What Are Sampling Techniques?
Sampling is the process of selecting a portion of the population to participate in your study. Since you usually cannot survey every community college student in the US, you select a sample that represents them.
Sampling answers: Who exactly will be in my study, and why them?
Why Sampling Matters
Good sampling ensures:
- Your findings are credible
- You can answer your research questions
- You do not waste resources on inappropriate participants
- Your study is feasible within time and budget
Types of Sampling
Sampling is broadly divided into two categories: probability (random) and non-probability (non-random).
Probability Sampling Table
| Technique | Description | US Dissertation Example | Advantages | Disadvantages |
|---|---|---|---|---|
| Simple Random | Every person in the population has an equal chance of being selected | Randomly selecting 500 registered voters in Ohio from a state database for a political poll | Eliminates selection bias; highly generalizable | Requires complete population list; can miss subgroups |
| Systematic Random | Selecting every nth person from a list | Selecting every 10th student from the university registrar's list of 20,000 students | Easier than simple random; still unbiased | Periodicity can bias if list has patterns |
| Stratified Random | Population divided into subgroups (strata), random samples drawn from each | Dividing a university population by class year (Freshman, Sophomore, etc.) to ensure all years are represented | Ensures subgroup representation; more precise | Requires knowing subgroup proportions; complex |
| Cluster Random | Randomly selecting groups (clusters), then sampling within them | Randomly selecting 10 school districts in Texas, then surveying all teachers in those districts | Cost-effective for large areas | Less precise than simple random; higher error |
Non-Probability Sampling Table
| Technique | Description | US Dissertation Example | Advantages | Disadvantages |
|---|---|---|---|---|
| Convenience | Selecting participants who are easily available | Surveying students in your own psychology class at an Illinois university about study habits | Easy, fast, inexpensive | Low generalizability; high bias |
| Purposive | Deliberately selecting individuals based on specific characteristics | Interviewing 15 school principals in Florida who have successfully implemented a new reading program | Ensures relevant participants; rich data | Researcher bias in selection; not generalizable |
| Snowball | Existing participants recruit future participants from their network | Studying undocumented immigrant entrepreneurs in New York City; one participant introduces the researcher to others | Accesses hidden populations | Sample may be biased; limited diversity |
| Quota | Selecting a predetermined number from subgroups | Ensuring 50 male and 50 female participants in a study about commuting habits | Ensures diversity; simpler than stratified | Not random; may still have bias |
Data Collection Methods
Your data collection tools must be reliable and valid. The sections below show common methods aligned with each methodology.
Quantitative Data Collection Methods
| Method | Description | US Context Example | Strengths | Weaknesses |
|---|---|---|---|---|
| Surveys & Questionnaires | Collecting standardized data using Likert scales, multiple-choice, or closed-ended questions | An online survey sent to 1,000 nurses in Massachusetts hospitals to measure job satisfaction using a validated scale | Large samples; standardized; efficient | Surface-level; no depth; response bias |
| Experiments | Manipulating one variable to see its effect on another | A lab experiment at a Michigan university testing if a new teaching app improves reaction times in students | Establishes causality; controlled | Artificial; may not generalize to real world |
| Existing Data Analysis | Using pre-existing datasets | Analyzing NCES (National Center for Education Statistics) data to study dropout rates in rural US high schools | Large samples; often free; longitudinal possible | Limited to existing variables; may not fit your question |
| Tests & Assessments | Standardized instruments measuring knowledge or ability | Administering the GRE to measure graduate school readiness across different demographic groups | Objective; norm-referenced | Cultural bias possible; expensive |
| Physiological Measures | Biological or physical measurements | Measuring heart rate and cortisol levels in New York office workers to study stress | Objective; precise | Expensive equipment; intrusive |
Qualitative Data Collection Methods
| Method | Description | US Context Example | Strengths | Weaknesses |
|---|---|---|---|---|
| Interviews | One-on-one, in-depth conversations. Can be structured, semi-structured, or unstructured | Conducting 20 semi-structured interviews with military veterans in Virginia about their transition to civilian careers | Rich, deep data; participant voice | Time-intensive; small sample; interviewer effect |
