Developing grounded theory:
Qualitative analysis for hard-nosed researchers.
Presentation by Langbourne Rust to NY Association of Public Opinion Research 3/18/93
Study 1: A grounded analysis of the differential appeal of classroom toys to boys & girls
Study 2: An analysis of CBS sitcoms pilot shows - grounded on program analyzer data
Study 3: Age changes in shopping behavior
Readings on Grounded Theory
The problemsSuppose you had a file cabinet full of commercials that had all been tested with a like-to-watch scale. How could you find out what attributes or elements make people like some commercials and not others?
Suppose you had a classroom full of toys, and a notebook full of data on how often boys and girls played with each one. How would you find out what made boys like some, and girls like others?
Suppose you had several hundred observations of parents shopping with their kids - and estimates of how old the children were. What could you find out about how shopping changes as children get older?
Suppose you had the verbatim responses of consumers to an open-ended question that asked them what happened in a commercial they had just seen. How could you develop codes for content coding that would differentiate the consumers most interested in buying the product.
For tackling problems like these, I have found grounded theory method to be quite useful. In 1967, two academic sociologists, Barney Glaser & Anselm Strauss (1967) at UC-San Francisco put forth a systematic approach to generating new conceptualizations of what is going on in newly emerging areas of study. Their seminal work, "The Development of Grounded Theory," moved researchers past the hypothesis-testing uses of raw data into the hypothesis-generating potential of their observations. The approach has been steadily expanding its reach within academia - through sociology and social anthropology and, more recently into applied disciplines like educational research.
"Grounded Theory" is theory that is "grounded" in data of the kind that it seeks to describe or explain. The approach has led me to some intriguing learnings about consumer behavior: learnings with an encouraging degree of predictive validity.
A grounded analysis of the differential appeal of classroom toys to boys & girls
The first project I ever did along these lines was a study of the differential appeal of classroom materials to boys and girls. Here is what I had to work with:
First. Two preschool classrooms full of educational materials and toys. Both classes were set up along open-classroom lines - physically, it meant there were products set out on low shelves where children could get at them - and socially, it meant there wan an understanding that children could play with whatever materials most interested them for as long as they wanted - provided certain ground rules were observed. The environment was rather like an educational supermarket for kids.
Second. One notebook, listing the 450 odd materials by name and scoring them by how often they were played with by boys and by girls over a several-week period.
The children were 3 and 4 years old: 24 boys and 24 girls.
What would you do, given 450 products and a single score on each?
In essence, I began with two lists, one with all the toys that boys played with more, and the other with all the ones girls played with more, and looked to see if there were any consistent differences. Two-hundred or more toys in each list. Try this and boggle. Perhaps someone with a photographic memory could examine two lists this long and see patterns, but my own channel capacity is much too limited. To get around the channel capacity problem, I needed to chop the task up into smaller units.
The next question was, what size should those units be? If the lists become too short, a new set of problems arises. The toys with the strongest sex-skewing were Lego Blocks, which boys played with 3 times as often as the girls did, a low climbing apparatus with two walking boards and a gangplank which was favored by the girls at a 7 to 1 ratio.
Number of choices by each sex
Start listing the differentiating attributes and you will see that the list never stops. Furthermore, most of these differentiating attributes are false leads, with no relevance to the sex-typed preference we are interested in. The task becomes solvable if you work with an Extreme-Contrast list that has something between 7 and a dozen items at each end of the contrasting dimension.
Products showing greatest boy- or girl-skew in usage
Most boy-used items:
Most girl-used items:
Lego Blocks share some similarities with Crystal Climbers, Unifex Cubes, Build-a-Fun and Baufix in ways that also differentiated them from the girl-skewed items. The first attempt at defining this quality was:
Hard building materials that have some means of holding the pieces together.
Applying this prototype concept to the whole list of 450 toys, it proved to describe 26 of them. Boys played with them 119 times and girls only 65.
