Weekend batch
Avijeet is a Senior Research Analyst at Simplilearn. Passionate about Data Analytics, Machine Learning, and Deep Learning, Avijeet is also interested in politics, cricket, and football.
YouTube Keyword Research Ideas
Free eBook: Essentials of Social Media Marketing
Quantitative Methods: A Complete Overview
A Beginner's Guide on How to Do Keyword Research
Research Study: Trends & Best Practices in Onboarding
Can't find what you are looking for.
Feel free to get in touch with us for more information about our products and services.
Written by Ruchir Dahal on October 18, 2021
This is one piece of a three-part series that looks at the various methods, techniques , and essential steps to ensure superior data analysis.
The majority of leaders from high-performing businesses attribute their success to data analytics. According to a survey done by McKinsey & Company , respondents from these companies are three times more likely to pin their accomplishments on data analytics.
That being said, although 1.145 trillion MB of data is created every day, stats show only 0.5% of it is analyzed to get results. This highlights a huge gap between companies that use data analytics to get ahead and those that don’t.
Grepsr has helped businesses bridge this gap, which in turn has given them an upper hand over their competitors. If you haven’t swiveled to data analytics yet, there is no better time to get started.
In the last post , we talked about the ‘why’ of data analysis, this time we will delve into the ‘how’.
The internet has enabled us to create large volumes of data at a staggering pace. Moreover, the way you analyze it depends on the type of data you are working with. We broadly classify data into two forms — qualitative and quantitative .
With so much data being created every day, it becomes imperative to go beyond the traditional methods to analyze this huge chunk of invaluable information.
Qualitative and quantitative data each have their own ways of being processed.
Learn the key differences between qualitative and quantitative research from our recent blog.
Quantitative data analysis is a more traditional form of analysis. As mentioned earlier, this process crunches numbers to get results.
Since one of the major functions of this process is to run algorithms on statistical data to obtain the outcome, the methods used in quantitative data analytics range from basic calculations like mean, median, and mode to more advanced deductions such as correlations and regressions.
Some of the scopes of quantitative data analysis include:
Qualitative data analysis is used when the data you are trying to process cannot be adjusted in rows and columns. It involves the identification, examination, and elucidation of themes and patterns in data (mostly textual) to bolster the decision-making process.
Unlike quantitative analysis, qualitative data analysis is subjective . This method of analysis allows us to move beyond the quantitative traits of data and explore new avenues to make informed decisions.
The following are some of the scopes of qualitative data analysis:
Learn more about qualitative data analysis in detail:
Quantitative and qualitative data analysis when used together can help you generate deeper insights . More often than not, quantitative and qualitative data can be collected from the same data unit as you can see below.
To gain richer insights you can even pair these two methods in different domains. Best use cases include Google analytics & user interviews, social media & community engagement, marketing & surveys, and so on.
Bear in mind that your way of analysis completely depends on your requirements. Sometimes, quantitative analysis will be more than enough, and other times, only qualitative will do just fine. When you want to dig deep into the data at hand, it is advisable that you go with both qualitative and quantitative data analysis.
If you want to learn more about the different techniques to perform qualitative and quantitative data analysis, click here . To add to that, if you ever need to analyze large amounts of data or need expert help, you can hire the top freelance data analysts to augment your development teams for data analysis.
We understand that your business has very specific needs when it comes to data. Be it data in the form of hard numbers or just images, we specialize in extracting data from far-flung areas of the internet.
Let us know about the data you need for effective analysis, and we will get back to you in a jiffy!
Are you drowning in, or swimming through your data? Your business is likely flooded with data: customer intel, operational data, and market insights, pouring in like a torrent. And most enterprises, George Kobakhidze of ZL Tech says, “…are not drowning in data because of its depths, they are drowning because they don’t know how to […]
By the time you’re done reading this post, human activity on the web and across devices will generate 27.3 million terabytes of data. According to Bernard Marr, author of Data Strategy, in the 21st century, “every business is a data business.” What information do you want to collect? Where are you going to store the […]
What if you could predict the next sleeper hit, build your own personalized recommendation engine, and forecast trending travel destinations? This isn’t science fiction. This is the power of IMDb data scraping. IMDb is perhaps the most authoritative voice in movie and TV content for good reason — with 200+ million unique monthly visitors and […]
Netflix, Spotify, Walmart, and other giants haven’t bet on their billion-dollar fortunes by shooting in the dark. These companies’ proactive analytics allow them to curate hyper-targeted services that offer a core feature to their customers: personalization. The question is — are you still relying only on historical data to drive your business? We’re living in […]
With the rise in Global Big Data analytics, the market’s annual revenue is estimated to reach $68.09 billion by 2025. Like the vast and deep ocean, Big Data encompasses huge volumes of diverse datasets that gradually mount with time. It refers to the enormous datasets that are far too complex to be handled by traditional […]
Data is ubiquitous and plays a vital role in helping us understand the world we live in. Quantitative data, in particular, helps us make sense of our daily experiences. Whether it’s the time we wake up in the morning to get to work, the distance we travel to get back home, the speed of our […]
This is one piece of a three-part series that looks at the various data analysis methods, techniques, and essential steps to ensure its superiority. According to Wikipedia, data analysis is a process within data science of inspecting, cleansing, transforming, and modeling data with the goal of discovering useful insights, informing conclusions, and supporting decision-making. Data […]
This is one piece of a three-part series that looks at the various methods, techniques, and essential steps to superior data analysis.
Objectivity vs subjectivity The stories we hear as children have a way of mirroring the realities of everyday existence, unlike many things we experience as adults. An old folk tale from India is one of those stories. It goes something like this: A group of blind men goes to an elephant to find out its […]
There are several reasons why we believe that visual representation of data is becoming an integral part of Big Data analytics or any other kind of data-driven analytics, for that matter
So you’ve set up your online shop with your vendors’ data obtained via Grepsr’s extension, and you’re receiving their inventory listings as a CSV file regularly. Now you need to periodically monitor the data for changes on the vendors’ side — new additions, removals, price changes, etc. While your website automatically updates all this information when you […]
How web scraping and data mining can help predict, track and contain current and future disease outbreaks
Our stats since the start of 2018
Advanced information technology has brought a massive paradigm shift in every aspect of human life We spend more and more of our working hours on the digital screens, either generating or aggregating digital data. Internet, what would have seemed something unimaginable only a few decades ago, has become an essential part of our daily businesses. […]
Digital Technology and Rediscovery of Geography A substantial amount of data that Grepsr processes and provides to its business partners worldwide contains location-specific information. According to IDC, an American data research firm, 80% of data collected by organizations has location element, and according to ABI Research, location analytics market will rise up to $9 billion by […]
In the recent years, data mining has become a prickly issue. The big controversies and clamors it has gathered in the political and business arenas suggest its importance in our time. No wonder, it is used as a household name in the business world. Data mining, in fact, is an inevitable consequence of all the technological innovations […]
Grepsr is what we like to call, “Managed Data Extraction Service”. Here are some of the reasons why we call it “managed”: We let you focus on your business and use the data — worrying about technical details of extraction is our job, and we will do it for you. We let you describe your […]
Root out friction in every digital experience, super-charge conversion rates, and optimize digital self-service
Uncover insights from any interaction, deliver AI-powered agent coaching, and reduce cost to serve
Increase revenue and loyalty with real-time insights and recommendations delivered to teams on the ground
Know how your people feel and empower managers to improve employee engagement, productivity, and retention
Take action in the moments that matter most along the employee journey and drive bottom line growth
Whatever they’re are saying, wherever they’re saying it, know exactly what’s going on with your people
Get faster, richer insights with qual and quant tools that make powerful market research available to everyone
Run concept tests, pricing studies, prototyping + more with fast, powerful studies designed by UX research experts
Track your brand performance 24/7 and act quickly to respond to opportunities and challenges in your market
Explore the platform powering Experience Management
Popular Use Cases
Market Research
The annual gathering of the experience leaders at the world’s iconic brands building breakthrough business results, live in Sydney.
