HCOM 300- FINAL EXAM

Ordinal Level Measurement (Rank order matters)

=data are measured by ranking element in logical order from high to low, or low to high (make sense to order in a specific way like quality of diamond; doesnt make sense to put hair color in order; but for quality it makes sense) Ex. Ranks order or speakers in terms of communication competence, ranking hardness of material (diamonds, glass, etc.) -Doesn't indicate degree of difference, it just gives order (we don't know how big the leap of difference between items is just that one is higher than the other)

Ordinal Level Measurement

*This is not frequently used in communication (in other types of research it is used more frequently)

realiable

-does the instrument CONSISTENTLY YEILD THE SAME RESULTS ex. if you maintain the same weight, does your bathroom scales say the same thing each morning? -even if the scale is inaccurate, but your weight never changes and the scale shows the same inaccurate weight day after day then that means it is still this ____

face-to-face surveys (disadvantages)

-introduces interviewer bias (hard to make the interview the same for every participant) -costly in terms of time and personnel

disadvantages of questionnaires (self-administered surveys)

-participipants can go at their own pace (can take longer for data collection) -requires literacy skills -generally limited to 1 to 2 open-ended questions

testing/retesting effect

...

validity

= the degree to which a measure ACTUALLY MEASURES WHAT IT IS SUPPOSE TO MEASURE or what you want it to measure and not something else Ex. Does your bathroom scale report and ACCURATE weight? -if you weigh 400 lbs. does it reflect that or does it show you weigh 200lbs instead

Ratio Level Measurement (equal spaces with true zero)

=assignment of numbers to items such that any adjoining values are the same distance apart and in which there is a true zero point Ex. number of children, number of times married, number of hours TV watched during week, distance

Interval Level Measurement (equal spaces with no true zero)

=data are measured based on specific numerical scores or values; intervals between numbers are equal in size -indicates amounts or degrees, but has no "true zero" Ex. temperature on Fahrenheit or Celsius scale (If it is zero degrees that does not mean there is a complete absence of temperature) -we often use it for liquert type of scale -this is equal spacing between intervals, but no true zero point

Nominal Level Measurement (Name)

=describes the presence or absence of some charactertistic of attribute (there are things while you dont have to put in any order or they do not have a meaning to them) also called DISCREET VARIABLES- you either have it or you don't have it (not a magnitude, intensity, or level- it is just the presence or absence) -no measure of the degree to which a variable is present (merely offers names or labels for charactertistics) Ex. Gender, Birthplace, Major, Hair Color, Eye Color (all are either there or not there)

realiability coefficient

A ________ _______ (0 to 1) tells us HOW CONSISTENT (reliable) A MEASURE IS. ex. IQ tests or Personality tests (usually these things do not change from day to day so these tests should be realiable day to day when testing)

online surveys (advantages)/ (disadvantages)

ADVANTAGES cost effective time efficient participants can take it where they want and take it when they want DISADVANTAGES may results in low quality data (people are not paying full attention to the survey and their attention is divided)

why sampling issues matter

An Example... (what could go wrong if you use the wrong methods or exclude a specific demographic) • Literary Digest correctly predicted outcome of presidential election in 1924, 1928, and 1932 • Before 1936 election, sent 10 million surveys to people listed in phone books and on lists of automobile owners o Predicted Alf Landon to be the winner by a large margin...Then Franklin Roosevelt won the election by a landslide o They used sampling methods that basically excluded people from low economic status (excluded large percent of population-mostly democrats excluded) so it was not the actual reality ---they used automobile lists when only rich people had automobiles

surveys

Lots of communication research and media studies uses s________

Threats to Validity

Maturation Experimental Mortality History Testing/retesting effect Placebo effect

experiment

a study of the effects of variables manipulated by the researcher in situations where all other influences are held constant

What is survey (i.e., survey research)?