This was encouraging feedback, but it was not time to quit. A number of the toys described by the concept were, in fact, played with more often by girls. So another contrast analysis was done - comparing the "deviant" cases with the confirming cases. Two of the deviant cases were stacking cups and bracelet-making materials, and unlike the other toys in the set, their pieces did not link rigidly together. So the concept would account for the data better if the word "rigidly" were added to its definition.
A look at the remaining deviant cases showed that several of them involved building tools - screwdrivers and tack boards, but they were not building materials in the sense that larger units were not assembled out of smaller pieces. So the loose term "building materials" was replaced and the definition ultimately read:
Hard materials that can be used for building larger units by linking the elements rigidly together.
The net effect was a tighter definition of the concept and overall improvement in performance. For the first, there were 19 confirming cases (items used more by boys), and 7 deviant cases (used more by girls). Chisquare = 4.2. For the second definition there were 16 confirming cases and 2 deviant ones. Chisquare = 9.4
The analysis proceeded along these lines - going to the Maximum-Contrast list to induce prototype concepts, to the full sample to test them, to the deviant cases to induce ways to improve them, and so on until no new concepts with any power can be found, and the provisional concepts can be improved no further... what the grounded theorists refer to as the saturation point.
Basic steps to generate grounded concepts
Then an effort is made to compare and integrate the concepts - once again, by referring back to the data - dropping the ones which prove to be redundant, clarifying the way the others are expressed conceptually.
Ultimately 12 sex-typed concepts, or attributes, were defined:
1. Linking-building - Hard materials that can be used for building larger units by linking the elements rigidly together
2. Blocklike - unstructured. Including rectangular prisms or cubes of rigid construction and not including any physical constraints or guidelines to pattern the way they are put together.
3. Learning about Animals. Animals or materials that tell about or picture single kinds of animals.
4. Unstructured Numerals and Letters. Focusing on letters or numerals in a relatively unprogrammed manner. 23 items were described by this concept. Four deviant cases remained at the end of the development process.
5. Dry Food. Involving food, but not involving liquid or sticky substances. 19 items, 5 deviant cases
6. Climbing - no walking. Apparatus conducive or climbing or sliding on, but on which walking or sitting would be difficult. 27 items, 8 exceptions
1. Art or Arts & Crafts. Materials for drawing, tracing, painting, collage, ornament making, etc. Five of the girl-favored items on the Hi-contrast list have this attribute. 77 items were denoted, with 21 deviant cases.
2. Clothing related (not male role clothing). Related to clothing that is not exclusively masculine. 14 items denoted, 1 deviant case.
3. Pourables. Items involving direct interaction with freely pourable substances. It denoted 26 items, with 4 deviant cases.
4. Cleanliness-related. Materials for achieving or maintaining cleanliness or dryness. 18 items, 1 deviant case.
5. One-solution materials (excepting picture puzzles). Materials offering structures or problems to which there is only one correct response or solution - excluding picture puzzles. 67 materials, 18 deviant cases.
6. Climbing with walking. Apparatus conducive of climbing, and on which sitting and walking is easily done. 14 items were denoted, with 2 deviant cases.
Ultimately, the concepts developed by this grounded process were found to apply to 70% of the 450 items in the classrooms. Children were twice as likely to play with materials having an attribute characteristic of their own sex than with materials characteristic of the opposite sex.
A follow-up study, 6 months later, tracked what happened with 50 new materials that were introduced into the original classrooms. Before data were collected, predictions were made (based only on the attributes defined earlier) as to which sex would play more with each of the new toys. The predictions were correct for 38 toys, incorrect for 12.Study 2 An analysis of CBS sitcoms pilot shows - grounded on program analyzer data
In the mid 70's I approached CBS with an idea. They had their file cabinets full of old pilot shows that they had tested over the years, using their program-analyzer technique that gave them, for each show, scene-by-scene scores of how much test audiences said they liked what they were watching.
