Skip to main content
Search form
  • 00:00

    [MUSIC PLAYING][RESEARCH METHODS tutorial][An Introduction to MANOVA & MANCOVA for Marketing Research][Saurabh Bhattacharya, PhD, Lecturer in Marketing,Newcastle University]

  • 00:14

    SAURABH BHATTACHARYA: I am Saurabh Bhattacharya.I am a Lecturer in Marketing in Newcastle University BusinessSchool.It's a Russell Group university in the UK.I primarily teach research methodsand marketing research at the undergraduate and postgraduatelevel.My research focus is interdisciplinary, and--means that I do work both in marketing, as wellas in international business and management.

  • 00:38

    SAURABH BHATTACHARYA [continued]: In marketing, my primary focus is celebrity endorsementand online reviews.In international business, my primary focusis on acquisitions, TMT, or operation law,and emerging market forms.The method which I will be discussing todayis known as MANOVA, which is MultivariateANalysis Of VAriance, [MultivariateANalysis Of VAriance] and a bit about its extension, whichis Multivariate ANalysis of COVAriance.

  • 01:05

    SAURABH BHATTACHARYA [continued]: So if you know MANOVA, you are very comfortable with MANCOVA.[How did you become interested in these methods?]A lot of my marketing-based researchis primarily related to celebrity endorsements,and there is no escaping experimental design these days.

  • 01:27

    SAURABH BHATTACHARYA [continued]: If you are working--if you are doing a consultancy project for industry,if you are doing academic piece of a publishable paper,whatever you've been-- in either of these, in both,in academics, as well as in consultancy work,you require the knowledge of experimental design.Experimental design primarily workson different means in two ways.

  • 01:50

    SAURABH BHATTACHARYA [continued]: One is, you need to have a regression framework.Another is the ANOVA framework.In the ANOVA framework, researchers these daysprimarily try to find the effect of an independent variableor multiple number of dependent variables.So the moment you have multiple number of dependent variables,it is MANOVA or MANCOVA which is of interest to you.

  • 02:13

    SAURABH BHATTACHARYA [continued]: So my primary interest into this fieldcame from picking up projects which are primarilyexperimental design-based.[What kinds of research questions can you answer usingthese methods?]To answer this, we need to just-- we needto understand a few things.

  • 02:34

    SAURABH BHATTACHARYA [continued]: The first one is that, whether you select MANOVA or MANCOVA,the independent-- there are three typesof variables in this method.The first one is a predictor variableor the independent variable.The second one is a moderator, and the third oneis called a control.So if you-- and, obviously, the fourth oneis the outcome variable.

  • 02:54

    SAURABH BHATTACHARYA [continued]: Now, if you are selecting MANOVA,then you will have multiple number of predictor variables,but the independent variable will be categorical in nature.So the research questions or the hypothesiswhich you should have are those whereyou're comparing across groups, mean values across groups.

  • 03:16

    SAURABH BHATTACHARYA [continued]: So for example, if you are comparing whether a Walmart$25 gift card is better than a $25 gift card of Walmartand Sam's Club, and what do you mean by better?

  • 03:38

    SAURABH BHATTACHARYA [continued]: Do you mean higher purchase?Do you mean more loyalty?So if there is only one type of dependent variable, whichis higher purchase, then you use ANOVA.If you have more than one, like better means higher purchaseor more-- and more loyalty, if, since there are two,so you use MANOVA.

  • 03:59

    SAURABH BHATTACHARYA [continued]: ANOVA is when you compare groups for one independent--dependent variable.MANOVA is when you compare groupsacross multiple number of dependent variables.So it's basically comparing means,but how many dependent variables are there?

  • 04:23

    SAURABH BHATTACHARYA [continued]: [What kinds of data can you analyze using these methods?]As regards to independent variables,since you will be doing experimental design,it's primarily stimulus.For example, one group is shown a stimuluswhich is $25 Walmart card--gift card.

  • 04:43

    SAURABH BHATTACHARYA [continued]: Other group is shown--given a card which is $25 Walmartor Sam's Club gift card.So the first group, we will call it as one, and the second groupyou can call it as two or zero.So these type of variables are called categorical variables,so that independent variable, it is primarilycategorical in nature.

  • 05:04

    SAURABH BHATTACHARYA [continued]: The dependent variable has to be continuous.For example, how likely are you to buy this card?It is on a scale of 1 to 7, 1 to 10, 1 to 11,so it will be primarily continuous in nature.So data for ANOVA means you shouldhave categorical variables as an independent variable,and dependent variable data should be continuous in nature.