a study that uses questionnaires to discover descriptive characteristics of phenomena

sample

a subset, or portion, of a population ADVANTAGES-less expensive, time consuming, more efficient DISADVANTAGES-might not be aware of the entire group you are trying to examine (can be missing info) & sacrificing some accuracy because not reaching everyong

survey

a system for collecting information (by asking questions or having participants respond to stimuli statements, researchers can collect data that can be used to describe, compare, or explain knowledge, attitudes, or behavior

pretesting

also called pilot testing (basically doing a trial run of the survey) -this should be done if you develop a questionnair or modified an exisiting one to see if there are major errors you will run into later

pilot testing

also known as pretesting; researcher's trial of a survey or questionnaire with a small group of participants who are similar to those individuals who constitute the population before data collection actually begins

challenges to creating questionnaires

ambiguous words (leads person on different route) a trigger word asking questions that do not push toward your hypothesis not asking bias questions

self-reports

another name for self-administered surveys; these can be done onsite, in the research setting, mailed and emailed to individuals at work or home, or administered through online software; advantage of this type of survey is that it can be more anonymous and done at the participant's own pace

closed questions

are short and concise disadvantages (you may miss information you want)

systematic random sampling

choosing a random number and choosing every 10th person to question or survey

categorical/discreet data

composed of qualitative kinds of variables (anyting being named without needing to know the degree it is expressed) -Nominal

population

consists of all unites (the universe) possessing the attributes or characteristics in which the researcher is interested (this is hard to do or it is unrealistic)

Ask "good" questions

construct open and closed questions appropriately depending on your purpose and desired data -most questionnaires use closed questions (answers must be exhaustive) -construct clear questions that are unambiguous -construct questions simple and concise questions (not too long or confusing) -avoid biased wording

What are factorial designs?

experimental designs in which more than one independent variable is used -Do doctors' sex and patients' sex affect patients' perception of doctor credibility? -Independent variable = doctor's sex & patients' sex -Dependent variable = percention of doctors credibility

continuous variables

expressed numerically (quantiative) to indicate matters of degree -ordinal, interval, ratio

nonresponse

failure to obtain data from individuals in the sample

semantic differential scale

form of interval measurement; using a stimulus statement, participants are asked to locate the meaning they ascribe to the stimulus on a response scale anchored by two opposites, usually bipolar adjectives

expert panel

form of questionnaire pretesting in which experts in research methology or in the survey's content read through the questionnaire together and discuss potential problems with the survey

panel

longitudinal design for a survey in which data are collected from the same participants at more than one point in time

open questions

more in depth and takes longer for them to answer

statistics

number computed from a sample

Ratio Example

number of children, number of times married, number of hours of TV watched last week, distance

parameters

numbers computed from populations (usually living in a world of statistics rather than ___________)

exhaustive

offering the partcipants all the possible answer choices (so that people cannot look at it and think well none of those apply to me) can often add "other" box so that people aren't forced into answering just for answering, but also dangerous because you do not want lots of people to answer that because you do not get your data then if all the answers are "other" or "neutral" -mutally exclusive

Test-retest method

one way of assessing realiability =giving the measure to the same group twice and reporting consistency between scores Ex. A medical study asked a group of 200 participants questions about their health background; asked the same questions three months later -only 15% reported the same information (very low reliability) -there could have been significant changes in their health background (they found out new info about health background, health code changing with their company) -the same person could have interpreted the question differently each time

experimental mortality

participant dropping out of experiment or dying

acquiescence response bias

people may want to say "yes" -they are hurring up and not paying attention -how people will agree no matter what is put in front of them -they just want to finish so will not pay attention to the question

open question

question for which respondents use their own words to formulate a response

closed question

question form in which respondents are asked a question (or given a statement) and then given a set of responses to select from

placebo effect

refers to the idea that participants may show changes even in the absence of experimental treatments, in reaction to the mental suggestion that they may have been given some stimulus ex. clip shown in class to wear velcrow to help balance or to put snails on face to help their skin

challenges to designing reliable and valid questionnaires

reliance on self-report can be problematic (people may forget information) (people may "reframe" information depending on current circumstances) (people may lie to give socially acceptable responses-social desirability effect) (people may want to say "yes"-acquiescence response bias) Question wording or format can affect respondent's answers Question order can affect respondent's answers (asking a question about satisfaction of relationship early and they say yes, might make them want to answer positively later on because they answered yes in the beginning, or you cannot ask if they like one place over another because you need to ask if they have been to that place first) Questions can be interpreted differently depending on culture

