DESIGN & PREDICTIVE DESIGNS MARGARET LAMAR, PH.D. SURVEY RESEARCH What is survey research?
Why use survey research?
TYPES OF SURVEY DESIGNS Longitudinal • Trend Studies • Cohort Studies • Panel Studies Cross-sectional DATA COLLECTION Questionnaires • Mailed • Electronic Interviews • One-on-one • Focus Group • Telephone CONDUCTING SURVEY RESEARCH Step 1: Decide what you want to learn Step 2: Determine your population & sampling approach Step 3: Determine methodology and data collection approach Step 4: Develop a questionnaire or find existing one Step 5: Pilot test questionnaire Step 6: Administer questionnaire Step 7: Analyze data Step 8: Write report INSTRUMENT DEVELOPMENT: PART 1 Identify the primary purpose Identify behaviors/define the domain Prepare table of specifications Construct initial pool of items Have items reviewed (and revise when appropriate)
(Crocker & Algina, 2008)
LIKERT SCALE ITEMS Put statements/questions in present tense Do not use factual statements Avoid statements with multiple interpretations Avoid items likely to be endorsed by everyone Use equal number of statements with positive and negative feelings Avoid universal language (e.g., all, always, none, never) Use simple sentences, not compound (and keep ‘em short!)
(Crocker & Algina, 2008)
OTHER SURVEY CONSIDERATIONS Response rate • Maximize – Why? How?
Cover letters
Limitations of survey research
SURVEY RESEARCH QUESTIONS Language: explore, understand, etc EVALUATING SURVEY RESEARCH Limitations? Predictive Research Designs What’s the difference between experimental and predictive designs
What’s the difference between DIRECTION of the relationship
and DEGREE of the relationship? RELATIONSHIPS WHAT IS THE DIRECTION AND STRENGTH OF THE FOLLOWING CORRELATIONS? .10 .99 -.76 .50 -.09 -.43 SPURIOUS CORRELATIONS https://www.youtube.com/watch?v=g-g0ovHjQxs PREDICTIVE MODELS Bivariate (2 variables) Multivariate (more than 2 variables) Multiple regression (Find the most important group of variables to make a prediction) Factor Analysis (how do independent variables relate to the dependent variable and with each other?) RESEARCH QUESTIONS & HYPOTHESES • Should include relationship language (not causal) • Degree to which 2 or more variables are related to each other • Avoid yes or no questions, if possible • Ensure that each question is just asking 1 question • Include multiple questions to capture multiple ideas • Exploratory – may not need a hypotheses • Confirmatory – need a hypotheses • 1 hypothesis per question • Be as specific as possible RQ1: As the number of hours a student spending studying increases, does the earned grade point average increase? • H1: There is no relationship • Ha: As the time spent studying increases, GPA increases
RQ2: Are self care practices associated with posttraumatic
growth for counselors responding to hurricane victims during deployment? • H1: There is no relationship • Ha: There will be a positive relationship between practice of self care and posttraumatic growth for counselors deployed to help hurricane victims EVALUATING PREDICTIVE RESEARCH Limitations?