Susan R. Hutchinson, Ph.D. Department of Applied Statistics & Research Methods CETL Workshop, July 10, 2012 First Steps in Designing Quantitative Research: Asking the Right Questions 1 • Overview of purposes for conducting quantitative research • Role(s) served by research questions • Attributes of useful RQs • Group activity on evaluating RQs • Exercise on constructing RQs CETL Workshop, July 10, 2012 Workshop Outline 2 CETL Workshop, July 10, 2012 Pretest 3 False CETL Workshop, July 10, 2012 All quantitative studies include one or more research questions. 4 False CETL Workshop, July 10, 2012 Research questions and research hypotheses are the same thing. 5 True CETL Workshop, July 10, 2012 Research questions should “foreshadow” the data analyses. 6 False CETL Workshop, July 10, 2012 Research questions should be posed after data are collected. 7 False CETL Workshop, July 10, 2012 Research questions should be written using non-specific language to allow flexibility in designing the study. 8 True CETL Workshop, July 10, 2012 All of the following are types of variables that might be included in research questions: dependent, moderator, mediator, extraneous, independent 9 True CETL Workshop, July 10, 2012 Writing quantitative research questions requires knowledge of statistics. 10 CETL Workshop, July 10, 2012 What are the major purposes for conducting quantitative research? 11 Quantitative research is intended to examine: • Differences between groups • Change over time • Direction, magnitude, and between-group differences • Relationships between 2 or more variables • Direction, magnitude, type/form, between-group differences CETL Workshop, July 10, 2012 • Statistical significance, magnitude, and direction of differences • Descriptions of populations and phenomena 12 CETL Workshop, July 10, 2012 As you’re constructing your research questions, keep in mind the overriding purpose of quantitative research is to understand differences 13 CETL Workshop, July 10, 2012 What role do research questions play in designing quantitative research? 14 • Indicate the target population • Identify the variables/constructs and what we want to know about them • Foreshadow the type of data analysis (and sometimes the research design) CETL Workshop, July 10, 2012 What information should research questions include? 15 What characterizes a “good” research question? • Clearly conveys the variables/constructs and the nature of their relationships • Must be answerable based on data from the study • RQs should not be rhetorical or require a value judgment to answer • RQs should match the research design • Should be worth investigating • A good RQ does not address a trivial issue; “so what?” • Answers to RQs should further the researcher’s understanding of the phenomenon CETL Workshop, July 10, 2012 • Includes independent, dependent, moderator, and mediator variables (where applicable) 16 CETL Workshop, July 10, 2012 “It is better to have an approximate answer to the right question than an exact answer to the wrong one.” John Tukey 17 • Dependent • Independent • Extraneous • Moderator • Mediator CETL Workshop, July 10, 2012 Types of variables that might appear in research questions 18 Dependent variable(s) • The dependent variable is a phenomenon one is attempting to explain or predict • In a survey study, often this is the variable or construct of primary interest to the researcher • It is something which the researcher wishes to understand more fully • For many researchers conducting nonexperimental research this is the starting point in planning a study SRM 627 Survey Research Methods, Hutchinson • Also referred to as the outcome or criterion variable 19 • In an experimental or quasi-experimental design, an independent variable would be the manipulated or treatment variable in your study • In a survey or other type of nonexperimental study, where no variables are manipulated, independent variables are the factors used to explain or predict the DV SRM 627 Survey Research Methods, Hutchinson Independent variable(s) 20 • Extraneous variables are also called nuisance variables • These are variables which are not usually of primary interest but are believed to be related to the independent and/or dependent variables • Their effects need to be controlled in order to obtain meaningful results • Extraneous variables are of two broad types: • dispositional and situational SRM 627 Survey Research Methods, Hutchinson Extraneous variables 21 salary job satisfaction A researcher is interested in examining the relationship between salary and job satisfaction. However, the researcher suspects that factors other than salary (such as job autonomy) are also responsible for an employee’s level of job satisfaction. SRM 627 Survey Research Methods, Hutchinson Illustration of controlling for extraneous variables 22 job autonomy Relationship with “job autonomy" taken into account or controlled SRM 627 Survey Research Methods, Hutchinson salary job satisfaction 23 Moderator variable(s) • A moderator variable influences the strength of the relationship between the primary independent variable and the dependent variable • A moderator variable would be the variable that interacts with the independent variable SRM 627 Survey Research Methods, Hutchinson • Usually a moderator variable is of secondary interest 24 SRM 627 Survey Research Methods, Hutchinson Illustration of moderator effect Computer skills/comfort 25 • Also referred to as an intervening variable • A mediator (or mediating) variable is some construct through which the independent variable indirectly affects the dependent variable SRM 627 Survey Research Methods, Hutchinson Mediator variable 26 Parents’ educational level Achievement motivation Performance on standardized tests SRM 627 Survey Research Methods, Hutchinson Illustration of mediator variable 27 • What are psychologists’ perceptions of professional development? • Are university faculty depressed? • What is the relationship between self-esteem and happiness? • How many undergraduate students dropped out of college in 2011? • What factors predict job satisfaction? • How are exercise and blood pressure related? • Is there a difference between gender and depression? • What is the impact of test anxiety on test performance? • Is religion good for society? CETL Workshop, July 10, 2012 Examples of less than ideal RQs 28 • What is the researcher hoping to learn? • What are the variables/constructs in the RQ? • What is the target population? • How might this RQ be improved? • Which aspects of professional development do psychologists rate most versus least positively? CETL Workshop, July 10, 2012 What are psychologists’ perceptions of professional development? 