those families who had not returned 1st Contact data and had not returned data in other studies between age 2 and age 9. These families were labelled as "inactive", and have not been contacted by TEDS since 2004.
In 2021, TEDS recognised that it was necessary to delete any remaining data for the inactive families: parent and twin names, old address records, twin birthdates and twin sexes, where recorded, were deleted. Hence, TEDS no longer holds identifiable records for these inactive families.
The TEDS sample may now be quantified as follows:
- ONS sample: 16810 families
- Inactive: 2865 families (records now deleted)
- TEDS sample: 13945 families
While there are 13722 families with 1st Contact data, there are around 200 further families who have provided later data but not 1st Contact, hence the total of 13945 families in the TEDS sample. By providing later data (for example at age 4 or age 7) such families gave their consent to take part in TEDS, even if they had not done so by completing 1st Contact.
1st Contact data entry
General data entry issues, applicable to all TEDS studies, are described in another page.
In the first phase of the 1st Contact study, for the 1994 cohort and some of the 1995 cohort, the booklets were entered in-house by TEDS staff. This data entry was done using Microsoft Access databases, which were specifically designed for this purpose. Later in the study, data entry was handled externally by a commercial company called NOP Numbers. They used their own data-entry software system, and returned the data to TEDS in Excel spreadsheet files. The data were extracted from these various raw data files, from within TEDS and from NOP Numbers, and were aggregated together into a single Access database. Information about the tables in this database is given in the 1st Contact data files page.
All raw data, entered by various means, are now cleaned and aggregated together and stored in the Access database file. Data cleaning is described in general terms in the TEDS data cleaning page. This Access database is now treated as the master copy of all 1st Contact raw data for construction of the dataset.
Data from the later data collections (2004 to 2006, 2013) were directly entered into this same Access database by TEDS staff. The database incorporated forms and programs that enforced various data rules, to ensure that only clean data were entered.
In all booklets, the "job title" responses for the parents were numerically coded using SOC and social class coding before data entry. This was done by TEDS staff, even for the later booklets that were sent to NOP Numbers for data entry.
The coding for other questions in the 1st Contact booklet involved the conversion of tick-box responses to numerical codes. This coding was done by the data-entry staff themselves, both in TEDS and at NOP. A coding manual (pdf) was provided for coding staff (in this document, references to NUMERIC1 and NUMERIC2 are to names of tables in the original data-entry Access databases). Further guidance for coding particular questions was also written in a document of coding instructions (pdf). Some of the codings described in these original documents have subsequently been modified during data cleaning.
The annotated 1st Contact booklet (pdf) provides detailed documentation of the variable names and value coding, both in the retained, cleaned raw raw data and in the current dataset.
The 1st Contact booklet included over 50 questions inviting text responses. Many of these followed other questions with instructions such as "if yes, please describe". The approach to data entry of verbatim text varied between questions, but generally fell into the following categories:
- Text data collected for admin reasons. These text were generally used at the time of data collection to update the TEDS admin database system. However, they have not been retained in the raw data and are not used in the dataset. Examples include names and contact details (consent form and page 1), sibling details (pages 4 and 5) and the medical records consent form (page 16).
- Text responses that were immediately coded into numeric form, before data entry. The verbatim text responses themselves were not entered and are not in the dataset. This applied to the respondent and partner job titles, numerically coded into SOC and social class groups (pages 9 and 10).
- Text responses that were entered verbatim in the original raw data. Due to changes in data entry arrangements, the text were entered from nearly all booklets from the 1996 cohort, a minority of booklets from the 1995 cohort, and almost no booklets from the 1994 cohort. Where the text were entered in the raw data, they have now been coded into numeric categories. Most examples can be found in the "Mother's pregnancy and birth" section (pages 11 to 14).
- Text responses that were not entered at all in the raw data. This included most instances in the "Your twins day to day" section (pages 2 to 5) and in the "Identical or non-identical?" section (pages 6 to 8).
Documentation of which text responses were entered in the raw data and how they were coded can be found in the annotated 1st Contact booklet (pdf). As indicated above, the coded employment categories (SOC and social class) for respondent and partner are present in the data for all cohorts. Codes derived from text responses in other questions are only present for the 1996 cohort and a small part of the 1995 cohort, from text originally entered by NOP. For the other booklets of the 1994 and 1995 cohorts, entered within TEDS, the convention was simply to enter the coded value 1 to show that some text existed, or the missing code (-99) if no text existed.