Documente Academic
Documente Profesional
Documente Cultură
Irina Shklovski
Measurement
Populations
Measurement
Populations and Sampling Random Assignment Generalizability
Time
Cross-sectional studies & single experiments Longitudinal studies & repeated measures
Method
Experiments
& Quasi-experiments Behavioral Measures Questionnaires & Surveys Social Network Analysis Archival and Meta-Analysis
Measurement
Population
Method
Experiments
Measurement Sampling
respondents from population of concern Random sampling Systematic selection Stratified sampling Convenience sampling Snowball sampling
Sampling Biases
Non-response bias
Be
Volunteer bias
Some
Random assignment
Different from random sampling Mostly used for experiments or quazi-experiments Protects against unsuspected sources of bias Does NOT guarantee to balance out the differences between participants Chance is LUMPY
Generalizability
How do you know that what you found in your research study is, in fact, a general trend? Does A really, always cause B? If A happens, is B really as likely to happen as you claim? Always? Under certain conditions?
experiment
Pronunciation: \ik-sper--mnt also -spir-\ Function: noun Etymology: Middle English, from Anglo-French esperiment, from Latin experimentum, from experiri Date: 14th century
An operation or procedure carried out under controlled conditions to discover an unknown effect or law, to test or establish a hypothesis, or to illustrate a known law
deliberately vary something in order to discover what happens to something else later To seek the effects of presumed causes
An Experiment is
Two requirements:
Independent variable that can be manipulated Dependent variable that can be measured
Experiments in Research
Comparing one design or process to another Deciding on the importance of a particular feature in a user interface Evaluating a technology or a social intervention in a controlled environment Finding out what really causes an effect Finding out if an effect really exists
Remember
Experiments explore the effects of things that can be MANIPULATED (but there is a caveat)
Types of Experiments
Randomized units/participants assigned to receive treatment or alternative condition randomly Quazi no random assignment Natural contrasting a naturally occurring event (i.e. disaster) with a comparison condition
Experimental design:
Shadish W.R., Cook T.D. & Campbell P.T. (2002) Experimental and Quasi-Experimental Design for Generalized Causal Inference. Boston, Mass: Houghton Mifflin
Self-report measures
Questionnaires
Interviews Diaries
& surveys
Types
Structured Open-ended
Advantages
Sample
Disadvantages
Self-report
Response biases
Social desirability
Solution:
Yea-saying
Solution:
Sampling are respondents representative of population of interest? How were they selected? Coverage do all persons in the population have an equal change of getting selected? Measurement question wording & ordering can obstruct interpretation Non-response people who respond differ from those that do not
Design is KEY
Format booklet, printed vertical, one-sided Question ordering earlier questions can prime answers to later questions Page layout group similar items & use consistent fonts and response categories Pre-testing conduct think-alouds with volunteers demographically similar to expected participants
Common Problems
Navigation is paramount make sure the survey is EASY to follow Open-ended questions
The
size of the field allotted will determine the number of words amount differences have little impact
Incentive is key
BUT
Designing surveys:
Dillman, D. A., Smyth, J. D., & Christian, L. M. (2009). Internet, mail, and mixed-mode surveys : the tailored design method (3rd ed.). Hoboken, N.J.: Wiley & Sons. Fowler, F. J. (1995). Improving survey questions : design and evaluation. Thousand Oaks: Sage Publications.
Analyzing data:
Cohen, J., Cohen, P., West, S., & Aiken, L. (2003). Applied multiple regression/correlation analysis for the behavioral sciences (3rd ed.). Mahwah, NJ: Lawrence Erlbaum Associates.
So what?
Difference between quantitative methods is in the questions they can answer There are a LOT of methods and even more statistical techniques Regardless of the method, if its not an experiment, you CAN NOT prove causation