PSYC 2005 Exam 1

statistics

branch of math focusing on organization, analysis, and interpretation of a group of numbers

descriptive statistics

summarize/describe group of numbers from a study

inferential statistics

draw conclusions and to make inferences based on numbers, but goes beyond the data

numeric variable

aka quantitative variable

* can be equal-interval or rank-order

equal-interval variable

numbers stand for approximately equal amounts of what is being measured

ratio scale

equal-interval variable that has an absolute zero point or complete absence of the variable

rank-order variable

numbers stand for relative ranking, also called ordinal variables

Nominal variable

aka categorical variable

\-values are names or categories

discrete variable

has specific values and cannot have values between them

\-days since last appointment

continuous variables

infinite number of values between only two values

* weight, BP

frequency tables

way of visualizing how often a value occurs

\-interval frequency table when a range of values are grouped together

histogram

barlike graph of frequency distribution in which values are plotted along horizontal axis and the height of the bar is each frequency

Skewed distribution

clearly not symmetrical

* right skew is a positive skew (more low values)
* left skew is a negative skew (more high values)

Floor effect

many scores pile up at the low end of the distribution because it is not possible to have a lower score

* opposite is the ceiling effect

Normal curve

bell-shaped frequency distribution that is symmetrical and unimodal

Kurtosis

how much the distribution deviates from Normal curve in terms of whether its curve in the middle is more peaked or flatter

Pseudoscience

uses methods and conclusions that pretend to be scientific but are not

* use anecdotal evidence

Research ethics

set of guidelines designed to protect subjects from risks of participating in research

* reporting up to the researcher

Constructs

we use them as if they exist in fact, but they are tools used to facilitate understanding of human behavior or other internal conditions

* memory, anxiety, intelligence

Reification of a construct

confusing a construct for fact

Nominal Fallacy

mistake the naming of phenomena as an explanation for the phenomena

All-or-None Bias

see a statement as either true or false when it could fall in the middle

Similarity- Uniqueness Paradox

to view things as similar to one another or different when in reality they could be both

Barnum Statement

appear to be insightful comments, but they are true for almost all issues

Evaluative Biases of Languages

language blends the description and evaluation, distorting perceptions of behaviors

Inductive Reasoning

going from particular to general - generates theory from broad observation

Deductive Reasoning

going from general back to particular - specific predictions about future events from theories

Theory

formalized set of concepts that summarizes or organizes observations and inferences

* a tentative explanation that must be testable and falsifiable

\

Parsimonious Theory

most simple and straightforward theory which is preferred

Validity

methodological soundness of research and how applicable it is to real world situations

Inductive Theories

built on a strong empirical base- stays close to the data

* stays away from intervening variables( inferred factors that make connections between observed variables)

Deductive Theory

emphasis on constructs, makes testable predictions based on theory

Functional theory

places equal emphasis on induction and deduction

most theories in psychology

Model

simplified representation of a complex reality

Level of Constraint

degree of systematic control applied in research

careful planning and detailed procedures decrease bias

usually start at low constraint and work toward high

Naturalistic Observation

observing participant in their natural environment, lowest constraint

Case Study

moves single participant into a moderately limited environment and observes their responses

Correlational Research

quantifies relationship strength between two variables

\-when it is between two preexisting groups chosen with a preexisting variable, it is known as differential research

Experimental Research

highest constraint, assign participants randomly to groups under different conditions

Precision versus Relevance Problem

as the research gets more specific, it may get more detached from the real world

Properties of abstract number system

1. identity - each number has a particular meaning
2. magniutde - numbers have inherent order
3. equal intervals - difference between units is the same
4. true zero - zero level of variable is measures

Nominal Scales

lowest level of measurement, only has identity

Ordinal Scales

share identity and magnitude, but how much between intervals is unclear

Interval Scales

distance between scores is same, but a zero score is not true absence of variable

Ratio Scales

highest level of measurement, perfect match to number system

behavioral variable

observable characteristics representing certain aspects of organism’s behavior (most common)

stimulus variable

part of environment to which organism has the potential to react

* can be simple or complex
* researcher can control this

organismic variable

classifiable characteristics of the participant

independent variable

manipulated by the experimenter

* can be random or pre-existing

dependent variable

participant response to manipulated variable

causal relationship

changes in one variable results in changes in another variable, hard to draw conclusions without manipulating the independent variable in high constraint environment

extraneous variable

uncontrolled factors that can control can outcome

researchers try to minimize this to make conclusions more valid

constant

events that researcher prevents from varying

Validity

quality/ precision of a study, procedure, or measure to how well each does what it is supposed to do

