Community Education > Assessing Grower Adoption of IPM Systems in the Northeastern U.S.A.
Identification of Future Research, Training, and Extension Needs.

ACKNOWLEDGEMENTS

The authors wish to acknowledge the substantial contributions made to this research by the following cooperators in participating states: Steven Alm, Univ. of Rhode Island, Lorraine Berkett, Univ. of Vermont, Jude Boucher, Univ. of Connecticut, Michael Brownbridge, Univ. of Vermont, James Dill, Univ. of Maine, Galen Dively, Univ. of Maryland, Francis A. Drummond, Univ. of Maine, Alan Eaton, Univ. of New Hampshire, Heather Faubert, Univ. of Rhode Island, Shelby Fleischer, Penn. State University, Eleanor Groden, Univ. of Maine, Vern Grubinger, Univ. of Vermont, Glen Koehler, Univ. of Maine, William Lord, Univ. of New Hampshire, Lorraine Los, Univ. of Connecticut, Chris Maier, Conn. Agric. Expt. Station, William MacHardy, Univ. of New Hampshire, Donald Prostak, Rutgers University, Leanne Pundt, Univ. of Connecticut, Joann Whalen, Univ. of Delaware. Special thanks to the hundreds of growers throughout the northeast who responded to our survey.

On behalf of itself and 9 other New England states, UMass, Amherst submitted and received a Phase I IPM Planning grant through the USDA. The purposes of the project were to facilitate development of multi institutional and multi disciplinary IPM implementation teams, to define IPM systems ready for adoption for five crops, to determine the current extent of adoption of these systems, and to prioritize research and extension needs which may be deterring greater adoption.

The process used included execution of a Dillman Method mail survey to growers of the five crops in participating states. After development of an initial draft of a survey instrument, state collaborators participated in several revisions of the survey questionnaire. In several cases, participants felt strongly about expanding the data collected beyond the scope needed to strictly meet the terms of the grant, and every effort was made to accommodate their interests. In an effort to achieve a number of finished surveys adequate to achieve statistical validity, nearly 3 times as many surveys were finally mailed (N=3,693) than had originally been proposed.

Once survey work was completed, all data (sweet corn, bedding plant, strawberry, potato, apple) were entered as Lotus files and checked for errors. SAS-generated data sets and summaries were run for all crops, sorting data by farm size, by state, and by state and size. Macros were written to export SAS files, allowing them to be read into EXCEL , and generate reports.

Other routines were written to assign points to survey responses of each grower and calculate percent adoption of available "IPM Practice Points" based on the Massachusetts commodity-specific IPM Guidelines. This allowed comparison of growers in different states of IPM adoption along a continuum from Low (0% to 33% of possible points), Medium (0ver 33% to 66%), to High (over 66% to 100%), as well as categorization of growers meeting the Massachusetts criterion of IPM adoption (70% of possible IPM points). Where IPM definitions do not exist (i.e., greenhouse bedding plants) the current IPM system will need to be defined, and practice points assigned before this process can go forward.

Response rates (% of sample returned) on a state-by-state basis ranged from 45% to 82%. Unfortunately, responses from between 7% and 40% of state samples, and 7% and 35% of original state mailing lists were unusable because respondents did not grow the crop in question. In spite of attempts to increase sample size, including sending surveys to the entire mailing list in some instances, finished survey numbers received for some states and some crops resulted in reduced confidence that responses reflect the larger population. However, the number of usable responses for all states combined participating in the apple, bedding plant, strawberry and sweet corn surveys allows us to operate at the 95% confidence level, with a margin of error of ±5%. For the two states participating in the potato survey, and for the state of Maine alone, sampling error at the 95% confidence level is ±10%.

State and/or regional meetings (or conference calls, etc) with non-LGU contributors still need to be carried out in order to further refine the IPM systems, and to determine if any modifications to prioritized research/extension needs are required.

Pest managers in New England and the mid-atlantic states, as elsewhere, are faced with many daunting problems, including: increasing levels of pest resistance to pesticides, environmental contamination, increasing regulation, media and consumer concern about food safety, and reduced availability of key crop protection chemicals. In addition, especially in the six New England states, agriculture is often conducted at the urban-rural interface. This situation has resulted in high development pressure on the best agricultural land, and frequent, often bitter conflict over aspects of normal agricultural practices taken for granted in more rural states.

However, a beneficial aspect of this proximity is that it provides direct access to markets, a fact which many of the regions' farmers utilize through diversified cropping, typically producing a mix of tree fruits, small fruits, vegetables and bedding plants, and selling largely or completely at retail, often through farms stands and farmer's markets. Members of a regional IPM implementation team believe that future IPM adoption will be enhanced by providing farmers with a suite of IPM systems that recognize the diversified, direct-sale nature of much of the region's agriculture. Hence, the organization of this proposal into commodity-focused "modules".

The objectives of the project were twofold: 1. To identify the IPM systems which are ready for adoption for five crops , and, 2. to determine the current extent of adoption of the defined IPM systems in several New England and Mid-Atlantic states.

