Emotions & Sales Sutton & Rafaeli How to conduct an observational study Deductive Study (Study 1) Inductive Study (Study 2) Re-analysis of Study 1 data @ store & individual level Lessons learned
Theoretical Explanations Emotions can be used as a “control move” to influence behavior Positive, neutral vs. negative emotions Some can be reinforcing Positive emotions it may encourage customers to buy more, or to re-patronize store
Preliminary Hypothesis Amount of positive emotion displayed leads to increased store sales What is the predictor and criterion variable?
Study 1 Context Friendly behavior during transactions encouraged by Training & incentives for clerks Incentives for franchise store owners 25% Bonus over base salary for regional managers of corporate-owned stores
Participants 1319 clerks in 576 Convenience stores 8 stores from each of 72 districts that make up 18 divisions in 2 countries Primarily urban sample of stores 44% male clerks Does not state if the same clerk could have been observed multiple times (implications?)
Method Time of measurement 3 month period Does not specify how long after training Each store observed during one day & one swing shift 25% of stores observed during night shift 1-20 transactions per visit Up to 60 transactions per store 11805 clerk-customer transactions 75% male customers
Procedure “Mystery shopper” observers Observed clerks at pre-test stores w/research director before actual data collection period Compared & clarified behavior coding differences Corporate HR staff volunteers dressed according to the profile of a typical customer May not be adequately matched for SES of working class male customers 18-34 yrs bec. observers had a wide range of jobs
Procedure Observers Only coded clerk at primary cash register from magazine rack/coffee pots Visited store in pairs Selected small item, stood in line, paid for item Spent 4-12 min per store depending on number of customers in store 3% of observations excluded due to clerks’ suspicions
Procedure Reliability of mystery shoppers’ codings Director of field research Sample of 274 stores Accompanied by second original observer Allowed for computing inter-rater correlations w/ratings of first original observer (mean=.82)
Predictor Variable Positive emotion display Rated each transaction on 4 features Greeting, thanking, smiling, eye-contact Coded as 1 or 0 depending on display Transactions aggregated at store level Score for each of 4 features calculated as proportion of transactions in which behavior was displayed over total number of transactions Overall store index of emotion composed of mean of 4 aspects (reliability=.76)
Criterion Variable Sales Total store sales during the year of the observation obtained from company records Standardized across stores included in sample to preserve confidentiality
Control Variables Store gender composition Customer gender composition Proportion of women clerks observed over total number of store clerks observed at each store Customer gender composition Proportion of female customers over all customers present during all observations in that store
Control Variables Clerk image Store stock level 3 items rated on a yes/no scale Was clerk wearing a smock? Was smock clean? Was clerk wearing name tag? Store stock level Rated on 5-point Likert scales as to whether shelves, snack stands & refrigerators were fully stocked
Control Variables Average Line length Store ownership Largest number of customers in line at primary cash register during each visit Store ownership Franchise vs. corporation owned Store supervision costs Amount (in dollars) spent on each store Region Location of store in one of four geographical region (NOTE Coding method for regression)
Regression Analyses Hierarchical method using sales as dv Step 1 = 8 control variables Note: Adjusted R2 accounts for the increased likelihood of finding a large and significant R with a small sample, and/or with a several predictors (I.e., differences between R2 and adjusted R2 are greater in such cases) Step 2 = Predictor variable i.e., Display of positive emotions
Regression Results Sales are positively related to Average line length (store pace) Supervision costs Clerk gender composition Sales are negatively related to Display of positive emotions contrary to hypothesis
Study 2 Explain the negative relationship between store sales and display of positive emotion
Data Collection Methods Case studies of 4 stores Researcher worked for a day as store clerk Conversations with store managers Customer service workshop 40 visits to different stores Paper Organizational Issue: Ordering of descriptions (p. 472)
Case studies Clerks Typically Display Positive Emotion Clerks Typically do not Display Positive Emotion High Sales 1 Low Sales Two 1-hour observations in each case study store Clerk consented to observer, had informal conversations re: customer service
Case studies Semi-structured interviews with store managers of case study store 30-60 mins long 17 questions re: Manager’s prior experience Selection, socialization, reward systems used in store Employee courtesy and its influence on store sales Info on how responses were coded not provided
Data Collection Methods Researcher works as clerk for a day In store with low sales but frequent display of positive emotions Viewed 30 min training video on employee courtesy before working Conversations w/store managers 150 hours of informal conversations re: negative relationship b/w positive emotions & sales
Data Collection Methods Customer service workshop attendance 2 hour prg. focusing on methods for coaching and rewarding clerks for courteous behavior Discussion on the role of expressed emotions in the store 40 visits to different stores Qualitative measures of store pace Not much detail provided
Theoretical Explanations Store pace determined norms re: emotional expression that affected emotions displayed Busy time evoked norms for fewer positive emotions Slow times evoked norms for more positive emotions
Norms for Busy Stores Fewer positive emotions helped maintain store efficiency Discourage customers from prolonging transactions Were perceived as more efficient by other customers waiting in line Evoked feelings of tension among clerks leading to fewer positive emotions
Norms for Slow Stores More positive emotions displayed by clerks Low pressure for speed/efficiency on clerks Customers have different scripts for slow stores Clerks regarded customers as a source of entertainment
Revised Hypothesis Expression of positive emotion is negatively related to store pace (as measured by store sales & line length)
Regression Analyses Hierarchical method with display of positive emotions as dv Step 1 = 7 of 8 control variables (as in Study 1) Step 2 = line length & total store sales
Regression Results Display of positive emotion is negatively related to Store sales Average line length (store pace) Control variables Store ownership Stock level Display of positive emotions is positively related to store clerk gender composition
Individual-Level Data Analyses N=1319 (clerks) Hierarchical multiple regression Step 1=Control variables Step 2= Line length negatively predicted display of positive emotion Did not use store sales as predictor bec analyses is at individual level, whereas store sales info is at store level
Typically Busy Stores Clerks show fewer positive emotions during slow times Slow times provide ‘opportunities’ to catch up on other tasks, customers are not perceived as source of job variety or entertainment Measured as large amount of store sales
Typically Slow Stores Clerks show fewer positive emotions during busy times Less experience in coping with pressure of busy times and feel tense Therefore… Stronger negative relationship between line length & display of positive emotion for slow stores Measured as small amount of store sales
Individual-Level Data Analyses Hierarchical multiple regression Step 1 & 2 as previous analyses Step 3= Interaction b/w line length and total sales negatively predicted amount of positive emotion
Individual-Level Data Analyses Classified stores as busy/slow based on store sales being above/below mean Separate hierarchical multiple regressions for clerks at slow & busy stores Line length was Negatively (-19) related to display of positive emotions (for slow stores) Marginally (06) related to display of positive emotions (for busy stores)
Discussion Found negative relation b/w positive emotions and store sales Why? Stores sales reflect store pace which causes emotions Could be different In diff org’n with different ‘service ideal’ (e.g., Mcdonalds) For longer transactions (e.g., restos)
Discussion Emotions as control moves affect things other than sales Negative/neutral emotions as control moves to increase efficiency Positive emotions used to achieve individual rather than org’n goals
Discussion Relative strength of corporate norms vs. store norms & inner feelings in determining display of emotions Reduce stress to encourage display of positive emotions
Discussion Observational methods Ethics of secret/unobtrusive observation Benefits of non-reactive vs. contrived observations Clerks informed about mystery shoppers Anonymity of clerks observed But each store had only 8-10 clerks!
Discussion Presenting the research process Acceptability of inductive & deductive process in Organizational behavior research publication process Corporate environments Media presentations Reader friendliness Student learning