Week 7 - Hypothesis Testing PDF

Preview

Summary

Hypothesis Testing...

Description

Research in Marketing Week 7 – Hypothesis Testing LECTURE 7 – HYPOTHESIS TESTING

PART I – DESCRIPTIVES AND INFERENTIAL STATISTICS Descriptive Statistics •Summarise and describe information (e.g. tables to summarise data) Measures of Central Tendency (e.g. Mean) Measures of Dispersion (e.g. SD)

Descriptive Statistics: A practice in SPSSDescribe the age profile of the sample: Use the Santa Fe data set.We will describe the variable x34: Age Frequency table –One-Way Tabulation •A frequency table shows responses for a single variable.•This is called: One-way Tabulation

Cross-Tabulation• Simultaneously compares two or more variables in the study.•Primarily used on nominal and ordinal (categorical) scales

Cross Tabulation: A practice in SPSS: Interpretation 1.The first count shown in the table shows the number of males and females who cannot recall the ads (males=188; females=82) 2.The second count shown in the table shows the number of females and males who cannot recall the ads (males=76; females=59) 3.The third count shown in the table shows presents the total number of males and females who respondent to the X31-Ad recall variable. (males=264; females=141) 4.The row percentage shows that out of the 270 participants who do not recall the ad, the 188 males do not recall the add (188/270=69. 6)5.The column percentage shows that of the 264 males participants the 188 males do not recall the add (188/264=71.2%) 6.The cell percentage shows that out of the total responses (including males and females) the 46.4% are males and cannot recall the advertisement (SPPSS divides 188/405= 46.4%)

Inferential Statistics •Are used to study sample characteristics and make inferences about population parameters. Statistics Inferences

-

generalising from the sample results to a target population

Research in Marketing Week 7 – Hypothesis Testing

-

procedures which allow you to calculate how probable that your results, given your sample size, could have been obtained by chance.

Confidence Level -precision of your estimates of the population as the percentage that is within a certain range or margin of error - 95% confidence: if your sample was selected 100 times, at least 95 of those samples would be certain to represent the population.

PART II - CORRELATIONAL AND EXPERIMENTAL RESEARCH

Definition: theory A theory is an explanation or set of principles that is well substantiated by repeated testing and explains a broad phenomenon. Definition: hypothesis A hypothesis is a proposed explanation for a fairly narrow phenomenon or set of observations. It is not a guess! It is an informed, theory-driven attempt to explain what has been observed.

To test hypotheses we need to move from the conceptual domain to the observable domain Hypothesis operationalisation -develop predictions. To test hypotheses we need to measure variables To test hypotheses we conduct: •Correlational Research •Experimental (Causal) Research

Correlational Research Researcher observes what naturally goes on in the world without directly interferingwith 1.Cross-sectionaldesign take place at a single time 2.Longitudinal design measuring variables repeatedly at different points in time

Experimental (Causal) Research Most scientific questions imply a causal link between variables (dependent and independent variables). An effect should be present when the cause is present. When the cause is absent, the effect should also be absent

Research in Marketing Week 7 – Hypothesis Testing Experimental methods provide a comparison between situations(called treatments or conditions) in which the proposed cause is present or absent

Experimental (Causal) Research. Two methods of data collection: •Between-groups, between-subjects, or independent designs (participants are exposed to only 1 treatment to avoid cofound effect from being exposed to the second treatment) •Within-subject or repeated-measures designs(participants are exposed to both treatments)

Measures in SPSS Summative scores in SPSS Use the Santa Fe Dataset The restaurant perceptions variables include three measures related to satisfaction. They are variables:X22, X23 and X24.You are asked to create the variable'Restaurant Perceptions'by combining the 3 measures: X22, X23 and X24.

PART III – HYPOTHESIS TESTING

Statistical Models Testing hypotheses involves building statistical models of a phenomenon of interest

Model Fit The degree to which a statistical model represent the data collected (the observable data) is know as the fit of the modelOutcomei= (model) + error i

Importanceof forming and testing hypotheses: Mathematical difference -means that numbers are not the same; however, this does not necessarily mean that the difference is important or significant Statistical difference -is a difference large enough that it is not likely to have occurred due to chance or sampling error

Managerial implications -managers must be able to distinguish between mathematical differences and statistically significant differences in order to make accurate and informed decisions based on data analysis

Research in Marketing Week 7 – Hypothesis Testing Hypothesis Testing Aims to test the likelihood of a relationship occurringby chance alone if there really was no difference in the population from which the sample was drawn Alternative hypothesis (H1): an effect is present, there is a relationship Null Hypothesis (Ho): the opposite of the alternative hypothesis -the effect is absent, there is no relationship

Hypothesis Te s t i n g : A n exa m p l e Is frequency of purchase related to the gender? Alternative hypothesis (H1): The frequency of purchase and gender are significantly related Null Hypothesis (Ho): The frequency of purchase and gender are NOT significantly related Rejecting nullhypothesismeansthatweacceptthealternative

Type of Errors Type I error (falsely rejecting a null hypothesis) Type II error (falsely accepting a null hypothesis)

Hypothesistesting Process (1)1.Set up a null hypothesis (suggestingnosignificance) 2.Decide on the acceptable level of statistical significance (benchmark) 3.Use a statistical test 4.If benchmark attained, reject null hypothesis; If not attained, accept it.

Research in Marketing Week 7 – Hypothesis Testing

generally p...