Sample Collection Methods

If you work in UX research, you know that your insights are only as good as the sample you collect. Perfect random samples are rare in our field, but that doesn’t mean you have to settle for low-quality data. The real challenge is balancing speed and cost with validity, and there are practical ways to do it. The first step is understanding your sampling options. In an ideal world, you would run a simple random sample where every user has an equal chance of being picked. If you have a clean customer database or panel, you can randomize IDs and draw participants this way, but it’s costly and rare in UX. A more accessible variation is systematic sampling: sort your list randomly and invite every 10th or 20th user. It works if the list is truly random, but beware of hidden patterns like chronological ordering that can skew results. For teams that need reliable subgroup comparisons - say you want both iOS and Android users represented - stratified sampling is a better fit. Divide your population into meaningful segments, get the actual proportion for each, and sample within those groups. And when you’re dealing with a geographically dispersed or very large audience, cluster or multistage sampling helps reduce cost by selecting groups like cities first, then sampling users within them, though you need a larger sample to maintain precision. Most UX teams can’t do pure probability sampling, so they rely on non-probability methods. These include convenience samples of whoever responds, quota sampling where you fill set targets like a 50/50 device split, snowball recruiting through referrals for niche users, and in-product intercepts that capture feedback right in context. They’re fast and cost-effective but come with high bias risks. The good news is you can make these work better: use simple quotas to make sure you hear from new and power users, recruit through more than one channel so you don’t only reach forum regulars, trigger intercepts in ways that don’t miss those who drop off, and always document who you didn’t reach, like churned users. For large-scale or high-stakes projects, a hybrid approach combines the best of both worlds. You might recruit 500 people from a random sample and add 1,500 from an opt-in panel, then use propensity modeling and weighting to align the opt-in group to the random group. This balances cost and statistical validity. Weighting in general is a powerful tool to align your sample to known population benchmarks like census data or internal analytics. Post-stratification weights on key cells such as age by gender, and raking iteratively aligns marginal distributions when you don’t have full cross-cell data. Weighting adds variance, so it’s important to calculate your effective sample size for proper margins of error rather than assuming your raw n reflects precision.

After years of teaching and conducting different types of experience-related research, I noticed a recurring gap: students and early UX practitioners rarely get to work with data that captures the true complexity and messiness of real UX studies. So, I decided to change that. I’ve created a collection of 17 multi-method UX datasets covering surveys, usability testing, eye tracking, EEG, telemetry logs, chatbot data, game analytics, and even XR interactions. Each dataset includes three sample sizes (small, medium, large), allowing students to see how statistical power, data variability, and measurement noise shape the conclusions we draw. Altogether, the collection includes 51 CSVs that are synthetic but modeled after real-world research data using validated UX measures such as SUS, UEQ, and NASA-TLX. What makes it valuable: ▪️ Realistic statistical distributions (normal, log-normal, Poisson, beta) ▪️ Built-in relationships between variables such as workload and frustration ▪️ Authentic data challenges: missing values, outliers, and small measurement errors ▪️ Documentation for each dataset, including variable definitions and suggested analyses This collection helps instructors and researchers teach: ▪️ Descriptive and inferential statistics ▪️ UX metrics and psychometric reliability ▪️ A/B testing, funnel analysis, and regression ▪️ Eye-tracking and physiological data interpretation If you teach UX, cognitive psychology, or data analysis, or if you simply want realistic data to practice with, you can explore the collection here: https://lnkd.in/eAaV-izw Created at Perceptual User Experience Lab to make UX education more rigorous, data-driven, and grounded in cognitive science.

Sampling in qualitative research is a critical process that shapes the depth and reliability of study findings. Unlike quantitative research, which seeks statistical representativeness, qualitative sampling focuses on rich, in-depth insights from a carefully selected group. This document provides a detailed exploration of qualitative sampling strategies, emphasizing their role in ensuring meaningful data collection and analysis. The guide distinguishes between probability and non-probability sampling, explaining why qualitative research relies on purposeful selection rather than random sampling. It explores key strategies such as convenience, opportunistic, purposive, and theoretical sampling, detailing their strengths, limitations, and best use cases. Special attention is given to conceptually driven approaches, including maximum variation, homogenous, and snowball sampling, ensuring researchers select participants who best contribute to answering research questions. For qualitative researchers, evaluators, and field practitioners, this document serves as a valuable resource for refining sampling plans and improving research validity. It underscores the importance of iterative sampling, data saturation, and contextual adaptation in shaping study outcomes. Whether conducting interviews, focus groups, or case studies, these insights help professionals design rigorous qualitative research and enhance the credibility of their findings.

