Week 2 RESEARCH METHODS Flashcards
(38 cards)
Key components to a statistical investigation are:
Planning the study: ask testable research question & decide how to collect the data
Examining the data: how? Using what graphs?
Reliability: consistency of measure
Validity: degree to which a measure is assessing what it is intended to measure (how much smth is actually testing what it’s supposed to)
Inferring from the data: draw inferences “beyond” the data
Drawing conclusions: based on what you learned, what can you draw?
Cause and effect conclusion: whether one variables changes the other
Distributional thinking
Data varies, so think of meaningful ways to capture that data-
Pattern of variation:
the distribution of the data (how variated it is)
Statistical significance/Significance:
a result is important only if it is unlikely to be caused by chance but rather science
P-value
the amount of probability something might happen (want it to be low, so that it can actually be related to science)
Sample: (DEF)
subset of individuals
Population: (DEF)
large group of individuals that sample came from and represents
Generalized
Conclusions from a sample can be generalized to represent the larger population
Random sample: (DEF)
use probability to select subset of people for a sample
Margin of error: (DEF)
statistic expressing the amount of random sampling error in the results of a survey
Randomly assigning: (DEF)
using probability-based method to divide sample into smaller treatment groups (Balances all variables)
Correlational design
▪ When scientists passively observe and measure phenomena
▪ Do not interfere and change behavior.
▪Identify patterns.
Positive correlation (scatter plot)
variables move together (inr. slope)
as x incr. y incr.
negative correlation (scatter plot)
variables move in opposite directions (dcr. slope)
as x incr. y decr.
Strong VS weak correlation
strong: Higher absolute value
weak: if close to 0 variables are unrelated to eachother
If correlation found, can you assume there is causation?
NO. CORRELATION =/= CAUSATION
Just because 2 variables have a correlation, can’t say they are caused by one other because don’t know which one came first. perhaps confounding variable present.
Experimental design
▪ Researchers measure abstract concepts, like happiness, with operational definitions
▪ Random assignment so participants can choose their conditions
▪ Researchers manipulate independent variable to observe changes in dependent variable
Operational definitions: (DEF)
Should be clear, objective, and specific to be tested and quantified
Confounding variables: (DEF)
other things that affect influences both IV and DV
Examples of cofounding variables
▪ Placebo effect
▪ Experimenter expectations
▪ Participant demand
Double-blind procedure: (DEF)
experimenter nor participant knows participant’s conditions
Qualitative designs:
Allows to study topics that we cant experimentally manipulate.
Observation
often involves researcher embedding self into group- (ex. Wanted to study cult, pretending to be in cult)
Case study
examines specific individuals or contexts