A user’s guide to data analysis.
29 min read Data on its own is nothing but facts and figures. To be useful, raw data needs to be broken down, modelled, and interrogated to provide useful information. You'll discover how to do this and more in this complete guide.
Data on its own is nothing but facts and figures. To be useful, raw data needs to be broken down, modelled and interrogated to provide useful information. Especially in businesses and organisations where decisions are based on data, collected data must be analysed and presented correctly and simply.
But what is data analysis, what are the best data analysis techniques, and why is it useful? Read on to find out.
Data analysis is a broad term that encompasses structured and scientific data collection, analysis, cleansing and data modelling. Data analysis applies to any source or amount of data, and helps to uncover insights and information that supports decision-making. Data science, on the other hand, only applies to quantitative data.
Data analysis typically revolves around two types of data: qualitative and quantitative data:
Qualitative data is descriptive and typically unstructured. Examples of qualitative data include: interview transcripts, open-ended answers to surveys, field notes, recordings, questionnaires (but only if you have a small sample size) and so on.
Put simply, quantitative data is survey data with sufficient sample sizes. It’s essentially data that you can count and assign a numerical value, e.g. revenue in dollars, weight in kilograms, height in feet or inches, length in centimetres. Quantitative data is often structured and suitable for statistical analysis .
You can use both types of data to gain an understanding of the entire business landscape, from how your offering fits into the picture to what customers are looking for.
For example, it can be used to understand the marketplace, your business’ position within the marketplace, and provide customer insights around preferences, decisions and behaviours.
We’ve added a table below to provide you with an at-a-glance view of the differences between qualitative and quantitative research.
Qualitative | Quantitative |
---|---|
Gathered from focus groups, interviews, case studies, expert opinion, observation, social media (via scraping) | Gathered from surveys, questionnaires, polls |
Use open-ended and open text questions | Use closed-ended (yes/no) and multiple choice questions and open-ended responses (providing a coding scheme is applied to quantify the topics) |
Uses a ‘human touch’ to uncover and explore an issue (e.g. a customer complaint) | Cannot use a ‘human touch’ to interpret what people are thinking or feeling |
Helps to formulate a theory to be researched | Tests and confirms a formulated theory |
Results are categorised, summarised and interpreted linguistically | Results are analysed mathematically and statistically |
Results expressed as text | Results expressed as numbers, tables and graphs |
Fewer respondents needed | Many respondents needed |
Less suitable for scientific research | Tests and confirms a formulated theory |
Helps to formulate a theory to be researched | More suitable for scientific research as it is compatible with most standard statistical analysis methods |
Harder to replicate | Easy to replicate |
Less suitable for sensitive data: respondents may be biased, too familiar or inclined to leak information | Ideal for sensitive data as it can be anonymised and secured |
Find out more about qualitative and quantitative research
The use of data in decision-making has become more important than ever. Across the globe, businesses and organisations are basing decisions on facts and proven theories, supported by data analysis, rather than making critical decisions on the best guess. These approaches are helping them to plan more efficiently, sustain growth and deliver amazing products.
Here are a few examples of how you can use data analysis:
Using the right data analysis methods, you can gain a complete understanding of your customers.
You can learn everything from their brand, product and service preferences to which channels they use when researching those products and services. You can even uncover their shopping behaviours and how they change based on what they’re buying.
You can also use data analysis to better understand your customers’ demographics, interests and habits so you can better tailor your marketing and brand messaging around themes they’ll connect with.
Marketing is highly reliant on information and data analysis to measure its effectiveness and success — or failure.
From website analytics like measuring traffic and leads to analysing brand sentiment, data analysis is critical to understanding the full picture of your marketing strategy. Based on measurement requirements, you can implement data collection and analysis methods to plug the gaps across the entire buyer journey, enabling you to make specific changes at each stage to help drive growth.
Presentation techniques like data visualisation also form a huge part of marketing analysis (we’ll talk more about data visualisation later on).
Understanding the marketplace is an essential part of figuring out your positioning, how the market is changing and what your business needs to do to adapt to competitors or fast-paced market conditions.
All this can only be understood with the constant collection and analysis of data.
By collecting accurate data, analysing it closely and gaining insights you can ensure your organisation is ready and able to adapt when it needs to.
It’s important to understand that there are many different types of data analysis. Each type has its uses, and your choice ultimately depends on what kind of information you want to get out of the process — and if you want to get qualitative or quantitative data.
Here’s a quick breakdown of some the main types of data analysis you can use and what they’re best for:
Descriptive analysis is a form of data analysis that uses responses from a data set to establish patterns or understand the average of one variable (both in isolation and how it compares to other variables in the study).
Descriptive analysis is typically used as a means of understanding occurrences or preferences. For example, you would use descriptive analysis when trying to determine the most popular type of exercise people did in the last year.
In your survey, you’d present a range of responses (running, weight lifting, swimming). Once the survey results were collected, you could use descriptive analysis to understand the popularity of each activity, the averages among the variables and how each variable compares to the other.
Where descriptive analysis is about understanding trends, diagnostic analysis is the next stage in uncovering the causes of those trends and highlighting any correlation between the variables in your study.
Diagnostic analysis usually starts with a hypothesis that you’re trying to prove or disprove. For example, your hypothesis could be that the number of people signing up to a gym in August is caused by the hot weather.
You’ll use this hypothesis to guide your data analysis and keep you focused on what you’re looking at.
Another thing to keep in mind with diagnostic analysis is understanding the distinction between correlation and causation.
Misunderstanding these two concepts could lead to you making incorrect interpretations of your data. When we talk about correlation, there can are two types you likely see:
Let’s say we have two variables. A positive correlation means that as one variable increases, so does the other.
In this case, as one variable increases, the other decreases.
However, just because the variables are correlated, it doesn’t always mean that one change is caused by the other.
When doing diagnostic analysis, causation is the ideal insight to gain, but correlation can still provide useful insights too.
In research, causation is when one factor (or variable) causes another. In other words, there is a cause-and-effect relationship between the two. Correlation doesn’t imply causation, but causation always implies correlation.
For example, a lack of graphics card memory can cause visual processes on a computer to freeze or not work entirely. Visual process failures and an inoperable graphics card are correlated, but they don’t highlight the cause. Similarly, a lack of random access memory (RAM) might cause your phone to freeze or lock up — the lack of RAM is the cause and the two issues (the phone freezing and locking up) are correlated, but neither causes the other.
Causation is important because it brings you to the root of all issues and enables you to see how other variables are influenced in the process.