SURVEY METHODOLOGY - The Dillman Total Design Method

To assess current IPM implementation, the states of Massachusetts, Maryland and Maine were able to draw upon surveys which had previously been conducted. However, to ensure consistent high-quality data on current implementation of IPM practices and priority needs for all states in the project, and to achieve a high response rate, growers of each commodity in each state were surveyed in a manner which allowed justifiable extrapolation for a relatively small sample to the total grower population. Because previous mail surveys conducted by Massachusetts (Paschall et al. 1992, Hollingsworth et al. 1992, Coli et al. 1996) had achieved return rates ranging from 55% to 75% using the Dillman Total Design Method (Dillman 1978; Salant and Dillman 1994), this methodology was again used to carry out the survey reported on here.

To achieve high rates of return, the Dillman method requires that key elements be followed. Research has documented that, while variation from the protocol is possible, each variation results in loss of potential respondents. Key conceptual elements of the method include: showing appreciation and regard for the opinions of the respondent, making the questionnaire interesting and brief (hence, reducing the physical and/or mental efforts to complete and return), and establishing trust by identifying with a legitimate or known organization or group.

In constructing questionnaires, another important aspect has to do with whether questions are open-ended, close-ended with ordered response choices, or close-ended with unordered response choices. Each may be appropriate, although each has both strengths and drawbacks.

For example, an open-ended question causes difficulty for the researcher in categorizing and summarizing responses, and subjecting them to statistical analysis. A close-ended question with unordered response choices (such as asking for respondents to assign priority to choices by ranking them in importance) is relatively difficult to answer, and is likely to "turn off" the subject, reducing the likelihood s/he will complete the survey.

Strict adherence to the method requires that the questionnaire itself be in a "booklet" format with no questions on the front or back pages, printed in a photographically reduced form, and using white or off-white paper. Ideally, the first questions should be "easy" in order to overcome respondent fears of difficulty or complexity, and questions should be grouped according to content. Detailed instruction as to how the researcher would like each question completed is also important. Questions should not overlap pages in order to reduce confusion or errors in completion.

One of the most important aspects of the method involves creating cover letters that personalize the inquiry, that convince the respondent that completing the survey is useful to them or some group with which they identify (e.g., sweet corn or strawberry growers in New England who are members of the New England Vegetable and Berry Growers Assoc., etc.), that their opinion is important, and that their response will be kept confidential. Although development of this "social utility argument" is not easy, the text provides examples of key phrases that can be modified to fit the survey population under investigation.

A Dillman survey can be costly because it requires an initial cover letter with the survey booklet and pre-stamped and addressed return envelope, a one-week follow up postcard to the entire sample, and a three week follow up (with a stronger "sales pitch", and another copy of the questionnaire) to those who have still not responded. Depending upon funding availability (and the fortitude of the researcher), a third or even fourth follow up may be desirable. All mailings are sent first class postage with personalized address and salutation, and all cover letters must be signed with "pressed blue ink "original signatures.

A 1995 Dillman survey to Massachusetts fruit growers served as an initial survey template which Module Leaders modified and sent to state collaborators. After development of an initial draft of a survey instrument, state collaborators participated in several revisions of the survey questionnaire. In several cases, participants felt strongly about expanding the data collected beyond the scope needed to strictly meet the terms of the grant (e.g., to collect baseline information on pesticide use and on sources of information, etc.) and every effort was made to accommodate their interests. As a consequence, however, some questions were not "good Dillman questions", a fact which likely influenced response.

For this project, we felt it important that growers in each state be contacted by a person whose name they knew, rather than an individual in Massachusetts who they likely had never heard of. Consequently, all cover letters were printed on letterhead of the state's land grant, and signed by the state lead person. This was a logistical problem which necessitated close coordination among states, and heavy reliance on overnight mail. Rapid turnaround was particularly important since all cover letters were personalized and dated. Appendices F, G, and H contain examples of an initial cover letter, a post card follow up and a three-week follow up cover letter respectively from a 1995 Tree Fruit Extension Team survey as examples.

According to the Dillman method, a population size estimated at 10,000 growers (estimated number of growers of the five commodities covered) in the Mid-Atlantic and New England states requires 370 completed, usable questionnaires in order to be 95% confident that response estimates have a sampling error of no more than ±5% (Salant and Dillman 1994, p.55). This is based on the assumption that individuals in the population are likely to vary in their use of and/or their attitudes to IPM practices. That is, the likelihood of an individual using or not using IPM is 50/50. If the same size population was less varied, that is, was likely to contain more IPM users than non-users, a smaller finished sample size (240) would be needed to achieve equal precision. Because we did not know a priori the amount of variation in the population, we chose to use the more conservative assumption. Also, to ensure that we had an adequate final sample size, we assumed a low response rate, and initially planned to send out approximately 1,400 surveys.

After initial stages of designing the survey instruments, it became evident that state collaborators were not content with statistical validity at the level of all growers in all states. Hence, we chose larger sample sizes in an effort to achieve statistical validity for each state and each crop (Table 1).