Each pilot show is tested with a sample of 80 out-of-town tourists who are recruited off the street. Each is provided with two buttons to press while they watch the show. They are told to press the green button (in their right hand) if they like what they are watching, and to press the red button (on the left) if they dislike. They are free to press neither button, but are asked not to press both buttons at once.
I proposed to CBS that they give me the tapes and response scores on a sample of these shows, let me do a grounded analysis to discover the attributes of scenes that people were responding to, and then set up a test of the findings: give me a new sample of shows - without the data - and see how well I could predict the test scores.
They were especially interested, at the time, in half-hour sitcoms, so we confined the study to this domain. They sent me tapes and charts on 26 pilot shows. There were 357 discrete scenes, each with an analyzer score.
The analysis procedure was the same as the previous study. But instead of examining toys, I examined program scenes.
Instead of preschoolers, I studied adults. And instead of grounding it on spontaneous everyday behavior, I grounded it on the green-button judgments of people in an artificial test situation.
I started with a high-contrast list of the dozen highest-scoring and dozen lowest-scoring scenes, and looked at them, over and over, taking notes and jotting down hypotheses about possibly differentiating concepts. These prototype concepts were then tested against the full sample of all scenes. The survivors were put through a deviant case analysis and refined until the saturation point was reached.
Here were the kinds of things I found that characterized high-scoring scenes:
Goodness help & kindness. Respondents hit the green buttons when they see characters who are, or try to be, good to others in an open, direct, face-to-face way. They must be moved by a genuinely good-hearted feeling. It is not sufficient to be routinely or habitually polite, remorseful, sympathetic or affectionate.
Harmless embarrassment. Scenes where the characters appear or feel awkward, silly, embarrassed or ridiculous. The embarrassing incident may actually occur or be vividly recalled or anticipated.
Anticipated Surprise. Scenes where the audience or one of the characters is surprised or startled. If a character is surprised, he/she must show it clearly. The emphasis must be on something happening that the audience is in on: the kind of private surprise that is shown by characters when they get a sudden idea, for example, does not qualify.
Justice realized: where a character who has acted with questionable or negative motives sees the light and reforms, relents under pressure, gets comeuppance, retribution, put-down, makes a fool of him/her self or is resisted successfully. It is essential that the character be aware of what happened.
Catch & Chase/Rough & Tumble.
Physical Discomfort. These were scenes in which non-threatening physical discomfort was received with good-natured grumbling, gag-making, slow burn or blustering.
Four attributes were found to characterize low-scoring scenes:
Sudden scene changes: where a show cuts suddenly to new surroundings not anticipated by the audience.. This excludes scenes which have been led into y the plot or the dialogue or the lyrics, but the lead-in must be very clear.
New characters: where important characters have their presence felt for the first time, with not prior introduction or characterization.
First five minutes. Scenes that end in the first five minutes of a show get low green-button scores.
Sad or troubled characters. Where a sympathetic character feels genuinely troubled, worried, or unhappy - and expresses it openly, green button scores decline.
The prediction test involved 9 new shows, containing 305 scenes. High scores were predicted whenever there were more positive attributes than negative ones in a scene.
The predictions proved to be correct 83% of the time: 76% of the high-score predictions scored high. 86% of the low-score predictions scored low.
Chi-square = 118 with 1 d.f.
Study 3 Age changes in shopping behavior
This study was conducted by a committee that I chair for the ARF's Children's Research Council. We went into stores, and observed parents and children as they interacted with each other. And then we did a qualitative analysis that helped us get some conceptual handles on what goes on there
The data we gathered were essentially anecdotal narratives. An observer would visit a grocery store or a toy store. Posing as a shopper, she would wait in an aisle. On seeing a shopping party enter the aisle, she would estimate the child's age, record some basic information about who was in the party, and then take notes on what the shoppers said and did. 200 records were collected. Here is an example involving a mom with 2 preschool children. They were in a supermarket cereal aisle.
