Types of Surveys

Surveys are used to study things like events, behaviours, opinions, and attitudes within a specific group of people. There are two main types: descriptive and analytic.

Descriptive Surveys

These surveys describe what's happening at a single point in time. They collect data from a sample of the population and use statistics to summarise the findings. Because they gather data at one specific time, they're called cross-sectional surveys. They often ask people about recent events, feelings, or behaviours, making them retrospective. Statisticians sometimes call these "observational research," but this can be confusing as "observational methods" are a specific technique in social science. Descriptive surveys are sometimes referred to as correlation studies because they can show relationships between things, but they can't prove cause and effect.

These retrospective cross-sectional surveys look at a random sample of a population at one moment. They're labelled retrospective because they ask about past and present behaviours, attitudes, and events. They describe what's being studied, look for connections, and try to understand things like how common something is within a group. They can also test ideas and generate new ones about possible causes and effects. These studies can range from simple analysis of routine statistics to detailed surveys. They're useful for understanding population characteristics, how quickly social or attitudinal changes occur, and what influences those changes.

Cross-sectional surveys are relatively quick and inexpensive because they can survey many people at once, and the data is easy to analyse. They're often used in social sciences and in health studies to look at how common a disease is. A major drawback is the potential for "recall bias," where people don't accurately remember past events. Careful questionnaire design can help reduce this. Importantly, these surveys can't prove cause and effect; they only show associations. For example, finding a link between being overweight and breast cancer doesn't prove that being overweight causes breast cancer. It could be the other way around, or a third factor could be responsible for both. However, statistical techniques can help minimise this limitation, and the findings can lead to further research.

Analytic Surveys (Longitudinal Surveys)

These surveys look at changes over time to try to understand cause and effect. They're called longitudinal because they collect data at multiple points in time. Most are prospective (looking forward in time), but some are retrospective (looking back using records).

Prospective longitudinal surveys follow people over time with repeated data collection. They can be panel studies (following the same people) or trend studies (using different samples each time). These are also called follow-up studies. If the group being followed shares a characteristic like birth year, it's a cohort study. These are commonly used in social science and health studies to track how often new cases of a disease appear.

These types of studies require careful planning, regular data collection, and high participation rates. Losing participants over time (due to moving, dropping out, or death) can bias the results. The timing of data collection needs to be carefully considered, and the methods used need to be sensitive enough to detect changes. These studies are sometimes called "natural experiments" because they observe changes as they happen. They can calculate how often new cases of a disease occur in different groups and identify possible causes.

However, prospective longitudinal surveys are costly, time-consuming, and require a lot of organisation. Even with careful planning, it's hard to definitively prove cause and effect due to factors like the difficulty in timing follow-up periods and the long timeframes involved. It's also possible that observed changes are due to "regression to the mean," where extreme measurements are temporary and return to average over time. Longitudinal studies are often used after simpler cross-sectional studies have identified important variables.

Participants in longitudinal studies can become used to the study, potentially changing their behaviour or answers. They might remember previous responses or become more aware of the research topic, which can affect their answers.

Types of Longitudinal Surveys

• Trend Surveys: These survey different samples from the same population at each data collection point to track changes over time. They're common in market research and are used in health studies to track new cases of disease.

  
  

• Panel Surveys: These follow the same group of people over time, repeatedly collecting data to track individual changes. This allows researchers to see who changes their behaviour, attitudes, or health status.

  
  

• Cohort Studies: These follow a group of people who share a common characteristic (like birth year). They can be cross-sectional (data collected at one point in time about the past), longitudinal and retrospective (data collected at multiple points in time about the past), or longitudinal and prospective (data collected at multiple points in time moving forward). Even prospective cohort studies often include questions about the past.

  
  

• Cohort Sequential Studies: These involve studying multiple cohorts (groups with a shared characteristic) at different times to account for "cohort effects," which are the unique experiences of each generation. This helps researchers understand how social changes might affect different generations differently but can struggle to account for period effects such as economic or political changes.

A key challenge in cohort studies is the "cohort effect," where each generation experiences unique historical circumstances that can influence their values and behaviours. This needs to be considered when interpreting the results.