Planning and Conducting Investigations: Sampling (VCE SSCE Psychology): Revision Notes
Planning and Conducting Investigations: Sampling
Understanding populations and samples
When researchers design investigations, they begin by identifying the group of people they wish to study. This broader group is called the population. The population represents everyone to whom the researcher hopes to apply their findings. Importantly, a population does not necessarily mean everyone in the world. Instead, it refers to the specific group the researcher is interested in understanding, such as children under 10 years old, P-plate drivers, females in Victoria, or people with dementia.
The term "population" in research has a specific meaning that differs from its everyday use. A research population doesn't have to be large—it simply needs to be clearly defined based on the characteristics relevant to the study.
Population: the wider group of people that a study is investigating.
Once the population has been defined, researchers must select a sample from this population. The sample consists of the actual participants who will take part in the investigation. Since it is usually impractical or impossible to study every member of a population, researchers work with a smaller, manageable group that ideally represents the larger population.
Sample: the smaller group of people selected from the population who will be participants in the investigation.
When describing a sample, researchers must include specific characteristics such as the number of participants, their age, sex, location, and any other relevant details. For example, if an Australian researcher is studying the effect of sugar on attention in children, they might define their population as all Australian children under 10 years old. Their sample might consist of 100 children (50 males and 50 females) aged 5-10 years from Victorian public primary schools.
Identifying populations and samples
When analysing research scenarios, identifying the population requires thinking about who the researcher intends to apply their results to. This information is typically found in the study's aim. The population description should focus on broad characteristics that define the group, rather than specific numbers.
Sample descriptions, in contrast, must include precise details: the exact number of participants, where they were recruited from, and relevant shared characteristics such as age or sex. For instance, stating "150 two-year-old children with motor delays from Melbourne" provides the specificity needed for a sample description, whilst "children with motor development delays" appropriately describes the population.
The importance of sample size
A fundamental principle in research design is that samples must be representative of the population. This means the characteristics of people within the sample should mirror those found in the broader population. Achieving representativeness depends largely on sample size.
There is no universal rule for an ideal sample size, but researchers must ensure their sample is sufficiently large relative to the population. The sample needs to adequately represent variables such as age, sex, geographical location, and socio-economic status. Consider a study on Victorian children: Victoria has approximately 740,000 children under 10 years old. A sample of 50 children would be disproportionately small compared to this population, whereas a sample of 5,000 children would be more appropriate and representative.
Large samples serve two critical purposes:
- They increase the likelihood that the sample accurately represents the population
- They minimize the influence of individual participant variables that could skew the results
When samples are small, a single participant with unusual characteristics can disproportionately affect the findings.
How sample size affects results
Consider a memory test conducted on two samples of different sizes. In a sample of 10 people, if one participant has an exceptionally high IQ, their superior performance could significantly inflate the group's mean score. However, in a sample of 30 people, this same individual's results would be balanced by the additional participants, producing a more accurate representation of typical performance.
Worked Example: The Impact of Sample Size on Mean Scores
Consider how one outlier affects different sample sizes in a memory test:
Sample 1 (10 participants): Scores: 4, 5, 5, 5, 3, 4, 5, 5, 4, 10
Mean = = 5
Sample 2 (30 participants): Scores: 4, 5, 5, 5, 3, 4, 5, 5, 4, 10, 4, 5, 5, 5, 3, 4, 5, 5, 4, 3, 4, 5, 5, 5, 3, 4, 5, 5, 4, 3
Mean = = 4.5
Analysis: In Sample 1, the participant who scored 10 pushes the mean up to 5. In Sample 2, the same score of 10 has less impact because it is averaged across 30 participants, resulting in a more accurate mean of 4.5 that better reflects typical performance. This demonstrates why larger samples produce more reliable results.
Sampling techniques
Researchers employ different methods to select participants from their population. These methods are called sampling techniques, and they involve specific procedures for choosing who will participate in the study. Two main techniques are random sampling and stratified sampling. Each has distinct strengths and limitations, but both aim to produce samples that are representative of the population.
