research and development

basic research

contribues to theory

applied research

if a theory is applicable to solve real world problems

experimental replication

experiment is reliable

parsimony/occam’s razor/KISS

findings should be explained and interpreted in simple and most economical way

true experiment

• ideal format is comparative to others

• research has control in a lab or controlled setting

• group of subjects not being manipulated or experimented on

• control group & experimental groupd

control group

people who do not receive the IV

experimental group

group who does receive the IV

confounding variable

impacts both IV and DV

independent variable

• experimenter manipulates

• experimental varaible

• individual with __ looks like they are receiving experimental treatment

dependent variable

• the outcome data that is being measured in the experiment

• some form of human response

sample bias

uses concept of random sampling + random assignment

random sampling

every member of population has an equal change of being selected for the study

random assignment

randomly assign to experimental or control group

quasi-experimental research

study in which subjects are not randomly assigned to treatment group or control group

stratified sample

persons from subgroups

proportional stratified sampling

sample mimics population

cluster sampling

select sample from naturally existing group and then randomly select from existing cluster

systematic sample

• Nth sampling

• probability technique

• first person is picked randomly and then regular intervals (10th, 15th, etc)

• common in homogeneous populations

representativeness is

critical and important to research

nonprobablity sample

• subjects selected by methods not based in probability theory

• judgment sample

• convenience sample

• quota sample

judgement sample

judgement of the research to choose subjects that are thought to be representative of population

convenience sample

intact existing group used with no random sampling

quota sample

subjects have pre-specified characteristics so that the sample will mimic the same type of characteristics that is assumed to exist in general population by being studied

hypothesis

• researcher’s hunch/idea

• null hypothesis

• experimental hypothesis

• modern form

null hypothesis

no difference between control group and experimental group

experimental hypothesis (HI)

alternative hypothesis

modern form of hypothesis

written in present tense without “significant and without any mention of measurements”

test of significance

• used to determine relevance between control and experimental

• operates on principle of probability (p)

• significant difference

significant difference

• p ≤ .05

• probability is less than 5% that differences occurs by chance

the smaller the p

the more convincing the experiment

level of significance is set at

.05 = confidence level = alpha level

statistical power

• likelihood that a statistical hypothesis test will find a real/true effect if it exists

• greater the power, the more confidence you have regarding the validity of the research

error

• there is still a chance or a probability that the results were caused by chance factors

• type I & type II

type I error

• reject null when it is true

• i.e: conclude that REBT helps lower the amount spent on gambling when it did not

type II error

• accept null when it is false

• i.e: conclude that REBT didn’t help the persons who are gambling too much, money at all when indeed it was helpful

probability of making a type I error

• equal to the level of significance

• p value is .05 then .05 is the probability of making this error

p value exceeds .05

accept the null hypothesis

extraneous variable

• also known as errors

• internal and external validity

internal validity

attempts to answer the question “does the experiment really, truly demonstrate that the DV changes or lack of them were caused by the IV? does the experimental condition or treatment make a difference?”

external validity

attempts to answer the question can this be generalized to other groups of people or other programs

threats to internal validity

• occurs when the researcher cannot control procedures that impact the experiment

• maturation, instrumental, statistical regression, selection of groups, intact groups (groups not being the same at the beginning of the study)

• attrition/experimental morality

• demoralized or experiencing rivalry

instrumentation threat

• threat to internal validity

• in or measurement methods or observer’s judgement changes impact the experiment

maturation threat

• threat to internal validity

• implies that time, rather than the IV, impacts the results (becoming older, a person maturing; subjects/clients becoming too fatigued to benefit from treatment)

• single group time-series design can control this threat

statistical regression to the mean

occurs when extreme scores regress towards the mean or the arithmetic average when a task or test is readministered

longitudinal study/research/trend studies

• a study that goes on for a long time

• developmental research

• following the same group of people over a period of time

attrition or experimental morality

• threat to internal validity

• one group loses more people than another or the final group is not like the original group because people dropped out

demoralized or experiencing rivalry

• threat to internal validity

• control finds out the treatment group is receiving treatment

• true benefits are being hidden

reactive effect

• threat to external validity

• person performs better because individual is being observed

• hawthorne effect

environmental setting

• threat to external validity

• laboratory is not the same as the real world

a study cannot have good external validity

unless it has good internal validity. however good internal validity will not guarantee good external validity

t-test/student t-test

• tests a hypothesis between two normally distributed samples

• appropriate for studies using 30 subjects or more

dependent or correlated t-test

same group is measured on two occasions using a pre- and a post-test

independent sample or uncorrelated t-test

two separate groups are measured on the same/one single occasion and compared to each other

