Study context
The study focuses on farmers in the state of Iowa in the US upper Midwest. This region, also referred to as the Corn Belt, contains some of the most fertile agricultural land in the world. While the Corn Belt encompasses a fraction of US farmland, it produces over half of all US corn (sp. Zea mays) and soybean (sp. Glycine max) (NASS 2009, 2011). Iowa has long been the top producer of corn and soybean in the US (USDA ERS; NASS 2011). It is also a state that has increasingly experienced extreme weather events that have had adverse impacts on agricultural production and the natural resource base (ICCIC 2011).
Weather changes attributed to climate change in Iowa include warmer winters, longer growing seasons, higher dewpoint temperatures, higher annual streamflow, and more frequent extreme weather events (ICCIC 2011). Some of these changes—such as a longer growing season—may be favorable to Iowa agriculture. Others, such as increased early-season rainfall, dew-points, and temperatures, can lead to negative impacts such as increased soil erosion, delays in planting or crop damage, higher incidence of pests and disease, and heat-induced pollination problems (Hatfield et al. 2011; Rogovska and Cruse 2011). Iowa had experienced all of these in the years leading up to the survey, incidence of extreme rainfall and flooding, in particular, and thus represents an excellent setting for research on farmer attitudes toward adaptation and mitigation in agriculture.
Data
The data for this research were collected through the 2011 Iowa Farm and Rural Life Poll (IFRLP), an annual statewide survey of Iowa farmers conducted by Iowa State University Sociology Extension. Iowa Agricultural Statistics administers the survey following a survey-postcard-survey mailing protocol. The 2011 survey was mailed to 2,030 farmers in February 2011. Useable surveys were received from 1,276 farmers, for a response rate of 63 %.
Research questions, variables included in the model, and hypotheses
Two related research questions guide this research: (1) do Iowa farmers support actions aimed at climate change adaptation and mitigation; and, (2) are beliefs and concerns about climate change associated with support for or opposition to those activities. Analysis focuses primarily on the relationships between three measures of attitudes toward adaptive and mitigative activities and variables that measure farmer beliefs regarding climate change and concerns about potential impacts on Iowa agriculture. This section introduces the variables employed in the analyses, provides descriptive statistics, and outlines the hypothesized relationships between the dependent and independent variables.
Dependent variables
Adaptation
Two dependent variables measure support for adaptation to climate change and one measures support for mitigation. All three variables consist of statements that elicited farmer agreement or disagreement on a five-point scale from strongly disagree (1) to strongly agree (5). The two adaptation statements focused specifically on potential responses to intensification of rainfall, a primary impact of climate change predicted for Iowa (ICCIC 2011). The first, “Iowa farmers should take additional steps to protect their land from increased precipitation,” measures belief that farmers ought to prepare farmland for potential future increases in precipitation (PROTECT). In the Iowa context, such protective steps consist of establishment or improvement of conservation practices including grassed waterways, terraces, and riparian buffers (Kling et al. 2007) many of which are also promoted as adaptive and/or mitigative actions for agriculture (Lal et al. 2011). A majority of farmers (62 %) agreed or strongly agreed that farmers should take additional steps to protect land (Table 1).
Table 1 Percentage distributions for dependent variables
The second adaptation statement, “Iowa farmers should increase investment in agricultural drainage systems (tile, ditches) to prepare for increased precipitation,” (DRAIN) is particularly salient for Iowa farmers given the hydrological history and current profile of Iowa’s most productive agricultural lands. Prior to European settlement, much of Iowa consisted of humid prairie whose soils contained extraordinary amounts of organic matter (Mutel 2007). These rich soils were largely unsuitable for agriculture, however, until artificial drainage systems were constructed beginning in the late 1800s. The drainage of the state, an engineering feat that prior to WWI equaled the cost of the construction of the Panama Canal (Peterson and Englehorn 1946, 23), unlocked extraordinary productivity and made Iowa a leading agricultural state (Peterson and Englehorn 1946; SCS 1983). Thus, drainage represents a longstanding engineered solution to the problem of “excess” water in Iowa and a logical adaptive strategy for maintaining productivity in the face of future increases in precipitation. Forty-six percent of participants agreed or strongly agreed that Iowa farmers should increase investment in drainage to prepare for increases in rainfall (Table 1).
Mitigation
The third dependent variable measures farmer support for action to mitigate the causes of climate change (MITIGATE). There are a number of policy measures that are commonly considered and/or implemented to reduce GHG emissions in agriculture. Three of the most widely recognized options are legislative mandates to reduce emissions, emissions taxes, and market-based cap and trade (CAST 2011). Because each of these options requires substantial government intervention to implement, we use a general measure of support for government action to address the anthropogenic causes of climate change.
