How to Design a Mixed Methods Study by John W. Creswell, Ph.D. and Vicki L. Plano Clark, M.S. University of Nebraska-Lincoln Andrews University, July, 2004 2 How would you combine two types of data? Qualitative Text Data This is a sample of a text file of words that might be collected on transcripts through interviews, fieldnotes from observations, or from optically-scanned documents. Quantitative Numeric Data 2342543112232132 23322543 3122432432132433 32334441 2222111432143213 22111555 2331432432132433 32135432 3 Objectives of the workshop: Let’s design a mixed methods study Let’s study how people learn mixed methods research in this room? (or you can work on your own project and follow along at each step) Let’s start with a title. Write a title. What data will we collect? 4 What are types of quantitative and qualitative data? Quantitative data Close-ended scales Attitudinal/behavioral scales Behavioral checklists Census, attendance records Qualitative data Open-ended responses Semi-structured interviews Semi-structured observations Records/documents Videotapes 5 Let’s identify our quantitative and qualitative data collection 6 Now let’s consider some reasons for why we are collecting (and mixing) both forms of data Together quantitative and qualitative data provide both precise measurement and generalizability of quantitative research and the in-depth, complex picture of qualitative research To validate quantitative results with qualitative data We do not have an adequate instrument. Thus, we need to explore views and develop an instrument Our quantitative data provide a general explanation and we need to follow-up with participants and have them explain the quantitative results In our experiment, outcomes to be measured are not enough; they need to be complemented by understanding the process of participants 7 Let’s identify our reason for mixing 8 So… There are good reasons for gathering both forms of data But…there are certain requirements for this to work best 9 Requirement #1: Now let’s consider whether we have the skills, time, and resources? We need minimum skills in both qualitative and quantitative data collection. What do we need? We need time and resources for extensive data collection and analysis. How much time and resources do we need? 10 Write down the skills, time, and resources we will need 11 Requirement #2: The audience(s) Does our audience appreciate both numbers and stories? Are they familiar with this design? Do they need to be educated? Are examples of published studies available in our content area? 12 Let’s identify the audiences 13 But audiences may not recognize it yet because it is so new Increased use and acceptance of qualitative research from 1990’s to present The complexity of our research problems today requires understanding trends, differences, as well as individual stories, setting Individuals advocating for and writing about mixed methods research as a distinct, new procedure (e.g., books) 14 They may think that it is analyzing data separately Quantitative Data Qualitative Data Mixing: converging the data or connecting the data 15 But how do we mix? Converge data: Qual Results Quan Connect data: Qual Quan Results 16 Why our audience may recognize it The evidence Books Methodological articles Many published research studies using it Federal agencies Private foundations 17 18 Other writings, initiatives on mixed methods research: Research studies reported in journals Methodological articles exploring issues and procedures Website for bringing mixed methods writers together Conference sessions Handbook of Mixed Methods in Social and Behavioral Research Private foundation interest; federal agency interest 19 NIH Guidelines - Mentioned several approaches for combining qualitative and quantitative research - Considerations for deciding what model to use (literature available, prior studies, realistic design, expertise) - Need to describe each method thoroughly 20 Quotes: “Combining qualitative and quantitative methods has gained broad appeal in public health research. The key question has become not whether it is acceptable or legitimate to combine methods, but rather how they will be combined to be mutually supportive and how findings achieved through different methods will be integrated.” (NIH, Office of Behavioral and Social Science Research, 1999). 21 National Academy of Sciences Three major research questions in quality educational research: • What is happening? (qualitative designs) • Is there a systematic effect? (a quantitative experiment) • Why or how it is happening? (a qualitative followup) 22 But even if they recognize it, they may not appreciate or understand how to design a mixed methods study “We are interested in a randomized control trial with a non-experimental approach embedded within it.” (a private foundation officer) “We accept multi-method studies, but investigators mostly do not sort out the complexity of these projects so that we can understand them.” (a federal projects officer). 23 We need to define mixed methods research for our audiences Mixed methods research is a design for collecting, analyzing, and mixing both quantitative and qualitative data in a single study or series of studies to understand a research problem. The purpose of this form of research is that both qualitative and quantitative methods, in combination, provide a better understanding of a research problem or issue than either method alone. 