Current consumer behavior research in forest products.

Abstract

A tenet of the marketing concept holds that businesses exist to satisfy customer wants and needs. Firms can satisfy those wants and needs only to the extent that they understand their customers. Therefore, an understanding of consumer behavior is important. In this paper,

we describe consumer behavior research methods with respect to forest products. We note a trend toward increasing sophistication in the methods used to collect consumer data. However, the increasingly sophisticated methods have presented new challenges. As evidence of these trends, we provide descriptive examples of recent consumer behavior research conducted at Forintek Canada Corporation, Oregon State University, and Virginia Tech. Results suggest increased scrutiny is advisable with respect to sampling error in traditional mail surveys. In addition, there are myriad challenges to conducting consumer behavior research, especially when done in cooperation with large retailers. We strive to inform the field of methodological challenges and encourage further development of consumer research specific to the forest sector.

**********

Consumer behavior research is the scientific study of the processes consumers use to select, secure, use, and dispose of products and services that satisfy their needs. The study of consumer behavior can be categorized based on who does the buying--intermediary customers (e.g., wholesalers, distributors, retailers, and secondary manufacturers) or final consumers (do-it-yourselfers and homeowners). In either case, knowledge of consumer behavior directly affects marketing strategy. This is because of the marketing concept, i.e., the idea that firms exist to satisfy customer needs (Winer 2000). Firms can satisfy those needs only to the extent that they understand their customers. For this reason, marketing strategies must incorporate knowledge of consumer behavior into every facet of a strategic marketing plan (Solomon 2002).

In most cases, forest products marketing research has focused on the behavior of intermediary customers. Presumably, the attention given these customer groups is because about 40 percent of the demand for forest products is derived from new home construction (Rich 1970, Taylor 1999). However, the advent of forest certification, and its associated marketing implications, has shifted more research toward final consumers. Although general marketing principles apply to both final consumers and intermediary customers, several differences exist (Winer 2000): 1) demand for intermediate products is derived from underlying final consumer demand; 2) many intermediate products are more complex than consumer products; 3) unlike consumer products, intermediate products are part of a supply chain that requires timely inventory replenishment; 4) the buying process among intermediate buyers can be much more complex than that of final consumers.

Surveys are a method of collecting primary data in which information is gathered from a representative sample of people (Zikmund 1994). This method has been, and still is, the primary data collection method used in forest products marketing research. However, we have identified an "evolution" in the way researchers are using surveys. This evolution is the theme of this article and we develop the idea with descriptive examples of recent research. As more consumer research is conducted, it is critical that it be informed by previous efforts. This is true not only for theoretical issues, but also for methodological and practical logistics considerations. Accordingly, we describe our recent efforts at better understanding consumers of forest products. First, we discuss how the location/setting of data collection is changing and the impact this has on the results. Then, we describe two recent studies that combined surveys with experiments as methods of data collection.

Intercept/exit surveys in a retail outlet

Surveys are conducted using a variety of mechanisms--mail, telephone, face-to-face, and more recently computer-based (e.g., web, e-mail, or on standalone purchase)--and are commonly used because they provide a quick, inexpensive, and accurate means of assessing information about a population (Sheatsley 1974). The typical objectives of surveys are to: 1) characterize target markets in terms of the quantities and types of materials used (e.g., Smith 1991, Haas and Smith 1997, Eastin, et. al. 1998, Forbes et. al. 2001); and 2) measure buyer attitudes and describe purchase patterns (e.g., Bush et. al. 1991, Stureson and Sinclair 1991, Vlosky and Smith 1994, Michael and Smith 1995, Hansen and Bush 1996, Weinfurter and Hansen 1999, Smith 2002).

Clearly, surveys have been, and will continue to be, a mainstay in forest products marketing research. However, we have noticed a shift in the location where surveys are conducted. Rather than having a consumer complete a self-administered survey at home or in an office, several more recent studies have used the intercept/exit technique to survey consumers at retail outlets for building products (e.g., Forsyth et al. 1999, Pakarinen 1999).