Here is how this study differed from the previous one:
Record #171. Observer = Julie
Store = Waldbaums. Aisle = Breakfast cereals
Shopping party= 1 Mother, 1 girl/7yrs, 1 boy/1yr. Both children in cart. No shopping list visible, no coupons visible.
Party enters section, girl in bottom of cart, boy in top. Girl immediately points to Count Chocula box and exclaims something to the effect of "Look at those eyes!" Boy then joins in "Oh yes, lets get that one, can we get that one?" Mom says something in Spanish as she picks up the box of Count Chocula. Group talks in Spanish, a few things are said -- neutral in tone -- mom puts Count Chocula in cart and wheels away.
Field Observer's Thoughts:
Mom seemed willing to please, didn't look at side panels or anything. Children were obviously drawn to the box and seemed to make decision solely on eyes on box. (Count Chocula picture on the box had plastic piece over the face that made the eyes appear to move as the viewer walked by).
Another example, this one with an older child in a cereal aisle:
Record #49 Observer S. Adragna
Store= Shoprite. Aisle = Breakfast cereals
Shopping party= 1 Mother, 1 girl age 12. No shopping list visible, coupons visible.
Daughter holds coupon for STRAWBERRY WHEAT SQUARES - obviously a planned purchase. Coupon in hand, she searches the shelf, finds the brand, reaches for it on tip-toe off the top shelf, and continues holding it as they exit the aisle.
Another - with a 2 or 3 year old boy and his dad:
Record #25. Observer L. Rust
Store = Grand Union, Aisle = Breakfast cereals
Shopping party = Father, 1 boy 2 yrs No cart visible.
Dad holding list, wanders up and down aisle. Boy, too, independently. Boy keeps turning boxes, looking at their backs. Points at one. Comments "Puzzle". He passes FREAKIES - stops, says, "Neat!" Tilts box back & forth looking at the hologram. Dad exits aisle without taking anything - boy lingers, then skips along after. Hear Dad in next aisle -- "Adam! Which do you want, one of these or these?"
Here is one from a toy store
Record #168. Observer Julie
Store = TSW. Aisle = Action Figures
Shopping Party = Mom, 3 yr. old boy. No cart
Dyad enters space. Boy (age 3) puts foot on long plastic play table and says, "I have this." Mom is looking at figurines on wall in see-through packages. Child comes over to where mom is and sees Teenage Mutant Ninja Turtle box. Child gasps in surprise "Teenage Mutant Ninja Turtles, I have them." Mom points to TMNT sneaker snappers and says, "Who's that?" Child says something inaudible. Mom points to Beetlejuice figure and says, "Who's that?" Child looks and says, "Beetlejuice." Dyad moves down aisle. Mom points to Batman and says, "Who's that?" Child says "Batman."
And finally, here is one with a mom and a 12 year-old daughter in a toy store.
Record #144. Observer S Jahayer
Store = Toys R Us. Aisle = Videogames
Shopping Party= Mom, 12 yr. old daughter. No cart
A mother and a daughter walked into the Video Software section and the girl was looking for Tetris video. She asked her mother, "Are these in alphabetical order?" Then the mother said, "Yes, they should be." Then the girl said, "Here, I got it." and the mother said, "Oh, is it $35?" Then the girl took the video and they walked away.
Grounded theory development brings structure to the inductive process. It requires the researcher to go back repeatedly to the raw observations in the effort to create new conceptualizations. By grounding theory directly on raw observations, rather than pasting it together from some second-order coding of those observations, it avoids some of the garbage-in/garbage out limitations of more traditional quantitative approaches.
For keeping track of the records, searching through them, rating them vis a vis various attributes, and tracing the patterns that emerge, a specialized text and data base program, SOLID GROUND, was used. The tasks could have all been handled with physical printouts and rating forms, but the computer environment greatly simplified the process.