29 CETL Workshop, July 10, 2012 Or, instead of posing a descriptive question about a particular construct, consider rewording so that you can explain differences on the construct. 30 CETL Workshop, July 10, 2012 • To what extent are private practice psychologists’ perceptions of the need for professional development explained by their years of experience, area of specialization, and professional degree? 31 • What are the independent and dependent variables? • What assumption about university faculty is implied by the research question? • What are possible revisions? • Does level of depression among university faculty differ based on their gender, departmental affiliation, and tenure status? CETL Workshop, July 10, 2012 Are university faculty depressed? 32 • What are the independent and dependent variables? • What assumption about university faculty is implied by the research question? • What are possible revisions? • Does level of depression among university faculty differ based on their gender, departmental affiliation, and tenure status, after controlling for marital status and life satisfaction? CETL Workshop, July 10, 2012 Are university faculty depressed? 33 CETL Workshop, July 10, 2012 Hypothetical depression scores 34 CETL Workshop, July 10, 2012 For complex constructs that are likely to be multiply determined, you should attempt to control for potentially extraneous variables. 35 • What do you think the researcher wants to know? • How might the question be revised to more clearly convey the type of information sought? • What is the strength of relationship between self-esteem and happiness? • Is self-esteem linearly related to happiness among college-educated adults? CETL Workshop, July 10, 2012 What is the relationship between selfesteem and happiness? 36 How many undergraduate students dropped out of college in 2011? • Which of the following variables predicted undergraduate college student dropout in 2011? (grade point average, perceived social support, ….) CETL Workshop, July 10, 2012 • What is wrong with this question? • How might it be improved in order to provide more useful information? 37 • What are the independent and dependent variables? • What is the target population? • How might you reword this RQ so that it provides a clearer focus the researcher? • To what extent do salary, autonomy, …. CETL Workshop, July 10, 2012 What factors predict job satisfaction? predict job satisfaction among full-time employees? 38 • What is the independent and dependent variable? • Target population? • How would you reword this RQ if you had reason to believe gender could be a moderator variable? • Does the strength of relation between amount of daily aerobic exercise and blood pressure differ for male and female adults? CETL Workshop, July 10, 2012 How are exercise and blood pressure related? 39 • What flaw do you notice in this RQ? • What do you think the researcher actually wants to know? • What is a possible revision? CETL Workshop, July 10, 2012 Is there a difference between gender and depression? 40 • Assuming the researcher plans to collect data via survey, what part of this RQ needs to be reworded and why? • How might you revise this item if you wanted to determine if study strategies was a mediator variable? • Do study strategies mediate the relationship between test anxiety and test performance among doctoral level social science majors? CETL Workshop, July 10, 2012 What is the impact of test anxiety on test performance? 41 Is religion good for society? • Are rates of violent crimes lower in communities with higher attendance at formal religious services? CETL Workshop, July 10, 2012 • Why is this not a suitable research question? • How could you revise the question so that it is suitable? 42 Other considerations when writing RQs • Be careful about your choice of words • Think about how you plan to answer your research questions as you write them • Consider the research design, measurement, and data analysis CETL Workshop, July 10, 2012 • Verbs such as improve, impact, affect, and predict imply certain types of research designs 43 Research question templates • Do male and female undergraduate students differ in their level of statistical test anxiety? CETL Workshop, July 10, 2012 • Do [groups of interest] differ on [name of construct]? 44 Templates – cont’d • Does level of participation in Greeley Parks & Recreation Center classes differ by age, educational level, gender, or ethnic group? CETL Workshop, July 10, 2012 • Does [name of construct] differ by [variable]? 45 Templates – cont’d • To what extent do salary, working conditions, and types of job tasks predict job satisfaction? CETL Workshop, July 10, 2012 • To what extent does a set of variables explain (or predict) a given outcome variable? 46 Templates – cont’d • Is there a relationship between teachers’ job satisfaction and perceived level of administrative support? CETL Workshop, July 10, 2012 • Is there a relationship between [variable] and [variable]? 47 Templates – cont’d • Is the relationship between teachers’ job satisfaction and perceived level of administrative support the same for novice versus experienced teachers? CETL Workshop, July 10, 2012 • Is the relationship between [variable] and [variable] the same for [names of groups]? 48 Templates – cont’d • Does child’s achievement motivation explain performance on statewide standardized tests, after controlling for parents’ SES and academic self-concept? CETL Workshop, July 10, 2012 • Does a given explanatory variable (or set of variables) explain a criterion, after taking into account one or more extraneous variables? 49 Templates – cont’d • Does reading level improve during a twomonth summer intensive reading program for elementary school English language learners? CETL Workshop, July 10, 2012 • Does [dependent variable] change [or improve or decline] across time among [population of interest]? 50 Templates – cont’d • Is the rate of improvement in reading level the same for males and females? CETL Workshop, July 10, 2012 • Is the pattern of change [or improvement or decline] on [name of dependent variable] the same for [names of groups]? 51 • Consider a quantitative study you might conduct • Think about the overall purpose and how the research questions will help support the purpose • You are to compose at least two research questions, each of which should include a dependent variable and one or more independent variables. • One of your RQs will include one or more extraneous variables and the other will include a moderator variable. CETL Workshop, July 10, 2012 Question Writing Exercise 52