Psych Research Methods Exam 2

Measures of Central Tendency

Mean, median, mode

Measures of Variability

Variance, standard deviation, range

Linear Regression

• line of best fit; y=mx+b

• positive slope = direct/positive relationship

• negative slope = indirect/negative relationship

Pearson Product Moment Correlation (r)

• sign of r indicates if relationship is positive or negative

• r²= proportion of variance in Y that is due to X (relationship strength)

• if r does not equal 1.0 only a portion of variability is related to the change in the other variable = imperfect predictions & error variance

• sample statistic

rho

population parameter for r

Magnitude (Measurement Scale Properties)

ability to order measurements

Equal Unit Size (Measurement Scale Properties)

Change of 1 is equal size across the scale

Absolute Zero (Measurement Scale Properties)

Score of zero means there is none of what is being measured

multiple correlations and regressions are 

nonlinear

Spurious Relationships

2 variables are correlated by chance

Ratio Measurement Scale

• Magnitude 

• Equal Unit Size 

• Absolute Zero 

Interval Measurement Scale

• Magnitude

• Equal Unit Size

Ordinal Measurement Scale

• Magnitude (rankings)

Nominal Measurement Scale

Classify groups that don’t have a magnitude relationship (category names)

Reliability

• repeatability/consistency of a measure

• extent that measurement represents the true value of what’s being measured

Validity

initial measurement’s ability to serve as a good operational definition of construct

2 Components to Measurement

1. True Score

2. Measurement error

Assessing Reliability 

• Test-Retest

• Internal Consistency (split half reliability - compute scores of 2 halves and compute correlation)

• Alternate & Parallel Forms Reliability (test equivalence of 2 different forms of the same test)

• Interobserver Reliability 

Construct

• idea developed to permit categorization & description of directly observable behavior

• psychological constructs are not directly observable and must be operationally defined

Relationship between Reliability & Validity

• Measurement can produce measure reliably but it could be incorrect (lack validity)

• Measurement can’t be the true value (lacking validity) if a measurement instrument is not reliable

Criterion Validity 

extent to which a measuring instrument accurately predicts behavior 

• Predictive Validity: give test and then give same test later on in time; compare scores between past and future in terms of prediction accuracy 

• Concurrent Validity: give test and immediately evaluate 

Evidence for Construct Validity

• show measure is new construct unrelated to existing (discriminant validity)

• show measure is related to other measures of same construct (convergent validity)

• develop theory of construct

Reactivity

Subjects behavior changes due to awareness of observer

FIXES:

• allow subject to acclimate to environment

• conceal observer

• allow for anonymous responses

Nonreactive/Unobtrusive Measures

indirect observation of behavior (may or may not involve researcher intervention)

Independent Groups T-Test

determine if means are significantly different between groups (RATIO or INTERVAL scales)

Chi-Square Test of Independence (x²)

if equal to 0 there is no difference in proportions between groups 

larger values indicate a greater difference which would be significant

NOMINAL data  

Effects due to participant characteristics

Demand Characteristics: cues or information about experiment that could guide behavior

Combat with placebo, deception, or “filler” questions

Experimenter Bias and/or Expectancy

experimenter acts in way that causes behavioral outcome they expect

Combat by training experimenter, using paper instructions, automation, or blind/double-blind experiment

Types of Dependent Variable Measures 

• Behavioral 

• Physiological (noninvasive, invasive)

• Self-Report (*could have reliability/validity issues*)

Be cognizant of Ceiling Effect (too easy) or Floor Effect (too difficult)

Naturalistic Observation

Observation without intervention

Advantage = external validity

Disadvantage = no control

Participation Observation

researcher is apart of observation group

(participating can influence behaviors)

Concealment

disguised vs. undiguised observation

Testimonials

systemic observation

study 1+ specific behavior in certain setting