Research terminology for studying families

Turning questions about families into research

Sociological research is a systematic investigation using evidence, rather than relying on assumptions about how families live. Researchers might investigate how couples share housework, how parents experience childcare or how family relationships change over time.

A hypothesis is a statement that can be tested against evidence. For example, a researcher could propose: ‘Couples in which both partners work full-time share housework more equally than couples in which only one partner works full-time.’ This is a starting point for investigation, not an established fact. The researcher would need to decide what counts as housework and how to measure its distribution.

The information gathered is called data. Data analysis means organising and examining that information to identify patterns and reach reasoned conclusions. A trend is a general direction of change over time, such as an increase or decrease in a measured family characteristic. A pattern linking two characteristics does not, by itself, establish that one causes the other.

Two different ways of describing data

Quantitative data are numerical: counts, percentages and statistics. Asking people how many hours they spent on housework could produce quantitative data. Numbers make it easier to compare groups and identify patterns, but may reveal little about why people behave as they do.

Qualitative data describe meanings and experiences, usually through words or quotations. Asking someone to explain how their household decides who does each task could reveal ideas about fairness, obligation or conflict. These detailed accounts can deepen understanding, although comparing many different answers is harder than comparing standardised numbers.

A separate distinction concerns where information comes from. Primary data are collected first-hand by the researcher for their investigation, for example through their own interviews. Secondary data already exist and were collected or produced by others, such as published statistics, autobiographies or earlier sociological studies.

These distinctions can be combined. A researcher’s own numerical questionnaire results are primary quantitative data. Quotations taken from an existing autobiography are secondary qualitative data.

Surveys, questionnaires and questions

A survey collects answers from a population or sample. An attitude survey specifically investigates opinions, beliefs or feelings: for example, views about whether parents should share childcare equally. A respondent is someone who answers questions.

A questionnaire is a written set of questions. It can include two types of question:

• A closed question offers fixed answers. For example: ‘Who usually prepares the evening meal: you, your partner, both equally, or someone else?’ Answers can be counted and compared.
• An open question allows an answer in the respondent’s own words. For example: ‘How did your household decide who would prepare meals?’ This allows explanations that the researcher may not have anticipated.

Questionnaires can reach many people relatively cheaply, and standardised questions help comparisons. However, fixed options may not fit someone’s experience. Respondents may misunderstand questions, rush their answers or not reply at all. If particular groups are less likely to respond, the final results may give a distorted picture.

Interviews and group discussion

An interview involves a researcher asking questions and recording answers. In a structured interview, respondents receive the same prepared questions, normally in the same order. This supports comparison. An unstructured interview is more flexible: the researcher develops questions during the conversation and can ask respondents to explain what they mean.

For a study of family responsibilities, an unstructured interview could explore why someone describes a division of work as ‘fair’, even when the partners do different amounts. Its flexibility can reveal meanings that fixed questions miss. However, interviews take time, and different conversations may be difficult to compare.

The interviewer can also influence answers through their wording, reactions or expectations. Respondents may give answers that make their family appear more harmonious or equal than it is. This is a reason to assess answers critically, not to assume that every interview account is dishonest.

A focus group is a group interview about a particular topic. Researchers can examine both participants’ views and how they respond to one another. A discussion about childcare might reveal disagreement over what makes a ‘good parent’, but some participants may dominate while others feel reluctant to disagree.

Observing everyday family life

Observation means watching and recording behaviour. It can show what people do rather than relying only on what they say they do.

Two distinctions describe the researcher’s role:

• In participant observation, the researcher joins the activities of the group. In non-participant observation, they watch without joining in.
• In overt observation, people know they are being studied. In covert observation, they do not.

These are separate distinctions: taking part does not necessarily mean concealing the research. Participation can help a researcher understand a group’s point of view, but becoming closely involved may make it harder to remain detached. Non-participation provides greater distance, although it may offer less insight into people’s meanings.

When people know they are being observed, they may alter their behaviour. For example, family members might make a special effort to share tasks during a visit. This is often called the observer or Hawthorne effect. Covert observation may reduce this problem, but raises serious questions about deception, privacy and consent. Secretly recording behaviour in a family’s home would be particularly intrusive.

Ethnography studies a group’s culture and everyday practices, usually through qualitative methods such as participant observation and unstructured interviews. It seeks detailed understanding of life in everyday settings rather than just a numerical summary.

Studying cases, change and existing material

A case study investigates a particular case in depth, such as an institution or a series of related events. It can offer rich detail, but one case cannot automatically stand for all families.