Sampling technique: involves procedures for selecting participants from the population.
Random sampling
Random sampling ensures that every member of the population has an equal opportunity to be selected for the study. This equal chance of selection is the defining characteristic of random sampling. When implemented correctly with a sufficiently large sample, random sampling tends to produce samples that are representative of the population, allowing researchers to generalise their findings.
Random sampling: selecting participants from the population in a way that means each member of the population has an equal chance of being selected to participate in the study.
The word "random" in everyday language often means "haphazard" or "without pattern." In research methodology, however, random sampling has a precise technical meaning: it ensures that selection is unbiased and that every population member has an equal probability of being chosen.
The process of random sampling begins with identifying all members of the population. In small-scale studies, this might involve writing everyone's name on paper, placing the names in a container, and drawing the required number. For larger studies, researchers typically assign a unique number to each population member and use random number generator software to select participants.
Worked Example: Random Sampling Process
Imagine a population of 40 people, each assigned a number from 1 to 40. To select a random sample of 12 participants:
- Assign each person a unique number (1-40)
- Use random number generator software to generate 12 random numbers between 1 and 40
- The individuals assigned those numbers form the sample
This process ensures that each person has exactly a 12 in 40 (or 30%) chance of being selected, meeting the equal opportunity requirement of random sampling.
Evaluation of random sampling
Random sampling offers important advantages but also presents practical challenges:
| Strengths | Limitations |
|---|---|
| A large enough random sample is probably representative of the population, improving external validity. | Small random samples may not be representative of the population, reducing external validity. |
| It may be difficult, time consuming, impossible or unethical to obtain names of all members of the population. |
The primary strength of random sampling lies in its potential for representativeness. When the sample is sufficiently large, the equal chance of selection means the sample will likely reflect the population's characteristics, enhancing external validity—the extent to which findings can be generalised.
Common Limitation: Random sampling has a significant practical constraint. Obtaining a complete list of all population members can be difficult, time-consuming, or even impossible. In some cases, accessing such information raises ethical concerns about privacy and consent.
For example, obtaining a list of all people with depression in Australia would be practically impossible and ethically problematic due to privacy laws and the sensitive nature of the information.
However, random sampling has notable limitations. Small random samples may not adequately represent the population, compromising external validity. Furthermore, obtaining a complete list of all population members can be difficult, time-consuming, or even impossible. In some cases, accessing such information raises ethical concerns about privacy and consent.
Stratified sampling
Some research questions require ensuring that specific subgroups within a population are fairly represented in the sample. Stratified sampling addresses this need by dividing the population into subgroups before selection. This technique produces samples that more accurately reflect population diversity than random sampling alone.
Stratified sampling: first dividing the population into subgroups, and then randomly selecting participants from each subgroup in the proportion that they appear in the population.
The stratified sampling process involves several steps. First, researchers identify relevant characteristics for dividing the population into subgroups (also called strata). These characteristics might include age, sex, geographical location, or socio-economic status. Second, researchers determine the proportion of each subgroup within the population. Finally, participants are randomly selected from each subgroup, maintaining the same proportions found in the population.
Worked Example: Stratified Sampling with Income Groups
Consider a study examining income effects on health outcomes. If the population consists of:
- One-quarter (25%) high-income earners
- Three-quarters (75%) low-to-middle income earners
A stratified sample of 40 participants would maintain these proportions:
- High-income earners: participants
- Low-to-middle income earners: participants
This ensures that both income groups are represented fairly according to their actual proportions in the population, rather than relying on chance selection that might over- or under-represent either group.
This approach ensures that important subgroups are not over- or under-represented. In schools, for instance, there are typically more students in Years 7 and 8 than in Years 11 and 12. A stratified sample of school students would include more Year 7 and 8 students, reflecting their greater numbers in the student population. Similarly, since approximately three times as many females as males study VCE Psychology, a stratified sample of Psychology students should maintain this 3:1 ratio.