Advantages of questionnaires (self-administered surveys)

respondent does not need to share information with interviewer (prevents socially desirable effect & other biases because the reader isnt saying the question in a particular way, people are just reading it how they would read it not with any other nonverbal cues by interviewer) participants can go at their own pace (they can take the time they need to process the info)-can also be disadvantage because it takes longer for data collection

representative sample

sampling that accurately reflect the population from which it was drawn (this is ideal to use)

rival explanations

these treaten validity of experiments

social desirability response

this can happen due to the presence of an interviewer where the participant to respond with answers he or she believes the interviewer will perceive as favorable

Ratio Level Measurement

this type of meausrement level offers the most power statistically

pretesting

trial of a survey or questionnaire with a small group of participants who are similar to those individuals who constitute the population before data collection actually begins

Likert-type scales

typically 5- or 7- point scale (typically with strongly agree, agree, neutral, disagree, strongly disagree: often, very often, occassionally, never) -sometimes people will give and even number so they force people to choose a side whereas if you do odd then people may choose the neutral number [number in middle] too often

Semantic differential scales (response sets for closed questions)

using bipolar adjectives to anchor a scale Ex. Clean : : : : : Dirty Fair : : : : : Bias

What is an experiment?

when variables are manipulated or introduced by experimenters for the purpose of establishing causal relationships

socially desirability effect

will give responses that they believe is more socially acceptatble or desirable -like a person not wanting to admit that they are burded by their boyfriend/girlfriend's illness just because it makes them feel bad that they feel that way and that their boyfriend/girlfriend should deserve all the sympathy

Rely on more than one item to measure (most) variables

you can use a scale of introversion (principles of what introversion includes, like do you blush easily, at parties do you tend to be a wallflower, etc.) better ask several questions to make sure they apply to a variable -consider including check questions

Face Validity

does the measure "seem" sound by looking at the content? -if it looks like a duck, sounds like a duck, then it's a duck -do the questions you are asking ask the proper questions towards what you are trying to find. Can't say I want to measure happiness and ask questions about sadness

face-to-face surveys (advantages)

effective for building rapport (interviewer can give encouraging nonverbals, the participant will be more thoughtful or more drawn to answer a question because the person is right there waiting for the answer) allows probing

sampling error

the degree to which a sample differs from the population on some measure (larger the sample, the less the errors)

external invalidity

the degress to which experimental results may not be generalized to other similar circumstances

maturation

the idea that people are continually CHANGING whether they are in an experiment or not -the particpants mature and we do not know if it is the experiement or the maturity that has changed the participant (it is hard to know if the experiement caused changes or if the person would have changed anyway because time has passed) -something to consider in longitudinal studies

self-administered survey

the most common survey type where individuals read and select a response on their own

internal invalidity

the presence of contamination that prevents the experimenter from concluding that a study's experimental variable is responsible for the observed effects

important considerations in questionnaire design

1. Ask "good" questions 2. Rely on more than one item to measure (most) variables 3. Consider the order of items 4. Give clear instructions and use an easy-to-read format

three types of experiments found in the communication literature

1. How does communication affect people's attitudes, beliefs, or behaviors? 2. How is communication affected by some phenomenon or factor? 3. How does some aspect of communication affect another aspect of communication?

questionnaires

Best survey type for complex or long response sets

Predictive Validity

Does the measure predict what the researcher intended to?

History

During a month-long experiment designed to test the effects of a PSA about sunscreen on a group of high school students, the school principal publicly announces that she has skin cancer. This is an example of what type of threat to validity?

cause, effect, independent, dependent

Experiments examine _______-and-_______ relationships between i_________ variables and d_______ variables

Nominal, Ordinal, Interval, Ratio

Four levels of Measurement 1.) N_________ level measurement (name) 2.) O_______ level of measurement (rank order matters) 3.) I______ level measurement (equal spaces with no true zero) 4.) R_______ level of measurement (equal spaces with true zero)

True experimental design

Pretest-postest Control Group Design Posttest-only control group design Solomon four-group design

False (want to go for saturation of data-answers are becoming redundant)

