unit 1 methods, stats

hindsight bias

knew it all along, rooting for one result

applied research

practical and specific problems, what method is better

basic research

just to gain understanding, broad

longitudinal

over a long period of time, understand changes over time, but drop out over time, expensive

cross sectional

different ages in one point in time

explain population during that time

look at differences in age groups

look at many variables at once, cheap and fast

but not casual, response rate low due to low sample size

hypothesis

statement of relationship between two variables

dependent variable

results dependent on independent variable

independent variable

variable being manipulated

operational definition

explain how measure a variable

validity

accurate, when measures what researchers wanted to measure

reliable

consistent, can be replicated

sample

group of participants that represents larger population

population

anyone/anything that can be in a sample

random selection

every member has an EQUAL chance of being selected. help increases likelihood sample represents, generalize findings, best with computer random generate

stratified sampling

process to ensure sample represents population 

meta analysis

process of looking at results of studies that measured the same variables

experimenter bias

unconscious to treat a group better to increase chance of confirming their hypothesis

sampling bias

samples that do not accurately represent the entire group

lab experiments

in lab, very controlled

field experiments

in the world, more realistic like social experiments

experiment

carefully controlled show CAUSAL relationship, cause and effect, control or manipulate variables to be more valid, but placebo and confounding variables

confounding variable

any difference between experimental and control that can affect dependent

random assignment

each person has equal chance of being placed in a group

participant relevant conducting

participant can't choose their own group

situation relevant conditioning

need to have same controlling enviro

double blind

researcher blind and participant blind (ex. researchers don’t know which is placebo and which one is real medication)

single blind

participants don’t know what group they are in

demand characteristics

cues to think participants are in certain group to respond in certain way

social desirability

give answers that look good to them, not true answer

participant bias

demand character, social desirability

control group

no independent variable

hawthorne effect

participant change behavior because being evaluated

placebo method

given medication to control but doesn’t have effect in it, just think they took drug

correlation

any variables correlation, how one can predict the other

make logical predictions, NO cause and effect

unknown variable can influence relationship

(ex. alcohol consumption and unemployment)

positive correlation

dots up and up

negative correlation

dots one down one up

no correlation

dots scattered

survey method

if there is relationship ask questions, no independent or dependent variable

get data from people fast, cheaper

response rate can be low, self reported answers are inaccurate

naturalistic observation

just observe, no manipulation, realistic, no cause and effect, no mental processes

case study

one specific person or small group, detailed picture, cannot generalize to large population

correlation coefficient

-1 and 1 strong, 0 weak, show if correlation is strong or week

illusory correlation

perception of relationship when it doesn’t exist

descriptive stats

to describe a set of data

frequency distribution

a table or graph showing the distribution

central tendency

MMM mean median mode

mean

average, add all divide by # of scores

median

central, middle number, good for outliers

mode

most freq, can have more than 1

outliers

very extreme score, leads to skewed distribution

positive skewed distribution

TO THE LEFT, caused by low score

negative skewed distribution

TO THE RIGHT, caused by high score

measures of variability

diversity of distribution, range, variance, standard deviation

standard deviation

distance from the mean (root of variance)

z score

distance of a score from mean, neg and pos

normal distribution curve

2 13 34 34 13 2

bimodal distribution

data distribution with two peaks

inferential stats

can be applied to larger population

sampling error

how much different sample is to population

sampling bias

sampling process that is not representative sample

p value

probability is due to chance, lower value the better (percentage decimal)

statistically significant

p value lower than 0.5 (or 5%)