Evaluation of stratified sampling
Stratified sampling provides enhanced representativeness but requires additional effort:
| Strengths | Limitations |
|---|---|
| A large enough stratified sample is probably representative of the population, improving external validity. | It may be difficult, impossible or unethical to obtain names of all members of the population. |
| Important subgroups of a population are ensured fair representation. | It is more time consuming than using a random sampling technique because of the need to form subgroups and any pre-testing required. |
The key advantage of stratified sampling is its guarantee that important subgroups receive fair representation. This reduces bias and improves external validity, making it more likely that findings apply to the entire population. The technique is particularly valuable when certain subgroups are small but important to include, as random sampling might miss them entirely.
Stratified sampling is especially useful when:
- The population has distinct subgroups that differ on important characteristics
- Some subgroups are small in number but critical to include
- The researcher wants to ensure conclusions apply to all population segments
- Previous research suggests that certain characteristics significantly affect the variable being studied
However, stratified sampling shares one limitation with random sampling: obtaining complete population lists can be difficult, impossible, or unethical. Additionally, stratified sampling demands more time and resources than random sampling. Researchers must identify appropriate subgroups, determine their proportions in the population (which may require preliminary testing or research), divide the population accordingly, and then conduct random selection within each subgroup.
Identifying sampling techniques in research
When analysing research scenarios, you may need to identify which sampling technique was used and evaluate whether it was appropriate. Both random and stratified sampling require researchers to have access to every member of their population of interest. When properly implemented, either technique can produce representative samples with good external validity.
Recognising random sampling
To determine whether random sampling was used:
- Identify the population of interest: Look at the investigation's aim to determine which group the researcher wanted to study.
- Assess equal opportunity: Determine whether the sampling technique gave each member of that population an equal chance of selection. Look for methods such as lottery-style selection (picking names from a container) or random number generator software.
If the population was defined very broadly (such as "all people" or "all children"), the researcher likely could not access every member to implement true random sampling. This is a common mistake when evaluating research—always consider whether complete population access was realistically possible.
Recognising stratified sampling
To determine whether stratified sampling was used:
- Identify the population of interest: As with random sampling, examine the investigation's aim to determine the target population.
- Check for subgroup division: Determine whether the sampling technique involved dividing the population into specific strata or subgroups based on characteristics such as age, sex, geographical location, or personal attributes. For example, researchers might divide the population into age brackets and determine the proportion of each bracket.
- Verify proportional representation: Confirm that the final sample was drawn from these subgroups in the same proportions they exist in the population.
When neither technique is used
If neither random nor stratified sampling was employed, the sample may not adequately represent the population. This can significantly reduce the investigation's external validity, limiting the extent to which findings can be generalised.
Warning Signs of Poor Sampling:
Common indicators that proper sampling techniques were not used include:
- Convenience sampling—selecting whoever is easily available
- Volunteer bias—relying only on people who choose to participate
- Biased selection—choosing participants based on researcher preference
- Inadequate population access—inability to reach all population members
When you encounter these approaches in research, be critical about whether the findings can truly be generalised to the broader population.
When evaluating research, consider whether the sampling technique supports or undermines the study's conclusions.
Remember!
Key Points to Remember:
-
Populations and samples are distinct: The population is the broader group a researcher wants to understand, whilst the sample consists of the actual participants in the study.
-
Sample size matters for representativeness: Larger samples are more likely to accurately represent the population and minimize the influence of individual participant variables on results.
-
Random sampling requires equal opportunity: Every member of the population must have an equal chance of being selected, typically achieved through lottery methods or random number generators.
-
Stratified sampling ensures proportional representation: The population is divided into subgroups, and participants are randomly selected from each subgroup in proportions matching the population.
-
Both techniques require population access: Random and stratified sampling both depend on researchers having access to all members of the population, which can be difficult, time-consuming, or impossible to achieve in practice.
-
Evaluate external validity: When assessing research, always consider whether the sampling technique used supports the researcher's ability to generalise findings to the broader population.