24 Now we could mix within single studies or multiple studies Single Study: Quan Qual Results Multiple Studies (called multimethod research): Quan Qual Study 1 Study 2 Qual Quan Study 3 Study 4 25 So how do we design a mixed methods study? The model Worldviews, theoretical frameworks, problem and research question, skills, resources Type of mixed methods design Procedures for: •designing the title •writing the introduction to a study •writing the purpose statement and research questions/hypotheses •data collection •data analysis •writing the mixed methods report •evaluating the mixed methods research 26 What is a worldview? Philosophy about your preferences for how you learn about something through research You prefer the quantitative worldview: you are the expert, you decide what needs to be learned, you build in objectivity You prefer the qualitative worldview: participant is the expert, participant helps you build knowledge, you bring personal bias in You prefer both the quantitative and qualitative worldview 27 The next steps in planning our study Let’s write the overall research question for our study 28 Then let’s choose a type of mixed methods study to conduct What designs are possible? 29 Types of mixed methods designs I. Triangulation Mixed Methods Design QUAN Data and Results + QUAL Data and Results Interpretation II. Nested Mixed Methods Design QUAN Pre-test Data and Results Qual Process QUAN Post-test Data and Results 30 Types of mixed methods designs III. Explanatory Mixed Methods Design QUAN Data and Results Follow-up qual Data and Results IV. Exploratory Mixed Methods Design QUAL Data and Results Building quan Data and Results 31 Triangulation Design: Characteristics Collecting both quantitative and qualitative data Collecting these data at the same time in the research procedure Analyzing the quantitative and qualitative data separately Comparing or combining the results of the quantitative and qualitative analysis Example: collect survey data (quantitative) and collect individual interviews (qualitative) and then compare the results 32 Triangulation Design: When is it used? When you want to combine the advantages of quantitative (trends, large numbers, generalization) with qualitative (detail, small numbers, in-depth) When you want to validate your quantitative findings with qualitative data When you want to expand your quantitative findings with some open-ended qualitative data (e.g., survey with closed- and openended data) 33 Nested Design: Characteristics Collecting both quantitative and qualitative data Collecting both types of data at the same time Having ONE form of data play a smaller role in the study than the other form of data Also, Using one form of data to answer one question; the other form another question Collecting one form of data at one level of analysis and another at another level of analysis Example: You conduct an experiment and during the experiment you gather qualitative interview data. The outcomes of the experiment assessed quantitatively address different questions than the process of the experiment explored qualitatively. 35 Nested Design: When is it used? When you do not have time or resources to commit to extensive quantitative and qualitative data collection When you want to study the process of an experiment as well as the outcomes When you want to examine different levels in an organization 36 Nested Research Design Experiment Quan Data collection Pre-test Intervention Quan Data collection Post-test Process – collection and analysis of qualitative data 37 Explanatory Sequential Design: Characteristics Viewing the study as a two-phase project Collecting quantitative data first followed by collecting qualitative data second Typically, a greater emphasis is placed on the quantitative data in the study Example: You first conduct a survey and then follow up with a few individuals who answered positively to the questions through interviews 38 Explanatory Sequential Design: When do you use it? When you want to explain the quantitative results in more depth with qualitative data (e.g., statistical differences among groups, individuals who scored at extreme levels) When you want to identify appropriate participants to study in more depth qualitatively 39 Here is an example of an explanatory design: Quantitative Data Collection (quan) Quantitative Data Analysis (quan) Case Selection Qualitative Data Analysis (QUAL) Quantitative Analysis Case Selection Qualitative Analysis Interpretation – based on quan and QUAL results + Qualitative Data Collection Quantitative Data* Number of cigarettes CES-D6 Qualitative Data* Semi-structured interviews, audio recorded and transcribed Graphic plot of CES D6 scores over time for each participant Graphic plot of cigarettes/day values over time for each participant * Data collected 10 times over the course of a calendar year for 40 participants Creswell et al. (in progress) Selected 5 cases maximally varying Identified critical months in which smoking varied Description of each case Identification of life events occurring during critical months where smoking increased or decreased Thematic analysis of life events for each case Cross-case thematic analysis Interpretation Why did changes in smoking occur? Exploratory Sequential Design: Characteristics Viewing the study as a two-phase project Qualitative data collection precedes quantitative data collection Typically, greater emphasis is placed on the qualitative data in the study Example: You collect qualitative diary entries, analyze the data for themes, and then develop an instrument based on the themes to measure attitudes on a quantitative survey administered to a large sample. 