T or F: Twenty is the minimum number of participants for an acceptable interview study

rival explanations

There are often _______ ______ to account for chances found in a study. Ex. • Cannot assume A caused B, instead we need to show that A occurred before B, that there is no other explanation (rival explanation for the changes), and need to show the relationship between the two  Ex. That ice cream sales go up and drowning goes up...does ice cream cause drowing... (need to consider that during summer ice cream sales will go up because of heat and drowning occurs more because of more swimming in the summer)

False

True or False? Survey answer choices should be mutually exclusive, which means that all possible answer choices should be listed

Face; Predicative

Two ways of assessing validity... F________ validity & P______ validity

communication, independent, dependent

Types of Experiements found in Comm. Literature 3. How does some aspect of communication affect another aspect of communication (in this case ________ is in the _______ variable and the ________ variable) Ex. How telling a truth or a lie affect nonverbal communication

communication, independent

Types of Experiments found in Comm. Literature 1. How does communication affect people's attitudes, believes, or behaviors (in this case _________ is the ______ variable) ex.?

communication, dependent

Types of Experiments found in Comm. Literature 2. How is communication affected by some phenomenon or facor? (in this case _______ is the ________ variable) ex.?

cause, effect

We are about to establish _______ and ______ with experiments

three main symbols used in notations for experimental designs

X: experimental treatment O: Observation of the dependent variable R: Randomization

pretest-posttest control group design; true

_____ experimental design that includes a randomly selected experimental and control groups, each of which is given a pretest and a postest R O X O ? ? ? ? R O O ? ? ? Notes Will want a pretest so that you can test to see if the groups are comparable or similar

one-group pretest-posttest design; pre

______ experimental design in which a pretest is added to a one-shot case study O X O -so from this we can now match pre-exisitng attitudes and the attitudes effected after, but there are other factors that could still skew our results

static-group comparison; pre

_______ experimental design in which a control group is used, but the groups are not know to be comparable X O O

one-shot case study, pre

________ experimental design in which an experimental treatment is introduced and researchers look at effects on some output (dependent variable) X O Ex. Showing anti-smoking PSA to a group of teens for a two-week period and then administering a survey about attitudes about smoking X- anti-smoking PSA O- attitudes about smoking However; we cannot say that anti-smoking PSA ads caused negative attitudes about smoking, because those attitudes may have already been there (we have no baseline of the independent variable), those who don't have negative attitudes could have dropped out, or people answering might not want to say they have positive attitudes about smoking because they fear their answers might be judged

check questions

a check on test-taking behavior that involves asking the same question twice at different locations in the questionnaire, usually once positively worded and once negatively worded -ex. asking if someone will take illegal substance to work or how their feelings of theft are in bby survey

reliability

a measurement's degree of stability, trustworthiness, and DEPENDABILITY

simple random sampling

a method by which researchers select participants or events such that each event in the population has an equal chance of being selected

stratified random sampling

a method by which researchers select participants or events to represent known proportions of characteristics in the population after the population characterisitcs are identified, a random sample of a given size is drawn from each homogeneous subgroup consistent with the population proportions

Consider the order of items

best to begin with the most interesting items, rather than demographics (because you want people to take the survey)

example survey topics from communication research

body images and mother-daughter communication: perceptions about people who stutter

control group

group in which assigned participants receive no treatment or stimuli or receive the standard form of the treatment

measurement

involves DECISIONS ABOUT HOW MUCH PRECISION THE RESEARCHER WANTS [ex. income, age, levels of (dis)agreement]

recall cue

statement preceding a survey or questionnaire designed to direct participants to recall episodes or past interactions in which they participated

questionnaires

survey forms in which inidividuals respond to items they have read

X

symbol for experimental treatment (independent variable; what the experimenter is trying to do to create a cause)

O

symbol for observation of the dependent variable (dependent variable; the way we observe something)

R

symbol for randomization (has to do with multiple groups and that each group has a chance to be assigned to a certain group-the treatment group or placebo group

Interval Example

temperature on Fahrenheit or Celsius scale

history

that people are not responding to the actual stimulus but are reacting to a current event (something in the news or going on in daily life) ex. political candidates can lose so much support in the polls if a sex scandal (current event) comes out about them -even though this has nothing to do with the political campaign, it effects the campaign more than effective speeches (actual stimulus)