DNA under the nails: Swabs vs scrapings When forensic evidence is collected at a crime scene, the smallest details in documentation can have the biggest impact later in court. One area where this often arises is in the collection of DNA from fingernails. Police officers may record that they collected either fingernail swabs or fingernail scrapings At first glance, these terms might look interchangeable. In reality, they may describe two different approaches to sampling — and understanding the distinction is critical. Fingernail Swabs ✅ A moistened swab is rubbed across the surface of the nail ✅ Captures DNA present on top of the nail or fingertip ✅ Useful for touch DNA or transfer material ❌ Does not usually target underneath the nail, where key trace DNA may be lodged Fingernail Scrapings ✅ A sterile tool, or mini pointed swab, is used to scrape material from underneath the nail ✅ Targets DNA that may be trapped during scratching or violent contact ✅ Can provide highly probative evidence of an assailant ✅ More invasive, but directed at a different and potentially richer source of DNA ⚖️ Why the Terminology Matters 'Swabs' may not include sampling under the nail 'Scrapings' = material was actively collected from underneath Misinterpreting these terms risks misleading the court about what evidence was (or wasn’t) collected.

Simple Random Sampling vs. Stratified Sampling! In statistics, selecting the right sampling method is pivotal, especially when dealing with varied population characteristics that could influence your results. Probabilistic techniques like simple random sampling and stratified sampling both produce unbiased estimates of the population mean, yet they differ significantly in their impact on data variation. Therefore, choosing wisely between them can dramatically enhance your data analysis outcomes. 🟢 For example, the benefit of stratification is clearly shown in the simulation below. Stratified sampling produces a tighter distribution of sample means around the population mean, compared to simple random sampling. This method not only maintains the unbiased nature of your estimates but also narrows confidence intervals, enabling more powerful statistical testing! 🟢 Namely, both methods produce an unbiased estimate of the population mean (41.2), but the key difference lies in the variation. Stratified sampling significantly reduces the variation, thereby increasing the power of the statistical testing. 🟢 So, recognizing distinct characteristics in the population (such as minority and majority groups in our case) and addressing them in sampling reduces the overall variation! This concept extends to machine learning as well, particularly in how data is handled during model training. Similar to how stratified sampling can improve statistical tests, stratified k-fold cross-validation ensures that each fold reflects the overall class distribution, which is crucial for training robust models in cases of class imbalance. When your data exhibits significant variability or class imbalance, opting for stratified techniques over simple random sampling can lead to more reliable and insightful outcomes. PS: When using stratified sampling, it is crucial to preserve the population structure. For instance, if your population consists of 20% from Class A and 80% from Class B, your sample should reflect these proportions accurately. In fact, this is the advantage of stratification over simple random sampling. #Statistics #DataScience #MachineLearning #SamplingMethods #DataAnalysis #StratifiedSampling #StatisticalTesting #Imbalancedata

Bridging the Data Gap in Artisanal and Small-Scale Mining (ASM) The vast discrepancy between official data and the reality observed on the field and even from pictures as well as videos, highlights the critical need for more accurate information on artisanal miners. This data gap hinders the development of practical and impactful solutions for the ASM sector. The following are some strategies to acquire more accurate data: 1. Utilizing Technology: ● Satellite Imagery: Analyze high-resolution satellite imagery to identify and estimate the number and size of mining sites. ● Mobile Phone Surveys: Conduct surveys via mobile phones, targeting specific regions with known ASM activity. This allows for wider outreach and real-time data collection, potentially reaching unregistered miners. ● Drone Surveys: Utilize drones for more precise mapping and estimation of mining activity in remote areas. 2. Community Engagement: ● Partner with Local NGOs: Collaborate with Non-Governmental Organizations (NGOs) with established trust within mining communities. They can facilitate data collection and ensure cultural sensitivity. ● Community Leader Involvement: Engage community leaders to act as liaisons and encourage participation in data collection efforts. ● Incentivize Participation: Offer incentives such as basic healthcare services or educational opportunities in exchange for participation in surveys or data collection processes. 3. Improved Data Collection Methods: ● Standardized Data Collection Tools: Develop standardized questionnaires that capture relevant information on demographics, mining practices, and environmental concerns. ● Training Enumerators: Train enumerators on proper data collection methodologies and cultural sensitivity to build trust with miners. ● Focus on Accessibility: Offer surveys in multiple languages and consider alternative data collection methods like focus groups. 4. Collaboration and Data Sharing: ● Government-NGO Partnerships: Foster collaboration between governments and NGOs to share resources, expertise, and data for a more comprehensive picture. ● Public-Private Partnerships: Encourage partnerships with private mining companies to share anonymized data on interactions with artisanal miners. ● Centralized Data Repository: Establish a centralized data repository for ASM information, accessible to stakeholders for informed decision-making. With the successful implementation of these strategies, we can bridge the data gap and acquire a more accurate understanding of the ASM sector. This will allow policymakers, NGOs, and other stakeholders to develop targeted interventions and practical solutions that truly address the needs and challenges faced by artisanal miners.

