Starts with background experience.

Starts blindly and then comes up with a theory to explain what happens.

Comes last.

Use data to framework theories.

Moves from specific data to more general.

Likes Objectivity because it starts blind and tries to make sense of anything. 

coming up with theories from hisotry. 

Theory driven

Comes first.

Data is used to determine whether theory is correct.

research stems from known position as to how the data will develop.

broad group of people

All unites possessing the attributes or characteristics in which the research is interested

 Determines by researcher where primary interest lies

Goal is to understand this population by viewing a subset of it

those members of the population that we are capable of studying.

the set of units that have a chance to be sampled.

Rarely exactly the same as the population, some people are unavailable. 

people that we care about, that we can reach, that are part of the study.

data collected from sample used to make inferences about population

Stats allow us to do this.

mess with what we care about. Stands in the middle of our two variables and bridges them together. 

- job performance leading to job satisfaction, there is another variable, rewards, that actually connects these two. 

part of the process. 

Not really an independent variable.

Not something we manipulated in the study

i.e- age

Studying aggression, looking at stress levels, anger levels..ect. 

How directly things should correlate with each other. 

Describes: outcomes processes or relationships between variables

Determines Causes: of communication behavior. intervention programs, real world application of research

Prediction: improve decision-making and precautionary measures.

Explanation: Understanding why a behavior occurs can help us adjust or modify behaviors

Control Outcomes

1. personal experience

2. intuition "it makes sense"

3. Authority "someone told me so"

4. democratic dialogue (the best ideas emerge when you assume all voices have equal weight.)

5. appeals of faith: lack logical proof or material evidence. 

1. based on evideence

2. propositions are testable

3. explores all possible explanations

4. results are replicable

5. results are made public

6. scientific research has a self correcting nature

7. relies on measurement and observation

8. recognizes the possibility of error- tries to control it

9. objectivity requires minimization of bias and distortion

10. scientists are skeptical 

11. interested in generalizability

12. heuristic in nature (the ability to suggest new questions or new methods of conducting research)

Error in instruments

anything that can happen can happen by chance

All measurement contains error

you will never perfectly capture something through an experiment

Monkeys typing Shakespeare. 

- Look out for authors of multiple papers; publication date. 

- Read the abstracts

- Read the literature reviews- identify theoretical foundations (assembly of different articles on the topic with synthesize)

- Read discussion Sections- what was found? what still needs to be found? is the topic worth further exploration? where can you fit in?

1. describe an event or observation that needs understanding

2. create logical explanation

3. move from specific explanation to more general application

4. derive predictions from the theory built

5. select a focus and test theory

6. use results to confirm, revise, expand, generalize, or abandon developed theory- theory flawed? or method flawed?

Breaking Down The Lit Review:

- Frames the research investigation; puts study in perspective

- Breakdown and analysis of variables studied (including history)

- Not just summary: Analysis, Synthesis, critique (acknowledge drawbacks)

Justifies study

Near beginning of lit review

Includes empirical repot findings, theoretical articles, previous similar studies and broad examination of an entire problem area. 

Beware of stylistic flaws

Beware of unclear careless writing; unspecified assumptions

Make sure report fits within context of body of research

Make sure all hypothesis are clear and specific

Make sure all key terms are defined theoretically and operationally

the variable that is manipulated or varied to see the effect on other variables. 

- used when researchers manipulates variables, when researcher does not manipulate the variable we call it the predictor variable. 

Variable influenced or changed by independent variable- this is the variable of prime interest.

- when researcher does not manipulate 1st variable it is called the outcome variable. 

- describe a unit of measument

- specify a level of measurement

- provide a logical/mathematical statement about how measurement is supposed to be made and combined. 

- Space and time contiguity

- Covariance

- Temporal Ordering

- Necessary connection 

when the presence of a 3rd variable obscures the nature of the relationship between two variables.

It can...

create a relationship that doesnt exist

inflate a relationships strength

deflate a relationships strength

1. identify the potential confound variable

2. hold it constant

- equalize assignment of cases based on the value of the confound variable across groups- (experiment control groups have equal proportions of the confound variable)

Controlling Statistical control:

- identify potential confounding variable

- measure the variable and include it as a control variable

- mathematically remove the effect of the variable from the variation of the dependent variable.