\-reduced by extraneous variables

* controls reduce confounding effects

WEIRD

western, educated, industrialized, rich, democratic

Mean

equal-interval, an average of all scores

* skewed measures in nonsymmetrical distributions
* M= sum(x)/N

Mode

used with nominal variables, most common value in a distribution

Median

middle score when ordered from low to high values

\-used with very skewed distributions, with outliers, or small sample sizes

Variance

how spread out scores are, the average of squared SD around the mean

* SD^2 = sum(X-M)^2/N

Standard Deviation

square root of the variance, most common measure of variability

SD = sqrt(sum(X-M)^2/N)

Boxplots

a way to visualize variability

* interquartile range is the middle 50% and the box in boxplot

Z-score

number of standard deviations a score is from the mean

z=(X-M)/SD

Population

entire group of people which is of interest

sample

scores of a particular group of the population

* use to infer things about population as a whole

Population Parameters

actual population value

mean= mu

sd = sigma

* predicted from sample statistics (mean=M, sd=SD)

Hypothesis Testing

procedure for deciding whthewhetherr the outcome of a study supports a particular theory or innovation

\-drawing conclusions about a population through a sample

Statistical Significance

\-usually p< 0.05 or 0.01

result is unlikely if there is no effect or if the same is no different than the comparison distribution

Null hypothesis

situation where there is no effect or difference between comparison distribution and the sample

Cutoff Z score for one-tailed 0.05

\+/- 1.64

Cutoff Z-score for one-tailed 0.01

\+/- 2.33

Cutoff Z-score for two-tailed 0.05

\+/- 1.96

Cutoff Z-score for two-tailed 0.01

\+/- 2.58

Comparison Distribution

represents a population situation where the null hypothesis is true

Language for when failing to reject null

"not statistically significant, results are inconclusive”

Language for rejecting the null

results are statistically significant, research hypothesis is supported

Measurement error

inaccuracy that distorts the scores so that observations dont accurately reflect reality

Response-set bias

tendency to respond in specific ways regardless of the situation or experiences

Social desirability

type of response- set bias where participants respond in a socially accepted manner

Operational Definition

measures abstract constructs, definition of a variable in terms of the procedures used to measure or manipulate it

Convergent Validity

multiple lines of evidence support the same conclusion

Interrater reliability

consistency of ratings between separate researchers

test-retest reliability

consistency in scores over time

internal consistency reliability

several methods used to obtain the same score

* the more observations, the greater the internal consistency

effective range

range where dependent measure is accurate

Scale attenuation effects

limits ability of instrument to make discriminations at the top or the bottom of a scale (creates floor or ceiling effect)

Validity

methodological or conceptual soundness of research

* study could be reliable but not valid, can’t be valid without being reliable

concurrent validity

assessing a measure by correlation with criterion that has already been measured

Distribution of means

distribution of means of samples of a given size from a population

Standard Error

the standard deviation of the sampling distribution of the sample mean

sqrt(sigma^2/N)

Z-Test

hypothesis testing procedure where there is a single sample and the variance is known

* with comparison distribution, the standard deviation is the SE

Margin of Error

distance from the mean to the line which is the maximum likely difference between the sample mean and population mean

* gets more narrow/precise as sample size increases

Decision error

incorrect conclusions in hypothesis testing in relation to real, but unknown, situation

Type 1 error

conclude that the study supports the research hypothesis when the null hypothesis is actually true

* chance this will occur is significance level alpha

Type 2 error

failing to reject the null hypothesis when it is actually false and the research hypothesis is true

* chance this will occur is beta

Effect Size

standardized measure of difference between populations - lack of overlap

* increases with greater difference in means in terms of SD

Cohen’s D

d = (M- mu/SD)

small = +/- .2

medium = +/- .5

large = +/- 0.8

Confidence Interval

range of scores that is (%) likely to include the true population mean

* the greater the confidence, the broader the interval

Confidence Limit

upper and lower cutoff values of a confidence level

* placed around the sample mean

Field research

conducted outside the lab, can be low or high restraint

* “in the field”

naturalistic observation

events as they occur in natural settings