-
measurement
observable characteristics of our variables
-
triangulations
multiple approaches in one research study; means to enhance the quality of the work done
-
methodologically
using both quantitative and qualitative methods
-
data
instead of using methods, bring in data from different instituations or places
-
researcher triangulation
have multiple researchers collect data
-
theoretical triangulation
bring in different theories and triangulate betweent two to three of them
-
Variable measurement: nominal
simply asking a question that puts a variable into a category.
-
Nominal characteristics:
- must be mutually exclusive,
- must be equivalent
- must be exhaustive --> you leave no possible responses out
-
Ordinal characteristics
- must meet all the nominal characteristics, but is also rank order
- anytime you have to rank your sources, such as grades or team sports
-
Interval characteristics
- equal distances between two points on the scale
- we will know how much someone likes something
- zero rating: its a point on the scale, doesnt mean it doesnt exist
-
Likert Scales
- scale must be an interval
- low numbers are disagree, higher numbers are agreement (1-5)
- likert type scale uses any modification such as 1-7
-
semantic differential scales
polar opposites on both ends of the scale
-
ratio measurement scales
zero does not mean the variable does not exist
-
tunnel method
same types of questions throughout
-
funnel method
start with broader questions, then narrow them down
-
inverted funnel method
specific to broad
-
4 question order effects
- consistency: lying about something up front means you will probably lie the whole way through
- fatigue: questionnaire is too long
- redundancy: similar questions posed in different ways annoy the reader and they can start to check boxes
- response set: clicking one question the whole way through
-
-
-
internal validiy
- draw accurate conclusions from my research
- we want a high amount of internal validity
-
external validity
how generalizable are my findings, as in, how much can i apply what i found to other people
-
if you are valid you must be
reliable as well. you are not necessarily valid if you are reliable
-
measurement reliability
consistency of our measure and establishes high internal validity
-
Three ways to have measurement reliablity
- test/retest method: comparing when tests are taken twice
- alternative forms: give one group of people your measure and also give them a measure that is testing about a similar concept
- split half: compare each half of the quetions to each other
-
internal consistency method
- best method to have measurement reliability
- calculates all the possible split halves, all the possible outcomes to give the best one
- Chronbachs Alpha gives this to us, we want higher than .7
-
Threats to internal validity
- history effect: something external to the study affects the participants
- sleeper effect: effects of the study took longer than expected and you didn't wait long enough to observe them
- sensitization: give a group a measure and ask them to take the same measure at a later time; didnt give them long enough to forget their answers
- data analysis: certain procedures and statistical analysis should be used in certain instances
-
more threats to internal validity
- participants:
- must have good incentive
- Hawthorne effect: know they are being studied
- selection: who are the people being studied
- statistical regression to the mean: need a larger group of people if your average is unusually high
- mortatlity/attritition: people drop out of the study
- maturation: internal changes that are going on in the participant
-
researcher threats to internal validity:
- combat threats to ourselves
- blind procedure: participants do not know what group they are in
- double blind: neither participants or researcher know what group they are in
-
external validity:
ecological validity: does the research describe what is accurately happening in real life
-
replication:
- exact: repeat study in exact same way
- partial: keeps some of the study the same, but modifies others
- conceptual: keeps the concept of the study the same but uses entirely new procedures
-
what is a sample
a subgroup of people collected from both a population and target population that actually participate in the study
-
three problems with samples
- size: contingent on statistical distribution -need a minimum of thirty people
- bias: systematically exculded some people that could have been useful to the study
- representativeness: we want our sample to accurately represent our population
-
sampling frame
list of all possible participants
-
sampling unit
each person on the roster
-
sampling error:
the extent to which my sampling deviates from my population
-
confidence level
how confident i am that my sample is representative of my population --> want to be 95% or higher
-
confidence interval:
plus or minus range that i believe my sample will fall between
-
IRB guiding principle
the rewards of doing my research do not outweigh the costs
-
-
conceptualization:
the dictionary definition of what you are going to stuyd
-
operationalizaton:
how are you going to measure your variables
-
Data collection
what method are you going to use
-
data analysis:
what statistical tests are you going to use; qualitative or quantitative
-
reconceptualization
discussion about how does it fit in with the theory, is more research needed, etc
-
Epistemology
- Pos: researcher and work are seperate from one another
- Nat: your perspective affects your research
-
Axiology:
- pos: research should be value free, unbiased
- nat: how you view the world is important to your research
-
Methodological
- pos: deduction, reason down to a conclusion- cause and effect, quantitative methods
- nat: induction -specific to general, goal is a wholistic understanding at human behavior
-
Rhetorical assumption
- Pos: standardized, in the third person
- Nat: first person, more story telling type way
-
Research Question and the three types
- tends to be less defined- unsure of what to look for or find
- descriptive: describe something or something that there is no information about
- tests of association: purpose of there is a relationship between concepts being studied
- Tests of difference: looking for differences between groups
-
Hypothesis and two types
- hypothesis is a statement
- tests of association: purposing a relationship
- tests of difference: will state a difference between variables
-
Two tailed and one tailed questions
- two tailed: does not predict a direction
- one tailed: predicts a direction between the variables
-
independent variable
influences change in the dependent variable
-
dependent variable
is the variable being changed
-
recursive cuasal model:
- one way street: one cause --> one effect
- ie: smoking cigarettes cuases cancer, and not the other way around
-
non-recursive cuasal models
two way street, reciprocal cause and effect go both ways
-
non-causal relationship:
variables are related but changes in the one do not cause changes in the other
|
|