  • 05:28

    SAURABH BHATTACHARYA [continued]: Now, many a times, your independent variable datacan be continuous, but it is generallysuggested that you convert them into categories.For example, you can convert 0 to 10 as 1, 10 to 20 as 2.So what happens is that you're converting theminto categories, so an independent variable,categorical.

  • 05:50

    SAURABH BHATTACHARYA [continued]: Dependent variable should be continuous.So the data is separate.[When have you used these methods & what did you learn?]We used this method in a couple of our projects.One of the projects was related to test the endorsementeffectiveness of a CEO, founder, vis-a-vis a celebrity.

  • 06:14

    SAURABH BHATTACHARYA [continued]: So groups were shown these three types of ad.One group was shown an advert of CEO, another a founder,and the third one was shown as a celebrity.The first group say--I called them as one.The second group I called them as two, and the third group Icalled them as three.Now, this is our independent variable.

  • 06:35

    SAURABH BHATTACHARYA [continued]: The dependent variable was--one of the dependent variable was overall attitudetowards the ad, which was a continuous variable.Why?Because we measured it to a 1 to 7 scale.A Likert scale.And then, there were other dependent variables also.These dependent variables were conative, cognitive,and affective ads.

  • 06:56

    SAURABH BHATTACHARYA [continued]: So there are four dependent variablesand one independent variable.So the natural process was to use MANOVA.We also had a moderator, but, again, it doesn't-- it means,if--for the moderator, per se, it means--it doesn't mean we have to use MANOVA.We use MANOVA because we'll have multiple dependent variables.

  • 07:18

    SAURABH BHATTACHARYA [continued]: So this was one of the study wherewe would-- been one of the study where we used MANOVA.We again did another study recently,where we compared celebrities over Instagram posts, betterendorsers, then social media influencers.

  • 07:41

    SAURABH BHATTACHARYA [continued]: So one of the groups got celebrity as a stimulus.Another group got SMIs as a stimulus.So celebrities, suppose we called one--SMI, suppose we call two.So the independent variable is categorical.The dependent variable was willingness to buy.So it was incontinuous.

  • 08:02

    SAURABH BHATTACHARYA [continued]: So if you have one independent variable, we will do ANOVA.But we had multiple number of dependent variables,so we did a MANOVA.So most of the means research, whereI use experimental design, where Ihave multiple number of dependent variables,and a categorical variable as the independent variable,I use MANOVA.

  • 08:24

    SAURABH BHATTACHARYA [continued]: It's better these days.It is suggested that with-- along with MANOVA,you should also have controls, for example,product involvement, for example,attitude towards the brand.These act as controls.So when you are using MANOVA with controls,it just shifts or changes to MANCOVA.

  • 08:45

    SAURABH BHATTACHARYA [continued]: So it's the same thing, only with control variables.[What tools & resources would you recommend for those lookingto use these methods?]Tools, it means, if you want, meanstools are pretty standard here.It means you use SPSS or SAS.And resources, again, Andy Field's Discovering StatisticsUsing SPSS, or SAS, or he also has a book on R--you-- means any of these, because he extensivelydiscusses in these books of these topics.

  • 09:18

    SAURABH BHATTACHARYA [continued]: An important thing that that means here these days ishappening is that people-- means are actively using--researchers are actively using somethingcalled moderated mediation analysis in a MANOVA framework.The other resource which researchersshould look for, if they're mixing up MANOVA,MANCOVA, with--and also using in moderated mediation analysis is to--the book is titled as Introductionto Moderation, Mediation and Conditional Process Analysis.

  • 09:50

    SAURABH BHATTACHARYA [continued]: It is by Andrew F. Hayes.So that book is also very good.At the very basic level, if you are doing MANOVA and MANCOVA,then Andrew--means Andy Field's book is sufficient,and both his videos, as well as the way he runs those analysis,I think they are available with the SAGE video collection also.So these resources will definitely help.

  • 10:26



Saurabh Bhattacharya, PhD, Lecturer in Marketing at Newcastle University, discusses the use of multivariate analysis of variance (MANOVA) and multivariate analysis of covariance (MANCOVA) for marketing research.

Looks like you do not have access to this content.

An Introduction to MANOVA & MANCOVA for Marketing Research

Saurabh Bhattacharya, PhD, Lecturer in Marketing at Newcastle University, discusses the use of multivariate analysis of variance (MANOVA) and multivariate analysis of covariance (MANCOVA) for marketing research.

Copy and paste the following HTML into your website