41 Exploratory Sequential Design: When do you use it? To develop an instrument when one is not available (first explore, then develop instrument) To develop a classification or typology for testing To identify the most important variables to study quantitatively when these variable are not known 42 Phase I Qualitative Research - Year 1 Qualitative Data Collection Qualitative Data Analysis Qualitative Findings Phase II Quantitative Research - Year 2 Quantitative Instrument Development Unstructured Interviews 50 participants 8 observations at the site 16 documents Text Analysis: Using QSR N6 Development of codes and themes for each site Create approximately a 80-item instrument plus demographics Administer survey to 500 individuals Quantitative Test of the Instrument Quantitative Results Determine factor structure of items and conduct reliability analysis for scales Determine how groups differ using ANOVA test Sequential Exploratory Mixed Methods Design How will we analyze the quantitative and qualitative data (within the design types)? Types of analysis: Quantitative analysis Numeric data Descriptive trend analysis Hypothesis testing, effect size, interval estimates Qualitative analysis Text/image data Coding Themes Description Interrelated themes 44 Triangulation data analysis QUAN data collection • Separate QUAN and QUAL data analysis QUAL data collection QUAN data analysis • Two options • Data transformation (change QUAL to QUAN or QUAN to QUAL) • Comparison (keep separate and compare/contrast) QUAL data analysis Results 45 Table. Example of Data Transformation of Text Units into Numeric Data Count Adj.Count** Row Pct Column Pct Patients N=2 Physicians N=4 Medical Assistants N=4 Familiarity With the Form 13 6.5 41.9 5.8 17 4.25 27.4 3.0 19 4.75 30.7 5.3 49 15.5 100.0 Reactions to the Form 23 11.5 22.2 10.2 100 25.0 48.3 17.4 61 15.3 29.5 16.9 184 51.75 100.0 Use for Managing Depression 67 33.5 38.6 29.8 177 44.25 51.0 30.7 36 9.0 10.4 10.0 280 86.75 100.0 Changes to the Form 115 57.5 37.5 51.1 196 49.0 32.0 34.0 187 46.8 30.5 51.7 498 153.3 100.0 Situational Use of the Form 7 3.5 8.9 3.1 86 21.5 54.4 14.9 58 14.5 36.7 16.1 151 39.5 100.0 225 112.5 100.0 576 144.0 100.0 361 90.3 100.0 Themes Nested data analysis Quantitative Experiment Quan Data collection Pre-test Intervention Quan Data collection Post-test Qualitative Process Data Analysis Pre-test scores Themes/Codes/ Interrelated Themes Post-test scores or gain scores Compare/Describe Results 47 Explanatory sequential data analysis QUAN data analysis Qual data collection (purposeful sampling) • Statistical results • Outlier cases • Extreme cases • • • • Select Select Select Select cases cases cases cases based on s.d. variables to represent outliers to represent extreme cases to make group comparisons Qual analysis • codes • themes • cases 48 Exploratory sequential data analysis QUAL data analysis Quan data analysis instrument development Quotes Items on a survey Codes Variables on a survey Themes Scales on a survey 49 Let’s identify how we will analyze the data 50 Drawing our Design Let’s draw a picture of our design Identify the type of design Add in data collection Add in data analysis Show the flow of activities Add in “products” for our audiences 51 Helpful tips for creating this visual: 1. 2. 3. 4. 5. 6. 7. 8. 9. Give a title to the visual model. Choose either horizontal or vertical layout for the model. Draw boxes for quantitative and qualitative stages of data collection, data analysis and interpretation of the study results. Use capitalized (QUAN) or small letters (quan) to designate priority of quantitative and qualitative data collection and analysis. Use single-headed arrows to show the flow of procedures in the design. Specify procedures for each quantitative and qualitative data collection and analysis stage. Specify expected products or outcomes of each quantitative and qualitative data collection and analysis procedure. Make your model simple. Size your model to one page. 52 Now let’s rework our purpose statement using some scripts 53 Sample Script for a Concurrent Design (Triangulation or Nested) “The purpose of this concurrent mixed methods study is to better understand a research problem by converging both quantitative (numeric) and qualitative (text or image) data. In this approach, ___________ (quantitative instruments) will be used to measure the relationship between the ________ (independent variables) and __________ (dependent variables). At the same time in the study, the __________ (central phenomenon) will be explored using _____________ (qualitative interviews, documents, observations, visual materials) with _________ (participants) at ____________ (the research site).” 54 Sample Script for a Sequential Exploratory Design “The purpose of this two-phase, exploratory mixed methods study will be to explore participant views with the intent of using this information to develop and test an instrument with a sample from a population. The first phase will be a qualitative exploration of a _______(central phenomenon) by collecting ___________(data) from ____________ (participants) at _______ (research site). Themes from this qualitative data will then be developed into an instrument (or survey) so that the __________ (theory and research questions/hypotheses) can be tested that ________ (relate, compare) ____________ (independent variable) with __________ (dependent variable) for _________(sample of a 55 population) at _________ (research site).” Sample Script for a Sequential Explanatory Design “The purpose of this two-phase, explanatory mixed methods study will be to obtain statistical, quantitative results from a sample and then follow-up with a few individuals to probe or explore those results in more depth. In the first phase, quantitative research questions or hypotheses will address the relationship or comparison of __________ (independent) and ________ (dependent) variables with ___________ (participants) at ___________(the research site). In the second phase, qualitative interviews or observations will be used to problem significant _______(quantitative results) by exploring aspects of the ________ (central phenomenon) with _______ (a few participants) at ____________ (research 56 site).” Criteria for evaluating our plan: Use appropriate terminology for title and design Provide a rationale for mixing and include it early in the study (“when you use…”) Create a mixed methods purpose statement Identify types of qual and quan data to be collected and qual and quan data analysis steps Include a visual/procedural diagram of methods with timeline Use rigorous procedures for the quantitative data collection and analysis 57 Let’s share our drawings of our mixed methods procedures 58 How to Design a Mixed Methods Study by John W. Creswell, Ph.D. and Vicki L. Plano Clark, M.S. University of Nebraska-Lincoln Andrews University, July, 2004