Exploratory analysis is used by data analysts to assess raw data and understand its primary characteristics, and then determine how to manipulate the data sources to get different answers.
Exploratory analysis is primarily used to prove the validity of results gathered from data and that they apply to any goals or objectives. Essentially it’s used as a way to use data before making any assumptions about a situation.
Once the raw data is collected, data analysts can then manipulate the data sources to test the impact of changing variables on the overall results. It can be particularly useful when checking assumptions, uncovering anomalies and discovering patterns.
Exploratory analysis can also be used for more complex studies and modelling, like machine learning.
As the name suggests, predictive analysis is a data analysis process that uses historical data, algorithms and even machine learning to try to predict what will happen in the future based on previous trends.
Predictive analysis has been rapidly growing in popularity in businesses and organisations as the data analysis tools used to map the future predictions – and better advances in machine learning – have made predictions more accurate.
It also has multiple business applications, which is another reason it’s so popular.
For example, predictive analysis is becoming a key part of cyber security and crime prevention — particularly when it comes to fraud.
By compiling vast amounts of data in real-time and analysing results and trends to uncover patterns, predictive analysis can help data analysts detect abnormalities in behaviour or actions based on previous events and take the necessary actions.
Reducing consumer risk is another major area for predictive analytics and it’s often used when assessing credit scores and predicting whether customers can afford to take on expensive financial products – like a mortgage or loan – based on their previous history.
Leisure businesses like airlines and hotels also use predictive analytics by using data from previous years to predict usage when setting prices.
Find out more about predictive analytics
Prescriptive analysis is one of the more advanced data analysis techniques and is often used by businesses and organisations trying to work out what decisions to make, or what steps they need to take to make something happen, such as increasing sales in Q4 compared to the previous year.
It involves data mining from multiple sources like resources, historical data and past performances and using advanced analysis techniques like machine learning to model the data and provide insights to guide future decisions.
It’s a relatively new type of data analysis because it relies on the collection and analysis of huge amounts of data and advanced technology like artificial intelligence or machines to process and analyse the data effectively.
As more data is provided and analysed, the models will become more accurate and change based on the variables being input.
Prescriptive analysis is often accompanied by data visualisation tools to help decision-makers understand what they’re looking at and take appropriate actions.
It does, however, require businesses and organisations to know the types of questions to ask to gain the correct information and stop people from making decisions based on the wrong assumptions. It also requires the ability to make the right judgments based on the data models provided.
The data analysis process is a fairly straightforward one regardless of the data analysis techniques being used.
But you do need to follow it properly if you want to capture the right data and glean meaningful insights that you can use.
You can see the outline of the data analysis process in the graphic above, but this is a quick rundown of the data analysis steps you’ll take:
The first question to ask before embarking on any data analysis is why are you looking to analyse data in the first place?
Do you need it to guide strategic business decisions or update business processes? Are you trying to find an answer to a specific question or do you want to test a hypothesis? Or do you need to make improvements to operational processes but don’t know where to start?
Why you need data can help you figure out the right data analysis methods to choose and also guide key decisions like whether you need qualitative data analysis or quantitative data analysis (or a combination of the two).
Then there’s the what of your data analysis — what data should you be collecting? What conclusions do you want to draw from the data? Clearly defining the what will help you to select the appropriate tools and methodology to capture the data for the why.
Once you’ve decided why you need the data and what method of data analysis is best, you can start the process of collecting your raw data.
This could be devising and sending out surveys for quantitative data analysis, emailing customers questionnaires or running focus groups for qualitative data analysis.
Regardless of how you collect your data, you have to account for errors and put measures in place to prevent them. For example, preventing participants from submitting the same survey twice to avoid duplication. Instead, allow them to modify their answers (if it’s an online survey).
For incomplete surveys, questionnaires or interviews, ensure that participants can only submit them once they’ve filled in the required fields. If they don’t have an answer, encourage them to put N/A, for instance.
Incorrect, incomplete and duplicate responses can affect the quality, consistency and usability of your data, preventing you from making accurately informed decisions.
Once you’ve collected and cleansed your data, you can begin the process of data analysis. As you’re analysing your data you’ll be able to identify patterns in your data that you can use to inform your business decisions.
(Predictive) Regression analysis : used to estimate the relationship between a set of variables. The idea is to find a correlation between a dependent variable (the outcome you want to measure or predict) and any number of independent variables.
For example, if you’ve been planting trees in an area at the same rate every day for a year, it can predict how many trees you’ll have planted in 5 or 10 year’s time.
(Predictive) Time series analysis: Time series analysis is a statistical technique used to identify trends and cycles over time, e.g. sales forecasts such as weekly or monthly profits, or fluctuations based on seasonality.
For example, time series analysis is helpful when it comes to industry forecasts, weather data and rainfall measurements, and stock prices. This is because rather than collecting data over time, analysts record data points at specific intervals, giving a complete, constantly evolving picture of the data.
(Predictive and prescriptive) Monte Carlo simulation: this is a complex, computerised technique designed to generate models of possible outcomes and their probability distributions. It measures a range of possibilities and calculates their likelihood.
This simulation is used for outcomes that are difficult to predict due to the intervention of random variables. It helps to understand the impact of risk and uncertainty in predictions and forecasting models. It’s also referred to as a multiple probability simulation.
For example, you could use Monte Carlo simulation to determine which moves to play in chess (of which there are between 10 111 and 10 123 positions (including illegal moves), which is more than the number of atoms in the world. The computer essentially calculates
all these possible moves (with the most powerful computers doing trillions of calculations per second) and continuously plays until it satisfies a set end condition, e.g. a win.
(Exploratory) Factor analysis : Factor analysis is used to reduce a large number of variables to a smaller number of impactful factors. It condenses large datasets into smaller, manageable chunks and helps to uncover hidden patterns.
For example, say you conduct a survey of 500 townspeople, resulting in a dataset of 500 variables. You could work to find which variables are correlated and put them into groups, e.g. income, education, family size. These groups are factors. From there, it becomes easier to analyse the data.
(Exploratory) Cohort analysis: Cohort analysis is a subset of behavioural analytics that takes the data from a given dataset and breaks it into groups for analysis. These related groups (or cohorts) usually share common characteristics or experiences.
For example, you could use cohort analysis to understand customer expenditure over time. You can analyse trends in purchase behaviour and then gauge whether or not the quality of the average customer is increasing throughout the lifecycle.
(Exploratory) Cluster analysis : This exploratory technique seeks to identify structures and patterns within a data set. It sorts data points into groups (or clusters) that are internally similar and externally dissimilar.
For example, in medicine and healthcare, you can use cluster analysis to identify groups of patients with similar symptoms. From there, you can alter your method of care. You can also use cluster analysis in areas like marketing to identify homogeneous groups of customers that have similar needs and attitudes.
(Exploratory) Sentiment analysis : A qualitative technique that belongs to a broad category of text analysis. Sentiment analysis looks at, assesses and classifies emotions conveyed through text data.
For example, you can use it to determine how customers feel about your brand, product or service based on feedback.
Once you’ve finished analysing data from your study, you can begin your data interpretation and begin to apply actions based on what the results are telling you.