A longitudinal study follows the same group over time, collecting information at intervals. Researchers could examine how family responsibilities change as children grow older. Following the same people helps reveal change within their lives, but takes time and money. Participants may move away or drop out, potentially changing the composition of the remaining group.

Official statistics are numerical information produced by government bodies or other official organisations. They are secondary data for a sociologist using them in a new investigation. The UK Census is an important example: a census aims to collect information about the whole population rather than a selected sample. Official statistics can support comparisons between groups and over time. However, their categories may not fit the researcher’s question, and changes in definitions can make historical comparisons difficult.

Content analysis systematically examines documents or images using defined categories. A researcher studying family representations in advertisements could create categories for the family arrangements shown, code each advertisement and count each category. Clear definitions help different researchers classify the same material consistently. Counts reveal patterns in representations, but do not by themselves explain why advertisers chose them or how audiences interpret them.

Choosing whom to study

The population is the whole group a researcher wants to investigate: for example, parents of dependent children in a particular town. A sample is a subgroup selected from that population. Studying a sample is often more practical than studying everyone.

A sampling frame is a list from which participants are selected. It needs to cover the intended population accurately. A list containing only parents who attend one support group would exclude many other parents.

A representative sample reflects the relevant characteristics of its population. An unrepresentative sample does not. For example, a study of parents’ childcare experiences would be distorted if it included only parents who were available during weekday working hours. Representativeness matters for generalisability: whether findings can reasonably be applied to the wider population.

How sampling methods differ

A random sample gives every member an equal chance of selection. A researcher could use a random number generator to select entries from a complete list. This reduces the researcher’s influence over who is chosen, but an incomplete list or refusals to participate can still undermine representativeness.

A systematic sample takes every nth entry from a sampling frame after a random starting point. For example, starting at the third entry and selecting every tenth would select entries 3, 13, 23 and so on. The researcher must consider how the list is ordered: a repeating pattern could distort the selection.

Stratified sampling divides the population into relevant groups and randomly selects people from each in proportion to their numbers. This can help ensure that important characteristics are reflected in the sample.

A quota sample also uses categories, but researchers fill a specified number of places in each category without randomly selecting individuals. For example, they might be required to interview set numbers of parents from different age groups. Quotas ensure those categories are included, but the researcher’s choice of individuals can introduce bias.

A snowball sample grows through participants recommending other potential participants. It can help reach people for whom no suitable list exists or who are difficult to contact. However, recommendations may remain within the same social networks, leaving other experiences underrepresented.

No method guarantees that the people who ultimately take part will represent the whole population. The best choice depends on the question, available lists, access and resources.

Judging the quality of evidence

Validity concerns whether research captures what it is intended to study. A housework questionnaire would have limited validity if it omitted major tasks or if respondents consistently understated their contribution to conflict over chores.

Reliability concerns consistency when the same methods are repeated under comparable conditions. Standardised questions and clear recording categories help researchers repeat a procedure. However, a consistently misleading question could produce reliable answers without validly capturing people’s experiences.

Bias is a systematic influence that distorts research. It can arise in selection, question wording, recording or interpretation. For example, ‘Why is it better for mothers to provide childcare?’ assumes that it is better, steering respondents towards a particular view. Selective use of data can also mislead: presenting only quotations that support a conclusion while ignoring relevant contradictory accounts gives an incomplete picture.

Mixed methods research combines methods to obtain both quantitative and qualitative data. A family study might use a questionnaire to identify broad patterns and interviews to explore their meanings. Triangulation means cross-checking findings from different methods or sources. Agreement can strengthen confidence, while disagreement prompts researchers to investigate why the evidence differs. Simply using more methods does not automatically remove weaknesses.

Protecting participants and their information

Ethical considerations concern whether research is morally acceptable. They affect the design and conduct of a study, not just its final report.

Informed consent means that people voluntarily agree to participate after receiving enough information about the purpose, their involvement and possible risks. Participants must be free to refuse or withdraw. Research involving children requires particular care over consent and safeguarding.

Researchers should anticipate and prevent harm. Questions about separation, conflict or parenting may cause distress. Participants should not be pressured to discuss painful experiences, and researchers must consider whether reporting information could damage family relationships or expose someone to other risks.

Confidentiality means keeping information private and restricting access to authorised people. Reports should avoid identifying participants. Removing a name may not be enough if a detailed description makes the family recognisable. Data protection includes storing personal information securely and controlling who can access it.

Researchers must also report evidence honestly. A method that offers valuable insight is not necessarily acceptable if obtaining that insight involves serious harm or invasion of privacy. A justified choice balances the quality of the evidence, practical feasibility and participants’ rights.