Intercept surveys involve conducting face-to-face interviews or using self-administered questionnaires in the retail setting where the research project is located (McDaniel and Gates 2000). The advantages of this technique are that: 1) there is a concentration of the target population at the retail outlet; 2) the researcher can effectively screen respondents to assure that they are members of the population of interest; 3) the researcher can gauge consumer response to new products using actual product samples or other visual cues, rather than relying on a written description of a product; 4) data can be collected quickly; 5) potential respondents are generally not offended when asked to participate in a survey because most view a retail outlet as a more appropriate place for a survey than, for example, their home. The major disadvantage of this method is that the sample is not necessarily representative of the desired population. Other potential disadvantages include the possibility of interviewer bias, the potential for demographically skewed samples, and studies that are limited to metropolitan areas.

The following section compares results from two similar studies. In one study, data were collected using the "traditional" mail survey technique. In the other study, an intercept technique was used.

Recent research at Forintek Canada Corporation

Background

In 2000/2001, Forintek Canada Corporation conducted two separate studies of do-it-yourself (DIY) consumers in western Canada. These studies shared the same target population, time frame, geographical region, and a series of identical questions. A key difference between the studies was the method of data collection. Study A (Fell and Gaston 2001a) collected data via a mail questionnaire, whereas Study B (Fell and Gaston 2001b) collected data via an intercept/exit survey. The differing data collection techniques combined with the other similarities offers a unique opportunity to compare the influence of data collection method on respondent profiles, and ultimately, results.

Objectives

The broad objective of study A was to explore the repair and remodeling (R & R) activities of homeowners in western Canada, with special emphasis on DIY activities. The specific objectives included identifying the types of projects undertaken, the products used, and who performed the projects. Collecting demographic data about DIY consumers was also a very important part of this study.

The broad objective of Study B was to explore outdoor R & R projects. The specific goals of this study were to assess attribute preferences and tradeoffs made by consumers with respect to material selection in outdoor projects and to assess useful segments of DIY consumers. The identification of consumer segments is useful in allowing manufacturers to meet the needs of specific customer segments.

Methods

The key difference between the two studies from a methodological perspective was the method of data collection. Study A used conventional mail survey methodology for data collection. Study B, on the other hand, collected data using a retail store exit survey.

In study A, market researchers at Forintek Canada Corporation and the University of British Columbia's Wood Science Department drafted a survey instrument. Colleagues at both institutes, as well as homeowners fitting the target population, pre-tested the survey. This process resulted in several minor changes to the questionnaire, after which it was deemed ready.

The population of interest for project A was owner-occupied households in Alberta and British Columbia. Households to be surveyed were randomly selected from these two provinces. The survey procedure included an initial mailing of a personalized letter and survey to 2,354 households. Two weeks after the initial mailing, a reminder card was mailed. There were 259 usable responses obtained from the initial questionnaire mailing, and due to deadlines, no further mailings were made. The resulting response rate was 11 percent.

Study B collected data from consumers as they exited a DIY big-box type home improvement center. A computer-administered questionnaire was designed that included questions on outdoor projects as well as more general R & R and DIY questions. The most significant section of the questionnaire was a series of conjoint questions regarding materials selection for decks. Conjoint analysis is a method commonly used to model how consumers make trade-offs among the range of attributes found in a given product. Before actual data collection, the questionnaire was pre-tested with store employees and shoppers.

Data were collected by university students in four western Canadian cities during weekend shopping hours. Booths were set up near store exits so that consumers could be approached after they had finished shopping. Potential respondents were screened prior to completing the questionnaire; eligible respondents were homeowners. A total of 150 questionnaires were filled out in each of the four cities.

Different results by method

These studies were not intended to be analyzed as one dataset, but they included questions on common subjects, included respondents with similar demographics, and included respondents from similar regions. Therefore, results were compared and some notable differences were detected. Additionally, comparisons were made between the demographics of the respondents in both studies.

Early comparisons indicated very different levels of R & R activity between the exit survey and the mail survey groups. Whereas the mail survey respondents only needed to be homeowners, exit survey respondents were homeowners visiting a DIY store. Thus, it was believed that the exit survey respondents would report higher DIY and R & R activities and expenditures. However, the opposite was found to be true. Only 36 percent of exit survey respondents reported building or renovating a deck in the past 5 years, compared to 61 percent of mail survey respondents; 36 percent of exit survey respondents and 48 percent of mail survey respondents had built a fence. When asked to estimate the dollar amount of DIY product purchases in the past 5 years, exit survey respondents reported spending $9,153 compared to $12,885 for mail survey respondents (p-value<.05).