7 Different Types of Statistical Sampling and their Use Cases in Data Science 🧬 Sampling is a fundamental concept in statistics and data science used to draw conclusions about a population by examining a subset of it. Here’s a breakdown of different types of sampling methods and their use cases: 1. Simple Random Sampling Description: Each member of the population has an equal chance of being selected. This can be done using random number generators or drawing lots. Use Cases: • Surveys: Ensuring that every individual in a survey has an equal chance of being selected. • Quality Control: Randomly selecting products from a batch for testing to ensure quality. 2. Systematic Sampling Description: Members of the population are selected at regular intervals. For example, every nth member is chosen. Use Cases: • Manufacturing: Sampling every 10th item in a production line to check quality. • Polling: Selecting every 5th person on a list to participate in a survey. 3. Stratified Sampling Description: The population is divided into distinct subgroups (strata) based on a characteristic (e.g., age, income), and a random sample is taken from each subgroup. Use Cases: • Market Research: Ensuring that different demographic groups are represented proportionally in surveys. • Medical Trials: Ensuring that different age groups or health conditions are adequately represented. 4. Cluster Sampling Description: The population is divided into clusters (e.g., geographic areas), and a random sample of clusters is selected. All members within chosen clusters are then surveyed. Use Cases: • Epidemiological Studies: Selecting specific regions or cities to study health patterns. • Educational Research: Sampling schools or classrooms rather than individual students. 5. Convenience Sampling Description: Samples are taken from a group that is easy to access or convenient. This method is often used when time or resources are limited. Use Cases: • Initial Research: Pilot studies or preliminary research where resources are constrained. • Public Opinion Polls: Using readily available participants like social media followers. 6. Judgmental Sampling (Purposive Sampling) Description: The researcher selects the sample based on their judgment and specific criteria. It’s often used when specific characteristics or expertise are needed. Use Cases: • Expert Opinions: Consulting a select group of experts for in-depth insights. • Case Studies: Focusing on particular instances that are believed to be informative. 7. Snowball Sampling Description: Used for populations that are hard to access. Initial participants are selected and then asked to refer others, creating a “snowball” effect. Use Cases: • Social Network Studies: Researching hard-to-reach populations like marginalized communities or rare diseases. • Qualitative Research: Exploring relationships and networks within a specific group.

Sampling in Qualitative Inquiry OnlineClassHelp.Net Sampling in qualitative inquiry differs significantly from quantitative research, where sample sizes must be statistically representative. In qualitative studies, the goal is not generalizability but in-depth exploration of participants' experiences, perspectives, and meanings. This makes purposive sampling the most commonly used strategy. Key Features of Sampling in Qualitative Research ✔ Purposeful Selection – Researchers select information-rich cases relevant to their research questions. ✔ Flexibility – Sampling evolves throughout the research process as new insights emerge. ✔ Saturation-Driven – Data collection continues until no new themes emerge, rather than based on a predetermined number. 📌 Example: A study on workplace stress may involve HR managers, employees, and mental health professionals to capture multiple perspectives rather than using random sampling. Types of Qualitative Sampling Strategies 1️⃣ Purposive Sampling – Selecting participants based on expertise or relevance to the study. 2️⃣ Snowball Sampling – Relying on referrals from participants to find additional subjects. 3️⃣ Convenience Sampling – Using easily accessible participants (less rigorous). 4️⃣ Theoretical Sampling – Common in Grounded Theory, where sampling evolves based on emerging data. 5️⃣ Maximum Variation Sampling – Capturing diverse perspectives for richer insights. 📌 Example: In an ethnographic study of hospital culture, researchers might purposively sample doctors, nurses, and patients to obtain varied viewpoints. Determining Sample Size in Qualitative Research Unlike quantitative studies, there is no fixed rule for qualitative sample sizes. However, some guidelines exist: ✔ Phenomenology: 6–25 participants. ✔ Grounded Theory: 30–50 participants. ✔ Ethnography: 30–50 participants. ✔ Case Study: 1–10 participants. ✔ Narrative Research: 1–2 participants. 📌 Data saturation determines when no new insights emerge, signaling the sample is sufficient. Challenges in Qualitative Sampling ❌ Over-reliance on Small Samples – Risk of insufficient depth if too few participants are included. ❌ Bias in Selection – Subjectivity in choosing participants may influence findings. ❌ Ethical Concerns – Maintaining participant confidentiality is crucial. Conclusion 🎯 Sampling in qualitative research prioritizes depth over breadth, ensuring rich, meaningful data rather than statistical generalization. Qualitative researchers can ensure trustworthy and credible findings by using strategic sampling methods and continuing data collection until saturation is reached. #QualitativeResearch #SamplingStrategy #DataSaturation #PurposiveSampling #GroundedTheory #Ethnography #CaseStudy #NarrativeResearch #SocialSciences #ResearchMethods #Interviews #DataCollection #AcademicWriting #ResearchEthics #ThematicAnalysis 🚀

