Study Guides (390,000)

CA (150,000)

Carleton (5,000)

PSYC (800)

PSYC 3000 (20)

Bruce Hutcheon (10)

Midterm

School

Carleton UniversityDepartment

PsychologyCourse Code

PSYC 3000Professor

Bruce HutcheonStudy Guide

MidtermThis

**preview**shows page 1. to view the full**4 pages of the document.**Run-series plots: horizontal axis=time/sequence, vertical= value of each observation (detect

sudden changes, gradual change/trend, patterns, problems, changes in variance, plotted same

sequence as gathered, shows all data)

Scatter plots: Show all data points, 2 variables on 2 axes, reveals unusual relationships btw

variables, detect changes in experimental conditions

Histograms: not show all data points, just summary. Atleast interval data on x-axis, y axis starts

at 0. Distributions of values. Identify non-normal distributions.

Numeric tools: Symmetric and no severe outliers= mean and sd. Not symmetric= median and

semi-interquartile range (resistant to outliers and can be used for ordinal).

Mean= average, interval/ratio, easy but affected by outliers,

Median=ordinal but not for ANOVA,

mode=nominal, most frequent, affected by sampling fluctuations.

Variance= Avg. squared deviation from the mean,

Standard Deviation=square root of variance (affected by extreme scores but easy for

ANOVA),

Interquartile Range= Q3-Q1/2 (resistant to outliers but not sampling fluctuations).

Nominal data= probability and odds.

Boxplots=summary of data, 50% data in box, 99% inside fences, dispersion.symmetry/outliers,

compare distributions/ medians. Central part in box, outer part=whiskers.

Boxplots over histograms= uniform description, outliers, compare many distributions but

don’t show modality like histograms or normality.

Error bars= can be associated with any measure of dispersion (SD,SE, quantiles, confidence

intervals), allow comparison of btw. and within variation, comparing different distributions.

Conduct visual t-test.

Hypothesis testing: Pick statistic, take sample and measure said stat in sample, make a

hypothesis, find out how stat should vary across sample if hyp. true (construct sampling

distribution), compare sample stat to sampling distribution, reject if sample stat in tails of

sampling distribution.

Skew: 0=symmetric, positive= pos. skew, neg.= neg. skew. Skew from symmetric pop doesn’t

have to be zero. Find out how much it could jump around if pop. symmetric. Sampling

distribution= Distribution of values if draw bunch of random samples from specified

distribution.

In normal distribution, 5% of values more than 1.96 SD from the mean (mean should be 0

if normal). If sample stat in outer 5% of samp.dist. then reject, too rare to occur by accident

(significant). Therefore conclude population can’t be symmetric if sample stat significantly

skewed.

If a population is symmetric then sampling distribution for the skew is normal with mean

of 0. Sample skew/SD of samp. Dist. gives number we compare to 1.96 SD.

Standard Error= SD of a sampling distribution.

Hypothesis tests of the mean: Observed mean (sample) – expected mean (hypothesis)/ SE

(z score). If samp. Dist normal then +/- 1.96 SD gives outer 5%. Often samp. Dist of mean is

normal but not always. If a population is normal then so is the corresponding sampling

distribution of the mean so can use hypothesized properties of pop. to find mean and SE of

sampling distribution. Mean of samp dist for a stat is called expected value <M>. SE of samp

###### You're Reading a Preview

Unlock to view full version