After discovering these differences in respondent activity between the two studies, demographic variables were compared. The first variable that stood out was gender. Thirty five-percent of exit survey respondents were women compared to just 20 percent for the mail survey. The age breakdown of respondents also differed by data collection method. In general, the exit survey was skewed to younger respondents and the mail survey to older respondents. Twenty percent of exit survey respondents fell within the 20-34 age category compared to 9 percent for the mail survey. On the other end of the spectrum, 18 percent of mail survey respondents fell within the 65 and greater age category compared to only 3 percent for the exit survey. The final demographic variable that showed marked differences was population density. The mail survey had close to 20 percent rural respondents compared to only 7 percent for the exit survey.

Lessons learned

Comparing the results of these two studies allowed us to draw some conclusions about the results obtained when different sampling and data collection methods are used. We can also infer some of the possible biases introduced.

Using a mail survey had both advantages and disadvantages. The mail survey allowed for greater control over sampling. The random sampling method used in the mail study is commonly relied on for introducing the least amount of sampling bias into a study. Further, a mail survey has the ability to reach a very wide cross section of respondent demographics, e.g., there were more rural respondents to the mail survey compared to the exit survey.

The disadvantages of the mail survey were greater in this study than had been expected. First, the mail survey also had the potential to reach the non-shopping or non-active members of the population. Second, a considerable self-selection bias is suspected in the results. Although mail surveys have excellent sampling control prior to mailing, post mailing control of the sample is poor. Self-selection bias occurs when only the sample members who are interested in the study choose to participate. Those most interested in this study were the active DIY consumers. This is evident in the higher level of activities and expenditures reported in the mail survey, despite the assumption that the mail survey would better represent households that are not active in R & R or DIY activities.

The very nature of exit surveys meant that sampling control was compromised at the onset because rather than using random sampling procedures, a convenience sample was used. This is especially true if the population is defined as homeowners in western Canada. Only the small proportion of the population that shopped at one of four stores on a certain weekend had any chance of being in the sample. If shoppers at the store differed from the general population of interest then certain biases may have been introduced.

Direct observation and experimentation

Unlike surveys, which measure consumer attitudes, direct observation and experimentation allow a researcher to collect data about actual consumer behavior. Experimental research allows the researcher to control the research situation so that causal relationships among variables may be evaluated (Zikmund 1994). Although experiments are much more common in the physical sciences, marketing researchers occasionally use experimentation. Experiments are conducted either in a lab or in a field setting. Lab experiments have high internal validity because the experiment can be tightly controlled. However, the artificial nature of a lab experiment means the results lack external validity. Field experiments, on the other hand, often lack internal validity, but have high external validity, or realism. Another consideration when conducting experiments is what is being measured.

An experiment allows the researcher to observe whether manipulating marketing variables is associated with a change in consumer behavior. However, the researcher cannot determine with certainty why the observed behavior occurred if the consumer is not given a post-purchase survey.

Since forest products marketing research using experimentation/survey combinations is just emerging, the following section describes two studies that attempted to use this methodology to assess consumer response to ecolabeled forest products (EFPs). The logistical difficulties encountered and the practical lessons learned are discussed.

Research at Oregon State University

Study background

The concept of attitude as a means of explaining social behavior first occurred in 1918 (Ajzen and Fishbein 1980). Since then, the idea of attitudes explaining behavior has been challenged several times. Notably, LaPiere (1934) investigated racial prejudice and found that expressed attitudes did not match observed behavior. Similarly, Wicker (1969) found only weak correlations between measured attitudes toward a psychological object and measured behavior toward that same object. Given the shortcomings of attitude behavior research, Ajzen and Fishbein (1980) developed the Theory of Planned Behavior, which holds that a person's behavior is determined by their attitudes about the outcome of that behavior and the subjective norms operating in that person's environment (e.g., what will other people think about the behavior). A number of empirical studies in a variety of contexts have found the Theory of Planned Behavior accurately predicts behavior.

Despite the advances in attitude/behavior research, studies about consumer attitudes toward EFPs have called for research that measures actual behavior (Ozanne and Vlosky 1997). Thus, in an effort to bypass the apparent weak link between attitude and subsequent behavior, we designed a study to observe actual consumer behavior with respect to EFP's. (1)

Research objectives

The original objectives of the study were:

* Evaluate the effect of EFPs on final consumer behavior.

* Evaluate final consumers' price sensitivity to a premium for EFPs.

