• No se han encontrado resultados

DISPOSICIONES GENERALES

Whilst the corpus of research described in this thesis reveals novel and interesting results, there are a number of limitations to address in summarising this work. Firstly, although the CCDS as a whole employs a representative sample, the subsample featured in this thesis differs in demographic characteristics to the full sample. Specifically, mothers in the sample used in this thesis were generally older than the national average, were slightly better educated, and more of them were classified as middle class than the national average. Despite these differences, the representation of the UK population is not as skewed as that in other smaller studies of maternal anger, and every effort has been made to control for

demographic variables wherever necessary, to reduce their potential confounding effect. Of course, there may still be some patterns in change or association that would emerge in a sample that better represents the population. For example, as noted in Chapter 6, the incidence of antenatal depression is reduced in this sample as compared to the full CCDS sample, which may have reduced the possibility of finding significant associations with infant outcomes. The findings suggest that further investigation on a fuller representative sample is called for, together with specific work involving those families at particular high risk. Investigations into mothers and families who enter parenthood as teenagers, or who at high risk in other respects, would allow specific patterns of association in these groups; the risks conferred by such families may differ qualitatively and quantitatively from those in the broad sample in this thesis.

Two major factors influenced the choice of sample and its demographic representation. Firstly, it was necessary that mothers reported their levels of anger

contemporaneously for the necessary analyses in Chapters 3 and 4. A number of women were not able to complete questionnaires in the allotted time windows, and these women were younger, less well educated, and more likely to be working class. A decision was taken that the Adult Wellbeing Scale, as a measure of present mood state, should not be administered retrospectively in catch-up questionnaires. This meant that the key anger items were not administered to women who were traced and found after the time window had closed. The overall attrition in the CCDS is remarkably low, and whilst it is unfortunate that this variable was not available in all cases, the sample size remains large enough to detect the effects that were hypothesised.

Secondly, the nature of the research questions contained within Chapters 5 and 6 determined that parents be in stable relationships, and remain so at the time of the postpartum measurement. This consideration also reduced the size of the sample, and also changed the demographic characteristics; mothers in families who met the inclusion criteria appeared slightly older, better educated, and more likely to be classified as middle class than mothers who were single or in unstable relationships. However, these differences were not statistically significant between the two samples employed in this thesis (234 vs. 218), suggesting that the resulting findings that are taken from them are comparable.

Because of the focus on the course and influence of interparental anger, the analyses here do not consider the association between maternal anger and infant or toddler anger in single mothers. Reassuringly, there does not appear to be a statistically significant difference in maternal pre-childbirth anger when comparing mothers who were and were not in stable partnerships (although there may not have been the power to detect these effects). However, the manifestations of mothers’ anger in the postpartum environment, and the methods of its

transmission to the child, are likely to be qualitatively different. Furthermore, although the influence of interparental anger would not be comparable in single mothers, there may be other individuals in the mothers’ life to whom she may express anger, or receive anger from. For example, the infant’s father, grandparents, extended family or friends. This topic warrants further investigation.

A further limitation of the data concerns methodology, in particular the choice of measurement. Whilst the measures of anger contained in this thesis are fit for purpose, they could be improved upon. Given the exploratory nature of this work, and the process of conceptualising anger that occurred as the thesis developed, the measurements are not ideal, and further work is needed to further test the findings. Firstly, the measures of anger are comprised of small sets of items. Whilst they have proven efficient in measurement, and have allowed a number of significant findings to emerge, the small number of items may not capture a wider range of scores that a more fine grained measure may allow. For the purpose of exploring the topic, as has been the case in this thesis, the measures are viable.

Secondly, whilst efforts were made to validate each questionnaire-rated measure where possible, further work to ensure the validity of some measures would aid confidence that the ratings accurately depict emotion. Whilst confidence in the maternal pre-childbirth anger measurement is increased by to its association with observed behaviour, this validation exercise was not measured concurrently. Concurrent expressions of anger, measured before pregnancy, would be most suited to validate this self-report measure. Interparental anger, although not validated by observation, showed remarkably high agreement between partners. Finally, the measurement of infant anger was validated by its association with observed infant anger in response to distress. There was no available measure of toddler anger coded to allow the validation of the measure at this age. However, the toddler measure was significantly

associated with anger measured in infancy, suggesting it too provided a relatively accurate indication of true behaviour.

Further efforts to validate the anger and interparental anger self-reports would increase the confidence in the global measures employed in this thesis. Although maternal pre-childbirth anger was partially validated by its significant association with experimenter- observed anger in a random subsample, extending this validation exercise to the entire sample would be an ideal future step within the CCDS. For the validation of fathers’ pre-childbirth anger and interparental anger, a different approach would be required that is unavailable in the current study design. Firstly, observations of father behaviour were not made in the CCDS. Thus, a separate study would be needed to assess the validity of the paternal pre- childbirth anger measure by an observational coding scheme analogous with that designed for maternal anger and included in this thesis. Secondly, an observation of both parents in

interaction was not available; such a design was outside of the original aims of the CCDS. Methods for observing interparental conflict that have been used previously could inform similar measures of interparental anger. These include rating conflict-related behaviours using coding schemes such as the Interactional Dimensions Coding System (Julien,

Markman, & Lindahl, 1989), the System for Coding Affect Regulation in the Family (Lindahl & Markman, 1991), or measuring facial expressions of anger (Gottman, Levenson, &

Woodin, 2001) to observe partners in a marital discussion tasks, in joint parent-child tasks, and observing couples discussing their marital problems (e.g., Lindahl, Clements &

Markman, 1997). Heyman (2001) presents a series of suggestions for the validation of such observational measures. This future step has two advantages. While coding for instances of anger via observation presents a future step in validating the questionnaire measures, it may also provide the tools for observing behavioural processes by which anger may be conveyed to the developing child.

Documento similar