Sampling: Flashcards
Why are surveys used in psychology?
What do surveys measure?
Why surveys?
- most common of the self-report measures
- more cost-efficient way to collect data than experiments
- useful for exploratory analyses to frame your research question
- useful for pilot studies
- useful to observe trends
What they measure:
Attitudes and beliefs - Most difficult - raises questions of validity
Facts and demographics - Straightforward - but which categories? How defined? Sensitive questions?
Behaviour and states -Asking participants to reflect on past, current or likely future behaviour/states
Open ended vs close ended question in surveys/scales
Open-ended questions:
-Open-ended questions allow any answer from the participant (language allows infinite ways to get an idea across!) e.g. “How do you feel about this picture?”
-Good to investigate what people are concerned about and the quality of their experience
-Difficult to analyze (especially when lots of participants), but great for hypothesis generation and theory building.
-Avoid.
Close-ended questions:
-Odd numbered
-In most situations, having an odd number of categories is usually the more effective way of gathering accurate data.
-Because many people are legitimately neutral on a subject, forcing respondents to answer a question on an even scale will bias your end results as truly neutral people must select a category that does not represent their opinion.
What are the different ways of asking questions in a survey:
Even numbered - ‘Forcing’ participants to take a ‘side’ Example: Medical marijuana should be legal?
Graphic rating scale. Unlike the discrete ordinal Likert scale, this scale is a continuous interval scale: Presenting two opposites and asking the subject to place their response somewhere on the continuum. On this scale, respondents indicate a level of agreement or disagreement with each statement:
Composite (summary) scales. Very common in health sciences. Multiple items sum to one final score (or total score). A single inventory can consist of multiple “subscales”, such as the GHQ- 28: Somatic, Anxiety/Insomnia, Depression, Social Dysfunction. Sometimes reverse coded to avoid negative or positive wording.
What is a Likert-type scale:
On the ordinal scale
Researcher develops a series of items on a continuum, worded favorably and unfavorably regarding the underlying construct
Researcher must decide how many response categories to allow and if they should be:
- even numbered (forcing a response towards one end or another)
- odd numbered (allowing a neutral or middle response)
1= strongly agree, 2= agree, 3= indifferent ..
How the questions are worded obviously has an effect on the type of answers you get especially with sensitive issues:
Comment on:
- Simplicity
- Double-barred questions
- Loaded questions
- Negative wording
- yea/nay saying
Simplicity - keeping it simple and unambiguous - avoid very long questions
Double-barred questions - Asking two things in one question - e.g., “Shall the government decrease taxes and borrow more money to fund healthcare?”
Loaded questions - Questions that bias or imply an answer - e.g. “What was the effect of the recent illegal legislation?”
Negative wording - “Shall the city not approve the women’s shelter?” - Better: “Shall the city approve the women’s shelter?”
Yea-saying and nay-saying - vary questions slightly to avoid yea- and nay-saying - can also function as lie-detector questions
What are the different ways of administering a survey?
Ways to administer the survey
-Interviews
-structured (guided) versus unstructured
-face-to-face, phone, focus groups
-easy to ensure that all questions are answered
-ambiguities can be resolved
-But: social desirability, interviewer effect
Questionnaires
-personal administration, mail surveys, internet surveys , iPad
-easy to collect a lot of data – but it is rigid
-Structured questionnaires: limitations of closed-ended surveys
What are some of the assumptions of a survey?
Assumptions of survey
-truthful responses
response sets - any systematic tendency of answering questions that deviates from the truth, often as a result of:
- order effects (possible fatigue or needing practice)
- interviewer effect
- social desirability (remember the Hawthorne effect)
- fear of consequences
- perceived lack of confidentiality
- perceived lack of anonymity
DEFINE:
- Sampling
- Population
- Sample
- Sampling frame
Sampling:
selecting a group of elements (people, events, behaviors, etc.) to obtain information about a population
Population:
The entire group of elements that share a set of specific, common characteristics.
Sample:
a subset of the population who are supposed to be representative of the population they were selected from.
Sampling frame:
the group of individuals you can select a sample from (sometimes also called accessible population which is not always the same as your intended target population!)
What types of sampling methods are representation of the population?
- Random sampling
- Simple random sampling
- Stratified sampling
Random sampling - every element has equal chance of being selected
Simple random sampling
-Individuals are selected at random from the target population
-Usually involves the use of a ‘table of random numbers
Stratified sampling
-Sampling frame is divided or arranged into sub-groups
-Participants are randomly selected within strata e.g.: we know that, in the population of persons with Autism Spectrum
-Disorder, 20% are women and 80% are men. We will then split the sample into two strata (male and female). Of our sample, 20% will be female and 80% will be male. This ensures that gender is
-proportionately represented in the chosen sample
- Allows for a smaller sample to be used than simple random sampling, while ensuring all study characteristic are preserved.
Define and explain:
-Non-random sampling
-Convenience sampling
-Purposive sampling
Non-random sampling - unequal chance of being selected
Convenience (haphazard) sampling
-Individuals are selected on the basis that they are available and easily accessible
-Little opportunity for control of biases [-] Inexpensive and time-efficient [+]
Purposive sampling
-Conscious effort by the researcher to ensure that particular individuals are included to meet study criteria.
-Relies on researcher’s judgment
Define and explain:
-Quota sampling
-Snow ball sampling
Quota sampling:
-Form of convenience sampling where strategies are used to try and ensure the inclusion of groups that tend to be under-represented (e.g. minorities)
Snowball sampling
-Takes advantage of social networks etc. and asks participants to refer others they know who meet selection criteria
-Used to locate participants that may be difficult to reach
How do these processes effect sampling:
- Random variation
- Systematic bias
Random variation (occurs by chance)
-expected difference in values that occur when one examines different subjects from the same sample
-An individual subject’s values will vary from the value of the sample mean. Therefore a small sample will increase random variation.
Systematic bias
-variation in values due to sampling not being random but affected by biases in selection - therefore consistently selecting subjects who are different or vary from the population in some specific way
-worse if unaware of it
-When your respondents are not representative of the population due to a bias in the sampling process. Systematic under-sampling of: Culture; Sex and gender; Education; Attrition
These are serious research problems that prevents generalization - no external validity
What is self selection bias?
How does having a larger sample size effect generalisability?
Self-selection bias
-e.g., selecting people who are willing and not considering those who declined
-Interaction between setting and treatment
* e.g., hospitals that welcome research versus those that do not
The larger the sample, the more likely it is going to represent the population – i.e, population generalisability. Population generalizability is the same as increased external validity
Define and explain:
- Frequency distribution
- Normal distribution
Frequency distributions (also called histograms)
- Used for quantitative variables (opposite of bar graphs that use categorical variables)
- A histogram is a presentation of a frequency distribution in bar format
- Frequency distribution – is a systematic arrangement of data values in which the unique data values are rank ordered
Normal distributions
- Ideally, we want a symmetrical distribution - this is a normal distribution (or ‘bell-shaped curve’)
- Many naturally occurring things have a normal distribution – e.g. most men in NZ are 175cm tall
- Many real-life distributions are not normal