Wellsite Geologist: (Sample catching and preparation) • The cutting samples contribute the basic information on the well. • Ditch or cutting samples are the only source of information on lithology, porosity and hydrocarbon shows when unforeseen events preclude wireline logs, cores and sidewalls core samples. • Wellsite geologist will make sure to obtaining the most representative cuttings possible under existing conditions. • This will require particular care during periods of caving shales, air drilling or under balance drilling, lost circulation, and other hole problems. • Wellsite often finds himself at odds with rig personnel whose ultimate aim is to drill the hole as rapidly as possible, often at the expense of obtaining good cuttings. • It has generally been found that when the quality of the samples deteriorates to the point that they are unreliable, the drilling and mud program is not being followed by the contractor. This should be brought to the attention of the drilling supervisor. • When a compromise cannot be reach and it appears that hydrocarbon shows could be overlooked do to the poor quality of the ditch samples this should be brought to the attention of Management. Collection and Preparation: •Every drilling Rig has a shaker screen for separating the cuttings from the mud as they reach the surface. •The shaker screen may or may not be a good place from which to take cuttings samples. • If the mesh size is small enough to remove small cuttings and the well is an area where there is reason to believe that no unconsolidated sand will be encountered, the shaker screen will serve as a satisfactory source of samples. • If the shaker screen is used, a broad or box should be placed at the foot of the screen for collection of composite samples. • A settling box through which a small portion of the mud is diverted will generally serve to collect more representative samples than those caught from the shaker screen. The use of such a box insures that a composite sample is collected and afford the surest means of collecting small cutting and finely divided sand. • If the drill rate indicates sandstone but none is present in the cutting samples and an increase of loose sand is observe in the de-sanders or de-silter, a settling box should be used. • Through zones of lost circulation, such a box provides practically the only means of catching samples while the shaker is by passed. Cuttings will not settle out very satisfactorily, however, from drilling mud of very high density and gel strength. • Washing and preparation of the sample of cuttings to be examined is extremely important. • In hard rock areas, the cuttings are usually quite easily cleaned. Washing usually is matter of merely hosing the sample in a container with a jet of water to remove the mud film. #wellsitegeologist #drilling #cutting #geologist #geology #oilfield #oilfieldlife #oilwell #mudlogging #oilserv

**GEOLOGICAL SAMPLING TECHNIQUES (Mineral Exploration)* 1️⃣ *Grab Sampling* *Definition:* Grab sampling is a non-systematic sampling method where a single rock sample is collected from a selected location believed to be mineralized. *Method:* A sample is “grabbed” from an outcrop, trench, dump, or vein. No fixed length, width, or orientation is maintained. Sample size depends on lithology and mineral content. *Use in Mineral Exploration:* Early-stage reconnaissance Quick assessment of mineral presence Confirmation of visible mineralization (e.g., malachite, chalcopyrite, quartz veins) *Advantages:* ✔ Fast and inexpensive ✔ Useful for identifying mineralized areas 2️⃣ *Chip Sampling* *Definition:* Chip sampling involves collecting small rock fragments (chips) at regular intervals along an exposed rock surface. *Method:* Chips are collected using a geological hammer. Sampling is done along a line or traverse. Each chip represents a short section of rock. *Use in Mineral Exploration:* Surface grade indication Lithological and alteration studies Preliminary evaluation of veins or shear zones *Advantages:* ✔ More systematic than grab sampling ✔ Better surface representation 3️⃣ *Soil Sampling* *Definition:* Soil sampling is a geochemical exploration technique used to detect anomalies caused by buried mineralization. *Method:* Soil collected from a consistent depth (usually B-horizon) Samples taken on grid or traverse lines Avoid contamination from organic matter or human activity *Use in Mineral Exploration:* Detect concealed ore bodies Target generation for drilling Regional and detailed geochemical surveys *Advantages:* ✔ Covers large areas ✔ Effective for hidden deposits 4️⃣ *Float Sampling* *Definition:* Float sampling involves collecting loose rock fragments (float) that have been transported from their original source. *Method:* Samples collected downslope or in valleys Lithology and mineral content examined Direction of transport interpreted to trace source *Use in Mineral Exploration:* Identifying unknown mineralized zones Tracing ore bodies uphill Reconnaissance exploration *Advantages:* ✔ Useful where outcrops are scarce ✔ Helps locate hidden sources #Mineral #Exploration #Sampling