Untitled

Alternatives to Experimentation: Survey and Interviews

• Learn about nonexperimental methods used to study behavior like interviews and surveys
• Understand the factors that go into designing solid questionnaires and the creation of good questions
• Learn the pros and cons of different sampling strategies
• Understand how to administer questionnaires and conduct interviews

Survey Research

• It obtains data about preferences, behaviors, opinions, and attitudes through questionnaire and interviews
• Allows researchers to study private experiences that cannot be directly observed

Advantages of Survey Approach

• Efficient collection of large quantities of data
• Anonymous surveys can increase the accuracy of answers to sensitive questions
• Inferences can be drawn about behavior causes; complements field and lab experiments

Limitations of the Survey Approach

• Survey approach does not allow testing hypothesis about causal relationships
• Surveys are low in manipulation of antecedents
• Surveys are usually written or face to face

Constructing Surveys: Steps

• Identify specific research objectives
• Decide on the degree of imposition of units (Do you want low or high imposition?)
• Decide how the survey data will be analyzed
• High imposition leverages statistics; low uses qualitative methods.

Question Types

• Closed questions (structured questions) are answered with a limited number of alternatives with a high imposition of units
• An example of a closed question is, "How do you feel our president is doing?: very good, good, fair, poor."
• Open-ended questions necessitate participants to respond with more than a yes or 1-10 rating, with a low imposition of units
• An example of an open-ended question is, "Why did you choose your major?"

Difficulties with Questions

• Many survey questions are ambiguous, too complex, or double-barreled
• Ambiguous questions can be difficult to understand
• A question like, "Have you ever considered the idea of abortion?" leads to ambiguity
• Complex questions use a double negative or a word whose meaning is difficult to understand
• "I don't usually dislike not being alone" is a complex question
• Double-barreled questions contain two ideas
• "Do you feel the country is going in the right direction and the president is doing a good job?" presents two ideas

Data Analysis Methods

• The number or percentage of responses can be reported for closed questions
• Open-ended questions can be analyzed using content analysis with objective rules

Measuring Responses: Nominal Scale

• This is the simplest level of measurement
• Assigns items to two or more distinct categories
• Categories can be named using a shared feature
• Does not evaluate item magnitude
• Categories include true, false, male, female
• An example is sorting professors into exciting and dull categories

Measuring Responses: Ordinal Scale

• Measures the magnitude of the dependent variable using ranks
• Doesn't assign precise values
• Listing favorite sodas from favorite to least favorite is an example
• This provides no information on how much someone likes their third ranked soda

Measuring Responses: Interval Scale

• Interval scale measures the magnitude of the DV using equal intervals between values with no absolute zero point
• Examples include Fahrenheit/Centigrade and the Sarnoff and Zimbardo's (1961) 0-100 scale
• 0 temperature is not a true zero
• A question such as, "How much do you like this professor?", with responses "0 = not at all, 1 = very little, 2= a little, 3= like him, 4 = like him a lot" is an interval scale

Measuring Responses: Ratio Scale

• Ratio Scale measures the magnitude of the dependent variable using equal intervals between values and an absolute zero
• Allows us to say that a 2-meter board is twice as long as a 1-meter board
• Examples include distance (in meters) and time (in seconds)

Measurement Scales: Which to Choose

• Ordinal scales provide more information than nominal scales
• Knowing candidates' placement ie first, second is more insightful than simply knowing winner and loser
• The percentage of people who voted for each candidate (ratio scale) is the most informative

Measurement Scales: Selecting

• The scale depends on the variable and desired precision level
• Psychological variables like traits, preferences, and attitudes represent a continuous dimension
• Different scales can be used to measure continuous dimensions like interval or ratio

Measurement Scales: Selection

• When working with variables like sociability, psychologists often select the highest scale possible
• High scales provide more information and allow analysis with more powerful statistics

Survey Items: Important Considerations

• Engage subjects from the start by asking interesting questions to increase cooperation
• Subjects decide to refuse to answer surveys during the first few questions

Survey Items: The first survey question

• The first question should be relevant, easy, interesting, answerable by most respondents, and in a closed format

Understanding Response Style

• Response styles are tendencies to answer questions or test items
• These occur without consideration of wording
• Differences exist regarding willingness to answer, position preference, yea-saying, and nay-saying

Response Style Example: Willingness to Answer

• This is a potential to guess or omit items if unsure
• Informing subjects there are no right answers could yield more responses

Response Style Example: Position Preference

• Selecting answers based on position
• An example is choosing "c" on multiple-choice exams
• An example is consistently answering "true" without reading the question because all the other statements were "true"

Response Style Example: Manifest Content

• Manifest content is the plain meaning of words on the page
• Subjects may ignore manifest content when answering questions about feelings or attitudes
• Subjects might just answer "yes" to everything instead of paying attention to the question

Response Style Example: Yea-Saying and Nay-Saying

• Yes-saying is agreeing with an item regardless of manifest content
• Nay-saying is disagreeing with an item regardless of manifest content
• To control for this, mix up questions with some needing "no" and "yes" responses
• Include things like "I am happy most of the time", "I enjoy being with other people" and, "When I can, I avoid noisy places."

