Oefeningen Marktonderzoek:
Inhoud
Marktonderzoek: Oefeningensessie 1..................................................................................................... 2
Oefening 1: SOUP (One-way ANOVA): ................................................................................................ 2
Oefening 2: FRAGRANCE (ANCOVA): ................................................................................................... 5
Oefening 3: CONTEXT (2x2x2 Factorial Design): ............................................................................... 10
Oefening 4: COMPARATIVE (Within-subjects Desing, Repeated Measures): ................................... 16
Marktonderzoek: Oefeningensessie 2................................................................................................... 20
Oefening 1: Pleasure and planning (Factoranalyse):......................................................................... 20
Oefening 2: Pleasure and planning (Chronbach’s Alpha): ................................................................ 33
Oefening 3: Work survey (Factor anlysis & Chronbach’s Alpha): ..................................................... 36
Oefening 4: Holiday (Factor Analysis & Chronbach’s Alpha): ........................................................... 42
Oefening 5: SAQ (Factor Analysis & Chronbach’s Alpha):................................................................. 46
Marktonderzoek: Oefeningensessie 3 ................................................................................................... 52
Oefening 1: Pizza (lineaire regressie): ............................................................................................... 52
Pizza: .................................................................................................................................................. 53
Oefening 2: Record (Lineaire regressie): ........................................................................................... 63
Oefening 3: Pizza incl gender (Moderation):..................................................................................... 68
Oefening 4: Exam anxiety (Mediation):............................................................................................. 72
1
,Marktonderzoek: Oefeningensessie 1
Experimental design:
→ When to use which statistical test?
(*) or OLS Regression if a certain assumption is fulfilled (e.g. causality)
(**) once samples are divided over different ‘variables’
=> When the same respondents are present in multiple samples: repeated measures ANOVA, if not
then ‘regular’ factorial ANOVA
Analysis of variance (ANOVA)
→ zie hoofdstuk 4 – Analysis of variance (p. 71 – 136): handbook Marketing Research with SPSS
→ zie handboek Marktonderzoek p. 269- 286
Oefening 1: SOUP (One-way ANOVA):
What is ANOVA?
- Experiment: creation of different groups
- ANOVA (Analyis Of Variance): test of differences between these groups extension of t-test
o allows joint test of >2 groups
o allows joint test of multiple independent variables
▪ factorial ANOVA
- Special type of regression
Anova is altjid voor meer dan 2 groepen !
Research question:
- Introduction of new soup
o 3 types of promo: display, in-shop tasting, decorated shelf
o Each promo type is tested in 5 shops (total of 15 shops), with all the shops as identical
as possible
- Independent variable: promotion
- Dependent variable: one-month sales data (=metrische variabelen!)
- RQ: “What is the effect of the different promo types on sales? What is the most successful
promo type?”
Statistical test? → One-way ANOVA (zie tabel)
1) Number of samples? 3 steekproeven (soorten promo)
2) Independent or dependent samples? Independent = “factor”
3) Level of measurement of the dependent variable(s)? at interval/ratio level (VK cijfers)
2
,3
,Output: nummers → volgorde waarin je de tabellen moet interpreteren (ook op ex!)
P<0,05: kans zo klein dat het toeval
is → significante waarde
0,05 = 5% kans dat de gevonden
waarde toeval is
Zijn allemaal verschillende waarden
bij ‘mean’ kan effect zijn
Reporting main effect:
- Reporting should be short: only what is necessary but sufficient! (method, independent
variables, dependent variables, and statistics)
- RQ: What is the effect of the different promo types on sales?
- A One-Way ANOVA (method) indicated that promotion (independent variables: factors)
(Mdisplay = 5.20, SDdisplay = 1.30; Mtasting = 13.80, SDtasting = 2.28; Mdecoration = 7.80,
SDdecoration = 1.92) has a significant effect on sales (dependent variable), F(2,12) = 27.53,
p < .001 (statistics).
F (2,12) : 2 vind je bij 2 df between groups & 12 vind je bij within groeps
Schrijf p nooit als = maar altijd > of <
➔ Next: Which promotion has the highest effect? (post-hoc)
Output:
H0= geen effect : p < 0.05
HA= wel effect: p > 0.05
Gelijke spreiding → Bonferoni
Ongelijke spreiding → Tamhane
• Hier willen we geen sign zien omdat er anders een verschil is in de spreiding
• P > 0,05 : P = 0, 515 dus H0 aanvaarden dus we zitten hier met een gelijke spreiding
= BONFERONI POST HOC
4
, Vb. Display vgl met tasting:
significant verschil
Je kan zien dat tasting een
verschil aan toont maar is dit
positief of negatief? Kijk
hiervoor naar de descriptives
Tasting heeft grootste
gemiddelde in sales
Reporting post-hoc test:
- RQ: What is the most successful promo type?
o Bonferroni post-hoc tests (method) showed that tasting (M = 13.80, SD = 2.28)
significantly leads to more sales than display (M = 5.20, SD = 1.30), p < .001 (statistics),
and decoration (M = 7.80, SD = 1.92), p = .001 (statistics). There is no significant
difference between the display and decoration promo type, p = .148 (statistics).