The central question of this study was: how do behaviors in stores change with the age of the child?
The first step was to generate some ideas.
Since I was searching for ideas on how old-child interactions differed from young-child interactions, I took a sample of cases with 2-3 year olds and contrasted them with cases with 10-13 year olds. Whatever changed with age would be likely to show up here.
One hypothesis was "Joint discussion or compromise." It was characteristic of many of the old-child cases .. and none of the young children..
Another hypothesis, "Physical involvement with the product" was more characteristic of little children.
And a number of them, like "Commenting on the price" failed to discriminate the cases, and were dropped forthwith.
The concepts which successfully differentiated the high-contrast cases were tested against the full data base of all 200 cases.
The two attributes I mentioned before held up well to the full-scale test.
Physical involvement with the product characterized 28% of the young-kid cases but only 11% of the cases with older children.
Mutual discussion and compromise occurred in 27% of the cases with kids over 5 but only 10% of the young-child cases.
Hypotheses which reflected no trend at all were dropped from further consideration.
If a hypothesis survived the full-sample evaluation, an attempt was made to see if it could be improved -- to account better for the data.
The search for improvements was guided by deviant case analysis.
For example, a hypothesis that parents were more likely to deny/ignore requests from younger children had received only lukewarm support from the full data base. Looking at cases with this attribute, broken down over five age groups, shows incidence levels dropping only marginally between ages 6 and 9.
Study of the deviant cases (the ones with older children) showed that a number of parents would first deny their children's requests, but then relent after discussion. This pattern never occurred among the younger-child cases. So the definition was modified to read "Parent firmly denies/ignores request."
The result was an incidence profile that reflected a sharper drop-off above age 5 .. and an attribute that had more meat to it.
After all the hypotheses had been tested and shaped through deviant case analysis, 13 of them remained: 7 young-child attributes and 6 old child ones.
Taken one at a time, these attributes differentiated the cases by age with some degree of success. But it was possible that two or more of them were actually pointing to the same cases, so that some of the attributes might not be making any net contribution towards our ability to account for the old child/young child patterns.
To explore this possibility, a Logistic Regression was run.
This is essentially a kind of multiple regression which works on categorical data. It enabled us to identify the smallest set of hypotheses that could account for the data.
A number of the attributes dropped by the wayside.
Some of them had been my favorites.
Parent distress had been more frequent in young-child interactions.
Child emotion (both positive and negative) had been more evident with young children, too.
The other three attributes that got culled,
Accept first choice
had been more characteristic of old children.
I do not mean to say that these rejected hypotheses are definitely invalid. In future studies, on new data, one or more of them may show signs of independent life. But in the current body of data, the other attributes account for the observed age differences just as well without them.
FINAL LIST OF ATTRIBUTES
Let me describe the attributes that survived the final cut.
Child in cart. Not many kids older than 6 rode around in or on the shopping cart. Nearly a third of the youngest children did so.
Pointing. Young children were much more likely to point at products or other things in the store. This was not dependent on riding in the shopping cart. Both riders and walkers were more likely to point if they were young. Pointing gives younger children a way to indicate desire, even when they lack the symbolic skills or knowledge to communicate verbally.
Physical involvement. Younger children were much more likely to exhibit some sort of physical involvement with products or displays or packaging ... over and above the functional contact involved in picking up a package and carrying it or putting it in the cart. They would explore things tactually, play with them, open them, consume them or manipulate them in one way or another. This was often done while sitting in the cart, but by no means always. One little girl, for example, walked down the cereal aisle, systematically turning every box around backwards. It appeared to be pure physical/sensory play.
Parent firmly denies/ignores the child's request. Although parents often turned down purchase requests, whatever the age of the asker, they were more likely to be firm and unyielding with their younger children. Kids between 6 and 9 sometimes negotiated successfully, following an initial turn-down. With children aged 10 and older, parent acceptance appeared to be more automatic. This may have been a function either of altered power relationships or of more educated children - who knew ahead of time what mom would accept.
