Eye Tracking data / Applied Statistics for the Social Sciences

Post on 18-Feb-2017

380 views 1 download

Transcript of Eye Tracking data / Applied Statistics for the Social Sciences

EyeTrackingdataClasspresentationforthePhDCourse:AppliedStatisticsfortheSocialSciences

HugoGuyader,PhDCandidateinMarketingDepartmentofManagement&Engineering(IEI)

DivisionofBusinessAdministration(FEK)

2015-12-02

HugoGuyader—EyeTrackingdata

StudyPurpose

2

How can retailers attract consumers’ visual attention and increase sales of eco-friendly products through in-store practices?

HugoGuyader—EyeTrackingdata

Hypotheses

3

H1 Looking at eco-friendly products impacts the green premium. H2 Priming consumers to buy green products will increase the search for green products. H3 An “eco-servicescape” impacts visual attention on eco-friendly products. H4 Displaying Point-of-Purchase (PoP) information about green consumption increases the visual attention on eco-friendly products. H5 Shoppers use green price tags to find the green products. H6 Greenwashing practices reduce visual attention on eco-friendly products.

HugoGuyader—EyeTrackingdata

Experimentaldesign

4

• Eye-tracking participants: 66 students (mean age: 23). • Instructions: to buy coffee and fabric softener. • Mock up store: products (classic, eco-friendly/organic, Fair-Trade),

price tags (signalling green products) and PoP information from ICA Maxi

• Eco-servicescape: various items evoking the countryside or agriculture were displayed around the shelves

• 2 experimental conditions: a control group (46%) and a primed group. The priming was: “The person you do the shopping for is sustainable-oriented and prefers to eat organic food.”

• Post-experiment survey collecting socio-demographics and self-reported “greenness”.

HugoGuyader—EyeTrackingdata

5

HugoGuyader—EyeTrackingdata

6

HugoGuyader—EyeTrackingdata

7

Fullvideo:https://youtu.be/Mm0g8mVHffE

HugoGuyader—EyeTrackingdata

Variables

8

๏ Visual attention data:- Eco-friendly Products: Sum of dwell times on eco-friendly, organic, or fair-trade products.- Green Price Tags: Sum of dwell times on price tags for eco-friendly coffee products. - Greenwashing: Sum of dwell times on each fabric softener with a misleading color packaging.- PoP Information: Sum of dwell times on the information displays about eco- friendly coffee.

- Eco-Servicescape: Sum of dwell times on each element (posters, plants, carpet, and so on).

๏ Decision data:- Green Premium: According to the participant’s product choice (classic, ecological, or fair-trade), we calculated the green premium and transformed it into a categorical variable (median split).- Self-reported “greenness” using Roberts (1995) socially responsible consumer behavior scale (24 items).

HugoGuyader—EyeTrackingdata

9

Analyses5 outliers (pairwise deletion): too small nbr of fixations/saccades ‣ analyses for visual attention data based on 61 subjects

➡ chi-square tests, logistic regression, Kruskal-Wallis (K-W) tests

‣ K-W’s H statistic shows whether the ranked data in between-groups is significantly different

‣ non-parametric test because distribution of eye tracking data violates statistical assumptions of normality

HugoGuyader—EyeTrackingdata

DirectLogisticRegression

10

Assess the relationship between • visual attention on eco-friendly products

- transformed into a ratio variable: the dwell time for eco-friendly products on the dwell time for all products displayed

• green premium - transformed into a dichotomous variable, with a median split between high and low green premiums

hypothesis1

HugoGuyader—EyeTrackingdata

DirectLogisticRegression

11

model statistically significant X2 (1, n = 61) = 61.04, p = .000

R2 = 63.2% (Cox & Snell) and 84.3% (Nagelkerke) 86.9%of cases correctly classified

hypothesis1

participants who looked more at green products were 247 times more likely to pay a green premium

HugoGuyader—EyeTrackingdata

Manipulationcheck

12

“greenness”: median split on 24-items scale (α = .886) comparison between primed & control groups • chi-square test for

independence:

the primed group had a significantly higher greenness than the control group

X2 (1, n = 66) = 3.911, p = .048

hypothesis2

HugoGuyader—EyeTrackingdata

Kruskal-WallisTest

13

difference in visual attention (= dwell time) on eco-friendly products (= “green” products) between the primed and control groups • K-W test statistically significant

H (1) = 17.420, p = .000 • mean rank for visual attention on green products:

40.03 for the primed group (52.5%) 21.03 for the control group (47.5%)

• effect size: small (29.03% of variance accounted for) • the primed group looked longer at green products

hypothesis2

HugoGuyader—EyeTrackingdata

Kruskal-WallisTest

14

difference in visual attention on green products, depending on whether the participants looked “long enough” at the eco-servicescape (= whether they saw it or not) • median split on eco-servicescape at 550 ms • K-W test statistically insignificant

H (1) = .854, p = .355 • mean rank for visual attention on green products:

29.00 for the group (52.5%) that noticed the eco-servicescape 33.21 for the group (47.5%) that didn't

• opposite direction than predicted!

hypothesis3

HugoGuyader—EyeTrackingdata

Kruskal-WallisTest

15

difference in visual attention on green products, depending on whether the participants looked “long enough” at the eco-servicescape (= whether they saw it or not) • cut-off value on eco-servicescape at 385 ms • K-W test statistically insignificant

H (1) = .196, p = .658 • mean rank for visual attention on green products:

30.19 for the group (60.7%) that noticed the eco-servicescape 32.25 for the group (39.3%) that didn’t

• still opposite direction than predicted!

hypothesis3

HugoGuyader—EyeTrackingdata

Kruskal-WallisTest

16

difference in visual attention on green products, depending on whether the participants looked “long enough” at the eco-servicescape (= whether they saw it or not) • cut-off value on eco-servicescape at 782 ms • K-W test statistically insignificant

H (1) = .036, p = .850 • mean rank for visual attention on green products:

31.50 for the group (42.6%) that noticed the eco-servicescape 30.63 for the group (57.4%) that didn't

• same direction than predicted!

hypothesis3

HugoGuyader—EyeTrackingdata

Kruskal-WallisTest

17

difference in visual attention on eco-friendly coffee, depending on whether participants noticed PoP info displays • K-W test statistically significant

H (1) = 5.076, p = .024 • mean rank for visual attention on green coffee:

35.88 for the group (52.5%) that noticed the displays 25.62 for the group (47.5%) that didn't

• effect size: small (8.46% of variance accounted for) • Participants who saw the PoP information display

also looked longer at the eco-friendly coffee products

hypothesis4

HugoGuyader—EyeTrackingdata

Kruskal-WallisTest

18

difference in visual attention on eco-friendly coffee, depending on whether participants noticed green-colored price-tags • cut-off value on green price-tags at 200 ms • K-W test statistically significant

H (1) = 9.862, p = .002 • mean rank for visual attention on green coffee:

35.08 for the group (24.6%) that noticed green price-tags 18.50 for the group (75.4%) that didn't

• effect size: small (16.44% of variance accounted for) • Participants who looked at the green price-tags looked longer at

the eco-friendly coffee products.

hypothesis5

HugoGuyader—EyeTrackingdata

Kruskal-WallisTest

19

difference in visual attention on eco-friendly fabric softener (pink), depending on whether participants looked at the misleading green-colored softener • cut-off value on green price-tags at 666 ms • K-W test statistically significant

H (1) = 5.441, p = .020 • mean rank for visual attention on eco-friendly softeners:

34.43 for the group (29.5%) that looked at the “greenwashed” softeners 22.81 for the group (70.5%) that didn't

• effect size: small (9.06% of variance accounted for) • Participants who looked longer at the “greenwashed” softeners were

more likely to miss the true eco-friendly softener.

hypothesis6

HugoGuyader—EyeTrackingdata

Findings

20

Retailers can influence consumers’ green behavior by influencing their purchase intentions (priming), displaying relevant information (PoP info display), orienting them inside the store (green pricetags), offering a “true” green product assortment (greenwashing).

➡As a consequence of visual attention on eco-friendly products, consumers’ green premium increased.

Greenwashing practices distract consumers from finding the eco-friendly products they look for.

hypotheses supported: 5/6

HugoGuyader—EyeTrackingdata

AlternativeAnalyses

21

• All of these were preliminary analyses… Decisions yet have to be taken.

✦ Multiple Regressions (Pearson or Spearman’s Rho) ✦ T-test & ANOVAs ✦ PLS ✦ Multinomial regression

➡It all depends on what raw data is used to represent the theoretical concept… Some things work, others don't.

HugoGuyader—EyeTrackingdata

Priming

VisualA.en0on:Eco-friendlyProducts GreenPremium

VisualA.en0on:Greenpricetags

VisualA.en0on:Informa0on

VisualA.en0on:Eco-Servicescape

VisualA.en0on:Greenwashing

+

+

H3

?

H2

H5

H4

H1

H6

HugoGuyader—EyeTrackingdata

Priming

VisualA.en0on:Eco-friendlyProducts GreenPremium

VisualA.en0on:Greenpricetags

VisualA.en0on:Informa0on

VisualA.en0on:Eco-Servicescape

VisualA.en0on:Greenwashing

coffeesoBener

+

+

HugoGuyader—EyeTrackingdata

Priming

VisualA.en0on:Eco-friendlyProducts GreenPremium

VisualA.en0on:Greenpricetags

VisualA.en0on:Informa0on

VisualA.en0on:Eco-Servicescape

VisualA.en0on:Greenwashing

coffeesoBener

+

+

H3

?

H2

H5

H4

H1

H6

Thankyou!

Anyquestions?remarks?

Hugo Guyaderhugo.guyader@liu.se