There are plenty of business intelligence tools you can use to model your data to make this interpretation easier and ensure you can make decisions quickly. We’ll outline a few of those tools shortly — but first, here are a few mistakes to avoid.
Diligence is essential when it comes to data analysis — but when you’re running complex studies at scale, it’s sometimes hard to keep on top of the quality and assurance process. Here are just a few of the most common data analysis mistakes researchers make and how to avoid them:
Sample bias is when you choose a sample that is non-representative of the wider population you’re trying to assess.
Any bias in your sample can result in data skewing more to one side and not providing reliable results.
A simple example of this is sampling 1,000 people to assess political preferences but oversampling one political allegiance.
One of the most effective ways to avoid sampling bias is to use simple random sampling. This ensures that samples are chosen by chance — and every person in the population has an equal chance of being selected.
This happens in data science when data analysts try to fit their data to support a particular theory or hypothesis. It can occur by accident, but is typically an intentional act and can have a serious impact on the validity of the study. Data manipulation also applies when participants can submit a survey more than once, skewing the overall results if you don’t double-check duplicate contact data.
You can avoid data manipulation by:
The respondents of your study should never be aware of the metrics you’re measuring your study with, because once they do you could end up in a situation where they try to tell you what you want to know.
That said, it’s helpful to provide respondents with guidance and the context of the study — why it’s important and relevant and how their honest responses can contribute to its validity. This includes a preface about the questions that are going to be asked.
This is a very common problem in data science when you find that data only shows you a result because it fits the data modelling you’re using it with. If you move the data over to another model, you could find it doesn’t then show any results.
For example, you might focus so heavily on the accuracy of a particular model that it can only fit a particular use case, e.g. measuring sales trends based on seasonality. Analysts typically build and use machine learning models to apply them to general scenarios — not incredibly specific ones. Overfitting the model will mean that it only works for one situation and subsequently fails others.
This particular problem is like comparing apples with oranges. If you create a time series model specifically for seasonal sales trends, that model will only give you that data. You can’t suddenly apply the data to other models and expect to get the same results, because they won’t account for the same variables.
Numbers can only tell you part of the story, and making decisions on numbers alone can result in negative consequences.
While quantitative data can produce some useful insights it must also be used in the context of the wider market or environment and, ideally, aided with some qualitative insights.
Solution bias can be a big risk for businesses who are convinced they have a good product or service, and are trying to find data to support their theory at any cost.
In these circumstances, you could end up with data analysis that you’re using simply to confirm your own assumptions, rather than properly testing your theory.
There are several tools available that can make analysing raw data easier and improve your data visualisation so you can easily interpret your information and make decisions.
Here are some of the most common and best data analysis tools available:
R is a free, open-source data analysis tool that can be used for a range of data science, statistical analysis and data visualisation studies.
Using R, data analysts can set up and run data models, which can be automatically updated as new information is added.
SAS is one of the most widely used software packages for data analysis. It’s primarily used for statistical analysis and data visualisation.
As well as quantitative analysis, SAS can be used for qualitative analysis, is highly customisable based on the data analysis you want to use, and offers a range of statistical methods and algorithms.
Python is an effective tool for every stage of data analytics and is widely used by data analysts. It’s an easy software language to learn and use and is highly effective for data mining — particularly when scraping large amounts of data from multiple sources.
Java is one of the most versatile statistical programming languages as well as coding languages that can be used for numerous types of data analysis. Java can be connected to SQL and MySQL databases. It’s also highly customisable and can be used for mass data scraping, analysis, and visualisation.
SQL is a relatively simple processing language that can be used to interact with multiple data sources at once, making it highly effective. It can also perform complex data analysis which, when combined with the data sources used, make it a highly accessible and effective data analysis tool for data analysts.
Data analytics is a highly effective tool for improving ROI because you can be sure you’re making decisions based on data, rather than instinct.
This is particularly true when it comes to using data analytics to gain customer insights, but also for improving specific aspects of your business. For example, you could carry out an employee engagement survey and use data analytics to uncover trends and areas for improvement. Your analysis of the survey results might find that your employees want more remote working/hybrid working options — you can then implement new flexible and hybrid working policies to support them. This in turn will improve productivity and engagement, which in turn can support your bottom line.
Here’s another example — let’s say you run a conjoint analysis survey to determine the optimal bundle of benefits and features, including price. You can readily test product options, features and pricing to find out what customers are most likely to buy and therefore what product variants will contribute to your bottom line.
You can even apply data analytics to product concept testing to uncover whether or not your product ideas are up to scratch and marketable to your target audience. This ensures you spend less time investing in ideas that won’t work, and more time on ideas that will.
The possibilities are endless and data analysis applies to every area of your business. By taking the time to analyse trends within your data, you can start to create better experiences and outcomes for all.
Free eBook: Start maximising your research ROI
Market intelligence tools 10 min read, qualitative research questions 11 min read.
Analysis & Reporting
Primary vs secondary research 14 min read, business research methods 12 min read, ethnographic research 11 min read, business research 10 min read, request demo.
Ready to learn more about Qualtrics?
Choosing your data type, qualitative.
Data that is verbally based (words and concepts).
Offers insight into research questions. May identify emerging trends in the data not previously considered by the researcher. Provides more direct representation of subjects’ responses. | Requires multiple stages of data analysis. Introduces researcher subjectivity in data analysis (see ). |
Data is numerically based (numbers only).
Allows for the direct application of statistical models such as ANOVA or t-tests (see ) to identify general trends and patterns. Potential for large data sets. | Limits insights to what the data shows statistically. |
One survey data type is not necessarily better than another. As summarized by Ahmad (2019), “Quantitative data can help to see the big picture. Qualitative data adds the details.” What type of data you need is going to be dependent on what you are trying to analyze.
Open-ended questions.
Open-ended questions are those in which a survey respondent can generate a unique response using their own words. These are seen predominantly in qualitative data surveys. These types of questions are particularly useful when information is needed about individual-specific context that might not be accounted for in a multiple-choice type closed-ended format. A key benefit of open-ended questions is that they allow for respondents to give personalized responses that are not confined to the choice selection set by the researcher. This is simultaneously a key disadvantage, however as it introduces the need for qualitative coding and in turn, introduces a new source of error.
Some examples of open-ended survey questions would be:
Describe any study techniques you found to be beneficial.
Open-ended survey questions are generally useful for small sample populations. Though these questions offer great insights, on a large sample population scale they are often unfeasible due to the administrative planning and analysis required. If you want the insight from open-ended questions but are working with a large respondent pool, a small sample population can be given an open-ended question-based Pilot Survey to obtain information and inform the design of close-ended question surveys for your larger populations.
Closed-ended questions are those in which the response options are limited and provided with the survey. These can be used in qualitative data acquisition as well as quantitative. Some of the main benefits of closed-ended questions are the reduction of the need for communication skills on the behalf of the respondent and the ease of analysis. Conversely, some of the main disadvantages include lack of depth in responses and lack of emergent insights (Hyman and Sierra 2016). These question types help to eliminate sources of error in the data analysis by reducing or eliminating the need to code free response answers; however, the data obtained will be limited to response options generated by the researcher. Closed-ended survey questions can often be evaluated statistically and are easier to use when evaluating large sample sizes, which encourages studies that are more generalizable. Some examples of closed-ended survey questions would be:
Often, qualitative data are linked to open-ended questions while closed-ended questions are paired with quantitative data. The reality is that open-ended qualitative questions can be converted into quantitative data and conversely closed-ended quantitative questions can be used to glean qualitative data. The key is the coding, so write the questions in whichever way will give you the data you are most needing to see while keeping in mind the logistical elements that come along with delivering each type to the population you are studying.
