I. Purpose/Problem II. Conceptualization III. Design/Data Gathering IV. Results/Analysis V. Implications/Conduction
What am I studying?
different angles of my research idea?
interested in the difference maker, difference made or the process of making a difference
most important concepts?
III. Design/Data Gathering
how will you answer the research questions?
what methods will be used?
what did you find?
what do your results mean
Starting a Research V
Explore a pool of experiences (your experiences/experiences of those helping us)
Questions to ask as you begin research:
What is the condition? X?
What comes after it? X—?
What comes before it? ?—X
What comes inside of it? X (over boxed ?)
What comes inside of it? (situationally) ? (over boxed X)
What is outside? (generally) ? (over dashed boxed X)
1. What is the condition? X?
focus of attention? (specific as possible)
2. What comes after it? X—?
consequences of this condition?
3. What comes before it? ?—X
a consequence of this condition?
what produced this condition?
4. What comes inside of it? X (over boxed ?)
what does this condition include/contain
what is contained in this relationship
5. What comes inside of it? (situationally) ? (over boxed X)
what is the immediate context for this condition?
is there immediate context in which is occurs?
6. What is outside? (generally) ? (over dashed boxed X)
what is the nature of the world in which this condition is found?
X? ——— focus of attention
X—? ?—X ——— association, cause/effect relationship
X [?] ——— distinctions
? [X] ?[[X]] ——— context
3 parts of a concept
way to refer to something (public opinion, media use, people)
what is the verbal meaning attached to concept label?
sometimes the label is enough, you don't need it. (intelligence: quantitative, analytical, verbal)
takes verbal meaning from conceptual definition and specifies the steps needed to measure/experience that concept. (develop steps for measuring conceptual definitions and devise a formula)
Operational definition should do a good job representing the conceptual definition, if it doesn't we aren't measuring what we thought we were
4 Levels of measurement
classifies people, things etc.
label for a category
refers to presence/absence of something
equivalence: everything in that category has to be equal
categories: exhausted and mutually exclusive
rank on some dimension but not specific (horse race w/ no stop watch: 1st, 2nd, 3rd)
non specific measurements, you know the order but you do not know the distance between
has an order: any category can fall higher or lower to another
categories: exhausted and mutually exclusive
equal interval measurements
NO TRUE ZERO: when zero does not mean absence of the thing because you can go below it. (weather)
DOES have a true zero, [[must have everything stated in interval]] ——— speed, distance traveled
Why should you start at the highest level of measurement?
you can start high and go low, but you cannot start low and go high
How will 4 levels of measurements be used?
chi squared data ——
pew data —— nominal/ordinal
analysis of variants (calculating mean) —— interval/ratio
how well a measuring devise measures what is it supposed to.
how well the operational definition matches the conceptual definition
a measuring is reliable if it consistently gives us the same answer.
measured through stability, internal consistency and equivalency
is a necessary condition for validity
= operational definition
Kinds of error in Operational Definitions:
consistently not measuring what we think we are.
operational definition doesn't represent conceptual definition well
problem with validity
if random error in operational definitions is controlled, the measure (exam/survey) will be more reliable.
μ = POPULATION MEAN
x̄ or σ = SAMPLE MEAN
N = POPULATION
n = SAMPLE
entire group or class group being studied
subjects of population that represents its population well
value that describes a population
value that describes a sample
[mean] statistical measure that takes one score that represents the information as a whole
techniques that allow us to study a sample then generalize it to the population
—can help us generalize about the populations from which samples were selected
—can tell us how likely it is that the differences between samples reflect real differences between populations
Non Probability Sampling (cannot calculate sampling error, try to avoid)
chose who is available [doesn't necessarily represent population you are talking about]
unqualified volunteer sample
purposive sample [survey those who relate to survey]
snowball sample [survey those who know people who have taken it]
Probability Sampling (can calculate sampling error)
random sampling: everyone has an equal chance of being selected to respond
systematic random sampling
Central Limit Theorem
the average/sum of a large bunch of measurements follows a normal bell curve even if the individual measurements do not
applies to distribution of sample means for ANY population
don't need large sample size before distribution is almost perfect (n=30)
can tell how much sampling error we have based on the relationship between samples and the normal curve
Central Limit Theorem applies to distribution of sample means for any population
Standard Deviation - average distance of scores from population mean
Standard Error - average distance of sample means from population mean
Z Score - # of SD's you are away from the mean
Confidence Level - plus or minus 2 standard errors will give us a 95.44% confidence interval
SS - sum of the squared deviations around the mean
SD (population) = SS/N = (x-u)^2
SD (sample s) =
SE (P) =
used with NOMINAL level data
used to calculate percentage or agreement between two coders
run on each item in a code book
acceptable level of reliability: min 90% better 95%
^because formula doesn't consider agreement by chance^
M: # of coding decisions on which 2 coders agree
N1: # of decisions by first coder
N2: # of decisions by second coder
RELIABILITY = 2M / (N1 + N2)
used with NOMINAL level data
% of observed agreement between coders
% of expected agreement between coders (agreement expected by chance)
the agreement you REALLY get divided by the max you COULD get if you take agreement by chance out.
acceptable level of reliability: min 85% better 90%
RELIABILITY = % observed agreement — % expected agreement / 1— % expected agreement
Recognizing Quantitative Data
look for references to coders, a code book or interceder reliability
look for a # (frequencies and percentages) in tables or text
describes communication context
examines changes in context over time
answer research questions about message characteristics
compare media content to "real world"
establish starting point for media effects studies
content analysis alone cannot tell us effect content has on audience
findings limited to categories, definitions used in that analysis
can be time consuming/expensive
Research is a never ending process
1 study will answer 1 set of questions.
How would you use the 6 types of box questions?
choosing between two things
understand all of the factors of an issue
consume media responsibly
deepen your thinking
If we want to emphasize, we talk about # of people effected
If we want to deemphasize, we focus on the raw number of people effected
Use scale when talking about beliefs or values
lets audience express
Reliable ALWAYS = Valid
Valid DOESN'T always = Reliable
There is ALWAYS chance for sampling error
Round to the hundredths place