ANOVA

• compare more than two groups

• analysis of variance

• used when different levels of IV are used

• expressed using F value

MANOVA

• used when more than one DV is used

• multivariate analysis of variance

ANCOVA

• analysis of covariance

• used to adjust the groups so that a variable that might correlate with the DV will not throw off the study

• remove difference caused by extraneous variable

extraneous variable

any factor outside of the independent variable that can unintentionally affect the DV

correlational research

• does a relationship between two variables exist

  • If so, what’s the magnitude and direction of the relationship

pearson product moment correlation

• expressed by r

• can go from -1.00 to 0 to +1.00

perfect correlations

-1 and +1

positive correlation/association

between two variables as one grows up and so does the other

negative correlation

• indicated when one variable goes up while another goes down

• inverse correlations/relationships

zero correlation

signifies no relationship

correlation does not imply

causation; does not imply cause and effect

the higher the number

the stronger the correlation

normal/true bell shaped

• gaussian curve

• normal distribution

• curve where the mean, median & mode all fall in the middle of the curve

• mean, median, and mode are the same value

measures of central tendency

mean, median and mode

mode

• most frequently occurring score or category

• always going to be at the top or the high point of the graph for the distribution

positively skew

• the mode (most frequently occurring score, and the median (the middle score) are lower than the mean (average of all scores)

• curve tail goes to the right

bimodal curve

• two high points

• has two maximum areas of concentration

multimodal curves

distributions have more than two modes

median

• cuts distribution in half if you rank order the scores from the highest to lowest

• exact middle of distribution or 50th percentile

mean

• most useful measures of central tendency

• add up scores and divide the total by the number of scores

skewed distribution

• extreme values and curves lean to one side or other if you draw it

• media is the statistic of choice to not be impacted by extremes

negatively skewed

• curved tail goes to the left

• mode and median are higher than mean

frequency polygon

drawing a curve of scores

y-axis

• vertical line to left of the curve

• ordinant

• DVs are placed here

x-axis

• abscissa

• IVs are placed here

histogram

continuous numerical data

standard deviation (SD)

• square root of variance

• spread of scores

variance

measure of dispersion of scores

what percentage of cases fall within ±1 SD of the mean?

68.25%

what percentage of cases fall within ±2 SD of the mean?

95.4%

what percentage of cases fall within ±3 SD of the mean?

99.74%

z scores

• a way to analyze formal, normal distribution

• same as SD

• 2.5 = 2.5 SD above the mean

• -2.5 = 2.5 SD below the mean

T score

tells you how many SD a sample mean is away from the population mean when the population SD is unknown

• mean is 50, with every 10 points landing at SD above or below

  • one SD above mean = 60

  • one SD below mean = 40

T scores simplify things by

eliminating negative numbers that can appear when you use z-scores

stanines scores

• standard nine scores

• divide the distribution into nine equal intervals with a mean of five and a standard deviation of two

• lowest ninth of the distribution to the highest

descriptive statistics

• averages (mean, median, mode); range, vaiance, sd and any other statistical device

• not experimental

nominal scale

• uses numbers to identify or classify

• qualitative

• weak scale only used for identification

ordinal scale

described variables that can be rank ordered

likert scales

categorized as ordinal and interval

interval scale

numbers scaled at equal distances, but there is no real zero point

ratio scale

• true absolute zero point

• highest form of measurement

i.e: Kelvin temperature scale

survey

• simplest approach to research

• conducted by giving a questionnaire or a so-called poll to a sample population

• return rate of 30-50% is typical

ethnographic research

• looks at overll dynamics in cultural or situation

• holistic, inductive, qualitative

• observational research and case studies

qualitative

information, research or analysis focused on the characteristics, meanings, and descriptive qualities of a subject rather than its numerical measurement or quantity

• core characteristics: non-numerical, context-specific, inductive

• i.e.: in-depth interviews, focus groups, content analysis, case studies, ethnography

quantitative

refers to information, research or data that is measured, counted or expressed in numbers and statistics

• core characteristics: numerical format, objective focus, large sample size, measurable variables, deductive, replicable design

• i.e.: surveys & questionnaires, experiments, statistical analysis

inductive reasoning

process where you generalize based on specific observations

deductive reasoning

• top down process

• specific hypothesis or hypotheses are derived from general principles