Sharon's Study Is Ecologically Valid?

Question 1.
When deciding on the research design, one choice is whether to use an experimental or non-experimental design. This depends on the aim of the study. If the aim is to test a prediction that two or more variables are simply associated with each other, the researcher may decide to use a non-experimental or correlational design. However, if the aim is to test a prediction that there is a causal relationship between two variables, then an experimental design is needed.
The correlational design examines the relationships between variables as they happen to occur, without altering people’s experiences.The correlational coefficient often is used to measure the association between variables and the results are normally presented graphically
in a scatterplot.
Sharon aims to find out the link between postnatal depression and the development of mother-infant relationships. Sharon’s research is a non-experimental/correlational study as she doesn’t aim to find out what causes either of the uncontrolled variables (mother’s depression and child’s happiness) but is testing the relationship between them. Sharon plans to summarise the results by presenting the data in a scatterplot correlating the depression scores for each mother with the total ‘happiness’ score for the mother’s child.

Question 2.
To consider whether ‘Sharon’s Study’ is ecologically valid, we need to see if this study can be related to, or reflects everyday, real life.
Studies with high ecological validity take place when participants are within their usual surroundings and behaving naturally so that results can be generalised beyond the setting they were carried out in, whereas studies low in ecological validity cannot.
Based on this definition, ‘Sharon’s study’ is not ecologically valid due to potential biases related to the opportunity sampling technique, the questionnaire used and the environment of the study not being representative of a natural environment for the infants involved.
The sample group in this study is very small and is a poor representation of society. The researcher is using the easiest and most available opportunity sampling technique, risking to produce a biased sample. It consists of mothers and infants that Sharon works with in a local Children’s Centre.
Mothers who attend the classes belong to a very specific group and their responses to the questionnaire cannot be generalised to the wider population, unless Sharon specifies that her target population is the mothers with literacy
problems.
The questionnaire itself represents a source of potential bias as the respondents often tend to provide socially desirable answers to appear in a better light. Mothers in this sample have problems with literacy so credibility of their answers is compromised by possible lack of the question’s comprehension.
The researcher intends to analyse the video recordings of herself playing with the children using a time-sampling technique. This creates an artificial situation for the infants as Sharon is not their usual carer and the environment is not as familiar as their own homes would be. In addition to this, time sampling allows for some behaviours to be missed. This may make the observation not representative.

Question 3.
‘Sharon’s study’ is a correlational study where the researcher is measuring two existing and uncontrolled variables: the level of mother’s depression (based on a questionnaire that gives a ‘depression’ score ranging from 0 (no depression) to 30 (high depression) and matches it to a child’s ‘happiness‘ score (calculated by applying a time-sampling technique to analyse video-recordings of Sharon playing with each child).
A correlation study looks for a consistent relationship between two variables, such that if a change occurs in one variable, there will be a corresponding change in the other. Variables are thought of as already existing and cannot be manipulated.

Question 4.
The researcher used correlational analysis to test possible relationships between two co-variables that she doesn’t manipulate or control. An association or correlation may only be showing indirect affects, as there are likely to be other factors linked to what is being measured that can influence the relationship between the two variables.
There is a group of factors related to the research situation that need to be kept standard, such as: temperature, noise level and light in the room, time of the day, number of children in the creche, choice of the toys and overall layout of the room.
Another group of factors to control is related to the participants, such as: age, wellness of the child, potential distress due to physical cause (hunger, pain, dirty nappy, teething), mother’s confirmed diagnoses of depression and whether she is taking a prescribed medication, level of literacy in mothers and how well all the participants are familiar with the centre.
Another set of characteristics to be controlled is the demand and investigator effect. In this case, Sharon should be monitoring if there are any cues in the external environment and her own behavior that lead children to respond in a certain way or mothers to answer the questions in the way they feel they are expected to.
All the above have a direct impact on hypotheses, design and execution of the research. To test the link between mother’s depression and mother-infant relationships, the researcher needs to control any extraneous or confounding variables. The overall reliability and validity of the research can be improved by using more valid measures, such as clinically proven depression score for DSM-IV diagnosis and mother-child interaction score from the reliable observations of the videotaped parent-child playtime in their natural setting.