| Focus Groups | A moderated group discussion with 6–10 participants | Running three focus groups with single mothers in Atlanta to discuss barriers to accessing childcare services | Group interaction generates ideas; efficient | Group dynamics may silence some; less depth per person |
| Observations | Watching participants in their natural environment | An ethnographic study observing classroom interactions in a bilingual elementary school in California | Real-world behavior; contextual | Observer effect; time-intensive; interpretation bias |
| Document Analysis | Analyzing existing texts, documents, or artifacts | Analyzing university mission statements from all Ivy League schools to understand institutional values | Unobtrusive; historical; stable | May be incomplete; not designed for research |
| Diaries/Journals | Participants record experiences over time | Having 30 first-generation college students keep weekly journals about their adjustment to university life | Captures process over time; participant perspective | High participant burden; attrition |
Common Mistakes to Avoid in Sampling and Data Collection
- Sampling Bias: Selecting a sample that is not representative (e.g., only surveying rich neighborhoods in a study about poverty).
- Small Sample Size: Not collecting enough data to achieve statistical significance (quant) or data saturation (qual).
- Poor Survey Design: Using leading questions or jargon that confuses US respondents.
- Not Piloting: Not testing instruments before full deployment.
- Ignoring Ethics: Collecting data without IRB approval or informed consent.
- Overpromising: Promising participants something you cannot deliver.
Frequently Asked Questions
What is a reasonable sample size for a US PhD dissertation?
It varies by method. For a quantitative survey, you need a power analysis to determine this, but 100+ is common for a basic correlational study. For a qualitative interview study, sample size is determined by saturation (when new data stops providing new insights). This is often around 15–30 participants, depending on the scope. For experimental studies, 30–60 per group is typical. Your committee will expect you to justify your number, not just pick one.
Can I use Amazon's Mechanical Turk (MTurk) to find participants?
Yes, MTurk is a popular tool for US researchers, especially for quantitative social science studies. However, you must be aware of the limitations. Data quality can be variable, so you need to include attention checks in your survey. Participants may not be representative of the general US population. You must also clearly justify why an online convenience sample is appropriate for your research question.
Does my university's IRB need to approve my data collection before I start?
Absolutely yes. If your research involves collecting data from people (surveys, interviews, experiments, observations), you must get IRB approval before you contact any participants. This is a non-negotiable ethical and legal requirement at all US universities. Plan for this: IRB review can take weeks or months depending on your university and the level of risk. Submit early.
My topic is about a very specific group. Is purposive sampling valid?
Yes, absolutely. Purposive sampling is a hallmark of strong qualitative research. When you are studying a specific phenomenon, you do not want a random sample; you want people who have direct, relevant experience. The key is to be transparent about your selection criteria and justify why those specific participants are the right ones to answer your research question. In US qualitative dissertations, purposive sampling is not just accepted — it is expected.
Can I use surveys I find online instead of creating my own?
Yes, using validated instruments is often better than creating your own. If a reliable, valid survey already exists for your construct, use it. This strengthens your study because the instrument has already been tested. You must properly cite it and, if it is copyrighted, obtain permission. Many US dissertations use established scales for constructs like depression, job satisfaction, or student engagement.
How do I handle data collection when my population is hard to reach?
For hard-to-reach populations in the US (e.g., undocumented immigrants, homeless veterans, specific professionals), use snowball sampling. Start with one or two participants you can access, then ask them to refer others. Build trust. Offer appropriate incentives. Partner with community organizations that serve your population. Be prepared for a longer recruitment timeline. Document your recruitment challenges transparently in your methodology chapter.
Continue to the next guide: Part 3: How Methodology Aligns with Research Questions →