Ahmad, S., Wasim, S., Irfan, S., Gogoi, S., Srivastava, A., & Farheen, Z. (2019). Qualitative v/s. Quantitative Research- A Summarized Review. Journal of Evidence Based Medicine and Healthcare, 6(43). https://journals.indexcopernicus.com/api/file/viewByFileId/916903.pdf
Hyman, M., & Sierra, J. (2016). Open- versus close-ended survey questions. NMSU Business Outlook , 14 (2), 1–5.
Moody Library, Suite 201
One Bear Place Box 97189 Waco, TX 76798-7189
Published on 9.7.2024 in Vol 26 (2024)
Authors of this article:
RTI International, Research Triangle Park, NC, United States
Claudia M Squire, MS
RTI International
3040 East Cornwallis Road
Research Triangle Park, NC, 27709-2194
United States
Phone: 1 9195416613
Email: [email protected]
Background: In-depth interviews are a common method of qualitative data collection, providing rich data on individuals’ perceptions and behaviors that would be challenging to collect with quantitative methods. Researchers typically need to decide on sample size a priori. Although studies have assessed when saturation has been achieved, there is no agreement on the minimum number of interviews needed to achieve saturation. To date, most research on saturation has been based on in-person data collection. During the COVID-19 pandemic, web-based data collection became increasingly common, as traditional in-person data collection was possible. Researchers continue to use web-based data collection methods post the COVID-19 emergency, making it important to assess whether findings around saturation differ for in-person versus web-based interviews.
Objective: We aimed to identify the number of web-based interviews needed to achieve true code saturation or near code saturation.
Methods: The analyses for this study were based on data from 5 Food and Drug Administration–funded studies conducted through web-based platforms with patients with underlying medical conditions or with health care providers who provide primary or specialty care to patients. We extracted code- and interview-specific data and examined the data summaries to determine when true saturation or near saturation was reached.
Results: The sample size used in the 5 studies ranged from 30 to 70 interviews. True saturation was reached after 91% to 100% (n=30-67) of planned interviews, whereas near saturation was reached after 33% to 60% (n=15-23) of planned interviews. Studies that relied heavily on deductive coding and studies that had a more structured interview guide reached both true saturation and near saturation sooner. We also examined the types of codes applied after near saturation had been reached. In 4 of the 5 studies, most of these codes represented previously established core concepts or themes. Codes representing newly identified concepts, other or miscellaneous responses (eg, “in general”), uncertainty or confusion (eg, “don’t know”), or categorization for analysis (eg, correct as compared with incorrect) were less commonly applied after near saturation had been reached.
Conclusions: This study provides support that near saturation may be a sufficient measure to target and that conducting additional interviews after that point may result in diminishing returns. Factors to consider in determining how many interviews to conduct include the structure and type of questions included in the interview guide, the coding structure, and the population under study. Studies with less structured interview guides, studies that rely heavily on inductive coding and analytic techniques, and studies that include populations that may be less knowledgeable about the topics discussed may require a larger sample size to reach an acceptable level of saturation. Our findings also build on previous studies looking at saturation for in-person data collection conducted at a small number of sites.
In-depth interviews are commonly used to collect qualitative data for a wide variety of research purposes across many subject matter areas. These types of interviews are an ideal approach for examining individuals’ perceptions and behaviors at a level of depth, complexity, and richness that would be challenging to achieve with quantitative data collection methods. Typically, trained interviewers conduct interviews using a guide designed to address the study’s key research aims by asking a series of questions and probes ordered by topic. These interview guides can range from highly structured to completely unstructured (eg, loosely organized conversations). Following the completion of data collection, interview notes and transcripts generated from audio recordings of the interviews are analyzed to assess for patterns in responses among the interviewees or subsets of the participants [ 1 , 2 ].
During the COVID-19 pandemic, web-based data collection became increasingly common, as traditional in-person data collection was not possible, and researchers continue to use web-based data collection methods post the COVID-19 emergency, citing advantages such as accessing marginalized populations, achieving greater geographic diversity, being able to offer a more flexible schedule, and saving on travel expenses [ 3 ]. Potential concerns about web-based data collection, such as the inability to build rapport and data richness, have been largely unfounded [ 3 , 4 ].
While we do not expect web-based data collection to supplant in-person research, it continues to show signs of growth. To date, much of the research on qualitative methods has focused on in-person data collection. Consequently, it will be important to conduct research to determine if previous widely accepted findings hold true for web-based data collection.
Researchers typically make a priori decisions about the number of interviews to conduct with the aim of balancing the need for sufficient data with resource limitations and respondent burden. The concept of saturation is frequently used to justify the study’s rigor with respect to the selected sample size. To provide empirically based recommendations on adequate minimum sample sizes, researchers have conducted studies to assess when saturation occurs. However, multiple types of saturation exist—such as theoretical, thematic, code, and meaning—and within each type of saturation, the definitions and measurement approaches used by investigators vary substantially, as does the level of detail researchers report in publications about their methods for achieving and assessing saturation [ 5 ].
This study aimed to examine the number of interviews needed to obtain code saturation for 5 recently conducted studies funded by the Food and Drug Administration [ 6 ] involving web-based interviews. Specifically, how many web-based interviews are needed to obtain true code saturation (ie, the use of 100% of all codes applied in the study) and how many web-based interviews are needed to achieve near code saturation (ie, the use of 90% of all codes applied in the study)?
Multiple authors have defined saturation as the point during data collection and analysis, at which no new additional data are found that reveal a new conceptual category [ 7 - 13 ] or theme related to the research question—an indicator that further data collection is redundant [ 11 ]. Additionally, Coenen et al [ 14 ] specified that no new second-level themes are revealed in 2 consecutive focus groups or interviews.
Other authors have distinguished between various types of saturation. One of the most common types of saturation mentioned in the literature is theoretical saturation, which emerges from grounded theory and occurs when the concepts of a theory are fully reflected in the data and no new insights, themes, or issues are identified from the data [ 5 , 11 , 12 , 15 - 18 ]. Hennink et al [ 17 ] expanded this definition, adding that all relevant conceptual categories should have been identified, thus emphasizing the importance of sample adequacy over sample size. Guest et al [ 15 ] operationalized the concept of theoretical saturation as the point in data collection and analysis when new information produces little or no change to the codebook, and van Rijnsoever [ 19 ] operationalized it as being when all the codes have been observed once in the sample.
Some authors have defined theoretical saturation, thematic saturation, and data saturation as the same concept [ 16 , 18 ], whereas others have defined these terms differently [ 12 , 20 ]. For example, some authors have defined thematic saturation as the point where no new codes or themes are emerging from the data [ 12 , 21 ]. For thematic saturation to be achieved, data should be collected until nothing new is generated [ 20 , 22 ]. Data saturation has been defined as the level to which new data are repetitive of the data that have been collected [ 12 , 23 , 24 ].