Question 5.
Researchers in psychology have a responsibility to conduct their research in an ethical manner.The British Psychological Society (BPS) has a set of principles that ensures ethical conduct, one being to obtain informed and freely given written consent from participants involved in the study. Generally, children under the age of 16 are not considered competent to give full consent therefore informed consent of parents or guardians, as well as others who act on the child’s behalf (schools officials or in this case, Sure Start Centre Program Director and both parents), should be obtained, preferably in writing. The principle of assent is used when working with young children. Researcher must pay attention to nonverbal behaviour for the signs of any distress that can be taken as withdrawal of assent and signal to stop or modify the procedure. In Sharon’s study, the issues of consent are very complex due to the age of infants involved as well as the potential low literacy level and the mental state of the mothers. The researcher needs to find a way to explain to them, in language appropriate to their level of understanding, all the aspects of the research that may affect their willingness to participate.

Question 6.
Other ethics issues arising in this study are risk to the participants, confidentiality, data collection and protection. Data should be accurate, adequate and relevant. In ‘Sharon’s study’ an excessive amount of unnecessary data is collected: each of her video-recording last for 12 minutes whilst only 30 ten-second segments are used for coding. To maintain confidentiality, video recordings made during the study should contain no names or other ways of identifying children. Results of the questionnaire and audiovisual evidence is ‘data’ and good practice in data protection must be followed by ensuring that no personal information is disclosed to third parties, data is kept securely and is destroyed appropriately once the study is concluded. The researcher must consider any risk to, or harmful effects on all the participants. In this study mothers might feel insecure and exposed whilst filling up the questionnaire with their infants left in the creche. There is a risk of infants getting distressed not having the parent in an environment that is different from their home whilst play ing with the person who is not their regular caregiver.

Question 7.
For the research to be reliable it must be possible to be replicated consistently using the same measures and with similar results. It is not clear from the summary information if Sharon attempted to assess the reliability of her study.
For observations to be reliable, the results should be the same regardless of who, or when, is doing the observing. For the coding to be reliable, it has to be clearly defined, structured and repeatable. In Sharon’s case, the behaviours to be coded need to be defined clearly: what constitutes happy, neutral or unhappy?
To assess the reliability of coding, the researcher needs to compare a set of her coding with another set that was produced by another observer analysing the same piece of video (‘inter-observer reliability’) or repeat the coding herself again some days later (‘intra-observer reliability’).
To assess the agreement between two sets of coding produced, the results are arranged in the contingency table and statistical kappa is calculated.
To assess the reliability of the questionnaire the test-retest method is used. This involves testing the same participant twice over a period of time on the same test. Similar scores would suggest that the test has external reliability.

Question 8.
To follow best practice in protecting the data, the researcher should follow strict guidelines. Sharon should ensure that mothers have full knowledge of what her study aims to achieve and obtain their fully informed consent in writing. In fair practice the video-recording is obtained with the guardian present. Data should be used only for the current research and be adequate, accurate and relevant. The standardised diagnostic rating scale for depression could provide a more accurate measure for Sharon’s study. To avoid collecting excessive amount of unnecessary data, each of her video-recording needs to contain only the 30 ten-second segments that are used for coding. A researcher should not copy or share personal data and should use only secure means of data transfer. To ensure against accidental disclosure the researcher can anonymize each participant by assigning them a codename and use it on all paper documents and video labels. Secure data keeping measures include no computers left unattended, use of the password protection for hard discs and individual files. All paper materials, removable disks and media should be stored securely and destroyed appropriately (shred all paper material, wipe video- and audio-recordings) when no longer needed.

Question 9.
Correlation refers to the measure of how strongly two or more variables relate to each other.
In the correlational analysis it is important to operationalise the variables by stating a clear way that the two independent variables (or co-variables) are going to be measured as this can affect the usefulness of this method. The hypotheses should clearly state the two variables that are being measured and what the researcher expects to find out. Whilst Sharon aims to look for a potential link between postnatal depression and the development of mother-infant relationships, her hypotheses does not reflect it and looks at a different co-variable instead (child ‘happiness’).
To make her conclusions, the researcher correlates the depression score for each mother with the total ‘happiness’ score for the mother’s child. The correlation she finds from using the statistical test is -0.12 which indicates very weak negative correlation: as the level of the mother’s depression increases, there is an insignificant chance that the mother’s child becomes less happy.
Scatterplots are a very useful type of descriptive statistics because they provide a graphical representation of the correlation between the variables. From the scatterplot the researcher can describe and summarise the data but cannot make the conclusion about the
hypotheses.
In this case, there is a clear demonstration that there is a very weak correlation along with the inconsistency of the results. On Sharon’s scatterplot the highest depression score of 17 correlates to the second highest ‘happiness’ score of 25. At the same time the lowest depression score of 3 correlates to the highest ‘happiness’ score of 27. The evidence in this study is inconclusive and indicates the issues with the choice of the variables and the methods used to measure them.