- When you use a corrective measure (e.g. Tamhane), you have to indicate this briefly:
violation of assumption(s)
o Refer to the outputs where the information can be found e.g. “The assumption of
homogeneity of variance was violated; therefore, the Tamhane’s T2 is reported. (…)”
- When the assumptions are not violated, there is no need for further explanation (unless
added value)
o Optional: outputs in appendix
Oefening 2: FRAGRANCE (ANCOVA):
What is ANCOVA?
- Covariaat: extra variable die een effect kan hebben (bv de leeftijd)
- Including ‘covariates’ = including independent variable(s) with interval/ratio level of
measurement
- To test for differences between group means when we know that an additional variable
affects the outcome variable
- Used to control known extraneous and confounding variables
5
, - Example: the effect of “marketing expenditure” (€) and “product type” (FMCG vs durables)
on “purchase intention”
What is ANCOVA?
ANCOVA is gelijkaardig aan
ANOVA maar met als extra een
covariaat meenemen in je
analyse omdat je denkt dat het
invloed kan hebben op je effect
(bv. hier: grootte van de winkel)
Research question:
- Fragrance and sales
- Data of 30 shops
- Independent variable: fragrance, factor with 3 levels (between-subjects)
1) no fragrance (control condition)
2) subtle fragrance
3) prominent fragrance Each level of fragrance is tested in 10 shops
- Additonal independent variable: store surface (size) in m² (covariate = control var.)
- Dependent variable: weekly sales
- RQ: Does fragrance (indeed) lead to higher sales? What is the ideal dose (subtle or
prominent)?
Statistical test?
1) Number of samples? 3 (condities)
2) Independent or dependent samples? Independent (als ze afhankelijk zouden zijn dan zou je
een winkel hebben met subtiele en dominante geur door elkaar en dan klopt het niet meer)
3) Level of measurement of the dependent variable(s)? interval/ratio level
→Tabel: One-Way ANOVA, maar hier ANCOVA want je neemt een extra onafhankelijke variabele
(covariaat) mee in je analyse
6
Inhoud
Marktonderzoek: Oefeningensessie 1..................................................................................................... 2
Oefening 1: SOUP (One-way ANOVA): ................................................................................................ 2
Oefening 2: FRAGRANCE (ANCOVA): ................................................................................................... 5
Oefening 3: CONTEXT (2x2x2 Factorial Design): ............................................................................... 10
Oefening 4: COMPARATIVE (Within-subjects Desing, Repeated Measures): ................................... 16
Marktonderzoek: Oefeningensessie 2................................................................................................... 20
Oefening 1: Pleasure and planning (Factoranalyse):......................................................................... 20
Oefening 2: Pleasure and planning (Chronbach’s Alpha): ................................................................ 33
Oefening 3: Work survey (Factor anlysis & Chronbach’s Alpha): ..................................................... 36
Oefening 4: Holiday (Factor Analysis & Chronbach’s Alpha): ........................................................... 42
Oefening 5: SAQ (Factor Analysis & Chronbach’s Alpha):................................................................. 46
Marktonderzoek: Oefeningensessie 3 ................................................................................................... 52
Oefening 1: Pizza (lineaire regressie): ............................................................................................... 52
Pizza: .................................................................................................................................................. 53
Oefening 2: Record (Lineaire regressie): ........................................................................................... 63
Oefening 3: Pizza incl gender (Moderation):..................................................................................... 68
Oefening 4: Exam anxiety (Mediation):............................................................................................. 72
1
,Marktonderzoek: Oefeningensessie 1
Experimental design:
→ When to use which statistical test?
(*) or OLS Regression if a certain assumption is fulfilled (e.g. causality)
(**) once samples are divided over different ‘variables’
=> When the same respondents are present in multiple samples: repeated measures ANOVA, if not
then ‘regular’ factorial ANOVA
Analysis of variance (ANOVA)
→ zie hoofdstuk 4 – Analysis of variance (p. 71 – 136): handbook Marketing Research with SPSS
→ zie handboek Marktonderzoek p. 269- 286
Oefening 1: SOUP (One-way ANOVA):
What is ANOVA?
- Experiment: creation of different groups
- ANOVA (Analyis Of Variance): test of differences between these groups extension of t-test
o allows joint test of >2 groups
o allows joint test of multiple independent variables
▪ factorial ANOVA
- Special type of regression
Anova is altjid voor meer dan 2 groepen !