Furthermore, Hennink et al [ 17 ] distinguished between code saturation and meaning saturation. Code saturation is based on primary or parent codes and relates to the quantity of the data (“hearing it all”). Meaning saturation is based on sub or child codes and relates to the quality or richness of the data (“understanding it all”). Constantinou et al [ 7 ] made the point that it is the categorization of the raw data, rather than the data, that are saturated.
The literature reflects multiple methods that have been used to determine saturation [ 7 - 10 , 13 - 18 , 21 , 25 ]. Sim et al [ 26 ] discussed the four general approaches that have been used to determine sample size for qualitative research: (1) rules of thumb, based on a combination of methodological considerations and past experience; (2) conceptual models, based on specific characteristics of the proposed study; (3) numerical guidelines derived from the empirical investigation; and (4) statistical approaches, based on the probability of obtaining a sufficient sample size.
For example, Galvin [ 9 ] used a statistical approach based on binomial logic to establish the relationship between identifying a theme in a particular sample and within the larger population; for example, number of chances of detecting a theme if that theme exists within number of the population. Using the probability equation, the researcher can determine the number of interviews needed for a stated level of confidence that all relevant themes held by a certain proportion of the population will occur within the interview sample. This method assumes the researcher knows in advance the emergent themes from the study and at what rate they may occur.
Constantinou et al [ 7 ] used the comparative method for themes saturation, which relies on both a deductive and an inductive approach to generate codes (keywords extracted from the participants’ words) and themes (codes that fall into similar categories). Themes are compared across interviews, and theme saturation is reached when the next interview does not produce any new themes. The sequence of interviews is reordered multiple times to check for order-induced error. When exploring the various methods for determining saturation, researchers reached different conclusions on when saturation was achieved (findings on saturation by other authors are present in Multimedia Appendix 1 ) [ 7 - 10 , 13 - 17 , 21 , 25 , 27 , 28 ].
Most studies assessing saturation focused on in-person data collection or did not specify the data collection method. Given recent increases in web-based data collection, studies assessing saturation for web-based interviews are critical to ensure that recommendations regarding sample size are tailored to the mode of data collection [ 4 ]. While there is evidence to suggest that the content of data coded from in-person as compared with web-based interviews is conceptually similar [ 29 ], this is a relatively new area of exploration. Rapport may be higher with in-person as compared with web-based interviews [ 30 ], which may impact the amount and type of content generated. Additionally, participants in web-based data collection studies are more geographically diverse and may be more likely to be non-White, less educated, and less healthy than participants in in-person data collection studies [ 31 ].
This study was based on analyses from data collected for 5 Food and Drug Administration–funded studies conducted using web-based platforms, such as Zoom (Zoom Video Communications) and Adobe Connect (Adobe Systems), and focused on patients with underlying medical conditions or on health care providers who provide primary or specialty care to patients. All platforms used for these interviews offered audio and video components and allowed for the sharing of stimuli on screen. A brief description of each study is provided in Table 1 . Each study’s data had been coded and stored using NVivo software (version 11; QSR International).
Study name | Sample size, n | General eligibility criteria | Primary objectives | Summary of topics | Length of interview (minutes) | Number of interview questions | Regions and states covered |
Study A | 30 | Patients diagnosed with a condition treated by biologic medications (eg, cancer, inflammatory bowel disease, and diabetes) | Obtain feedback on multimedia educational materials about biosimilar biologic medications | 90 | |||
Study B | 48 | Patients diagnosed with vulvovaginal atrophy or type 2 diabetes | Explore how patients use boxed warnings when making decisions about prescription drugs and how well the warnings meet patients’ information needs | 30 | |||
Study C | 70 | Primary care physicians or specialists who write at least 50 prescriptions per week | Assess how primary care physicians and specialists access, understand, and use prescription drug labeling information, including information on labels for drugs that have multiple indications. | 60 | |||
Study D | 35 | Patients diagnosed with type 2 diabetes | Understand how patients weigh the potential benefits against possible risks and side effects, dosage and administration characteristics, and costs when selecting treatments for chronic health conditions. | 60 | |||
Study E | 35 | Patients diagnosed with psoriasis | Understand how patients weigh the potential benefits against possible risks and side effects, dosage and administration characteristics, and costs when selecting treatments for chronic health conditions. | 60 |
This project was determined to not research with human participants by Research Triangle Institute’s institutional review board (STUDY00021985). The original 5 studies that this project is based on were reviewed by Research Triangle Institute’s institutional review board and were determined to be exempt under category 2ii. Participants in these studies were provided information about measures used to protect their privacy and the confidentiality of their data in the study’s consent forms. All participants were provided compensation for their time (the amount and type varied by study).
We established and applied a systematic approach to analyze all 5 study data sets. Our analytic approach was organized into 2 stages—data preparation and data analysis.
First, because previous interviews sometimes influence moderator probes—for example, the moderator asks a follow-up question based on something they heard in a previous interview—we sorted interviews from each study by interview order. We then extracted code- and interview-specific data from the NVivo databases—including transcript name, code name, number of files coded, number of associated parent and child codes, and number of coding references—and compiled these data in an Excel (Microsoft Corp) file. We then updated the Excel file with important code and interview characteristics, including the order in which interviews were conducted, whether each code was directly (ie, child codes) or indirectly (ie, parent codes) applied to transcripts (in a tiered coding scheme, direct codes are those that have no child codes, whereas indirect codes function as “parents” that have additional codes nested beneath them), and the point at which each code was first applied to an interview. Finally, we created pivot tables within each Excel file to compile the data.
Once the data were compiled, the data summaries were examined to determine when true saturation and near saturation occurred during data collection. True saturation was defined as 100% of all applied codes being used; near saturation was defined as 90% of all applied codes being used. We calculated saturation separately for each study’s data set, and we calculated saturation separately for all codes (ie, parent and child codes) as compared with direct codes (ie, child codes only). True saturation and near saturation points were identified by calculating the cumulative percentage of new codes for each interview, flagging when 100% and 90% of applied codes had been used.
The number of web-based interviews used across the 5 studies ranged from 30 to 70 ( Table 2 ). True saturation (100% use of all applied codes) was reached in the final or near final interview ( Figure 1 ), suggesting that, even with a large sample size, additional interviews are likely to continue uncovering a small number of new codes or findings.
Study | Total interviews, n | Coding: total codes in codebook, n | True saturation: interviews needed, n (%) | Near saturation: interviews needed, n (%) |
Study A | 30 | 657 | 30 (100) | 18 (60) |
Study B | 48 | 313 | 47 (98) | 21 (44) |
Study C | 70 | 362 | 67 (96) | 23 (33) |
Study D | 35 | 205 | 33 (94) | 15 (43) |
Study E | 35 | 200 | 32 (91) | 15 (43) |
Across all studies, near saturation (90% use of all applied codes) was reached near—and often before—the midpoint of data collection. In other words, only a small number of new codes or findings were uncovered once the first half of the sample had been interviewed. In terms of absolute numbers, the point at which near saturation was reached occurred between 33% and 60% (n=15-23) of planned interviews ( Table 2 ). Despite the participants being more geographically, and possibly demographically, diverse compared with typical in-person participants, our findings were similar to previous studies on saturation [ 10 , 15 , 17 ].