* Test for correlates between final consumer characteristics such as knowledge, attitude, and values and their purchase behavior with respect to EFPs.

Methods

In this study, we sought to determine if the presence of an ecolabel was associated with increased sales. We accomplished this by placing certified and non-certified wood products (plywood) in side-by-side bins at home-improvement centers. The products in either bin were identical except for the following:

* Ecolabel -- The product in one bin had the Forest Stewardship Council (FSC) ecolabel, while the product in the other bin did not. This key difference was held constant throughout the study.

* Price -- We manipulated the price difference between the bins. First there was a period of time when the price of the products in each bin were equal, then a period of time when the FSC certified product was more expensive.

We also felt that few consumers knew what the FSC ecolabel is intended to represent (e.g., products that originate in well-managed forests). Therefore, we placed a point-of-purchase display near the FSC-certified bin. The display featured the FSC logo and the following text "The presence of this logo is your assurance from the Forest Stewardship Council that this wood product comes from a well-managed forest."

The products in each bin had unique SKU (stock keeping unit) numbers. Thus, each time a consumer purchased either product the SKU number was scanned and the sale was recorded. This allowed us to collect data by simply tracking sales of each type of plywood. This experiment was conducted under the assumption that the two variables we manipulated (price and ecolabel) would affect consumer behavior. Thus, any observed differences in the proportion of ecolabeled and non-ecolabeled products sold could be attributed to the effect of the variables. The chi-square goodness-of-fit test was used to test whether the observed proportion sold for each type differed significantly from the assumption of equal proportions (Ramsey and Schafer 1997).

Finally, we believed that each person's unique set of beliefs, values, and attitudes influence their purchase decision. Therefore, we designed a post-purchase questionnaire to measure those items. It was to have been completed by each person after they purchased a product from one of the bins.

Significant real world challenges

In the process of implementing the research design, we encountered a variety of distinct challenges:

1. It was difficult to secure a supply of FSC-certified product. Our initial choice was dimension softwood lumber because it is a low involvement product with high turnover. Green Douglas-fir is preferred in local Oregon markets. Over the course of 18 months, we could not locate a steady supplier of FSC-certified, green Douglas-fir.

2. We had difficulty finding a retailer willing to participate in the study. We contacted five different forest product retailers. None were willing to conduct the study in their stores.

3. During a pilot test of our research design, we could not get consumers to complete a post-purchase questionnaire. We placed a stack of questionnaires at each cash register and instructed the cashiers to give a questionnaire to each customer who bought the product. In over 20 transactions, not a single customer completed a questionnaire. Our initial idea to solve this problem was to collect contact information from each customer, so that we could mail or call him or her to complete a post-purchase questionnaire. However, such a protocol potentially compromises the confidentiality of the survey and was discouraged by OSU's Institutional Review Board.

Our response

It was clear that the spotty supply of certified product and a retailer's willingness to participate were related. Therefore, rather than first identifying a certified product and then approaching a retailer, we switched our focus to first finding a retailer willing to participate and then determining a product. Using this strategy, we were able to secure the participation of The Home Depot, the world's largest home improvement retailer. After discussion with personnel at their corporate offices, we chose an FSC-certified product that was already in their stores: 3/4-inch by 4-foot by 8-foot BCX grade, sanded plywood. Using the sanded plywood was a bit of a compromise because it introduced additional product differences, aside from ecolabeled and non-ecolabeled. For example, different manufacturers made the ecolabeled and non-ecolabeled plywood used in the study. In addition, there were different numbers of plies in the core of the ecolabeled and non-ecolabeled plywood. We accepted this compromise because we judged these differences to be relatively unimportant.

At this point, two new challenges arose. First, each type of plywood had a unique SKU number. Therefore, we anticipated being able to use electronic scanner data to track plywood sales by transaction. However, we were only able to obtain sales reports for each SKU number once per week. Thus, we were unable to access the number of pieces sold per transaction. This meant that we could not be sure if each piece sold was an independent event. Therefore, it was possible that a single consumer, who purchased a large quantity of either type of plywood, could have skewed the results. Second, although, we had secured participation from The Home Depot and had discussions with their corporate level personnel about how to carry out the study, we needed to establish contacts at the local stores. This proved to be another hurdle since local store personnel did not necessarily exhibit enthusiasm for the project. We quickly learned that it was vital to the success of the study to establish one store employee as the person responsible for carrying out the logistics of the project.