Research question:
- Introduction of new soup
o 3 types of promo: display, in-shop tasting, decorated shelf
o Each promo type is tested in 5 shops (total of 15 shops), with all the shops as identical
as possible
- Independent variable: promotion
- Dependent variable: one-month sales data (=metrische variabelen!)
- RQ: “What is the effect of the different promo types on sales? What is the most successful
promo type?”
Statistical test? → One-way ANOVA (zie tabel)
1) Number of samples? 3 steekproeven (soorten promo)
2) Independent or dependent samples? Independent = “factor”
3) Level of measurement of the dependent variable(s)? at interval/ratio level (VK cijfers)
2
,3
,Output: nummers → volgorde waarin je de tabellen moet interpreteren (ook op ex!)
P<0,05: kans zo klein dat het toeval
is → significante waarde
0,05 = 5% kans dat de gevonden
waarde toeval is
Zijn allemaal verschillende waarden
bij ‘mean’ kan effect zijn
Reporting main effect:
- Reporting should be short: only what is necessary but sufficient! (method, independent
variables, dependent variables, and statistics)
- RQ: What is the effect of the different promo types on sales?
- A One-Way ANOVA (method) indicated that promotion (independent variables: factors)
(Mdisplay = 5.20, SDdisplay = 1.30; Mtasting = 13.80, SDtasting = 2.28; Mdecoration = 7.80,
SDdecoration = 1.92) has a significant effect on sales (dependent variable), F(2,12) = 27.53,
p < .001 (statistics).
F (2,12) : 2 vind je bij 2 df between groups & 12 vind je bij within groeps
Schrijf p nooit als = maar altijd > of <
➔ Next: Which promotion has the highest effect? (post-hoc)
Output:
H0= geen effect : p < 0.05
HA= wel effect: p > 0.05
Gelijke spreiding → Bonferoni
Ongelijke spreiding → Tamhane
• Hier willen we geen sign zien omdat er anders een verschil is in de spreiding
• P > 0,05 : P = 0, 515 dus H0 aanvaarden dus we zitten hier met een gelijke spreiding
= BONFERONI POST HOC
4
, Vb. Display vgl met tasting:
significant verschil
Je kan zien dat tasting een
verschil aan toont maar is dit
positief of negatief? Kijk
hiervoor naar de descriptives
Tasting heeft grootste
gemiddelde in sales
Reporting post-hoc test:
- RQ: What is the most successful promo type?
o Bonferroni post-hoc tests (method) showed that tasting (M = 13.80, SD = 2.28)
significantly leads to more sales than display (M = 5.20, SD = 1.30), p < .001 (statistics),
and decoration (M = 7.80, SD = 1.92), p = .001 (statistics). There is no significant
difference between the display and decoration promo type, p = .148 (statistics).
- When you use a corrective measure (e.g. Tamhane), you have to indicate this briefly:
violation of assumption(s)
o Refer to the outputs where the information can be found e.g. “The assumption of
homogeneity of variance was violated; therefore, the Tamhane’s T2 is reported. (…)”
- When the assumptions are not violated, there is no need for further explanation (unless
added value)
o Optional: outputs in appendix
Oefening 2: FRAGRANCE (ANCOVA):
What is ANCOVA?
- Covariaat: extra variable die een effect kan hebben (bv de leeftijd)
- Including ‘covariates’ = including independent variable(s) with interval/ratio level of
measurement
- To test for differences between group means when we know that an additional variable
affects the outcome variable
- Used to control known extraneous and confounding variables
5
, - Example: the effect of “marketing expenditure” (€) and “product type” (FMCG vs durables)
on “purchase intention”
What is ANCOVA?
ANCOVA is gelijkaardig aan
ANOVA maar met als extra een
covariaat meenemen in je
analyse omdat je denkt dat het
invloed kan hebben op je effect
(bv. hier: grootte van de winkel)
Research question:
- Fragrance and sales
- Data of 30 shops
- Independent variable: fragrance, factor with 3 levels (between-subjects)
1) no fragrance (control condition)
2) subtle fragrance
3) prominent fragrance Each level of fragrance is tested in 10 shops
- Additonal independent variable: store surface (size) in m² (covariate = control var.)
- Dependent variable: weekly sales
- RQ: Does fragrance (indeed) lead to higher sales? What is the ideal dose (subtle or
prominent)?
Statistical test?
1) Number of samples? 3 (condities)
2) Independent or dependent samples? Independent (als ze afhankelijk zouden zijn dan zou je
een winkel hebben met subtiele en dominante geur door elkaar en dan klopt het niet meer)
3) Level of measurement of the dependent variable(s)? interval/ratio level
→Tabel: One-Way ANOVA, maar hier ANCOVA want je neemt een extra onafhankelijke variabele
(covariaat) mee in je analyse
6