We examined the types of codes applied after near saturation had been reached. In 4 of the 5 studies, most of these codes (n=8-33, 57%-62%) represented previously established core concepts or themes, such as a trusted source of information, a behavioral intention, or a recommended change to educational material. Codes representing newly identified concepts (n=2-8, 10%-15%), other miscellaneous responses (eg, “in general”; n=6-9, 13%-41%), uncertainty or confusion (eg, “don’t know”; n=0-6, 0%-11%), or categorization for analysis (eg, “correct as compared with incorrect”; n=0-3, 0%-4%) were less commonly applied after near saturation had been reached.
The overwhelming majority of codes applied after near saturation (n=9-41, 73%-82%) had already been established in study codebooks before analysis. Only a small number of codes applied after this point (n=4-20, 18%-27%) were conceptually distinct enough to merit updating the study codebooks by including them. Likewise, most of the codes used after near saturation (n=11-35, 44%-64%) were applied to only a single interview. Far fewer codes were applied to 2 interviews (n=0-13, 0%-27%), 3 interviews (n=0-6, 0%-21%), or 4 or more interviews (n=0-12, 0%-21%).
Study B was an outlier in terms of codes applied after near saturation. This study had fewer codes representing core established concepts (n=8, 28%) and more codes representing newly identified concepts (n=7, 24%) or providing categorization for analysis (n=3, 10%) than other studies. The study also had a much higher proportion of new codes (n=20, 69%) that were added to the study codebook during analysis. These differences may be because the study sampled 2 populations with very different medical conditions (ie, type 2 diabetes as compared with vulvovaginal atrophy), leading to a broader range of applied codes.
In examining the relationship between the number of codes in the codebook for each study, the study with the most codes (study A: 657 codes) required the largest number of interviews to reach both true saturation and near saturation. However, this pattern did not hold true for the remainder of the studies. The study with the next highest number of codes (study C: 362 codes) was third to reach true saturation and last to reach near saturation.
All 5 study codebooks included both parent (ie, top-level codes) and child codes (ie, subcodes). We examined saturation using two analytic lenses—(1) all codes (parent and child) and (2) parent codes only—to determine if there were differences in when saturation was reached. We found no differences in when true saturation was reached. However, near saturation was reached slightly later (ie, after an additional 3 to 4 interviews) when examining only parent codes ( Figure 2 ).
In total, 3 of the studies had codebooks that consisted almost entirely of deductive (ie, concept-driven) codes, whereas the codebooks in the remaining 2 studies contained a mix of both deductive and inductive (ie, data-driven) codes. Although the results were largely consistent across the 5 studies, as expected, the studies that relied heavily on deductive coding reached both true saturation and near saturation sooner. This finding suggests that studies using more inductive coding and analytic techniques may require slightly larger sample sizes to reach saturation.
Although all the studies used a semistructured interview guide, the level of structure varied across studies. The 3 studies (ie, studies C, D, and E) that had a more structured interview guide (eg, questions for which participants were asked their preference among discrete choices or the range of likely answers was limited) reached both true saturation and near saturation sooner. In fact, the study with the most structured guide reached near saturation the soonest, although it fell in the middle for true saturation. This finding suggests that studies using a less structured interview guide may need to conduct more interviews to reach an acceptable level of saturation.
Although true saturation was not reached until the final interview or close to the final interview, near saturation was reached much sooner, ranging from just below to just above the midpoint of data collection, with most of the studies falling just below the midpoint. Although additional interviews conducted after near saturation may result in new information, our findings suggest there may be diminishing returns relative to the resources expended. We have identified several study characteristics that researchers can consider when making decisions on sample size for web-based interviews.
Although our findings were mostly consistent across the 5 studies we examined, near saturation was reached sooner on the studies that consisted of largely deductive codes compared with those that had a greater number of inductive codes. Consequently, researchers should consider their analytic approach when determining sample size. Studies that intend for the coding scheme to be iterative throughout the coding process may want to err on the side of having a slightly higher sample size than if the codebook is expected to consist largely of deductive codes tied to the interview guide.
These studies ranged in length from 30 to 90 minutes, and a majority (n=3) lasted 60 minutes. Although the 90-minute study reached both true saturation and near saturation at the latest point, the shortest interview (at 30 minutes) required the second-highest number of interviews to reach both saturation points. Although the length of the interview may be a minor consideration, the level of structure of the interview guide and the types of codes used seem to be larger drivers.
Our findings point to the need for a slightly higher number of interviews to reach an acceptable level of saturation—categorized by us as near code saturation—than what has been found in other studies. For example, Guest et al [ 15 ] found that 6 interviews were enough to get high-level themes, reaching a plateau at 10 to 12 interviews. Similarly, Young and Casey [ 27 ] found that near code saturation was reached at 6 to 9 interviews.
Our findings also build on previous studies looking at saturation for in-person data collection conducted at a small number of sites. Data from our studies included participants from all US Census Bureau regions, which provides support that these findings may be more generalizable than previous studies.
Our study had several limitations. First, our analysis was conducted on a sample of 5 studies that had similarities. All the studies were related to the medical field, and our study populations (patients with an identified medical condition and health care providers) were knowledgeable about the topics discussed. Second, all the studies were conducted using semistructured interview guides that leaned toward being more structured (ie, interviewers largely stuck to scripted probes as compared with guides that allow for unscripted follow-up probes and unstructured conversations). Additionally, all the studies used a similar approach to coding by using a mix of both deductive and inductive codes (though to varying extents). Consequently, studies with a less structured approach to both the interview and coding process may yield different results. Finally, all our studies are broadly classified as social science research. The findings for other fields of inquiry, such as economic or medical studies, may differ.
Saturation is an important consideration in planning and conducting qualitative research, yet, there is no definitive guidance on how to define and measure saturation, particularly for web-based data collection, which allows for data to be collected from a more geographically diverse sample. Our study provides support that near saturation may be a sufficient measure to target and that conducting additional interviews after that point may result in diminishing returns. Factors to consider in determining how many interviews to conduct include the structure and type of questions included in the interview guide, the coding structure, and the population being studied. Studies with less structured interview guides, studies that rely heavily on inductive coding and analytic techniques, and studies that include populations that may be less knowledgeable about the topics discussed may require a larger sample size to reach an acceptable level of saturation. Rather than trying to reach a consensus on the number of interviews needed to achieve saturation in qualitative research overall, we recommend that future research should explore saturation within different types of studies, such as different fields of inquiry, subject matter, and populations being studied. Creating a robust body of knowledge in this area will allow researchers to identify the guidance that best meets the needs of their work.
Research Triangle Institute–affiliated authors received support for the development of this manuscript from the RTI Fellow’s program under RTI Fellow, Leila Kahwati, MPH, MD. All studies included in the analyses were funded by the Food and Drug Administration. The authors would like to thank the following Food and Drug Administration staff for their contribution to this research: Kit Aikin, Kevin Betts, Amie O’Donoghue, and Helen Sullivan.
The data sets analyzed during this study are available from the corresponding author on reasonable request.
None declared.
Achieving saturation in interviews: saturation type, methods for achieving saturation, and findings by other authors.
Edited by A Mavragani; submitted 22.09.23; peer-reviewed by K Kelly, G Guest; comments to author 24.10.23; revised version received 30.01.24; accepted 09.05.24; published 09.07.24.