Eventually, we were able to begin directly monitoring consumer behavior to test for an ecolabel effect on sales. However, assuming there was an ecolabel effect, we still needed to implement a survey to begin gaining insight into why there was an effect. Therefore, given our lack of success implementing a post-purchase survey, we instead chose to conduct a store intercept survey of consumers shopping in the lumber department. On the questionnaire, we used a conjoint analysis design that exactly replicated the product attributes of the 3/4-inch by 4-foot by 8-foot BCX grade, sanded plywood. The questionnaire also included questions about attitudes, beliefs, values, and demographics that we hypothesized to be related to the consumer's choices in the conjoint design. We did this because we felt that conjoint analysis could be used as a proxy for actual behavior. It should be noted that although conjoint is widely used in consumer behavior studies, most forms of it rely on the assumption that consumers evaluate product attributes additively (i.e., there is not interaction among product attributes). The development of the questionnaire was based on a review of the literature and a "pilot-test" conducted on nearly 300 undergraduate students at Oregon State University (Anderson and Hansen 2003).

Lessons learned

Clearly, this was a difficult project from a logistical standpoint. However, we believe future research using slight modifications to the methods described here is warranted. Perhaps the most important modification would be to continue to directly observe consumer purchase behavior, but have a self-addressed, postage paid, post-purchase questionnaire physically attached to the product. The advantages of such a method are:

1. The researcher can measure the attitudes, beliefs, values, etc. of the purchasers as opposed to intercepts of store consumers. This would allow a more precise determination of the psychological and sociological factors associated with each consumer's decision.

2. The consumer could complete the survey at his or her convenience.

3. It would be anonymous and therefore not compromise privacy.

4. An accurate response rate could be determined by comparing the number of units sold with the number of returned surveys.

When products are used where a consumer might buy multiple pieces, a question on the survey could determine how many pieces each consumer bought, thereby addressing the independent events issue.

Research at Virginia Tech

Study background

Without education and promotion efforts, consumers are unlikely to recognize the value of forest certification, or to understand its' significance (Hansen 1997). Lamson (1997) agrees with Hansen that there is a lack of consumer understanding about what "certified" products are or what a sustainable forestry logo on a product means. He argues, "The public needs more detailed information ... that empowers them to do the right thing." Although this was simply Lamson's opinion, Teisl (2003) provides empirical evidence in support of Lamson's position. Teisl found that consumer demand and willingness to pay for ecolabeled forest products is contingent upon the information displayed on the ecolabel. These findings point out the need to make additional information about forest certification available to consumers. Despite the agreement by leading researchers that promotion could increase the success of certified forest products in the marketplace, little research had been done in this area, so a study was undertaken at Virginia Tech. (2)

Research objectives

The objectives of this study were:

* Determine the influence of promotional material for certified forest products on consumer purchase decisions.

* Examine the impact of price premiums on consumers' purchase decisions.

* Compare the consumer demographic findings of this study to those of previous studies to identify demographic indicators of the "green" consumer.

Methods

The study concentrated on DIY and professional contractors purchasing high value, S4S hardwood boards at home improvement centers. This type of retail outlet was selected because of its growing popularity with consumers, reputation for carrying a large variety of products, and importance as a wood products distribution channel for the growing R & R market.

At 12 store locations of a major home improvement retail chain in 3 southeastern U.S. states, S4S, environmentally certified, FAS grade hardwood boards were offered for sale alongside equivalent, non-certified boards. The store locations were selected to represent a variety of community sizes and income levels among consumers. The stores were grouped into one of two blocks based on an urban or rural classification.

Two experimental variables--price and promotion--were manipulated across the four experimental treatments. Each treatment was replicated in two stores. In some stores, prices were equal between certified and non-certified, while in other stores the certified product was priced at a 20 percent premium. The premium amount was selected based on an examination of other environmentally oriented products (e.g., organic produce), as well as the premium levels identified in studies on the willingness to pay for certified wood products. In addition, the large difference in price meant that researchers could more safely conclude that observed differences in sales volumes were due to the amount of promotion. Promotion was the other variable. In some stores, the environmentally certified boards only displayed the GreenMark (trade name) logo and a sign stating the boards were FSC certified, while in the other stores the certified boards also had a tear-away information sheet explaining certified products. The message focused on the environmentally friendly and sustainable aspect of the product, as well as its high quality.