- access a unique effect of X on Y with the effect of the confounding variable on dependent variable removed.

no need to identify the potential confounding variables

- random assignment to groups representing levels of independent variable (ex. experimental/control groups) 

- assures that effects of other variables are equalized across groups or any cross in assignments are random. 

device/instrument used

How is the device used

Skill of person using device

Attributes or characteristics being measured

These dimensions of measurement can/will impact data

The proper use of measurement influences the accuracy. 

- level used to measure is based on your operationalization.

- determines how you will measure a concept

-must specify a level & unit of measure.

Nominal Data (lowest)

Ordinal Data

Interval Data

Ratio Data (highest)

- each level has all the characteristics of the preceding level

- higher levels offer greater precision. 

- describes presence of absense of some characteristic or attitude.

- no way to express partial presence.

- any value imposed on the categories is arbitrary

- minimum of two categories, must be mutually exclusive and exhaustive

- type or category based, categories have no logical rank or level (also known as categorical data EX. M&F)

- Measured by ranking elements in logical order from lowest to highest

- sequencing of data without precise measurement.

- rankings are relative (there can be no zero)

- logical rank order, but no logical distance between elements

- determined by "more" or "less", but we don't know how much "more" or "less"

EX- class standing

- measured based on specific numeric scores or values

- distance between any two adjacent points assumed to be equal

- zero is arbitrary, there is no absolute zero

-rank order, values equality spread apart, but no true zero point. (EX. using a 1-5 scale)

- interval level data qualities, but also have an absolute zero

- most variables that meet interval-level requirements also meet requirements for ratio level (EX.- Number of children in a household 0-5)

- allows us to compare two people on variable and determine the ratio of one person's value to another person's

- higher scale= raw numbers, lower scale= arbitrary continuum

-Level you decide to measure your variables at should be guided by analysis you will later do

- Certain stats require certain types (levels) of data

- normally best to measure at highest possible level

- can always modify to a lower level of data, but can never modify to a higher level of data.

consistency/stability of measurement

-reliability is necessary, but not sufficient for validity.

- reliability refers to how consistent your measure is at measuring what is supposed to be measured.

Administering the same measure multiple times and assessing how consistent the responses are

*the more time between responses the lower the reliability

Generate two versions of the same measurement using items from same pool. Administer both versions to the same  group.

*if results from both versions are similar, strong alternate form reliability. 

randomly split questionnaire in half, giving each half to different groups. Consistent results across both groups indicate reliable measure

(EX. even & odd)

the extent to which conclusions developed from data collected from a sample can be extended to the population

- the goal of sampling is to have a representative sample that is highly generalizable 

degree to which sample differs from populations characteristics on some measurement. 

Degree to which our sample is not representative.

the random error that occurs in any sample- unpredictable things happen. 

sampling error reduced by increasing sample size

some form of random selection ensures anyone can be sampled

probability for any element being selected is equal and known for every element in the sample.

randomization eliminates researcher bias in sampling. 

techniques do not use any form of random selection

used when randomizing isn't feasible or phenomena (believe to)  doesn't exist throughout population (IE- diseases.)

- no other technique can be reasonably used (ie- time or money)

- when we believe that a variable of interest are evenly distributed across the population of interested

- when looking at special cases. 

Do 2 groups differ after treatment is presented to only one?

-if yes, we attribute an effect to the independent variable

Remember, multiple groups can receive some form of the manipulation. 

-random assignment should provide equivalent groups.

- but w/ only a post-test we have no way of being sure.

- allows us to test equality of groups at a time.

- introduces some chance of testing effect.

- we can compare groups and prepost questions.

helps control for the testing effect and combo between posttest and pretest. 

researcher manipulates 2 or more independent variables. 

-used to explain complex cause-effect relationships, that cannot be explained by one independent variable. 

multiple measurements of the dependent variable over time

- can be applied to post-test only or pre-post design. 

-easy way to establish causality

- cannot be applied to all communication phenomena

- strong in terms of internal validity

- weak in terms of external validity

-researcher does not control the manipulation of the independent variables

- participants are not randomly assigned to the conditions. 

-causation cannot be determined.

- Predictor and criterion (do not imply causality) are better labels for variables, as the do not imply causality.

does it measure accurately predict future behaviors?

do those scoring high on apprehension scale behave apprehensively?

does measure correlate highly with theoretically similar measure?

ie- shyness scale with communication appregension scale

- depression scale with low self esteem scale.

-Instrument itself: is it outdate? bias?

-Pygmalion Effect: does instrument inadvertently suggest desired response?

-Maturation: people change or become familiarized with questionnaire. 

-Mortality/Attribution: people leave the study