Chapters 12, 13, 14, & 15

Plus de détails

Stat Terms

par Kyle Olson

chapter10

par Miralda Medina

SEBoK Table of Contents

par José Arévalo

Methodology

par Pablo López

Chapters 12, 13, 14, & 15

Davis, 2005

(14) Analysis of Variance and Regression Techniques

the formulas outlined in this chapter are going to important when analyzing data

Nonparametric ANOVA

Kruskal-Wallis ANOVA

Wilcoxon rank-sum test

analyze of the difference between:

an ind var of a categorical nature

a dep var that doesn't contain interval properities

Linear Models

Covariance

long and complicated

ANCOVA

2. fully understand the differences

1. remove extraneous variance - increase precision

Linear regression

multiple linear regression

used to estimate and to look at the intercorrelation

essentially the same as simple linear regression

>1 independant variable

Simple linear regression

analyze the change of the dep var by using info on 1 or more ind var

used to estimate to predict

Two-way analysis of variance (ANOVA)

more complex than the one-way analysis

interaction effect

2 classification variables and 1 dependant variable

One-way analysis (ANOVA)

dependant variable is maintained

ANOVA = analysis of variance

mathematical formulas distinguish: variance/ relationship

mixed effects model

fixed effects model vs. random effects model

The Nature of Variance Decomposition

finding relationships is crucial

the researcher tries to control the error variance

breaking down the variance into explained and error will help the study

variance - the spread of the results

(15) Advanced Multivariate Analysis

it should follow the 3 rules in order for the study to be successful

choosing a technique to fit our study will be important

Comment on the Techniques

rules for choosing a technique:

3. determine the final use of the outcome info.

2. know the data characteristics and requirements

1. acknowledge the assumptions

Analysis of Interdependency

Multidimensional scaling

involves the spatial rep of relations among the perceptions and preferences of individuals

Cluster analysis

attempts to subdivid or partition a set of different objects, variables, or both into similar groups

dimension-free classification procedure

Factor analysis

only performed for very large samples

follows a series of steps

generate a correlation matrix

4. theoretical considerations

3. the indeterminacy problem

2. the # of factors to keep

1. estimates of communality

uses a linear approach

Analysis of Dependency

Covariance structure analysis

researcher specifies the relationships between theoretical constructs and observables

causal modeling

Conjoint analysis

similar to variance

concerned with the joint effect of a combin or rank-ordered ind var on a dep var

stat technique

Multiple discriminant analysis

examines the relationship between:

& a set of exploratory or ind var

a nominally scaled dep var

stat technique

Multivariate anlaysis of variance

a more multivariate nrmal dist is used

uses general linear model & follows similar assumptions of ANOVA

2. takes into consideration that the dep var are correlation somehow

1. provides more total info

examines the effect of a treatment on 2 or more dep var

MANOVA

Selection of a Multivariate Technique

depends on the study and design

Introduction

allow the researcher to examine interactions, dependencies, and commonalities

2. fitting algebraic models to situations

1. means of analyzing phenomena

complex design structures emerged with the innovation of technology

(13) Basic Analytical Methods

these statistical analysis of the data will be important in order for us to interpret the results

Basic Methods of Assessing Differences

t-test for differences in means

mean differences used on interval-scale measures

Z-test for differences in proportions

divided by the standard error of the mean

illustrated by the normal dist curve

Chi-square (x^2) test

test of hypothesis for group differences

Basic Methods of Assessing Association

Pearson's r

summary stat of the strength of the relationship

Spearman rank correlation

the measurement must be ranked for each variable

Kendall tau

Contingency correlation

stronger as it approaches 1

C range 0 to 1

phi coefficient

range -1 to 1

calcs the strength of the relationship

Crosstabulation

shows whether there is a relationship

need the ability to calc percentages

uses categorical data

cross-break

Exploratory Data Analysis

visual displays

scatterplots

box plots

histograms

bar charts

frequency tables

graphical displays

Classification by Purpose

findings could be analyzed with inferential stats and a statistical test of hypothesis

can a discovery be made?

SBR can generalize results where NSBR can not

(12) Planning for Data Analysis

Feel Good Store (project)

SBR would be appropriate because we would want a generalization for the population

choosing analytical software or online research will come down to which is the cheapest and easiest to use

planning is extremely important

with improper planning we could be

developing the wrong conclusion

wasting time and resources

Basic Analytical Framework for Business Research

Scientific business research (SBR)

hypothesis testing

causal

Nonscientific business research (NSBR)

discovery

description/ prediction

exploratory

The Preanalytical Process

Step 7: Tabulation

can be used for crosstabulation and statistical purposes

the results are seperated by how they responded

Step 6: Preanalytical computer check

data is prep'd for computer analysis

Step 5: Data structure generation

data structure is finalized for storage

Step 4: Error check

the coded data is verified before it's transfered to storage

Step 3: Data coding

transfering data into numeric codes

Step 2: Variable development

specifying the variables from the data

Step 1: Data editing

clarity, readability, consistency, and completeness to a set of data

Analytical Software vs Online Research Supplier

Online Research Suppliers

outsourcing the research function

benefits > costs?

expected to grow because of new technologies

Analytical Software

Excel

Microsoft

most popular for it's statistical and analytical tools

SAS

widely used for data anlysis and report writing

SPSS

offers a full line of data analysis products

most widely used

improvements are expected

"point and shoot"

voice recognition

custom software for a specific purpose

availability and familiarity

Planning Issues

In the planning process, the researcher should also prepare for any potential problems

planning is important because it will help the validity of the research