Data were collected in two ways. First, certified and non-certified boards each had unique SKU codes. The sales volume of each board type was recorded on a monthly basis. This allowed for testing of associations between price, promotion, and sales. The other form of data collection was a consumer survey. The survey was designed to examine consumer attitudes regarding certified wood products, pricing, and promotion. The questionnaire included questions to determine promotional effectiveness, factors considered in the consumers' purchase decisions, price sensitivity, trustworthiness of various certifying groups, as well as demographics. The target population for the survey was purchasers of either certified or non-certified hardwood boards. Potential respondents were to have been selected as they passed through the check-out line. The sales cashier was to have collected the person's telephone number so that they could be contacted for a telephone survey.

Significant real world challenges

In the process of implementing this research design we faced several distinct challenges:

1. During the initial meeting with the homecenter corporate headquarters, we agreed that store cashiers would get the phone number of individuals purchasing DIY boards so that researchers could later call them to conduct a phone survey. However, just before the start of the project, the homecenter told us they were not allowed to collect phone numbers and provide them to us.

2. Certainly, one of the most difficult obstacles to overcome with this research was effective organization and communication among the large number of parties involved. We initially believed the project would include Virginia Tech, the lumber supplier, and the homecenter headquarters. However, once the project was initiated, it quickly changed to include the 12 individual stores (sometimes with more than one person involved at each store), the supplier of the non-certified hardwood lumber (sometimes from more than one location), the data collection company, 3 different individuals at the corporate headquarters, and the distributor of the DIY boards.

Our response

Not being able to collect data via telephone surveys was a difficult hurdle to overcome because we had little time to formulate another plan. After a number of discussions, the homecenter agreed to hire a professional survey group, Smith Research, Inc., to conduct mall intercepts after the purchase of the DIY boards. This cost the company approximately $20,000 that was not expected and/or budgeted for the research.

Consumers were intercepted as they were leaving the aisle and had already committed to the purchase of the S4S hardwood board product. The shoppers were offered a $5 discount coupon for completing the survey. An incentive is especially important for lengthy questionnaires such as the one used in this study, which took approximately 10 minutes to complete. Enhancing response rate with an incentive was critical to our success because the individual sales occurrence for these products is low. It was also important to survey the shoppers after they committed to their purchases, to minimize the influence on their purchase decision. Sudman (1998) suggested that sampling in stores should be done so that shoppers are chosen randomly. This includes sampling at different times of the day and various days of the week. Although the bulk of the research was collected during busier times, such as Saturdays, information was collected on Thursday, Fridays, and Sundays as well. This was done to minimize sampling error.

The target number of completed questionnaires per store was 50, resulting in a total of 600. Using a confidence level of 0.10, the optimum sample size is 656. However, financial constraints meant that 600 responses was the highest number feasible. Unfortunately, due to slower traffic volume than expected, many stores did not reach their expected goal. However, we feel the quality of the intercept data was much better than what might have been obtained using telephone surveys.

The logistical difficulties encountered in this project were more difficult to address. For example, problems associated with the delivery of 12 point-of-purchase (POP) display booths were an indication of the difficulties created by the large number of people involved. Early in the project, we were communicating directly with corporate headquarters about the delivery of 12 displays that were made at Virginia Tech's wood shop. We believed that since the corporate headquarters chose the 12 stores, they would inform them of the project, what it entailed, and why it was being conducted. Thus, we expected the stores to be ready to receive the POP displays. Not only did the majority of stores not know that displays would be arriving, a number of stores didn't even know about the project at all.

Another example of logistical difficulties was stocking the displays. The lumber supplier had contracted a dimension mill to manufacture certified lumber into S4S boards. The problems created by the addition of another party included: 1) the dimension mill had difficulty manufacturing enough product to keep the displays fully stocked; 2) in one instance, the wrong UPC codes were placed on a number of products; and 3) it was unclear who was responsible for stocking the displays. The policy at the homecenter was that a material supplier does the stocking and refilling of the shelves. Fortunately, the supplier of the non-certified boards was kind enough to stock the shelves as long as the lumber was there to stock. This series of foul-ups caused several of the 12 store managers to contact corporate headquarters to request an early end to the project. We responded by sending graduate students to each of the 12 stores to make sure the displays were set up correctly and to calm the worries of the store managers.