©Claudia M Squire, Kristen C Giombi, Douglas J Rupert, Jacqueline Amoozegar, Peyton Williams. Originally published in the Journal of Medical Internet Research (https://www.jmir.org), 09.07.2024.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research (ISSN 1438-8871), is properly cited. The complete bibliographic information, a link to the original publication on https://www.jmir.org/, as well as this copyright and license information must be included.
An official website of the United States government.
Here’s how you know
The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.
The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.
Publication info, research methodology, country, state or territory, description, other products.
For two decades, the U.S. Department of Labor (DOL) has invested substantial funding toward programs serving justice-involved individuals. Among its recent investments, DOL awarded over $243 million in Reentry Projects (RP) grant programs between 2017 and 2019 to improve participants’ employment and justice outcomes. DOL prioritized awarding grants to programs that were evidence-informed, and many went to experienced providers. They were awarded across a broad range of intermediaries and non-profit community-based organizations serving a total of 17,361 participants across 34 states, Washington DC, and Puerto Rico. RP grants were 36-39 months long and were at different phases when the COVID-19 pandemic began in March 2020. RP grantees served a total of 9,098 adults (individuals over 24) and 8,263 young adults (individuals between ages 18 and 24) after their release from jail or prison.
In 2017, the Chief Evaluation Office, in collaboration with the Employment and Training Administration funded the Reentry Project Grants Evaluation. This implementation and impact evaluation aims to identify and evaluate promising practices used in reentry employment programs, which are comprehensive strategies to address the range of challenges formerly incarcerated adults and young adults who have been involved in the justice system face in making a successful transition back to the community.
This issue brief describes the differences and similarities between adult and young adult grantees in terms of the services they offered, and the implementation challenges they reported. The analysis draws on quantitative data from a survey of all 116 organizations that received RP grants. Data from the grantee survey were analyzed using descriptive statistics as well as chi-squared tests to determine whether differences across grant types were statistically significant. The brief also draws on in-depth qualitative data from a subset of nine grantees that received both adult and young adult grants.
Some key findings from RP grantee survey include:
IMAGES
VIDEO
COMMENTS
The main difference between quantitative and qualitative research is the type of data they collect and analyze. Quantitative research collects numerical data and analyzes it using statistical methods. The aim is to produce objective, empirical data that can be measured and expressed in numerical terms.
When collecting and analyzing data, quantitative research deals with numbers and statistics, while qualitative research deals with words and meanings.
Learn how to craft effective research questions and hypotheses for scholarly articles, with examples and tips from quantitative and qualitative approaches.
Qualitative and quantitative research differs in terms of the methods they employ to conduct, collect, and analyze data. For example, qualitative research usually relies on interviews, observations, and textual analysis to explore subjective experiences and diverse perspectives. While quantitative data collection methods include surveys ...
Another major difference between quantitative and qualitative data lies in how they are analyzed. Quantitative data is suitable for statistical analysis and mathematical calculations, while qualitative data is usually analyzed by grouping it into meaningful categories or themes.
Get the plain-language low down on qualitative vs quantitative research, data and analysis. Plain-language explanation with clear examples.
The difference between qualitative and quantitative data and analysis - all you need to know. Qualitative vs quantitative data analysis: definition, examples, characteristics, contrast, similarities, and differences. Comparison chart in PDF.
Abstract Given the vast and diverse qualitative analytic landscape, what might be a generative starting point for researchers who desire to learn how to produce quality qualitative analyses? This question is particularly relevant to researchers new to the field and practice of qualitative research and instructors and mentors who regularly introduce students to qualitative research practices ...
When collecting and analysing data, quantitative research deals with numbers and statistics, while qualitative research deals with words and meanings.
Qualitative vs quantitative research Learn about the differences, see examples and find out when to use which methods!
Qualitative data is commonly used in survey research, interviews, and observational studies, as it dives deeply into participant motives, attitudes, and actions. Open-ended questions in surveys capture complex replies, whereas interviews allow for direct involvement for a deeper understanding.
Qualitative data analysis is a systematic process of examining non-numerical data to extract meaning, patterns, and insights. In contrast to quantitative analysis, which focuses on numbers and statistical metrics, the qualitative study focuses on the qualitative aspects of data, such as text, images, audio, and videos.
Qualitative research gains a better understanding of the reason something happens. For example, researchers may comb through feedback and statements to ascertain the reasoning behind certain behaviors or actions. On the other hand, quantitative research focuses on the numerical analysis of data, which may show cause-and-effect relationships.
The focus for Q3 of 2023 was on analyzing and interpreting qualitative and quantitative data. Find all the posts, interviews, and resources here!
Qualitative research is the opposite of quantitative research, which involves collecting and analyzing numerical data for statistical analysis. Qualitative research is commonly used in the humanities and social sciences, in subjects such as anthropology, sociology, education, health sciences, history, etc.
The difference between qualitative and quantitative data, how to conduct quantitative and qualitative research, perform data analysis.
Qualitative and quantitative research methods differ on what they emphasize—qualitative focuses on meaning and understanding, and quantitative emphasizes statistical analysis and hard data. Learn how they're applied.
This blog post will further explore different qualitative and quantitative analysis methods, their strengths and limitations, and how to apply them in various research and business contexts. Whether you're a researcher, analyst, or decision maker, understanding these methods will help you make informed decisions when analyzing data and deriving valuable insights.
Qualitative data can lay the foundation for quantitative analysis by creating defined categories in which to explore data later on. Deciding what category boundaries to set is a judgment call, but the category's relevance is only as good as the quality of the qualitative research that underpins it.
Quantitative research in psychology and social sciences answers "how much" questions. Qualitative research answers the "how" and "why" of a phenomenon. Learn more.
Discerning if quantitative vs qualitative research design is best for your needs can be challenging. Read on for help discerning between the two.
It explores the "how" and "why" of human behavior, using methods like interviews, observations, and content analysis. In contrast, quantitative research is numeric and objective, aiming to quantify variables and analyze statistical relationships. It addresses the "when" and "where," utilizing tools like surveys, experiments, and ...
Quantitative research, which includes gathering and analyzing numerical data for statistical analysis, is the antithesis of qualitative research. The humanities and social sciences frequently employ qualitative research in sociology, anthropology, education, history, health sciences, etc.
All forms of data analysis methods are broadly classified into two parts: qualitative and quantitative. Read on to learn more about it.
Data analysis typically revolves around two types of data: qualitative and quantitative data: Qualitative data Examples of qualitative data include: interview transcripts, open-ended answers to surveys, field notes, recordings, questionnaires (but only if you have a small sample size) and so on.
Often, qualitative data are linked to open-ended questions while closed-ended questions are paired with quantitative data. The reality is that open-ended qualitative questions can be converted into quantitative data and conversely closed-ended quantitative questions can be used to glean qualitative data.
Background: In-depth interviews are a common method of qualitative data collection, providing rich data on individuals' perceptions and behaviors that would be challenging to collect with quantitative methods. Researchers typically need to decide on sample size a priori.
Analysis of grantee survey and qualitative data suggest that adult and young adult services may differ in four key areas: (1) positive youth development components (2) legal services for child support and diversion, (3) educational services and requirements, and (4) program length.