This preview shows page 1. Sign up to view the full content.
Unformatted text preview: Ch. 9 Using Survey Research WHY CONDUCT SURVEYS?- Provides a methodology for people to tell about themselves Study relationships among variables and ways that attitudes and behaviors change over time Provides useful information for making public policy decisions Important complement to experimental research findings You can ask questions on a survey about behaviors that you would not want to observe What types of behaviors would you not want to/not be able to observe? CONSTRUCTING QUESTIONS TO ASK Defining the Research Objectives: What do you want to know? Attitudes and beliefs Facts and demographics Behaviors CONSTRUCTING QUESTIONS Question Wording Potential problems that stem from difficulty in understanding the question: Vague or imprecise terms e.g., "How conservative are you?" (What do we mean by conservative: how we dress, political ideals, religious beliefs, etc.) Ungrammatical sentence structure Phrasing that overloads working memory/cognitive ability e.g., A question that is so long the person can't remember the beginning by the time they get to the end. Embedding the question with misleading information e.g., "Don't you agree that the administration acted haphazardly when it hurriedly made the decision to stop the program?" Important considerations when writing questions: Simplicity make the questions as simple as possible Avoid double-barreled questions questions that ask 2 things at once "How much time each week do you spend outdoors and using the computer?" Avoid loaded questions "Do you favor eliminating the wasteful excesses in the public school budget?" A better alternative: "Do you favor reducing the public school budget?" Emotionally charged words: waste, immoral, rape, etc. Avoid Negative wording "Do you feel that the city should not approve the proposed women's shelter?" Avoid giving the participant the opportunity for Yea-saying or nay-saying (saying "yes" to every question) Response sets pattern of individual response that is not related to the content of the questions "I feel isolated from others" & "I feel part of a group of friends." If someone said "yes" to both of these, they might be yea-saying. yea-saying. Different Types of Questions Open-ended questions How would you describe a perfect weekend? Partially open-ended questions What is your favorite thing to do on a Sunday afternoon? ___ nap ___ watch television/movies ___ spend time outdoors ___ computer/video games ___ Other (please specify) __ RESPONSES TO QUESTIONS Rating Scales: Likert scales give numerical-based response options Rating Scales: Graphic rating scale How would you rate the movie you just saw? Not very enjoyable--------------------------------------------- Very enjoyable The participant makes a mark on the line. To get the "answer", the researcher measures (with an actual ruler) how far the mark is away from one of the anchors. This allows for more variability in responses than a Likert scale Rating scales: Semantic differential scale Rate items on a series of bipolar adjectives 1. Jacksonville Jaguars: Good __ __ __ __ __ __ __ Bad Strong __ __ __ __ __ __ __ Weak Active __ __ __ __ __ __ __ Passive Rating Scales: Nonverbal scales for children Circle the face that matches how you feel? h h h When you are creating questions, you have to think about the response alternatives that you are giving the participant. For example, who are you asking/what are they rating? In comparison with other graduates, how would you rate this student's potential for success? ____ Lower 50% ____ Upper 50% ____ Upper 25% ____ Upper 10% ____ Upper 5% RESPONSES TO QUESTIONS: Labeling Response Alternatives How often do you exercise for at least 20 minutes? __ less than twice a week __ about twice a week __ about four times a week __ about six times a week __ at least once each day ASSEMBLING THE QUESTIONNAIRE Formatting the Questionnaire Demographics should not be presented first (participants find these questions boring so they quickly lose interest) Present related items together Should appear attractive and professional Neatly typed and free from errors Use point scales consistently ADMINISTERING SURVEYS Questionnaires In-person administration to groups or individuals Mail surveys Internet surveys Nonresponse bias ________________ ___________________________________________________________________ ___________________________________________________________________ Solutions : Multiple contacts Prenotice alerting them that you will be sending the questionnaire, sending the questionnaire, a thank you postcard with a reminder to complete the questionnaire if they haven't already Include small token of appreciation e.g., $1 bill Interviews Face-to-face interviews Telephone interviews Focus group interviews The problem with interviews: interviewer bias the interviewer might inadvertently have an effect on how the participants answer SURVEY DESIGNS TO STUDY CHANGES OVER TIME Questions Are the Same Each Time Surveyed Tracks Changes Over Time Examples: President approval rating Number of high school students who drink/ use drugs/ etc. We have talked about constructing and administering surveys. Now we're going to talk about how to get participants for the surveys. SAMPLING TECHNIQUES Probability Sampling Each member of the population has a specifiable probability of being chosen Nonprobability sampling Unknown probability of any member being chosen Probability sampling Simple random sampling each person in the population has an equal probability of being selected Probability sampling Stratified random sampling population is divided into subgroups (strata) and random samples are taken from each strata Survey on sexual attitudes of people in Jacksonville Stratify: age, gender, education This will ensure we have representatives from different groups of people in Jacksonville (older AND younger, men AND women, those with higher education AND those without) Sample will accurately reflect the numerical composition of the subgroups Cluster sampling identify clusters and sample from the clusters The basic sampling unit is a group of participants rather than an individual Randomly select classrooms from an elementary school, then survey every student within the selected classrooms. Multistage sampling Identify large clusters Randomly select individuals from the clusters (instead of sampling every individual as in cluster sampling Nonprobability Sampling Cheap & convenient Haphazard sampling convenience sampling Stand outside of Publix and collect data from the first 100 people who agree Often biased Purposive sampling sample meets predetermined criterion If you want half male and half female participants, stand outside Publix and collect data from the first 50 men and 50 women who agree to participate Quota sampling sample reflects the numerical composition of various subgroups in the population Haphazardly collected If you are at a University where the student body is composed of 60% women and 40% men, then you would stand on campus and collect data from the first 60 women and the first 40 men who agreed to participate. SAMPLING FROM A POPULATION Sampling error ___________________ ___________________________________________________________________ ___________________________________________________________________ Sample Size A larger sample size reduces the amount of sampling error EVALUATING SAMPLES-Should be representative of the population SAMPLESSampling Frame Actual population from which the sample is drawn Rarely coincides with population of interest The actual population might be "college students" but the sampling frame might be "college students at UNF" Response Rate % of respondents who complete the survey out of the number who were asked Why would we use convenience samples instead of using probability sampling? Reasons for Using Convenience Samples Less costly than randomly sampling from every single person in population of interest Can be representative Most research is interested in relationship between variables, not actual amount in population ...
View Full Document
- Fall '10