Lessons learned

One lesson learned on this project was the need to follow through on the agreed-upon research plan. On a number of occasions, we assumed that what was said would be completed as stated and on time. As indicated, this did not occur on numerous occasions. These problems could have been minimized with better communication among the parties involved in the research. For example, early in the project we should have contacted each store and scheduled a visit to properly set up and stock the displays, instead of waiting until a crisis occurred. This points out another valuable lesson: deal with problems immediately, rather than waiting until they become critical. In some instances, when told that the displays were in the back shed, we assumed that the manager, when told what the project was, would take action and do as we expected. For 6 months, it seemed as if we had only problems or challenges to solve. If we had stayed in contact with every party in the project better, we would have created a better working atmosphere and ultimately had a better project when completed.

Conclusions

This article illustrates both theoretical and practical considerations for other researchers in the field of forest products marketing. As can be seen from our collective experience, the methods used presented significant challenges in both implementation and interpretation. For example, the differences found between the mail survey and mall intercept methods of data collection suggest that researchers need to carefully consider and control for the sources of bias in the forms of data collection typically used by researchers in the forest products business/marketing field. Further, non-response bias (or coverage error) may be a larger issue than is often currently considered. More advanced methods for detecting such error are advisable.

"Real world" experiments that directly observe consumer behavior introduce a myriad of practical challenges to conducting research. Perhaps most important among these is communication and coordination of logistics among all parties involved. This method requires a different skill set than that of traditional mail survey methods. Accordingly, researchers will need to adapt their efforts to incorporate the networking aspects of this method.

Collaborating with large retailers when conducting consumer research is both an asset and a liability. Although they present access to large numbers of consumers, they are often reluctant to allow research to be conducted in their stores. This reluctance comes from a number of concerns. For example, not wanting to harass their customers and not wanting to create extra work for employees are very real impediments to conducting research. The bureaucracy that comes with large retailers adds to the difficulty in conducting a research project. Once permission is obtained at the corporate level, the battle is only begun in facilitating the project at individual stores. Local store personnel may only see the project as an added hassle in their already hectic and demanding jobs. Finding a "champion" of the project in each store is critical for success.

Ideally, observations from an experiment should be tied to individual responses. While this was accomplished in the Virginia Tech study, it came at a very high cost. Other methods should be explored to successfully link the purchase decision with consumer data while at the same time keeping costs to a feasible level for academic research.

Despite the difficulties in conducting and interpreting results from the research projects just discussed, we feel these efforts, along with other similar efforts elsewhere, are an important contribution to the field. Learning from the challenges experienced by others should help facilitate future research and assure continued efforts in increasing the sophistication of research in the field.

(1) A complete report on this research can be found in a PhD thesis by Roy Anderson: An Analysis of Consumer Response to Environmentally Certified, Ecolabeled Forest Products. Oregon State Univ., Corvallis, OR.

(2) A complete report on this research can be found in a Working Paper by S. Gomon and R.L. Smith: Purchase of Certified Wood Products and Consumer Understanding of Forest Certification. Center for Forest Product Marketing, Virginia Tech, Blacksburg, VA.

Literature cited

Ajzen, I. and M. Fishbein. 1980. Understanding Attitudes and Predicting Social Behavior. Prentice-Hall, Englewood Cliffs, NJ.

Anderson, R.C. and E.N. Hansen. 200_. The impact of environmental certification on preferences for wooden furniture: a conjoint analysis approach. Forest Prod. J. (in press).

Bush, R.J., S.A. Sinclair, and P.A. Araman. 1991. Determinant product and supplier attributes in domestic markets for hardwood lumber. Forest Prod. J. 41(1):33-40.

Eastin, I.L., C.L. Lane, R.D. Fight, and J. Barbour. 1998. An assessment of the industrial markets for softwood clearwood lumber. Forest Prod. J. 48(11/12):48-54.

Fell, D. and C. Gaston. 2001a. Material selection in outdoor projects. Forintek Report. Forintek Canada Corp., Vancouver, BC, Canada.

Fell, D. and C. Gaston. 2001b. The do-it-yourself market in Western Canada. Forintek Report. Forintek Canada Corporation. Vancouver, BC, Canada.

Forbes, C.L., L.G. Jahn, and P.A. Araman. 2001. An investigation of hardwood plywood markets. Part 2. Fixture manufacturers. Forest Prod. J. 51(6):25-31.

Forsyth, K., D. Haley, and R. Kozak. 1999. Will consumers pay more for certified wood products. J. of Forestry. 9(2):18-22.