Labeling. A certain amount of the dialogue between parents and children in stores is involved in communicating the names of things. Name learning is a key developmental task of preschool children. They are hungry to learn the names of everything around them, and parents appear eager to nurture them this way. Research based of videotapes of children while they watch TV has shown that preschoolers consistently pay close attention whenever the TV material involves show-and-tell, or labeling. (Rust, 1972, 1986) The store environment is full of objects to learn about, and as parents browse along with their little children, they spend a certain amount of time identifying the things around them.
Teamwork. Parent interactions with older children often reflected a degree of teamwork: a division of labor with coordination and communication between the members of the shopping party, and a set of shared objectives.
Pre-planning. Shopping with older children more often showed signs of prior planning. This sometimes showed up in the dialogue, when they would refer directly to prior conversations and intentions, and sometimes in the fact that the child would refer to a list, or bring out a coupon that had been saved for use on this trip.
Reading. Children aged 5 or under were seldom seen reading things, either on displays or packaging. Although a fair number of 4 and 5 year olds in the population have some ability to read, not many of the ones we saw were spontaneously motivated to do so in the store environment. Older children were much more likely to orient to the text stimuli around them.
These attributes, as a set, correctly predicted the age of the children 77% of the time.
Consider some implications of these findings....
To reach little children, and their parents, consider the unique circumstances and interactions in the store.
Little children ride in carts, so....
Direct the information on the cart to the child inside. The child taking a ride is a captive, and sometimes restless, audience, hungry for focus and stimulation. Many moms would be pleased to have their children concentrating on in-cart media, rather than having them bored and meddlesome and frustrated at their captivity. Here is an opportunity to communicate marketing information to the shopping party in such a way that provides genuine benefits to both parent and child.
Display products at cart height. Although bottom shelves may be good for the walking or toddling child, children in carts may have a harder time seeing, or noticing, products there.
Incorporate the cart-riding child in advertising copy imagery. If the copy scenarios echo the shopping scenarios, more shoppers (parents and children) are likely to make the link between ad and store.
Make packaging and display materials for young children to stand out and be noticeable from a middle-of-the-aisle distance. Children riding in carts are confronted with a staggering array of stimuli in the average toy or grocery store. We know, from studies of child development, that little children have very limited abilities to scan broad arrays of novel stimuli and make much sense out of them. Recognizable elements are pivotal. Faces, particularly those of familiar characters, are remarkably effective at getting noticed in cluttered environments.
Give children packaging or promotional material that can keep them occupied inside the cart.
Little children do lots of pointing, so.....
Stimulate long-distance recognition to get the shopping party oriented and moving toward your product from a long way off.
Incorporate physical gesturing in advertising copy. Why not develop pointing as a routine or ritualistic part of what people do when they see your product? A silly or distinctive pointing gesture would be very salient to little children. Children aged 6 or older might be too self-conscious for such public behavior, but for products targeted at preschoolers (who are the prime "pointers" anyway) such a device might be very effective.
Point at them. Portray a product's spokescharacter pointing back at the viewer ... like Uncle Sam used to do. If a product "means" pointing to a little child, that child will point when she sees it.
There is an implication for research methodology here, too. In your tests, you might ask little kids to use pointing to indicate brand interest. Pointing is non-verbal, and guided by recognition and association ... not by verbal/cognitive mechanisms. Marketers should not aspire only to changing what kids say they want, when asked their preferences, they should also aspire to change what kids notice, and orient to and point at.
Young children show high levels of physical involvement, so ....
Design packaging to attract (and withstand) physical contact and interaction. Activities (fantasy or otherwise) that use the box as a prop could be very attractive. They would keep little children oriented to your product, (and away from your competition). If the activity were one moms approved of, both parent and child would regard the involvement as a genuine benefit. Handles or straps on the packaging might get them to ask mom to let them carry it.