Because breaking the total population into smaller subsets reduces accuracy, a larger percentage needs to be sampled in order to reduce sampling error. Consequently, we ultimately sent out more than twice as many surveys (3,693) as originally planned. In some cases, questionnaires were sent to the entire original mailing list. Table 1 shows the number and total of surveys sent by state and by crop. The largest number of surveys were sent to Maine growers, followed by Massachusetts, Connecticut, Pennsylvania, New Hampshire, Maryland, Vermont, New Jersey, Rhode Island and Delaware.

In spite of attempts to increase sample size, including sending surveys to the entire mailing list provided by some states, finished survey numbers received sometimes resulted in reduced confidence that responses reflect the larger population.

* Unusable = those who returned surveys and checked that they do not grow sweet corn.
** Original list reduced by 10% to reflect proportion of those on original lists who returned surveys but who do not grow sweet corn.

However, the number of usable responses for all states combined participating in the apple, bedding plant, strawberry and sweet corn surveys allows us to operate at the 95% confidence level, ±5%. For the two states participating in the potato survey, and for the state of Maine alone, sampling error at the 95% confidence level is ±10%.

The effect of mailing list quality can be readily seen in the high percentage of usable sweet corn surveys from New Jersey, where the Extension Specialist personally went over the list to ensure that known sweet corn growers were being contacted (Table 2). In some other crops and states, up to 40% of respondents checked that they did not grow the crop (Tables 3-6). This likely has to do with the nature of many mailing lists, which are sent to vegetable growers, or small fruit and vegetable growers, not all of whom grow the crop of interest. For future surveys, it would be useful if extension mailing list renewals could include a request to list crops grown so that information distribution or generation could be better targeted.

* Unusable = those who returned surveys as requested and checked that they do not grow bedding plants.
** Original list reduced by 13% to reflect proportion of those on original lists who returned surveys but who do not grow bedding plants.

If this information were then stored in a searchable data base of mailing labels, ability to specifically target desired populations for various reasons (e.g., specialized training sessions, minor use or special local needs registrations, etc.) would be enhanced. This ability should also reduce costs associated with mailing information to inappropriate individuals.

Should this not be possible, states wishing to conduct surveys which generate a suitable number of usable responses should plan to increase sample sizes to account, not only for lack of response, but for a sizable unusable percentage. As we discovered, however, even mailings to an entire list do not always result in achieving a desired number of usable surveys.

* Unusable = those who returned surveys as requested and checked that they do not grow strawberry.

Data in Table 4 illustrates this point. In this case, additional follow up contacts would have been the only way to potentially increase usable numbers. Even then, there would be no way to know in advance if the percentage of growers/non-growers would have remained consistent, or if a larger percentage of non-respondents were growers who just did not want to take the time to complete the survey.

Another significant problem for a state wishing to conduct a large regional survey, is that some existing mailing lists are not maintained in an electronic format. In the case of this survey, hard copy mailing lists provided by some states, necessitated a significant amount of data entry, and raised the possibility that some surveys will not be deliverable. Unfortunately, in a few instances, surveys were returned by family members reporting that the addressee was deceased.

The apple survey (Table 5) achieved the highest percentage of usable responses. This is probably because, although not all vegetable growers produce sweet corn, and not all floriculture mailing list members grow spring bedding plants, most individuals calling themselves tree fruit growers probably grow at least some apples. Based on previous experience surveying this group, many of the unusable surveys were likely sent to people who no longer grow apples, or who remain on free newsletter mailing lists even after death.

* Unusable = those who returned surveys as requested and checked that they do not grow apple.
** Original list reduced by 17% to reflect proportion of those on original lists who returned surveys but who do not grow apple.
⁺ List A=162, List B=91
⁺⁺ 100% from List A, 50% from list B

* Unusable = Those who returned surveys as requested and checked that they do not grow potato.
** Original list reduced by 9% to reflect proportion of those on original lists who returned surveys but who do not grow potato.

Remaining Steps
Additional computerized routines which will assign and total points from the Massachusetts IPM Guidelines to individual responses, and allow categorization of IPM adoption along a continuum from Low (0% to 33%), Medium (over 33% to 66%), to High (over 66% to 100%) have been completed for sweet corn. This task still needs to be done for apples, strawberries and potatoes. No Guideline currently exists for bedding plants. Because it is not our intention to suggest that the Massachusetts Guidelines be adopted by all states growing similar crops, we hope that participating states will develop their own, state-specific guidelines (i.e., IPM systems ready for adoption) based on survey responses and/or other feedback. Our experience from an earlier project (with Maine, New Hampshire, Connecticut and Rhode Island) developing state-specific apple IPM guidelines, indicates that states doing so will find many practices and pests in common. Hence, developing a regional IPM guideline, which still recognizes differences in practices among states, is also possible.

A further importance task is to schedule and conduct state and/or regional meetings with growers, grower groups, consultants, etc., to review and further prioritize research/extension needs described by surveys.

Attitudinal Questions Common to all Surveys
In order to better understand IPM research, extension and training needs, all surveys asked for information on the respondent's experience with and attitudes toward IPM, and on relative importance of various pest management information sources. Responses presumably will enable IPM efforts to target issues or information sources which enhance ultimate understanding and adoption.

Table 7. Percent of grower agreement with positive and negative opinions about IPM use, by crop.