Haas, M.P. and P.M. Smith. 1997. Global markets for U.S. hardwood components. Forest Prod. J. 47(3):45-51.

Hansen, E.N. 1997. Forest certification and its role in marketing strategy. Forest Prod. J. 47(3):16-22.

___________ and R.J. Bush. 1996. Consumer perception of softwood lumber quality. Forest Prod. J. 46(10):29-34.

LaPiere, R.T. 1934. Attitudes vs. actions. Social Forces 13:230-237.

Lamson, J. 1997. Bringing the message to market: What consumers think about certified wood products. Understory 7(2):1-2.

McDaniel, C. and R. Gates. 2000. Marketing Research Essentials. Southwestern College Publishing, Mason, OH. pp. 183-185.

Michael, J.H. and P.M. Smith. 1995. Information sources used by furniture retailers: The importance of furniture markets. Forest Prod. J. 45(4):25-30.

Ozanne, L.K. and R. Vlosky. 1997. Willingness to pay for environmentally certified wood products: The consumer perspective. Forest Prod. J. 47(6):39-48.

Pakarinen, T. 1999. Success factors of wood as a furniture material. Forest Prod. J. 49(9):79-85.

Ramsey, F.L. and D.W. Schafer, 1997. The Statistical Sleuth: A Course in Methods of Data Analysis. First Ed. Duxbury Press, Belmont, CA. 742 pp.

Rich, S.U. 1970. Marketing of Forest Products: Text and Cases. McGraw-Hill, NY.

Sheatsley, P.B. 1974. Survey design. In: Handbook of Marketing Research. R. Ferber, ed. McGraw Hill, NY.

Smith, P.M. 1991. The Washington State wood pallet industry. Forest Prod. J. 41(5):39-44.

Smith, T.M. 2002. An analysis of northern hardwood lumber buyers' use of electronic commerce. Forest Prod. J. 52(2):62-68.

Solomon, M. 2002. Consumer Behavior: Buying, Having, and Being, 5th ed. Prentice Hall, Upper Saddle River, NJ.

Stureson, F.N. and S.A. Sinclair. 1991. United States consumers' views on ready-to-assemble furniture. Wood and Fiber Sci. 23(2):207-219.

Sudman, B. 1998. Marketing Research--A Problem Solving Approach. McGraw Hill, NY. pp. 157-158.

Taylor, R.E. 1999. Wood Markets: The Solid Wood Products Outlook 2000-2004. Inter. Wood Markets Research Inc., Vancouver, BC, Canada.

Teisl, M.E. 2003. Eco-labeled forest products: Are consumers not listening, or are producers not communicating? Invited paper, presented at the New England Society of American Foresters Winter Meeting, Burlington, VT.

Vlosky, R.P. and P.M. Smith. 1994. The retail homecenter perspective on point-of-sale scanning of UPC barcoded wood products. Forest Prod. J. 44(7/8):27-33.

Weinfurter, S. and E.N. Hansen. 1999. Softwood lumber quality requirements: Examining the supplier/buyer perception gap. Wood and Fiber Sci. 31(1):83-94.

Wicker, A.W. 1969. Attitudes versus actions: The relationship of verbal and overt behavioral responses to attitude objects. J. of Social Issues 25(4):41-78.

Winer, R.S. 2000. Marketing Management. Prentice Hall, Upper Saddle River, NJ.

Zikmund, W.G. 1994. Exploring Marketing Research. 5th ed. The Dryden Press, Fort Worth, TX.

Roy C. Anderson*

David Fell*

Robert L. Smith*

Eric N. Hansen*

Stephanie Gomon

The authors are, respectively, Forest Products Marketing Specialist, Montana State Univ. Extension Service, Missoula, MT; Market Researcher, Forintek Canada Corp., Vancouver, BC, Canada V6T 1W5; Associate Professor, Dept. of Wood Science and Forest Products, College of Natural Resources, Virginia Tech, Blacksburg, VA 24061; Associate Professor, Dept. of Wood Science and Engineering, College of Forestry, Oregon State Univ.; and Millwork Sales Representative, Babcock Lumber, Pittsburgh, PA. Portions of this research were funded by the NRI Competitive Grants Program/USDA, grant number #00-35400-9141. This paper was received for publication in October 2003. Article No. 9778.

*Forest Products Society Member.