Advertising copy could establish the repertoire for what the child could do with the product in the store. This would help maximize the play value to the child, and the social or educational value from the mother's point of view.
Parents of the younger children tend to be firm when they deny a request, so ...
Don't encourage argument or back talk if you are promoting a product for little children. Moms are not likely to change their stance if they rejected the product initially. It is important that you, not the kid reach the mom first with whatever persuasive information is needed to overcome their resistance.
Moms use the store environment to teach and review the names of things, so...
Select the names of products with great care. If children can't pronounce a name, the product is in jeopardy. And if moms and kids tend to call it by different names, the "show & tell" activity may become muddled.
Educational and cultural material may draw attention and build positive associations with a product. So consider product names, displays and packaging that support the nurturant interplay between mom and child in the store. Products that give an opportunity for color-learning, or shape-naming, or story-telling, or animal identification, etc., are likely to get noticed, have time spent with them, stimulate positive feelings and build good memories.
Older children read, and like to do it, so...
Labels and packaging that relies on reading will skew strongly to the older children.
Consider the reading level of the text you use. Children are very motivated to read things that are at their level. Text that is too hard can be a turn-off. Text that is too easy is no challenge, and gives no accomplishment. But if you have a narrow target age for your product, you can have a big impact on kids' involvement with it by matching (or mis-matching) their reading level.
Also consider the content of the text. This can over-ride reading-level problems. Think of the interest of a young model builder in the text that describes the rocket inside the box and its history and what it can do. Interesting details or a good back-story can get kids very involved, even when it is a little hard for them to read. It can give them a great sense of accomplishment and identity to read it to themselves, or to tell their parent all about it. And parents are likely to be very reinforcing -- right there in the store. This can hardly hurt the product's chances of being bought.
Parents and older children often show real teamwork as they shop together, so...
Appeal to the positive sides of the joint shopping experience (for both mom & kid). Cast the kid as a helper and team player.
Advertising copy could give models of "grown-up" shopping behavior to kids: models in which your product plays a pivotal role. When kids emulate the example, they would feel a sense of accomplishment, and their parents would reinforce them. Think of ways your product could become an integral prop in this positive type of interaction. By giving them a concrete repertoire for teamwork in the store, you would bring parent and child closer together while enhancing the child's sense of autonomy and competence. And you would significantly add to the benefit profile of your product.
Older children and their parents often plan ahead, so ...
Provide supports to the planning process. Coupons, lists, mnemonics (memory gimmicks) will all increase a product's odds. We saw a number of kids who had brought along coupons for themselves. Systematic distribution of coupons to kids might pay off for some products .. especially to kids whose moms use coupons.
Reach them at home, where the planning gets done -- or on the road on the way to the store - via radio or outdoor.
Make product placement predictable for the pre-planned products. Older kids (and moms who have had kids for a while) are not open-eyed explorers. They know pretty much what they want and where to find it. If they don't see the first-choice brand where they expect it, they may take the second-choice brand that is there instead -- rather than initiate a search through the store.
Get into the routine. For families with older kids, many purchases are routine. The kids and their moms have been through the process hundreds of times before. To sell to them, the challenge is to find some way to become part of their routine.
Grounded analysis gives researchers a way to create powerful clustering schemes without relying on pre-established sets of ratings, scales, attributes or dimensions (which all too often put blinders on the analysis and prevent the discovery of discriminators as yet unthought-of). The analyst keeps going back to the fundamental phenomena under study -- the products, consumer behaviors, TV scenes, or whatever -- identifying contrasting cases and thereby stimulating the inductive process in an extraordinarily productive way. The key to the power of this procedure is the concrete level at which the phenomena are dealt with. It allows the human mind to do something it does extraordinarily well: develop abstractions from differential exposure to concrete phenomena i.e., create theory from experience.