Survey Issues: Context Effects

• Context effects are changes in question interpretation due to position within a survey
• Common when two questions are related and not separated by unrelated questions

Survey Issues: Context Effects Example

• If asked to rate Kerry Washington from not sexy to sexy, subjects may rate Miley Cyrus according the the "sexy" question before it
• A buffer question in between the two would resolve this issue

Social Desirability Response Set

• Representing ourselves in a socially appropriate way when responding to a question's implied meaning
• Subjects respond to sound good in the experimenter's eyes
• Ex: Question: Are you able to get along with loud-mouthed obnoxious people? Answer: "yes" to be more likeable

Social Desirability Scale

• Scales exist to demonstrate if a person has high social desirability
• Data may not be usable in analysis should if the individual does
• The Marlow-Crowne Social Desirability Scale can be used to measure it

Methods of Collecting Survey Data

• Written questionnaires
• Mail surveys
• Telephone surveys
• Internet surveys
• Interviews
• Focus groups

Collecting Survey Data: Written Questionnaires

• Should have clear concise instructions
• Privacy and anonymity should be given to control for reactivity
• Anonymity and privacy should be given if questions are embarrassing

Collecting Survey Data: Mail

• Always include a polite cover letter and stamped return envelope
• Include a small gift if possible
• Response rates are between 45%-75%
• Track who does not return it and consider another mailing
• High non-return rate compromises results

Collecting Survey Data: Computer and Internet

• SurveyMonkey.com and SurveyGizmo.com can be used
• Allowing for easy data collection and analysis
• There is less concern of social desirability because it feels more anonymous
• Cannot tell if person takes survey multiple times

Collecting Survey Data: Telephone

• Use random digit dialing
• Provides a wide sample
• The response rate is 60-90%
• Caller ID increases refusal to calls
• Male telephone interviewers are statistically more effective than females

Collecting Survey Data: Interviews

• Tend to the most expensive and time consuming
• Rapport should be established
• Female interviewers tend to be more successful than males
• Best results when interviewer matches the respondent on race, physical appearance, and socioeconomic status
• Interviews can either be structured or unstructured

Collecting Survey Data: Interview Types

• Structured- questions are asked the same each time from a script, yielding quantifiable data
• Unstructured interviews allows the interviewer to explore interesting topics as they arise, yielding data that is hard to use for content analysis

Collecting Survey Data: Focus Groups

• Small groups of people with similar characteristics
• The facilitator guides the group through issue discussions
• Open-ended questions are usually used

Evaluating Surveys

• Validity: Does the survey measure what it's supposed to measure?

Understanding: Sampling

• It's deciding who fills out your survey
• A population consists of all animals, people, or objects that share at least one characteristic (e.g. -All undergraduate students, all senior citizens)
• A sample is a subset the population of interest. -Data can be used to draw inferences about the population.
• Representativeness- sample mirrors the larger population
• Two sampling approaches exist: probability and nonprobability sampling

Probability Sampling

• The odds of their being in the study are known or can be calculated. If the target population is all undergraduate students from city colleges, it can be counted from city records.
• In this case we know the odds of anyone person being in the study
• The researcher must use an unbiased method of choosing participants, such as flipping a coin or by random selection
• Likely to represent the population

Probability Sampling Methods

• Simple random sampling is a portion of the whole sample selected in a random way
• Systematic random sampling means to list all the population in an unbiased way and take every nth one
• Stratified random sampling means to randomly sample from people in each subgroup at the same proportion they exist in the population
• Ex: If school is made up of 70% female and 30% male, then want your sample to also be 70% female and 30% male.
• Cluster sampling - sample entire clusters or naturally occurring existing group
• Works if the cluster has traits similar that of the population

Nonprobability Sampling

• Most studies using nonprobability sampling

Main Nonprobability sampling methods

• Quota sampling means the researcher has quotas to fill, needs 50 white and black men to answer questions.
• Convenience sampling is groups that already happen to be available and is commonly used
• Purposive sampling selects a sample needed for the purpose of a study
• Snowball sampling- The researcher gets a few people who fit criterion and ask them to locate additional individuals.