Field observations have a richness to them that make them ideal for qualitative studies. How people behave spontaneously, in the act of shopping or consuming products, is often different from the descriptions they give in interviews. They are much more responsive to social and environmental stimuli than they are consciously aware of. And the only way researchers can get a handle on all these influences (which are of paramount importance to marketers) is by observing the behaviors as they occur in the field.
So far, I have found Grounded analysis to be well adapted to conceptualizing:
1. Consumer perceptions of complex stimuli - like products or advertising copy,
2. Observational and behavioral data - where consumers are not in a self-report mode and, by extension, to a host of low-involvement consumer situations
3. Existing data bases where a lot of complex materials (products, ads, promotions) have been tested in a standardized fashion over a period of time: and where you want to find out the underlying stimulus attributes that people have been responding to.
Studies like these are essentially qualitative analyses using quantitative data. They use hard numbers to guide the design of new conceptual structures which can account for the raw observations.
It seems to me that the most useful distinction between qualitative and quantitative research is not in the style of data gathering , or in the degree of structure involved in questioning, but rather in the purpose of the analysis. All empirical statements have both a quantitative and a qualitative component: something happens to some degree. The quantitative component is an expression of the degree, the qualitative component is the definition of what it is that happens. Quantitative research puts its emphasis on how often things happen, qualitative research puts its emphasis on what those things are. Qualitative researchers, in marketing, are the ones who keep the mind of the industry on track - providing the conceptual structure for understanding the consumer.
Grounded theory development can use both qualitative and quantitative tools - as we did here - but its focus is on creating the most powerful conceptualizations possible - concepts that are grounded in the data. In that sense, it is essentially qualitative research. And I have every reason to expect that it will bring new power and importance to qualitative research in marketing.
Readings on Grounded Theory
Bogdan, R., and S.K .Biklin. Qualitative Research for Education: An Introduction to Theory and Methods. Boston: Allyn and Bacon, Inc. 1982.
Glaser, Barney G. "The Constant Comparative Method of Qualitative Analysis." Social Problems. (1965): 436-445.
Glaser, Barney G. and Anselm L. Strauss. The Discovery of Grounded Theory: Strategies for Qualitative Research. Chicago: Aldine Publishing Company, 1967.
Glaser, Barney G. Theoretical Sensitivity: Advances in the Methodology of Grounded Theory. Mill Valley, CA: The Sociology Press, 1978.
Strauss, Anselm, and Juliet Corbin. Basics of Qualitative Research: Grounded Theory Procedures and Techniques. Newbury Park, CA: Sage Publications, 1990.
Strauss, Anselm. Qualitative Analysis for Social Scientists. New York: Cambridge University Press. 1987.
Rust, Langbourne. Attributes that Differentiate Boys' and Girls' Preferences for Materials in the Preschool Classroom: A Systems Design Approach. Ann Arbor, MI: University Microfilms, 1971.
Rust, Langbourne. Attributes of The Electric Company that Influence Children's Attention to the Television Screen. New York: Children's Television Workshop, 1972 (a).
Rust, Langbourne. Attributes of Sesame Street that Influence Preschoolers' Attention to the Television Screen. New York: Children's Television Workshop, 1972 (b).
Rust, Langbourne. "Using test scores to guide the content analysis of TV materials." Journal of Advertising Research, 25, (1985): 17-23.
Tesch, Renata. "Software for Qualitative Researchers: Analysis Needs and Program Capabilities." in Fielding, Nigel G. and Raymond M. Lee. Using Computers in Qualitative Research. Newbury Park, CA: Sage Publications, 1991.
Richards, Lyn and Tom Richards. "The Transformation of Qualitative Method: Computational Paradigms and Research Process." in Fielding, Nigel G. and Raymond M. Lee. Using Computers in Qualitative Research. Newbury Park, CA: Sage Publications, 1991.