Farmers were provided with the statement, “Government should do more to reduce the nation’s greenhouse gas emissions and other potential causes of climate change,” and asked to rate their agreement on a five-point scale from strongly disagree to strongly agree. It is important to reiterate that support for mitigation implies an acceptance that humans are causing climate change (at least to some extent), and that changes in human behavior are needed to reduce those contributions. Support for government action implies agreement that collective action is necessary to incentivize, regulate, or otherwise induce those changes in behavior. Thirty-three percent of participants agreed that government should take action to curb human contributions to climate change (Table 1).
Independent variables
Perceived vulnerability and concern
Concern about the impacts of climate change on agricultural livelihoods is hypothesized to be a critical arbiter of farmer support for adaptive or mitigative actions (Howden et al. 2007). Attitudes can be strong predictors of behaviors or acceptance of ideas (Ajzen 1991; Dietz et al. 2005). Studies of agricultural conservation practice adoption have found positive associations between awareness of environmental problems, attitudes toward potential solutions, and willingness to adopt those solutions (Prokopy et al. 2008). Furthermore, it is when situations come to be perceived as problems that attitudes regarding potential ameliorative actions are more predictive of behavior change (McCown 2005).
Four variables measuring constructs of concerns about the potential impacts of climate change on a five-point agreement scale (strongly disagree = 1 and strongly agree = 5) are included in the model. Three of these variables are combined into a summative scale that measures overall concern about potential negative impacts of climate change on agriculture. The three statements and their percentage distributions are provided in Table 2. The summative scale for the three items, labeled “CONCERN,” has a Cronbach’s alpha reliability coefficient of 0.86, indicating a highly acceptable level of internal consistency. The scale ranges from 3 (very low concern) to 15 (very high concern), with a mean of 9.5.
Table 2 Means and percentage distributions for continuous covariates
Human ingenuity
A second measure of concern is a single statement: “Climate change is not a big issue because human ingenuity will enable us to adapt to changes” (INGENUITY). This item was developed from the “New Environmental Paradigm (NEP)” literature that has long examined human-environment relations (see Dunlap 2008). Specifically, the item draws on the “human exemptionalist” thread of that literature highlighting a widely-held belief that “…humans—unlike other species—are exempt from the constraints of nature” (Dunlap et al. 2000, 432). The item measures, on a five point scale ranging from strongly disagree (1) to strongly agree (5), the degree to which farmers believe that “human ingenuity” will reduce threats from climate change by facilitating adaptation to changes. Thirty-three percent of farmers agreed with the statement, and the mean score for the item was 3.0 (Table 2).
Beliefs about climate change and its causes should logically influence attitudes toward adaptive and/or mitigative actions to address it. Beliefs are generally conceptualized as understandings of the state of things in the world, or estimates of reality as individuals perceive it based on accumulated knowledge (Dietz et al. 2005; Wyer and Albarracin 2005). In turn, beliefs influence the formation of attitudes toward objects or actions (Kruglanski and Stroebe 2005) and behavioral decisions (McCown 2005).
A multi-dimensional variable that measures both beliefs about the existence of climate change as a phenomenon and attribution of cause is employed to model the relationship between beliefs about climate change and attitudes toward potential avenues of adaptation and mitigation. Study participants were directed to select one of five response options that best reflected their beliefs (variable labels in parentheses): climate change is occurring, and it is caused mostly by human activities (HUM); climate change is occurring, and it is caused equally by natural changes in the environment and human activities (HUMNAT); climate change is occurring, and it is caused mostly by natural changes in the environment (NAT); there is not sufficient evidence to know with certainty whether climate change is occurring or not (UNCERT); and, climate change is not occurring (NOCC).
A substantial majority of farmers (68 %) indicated that they believe that climate change is occurring (Table 3). Ten percent attributed climate change mostly to human activities. Thirty-five percent believed that climate change is occurring and caused equally by human activities and natural variation. Twenty-three percent attributed climate change primarily to natural causes. Slightly more than one-quarter (28 %) of farmers indicated uncertainty, and 5 % did not believe that climate change is happening.
Table 3 Observed frequencies for categorical covariates
Control variables
Three variables are included as controls: acres of row crops (i.e., corn and soybeans) farmed (ROWCROP), age (AGE), and education (EDUC). These are included because they are standard measures of capacity to adopt conservation practices (Prokopy et al. 2008). Acreage farmed has generally been found to be positively related to conservation practice adoption. For this study, row crop acreage might also be considered to be a measure of exposure to climate change-related risk (i.e., waterlogging, soil erosion). Age has been associated with lower rates of practice adoption, while education is generally seen as a positive predictor. On average, participants were 65 years old, farmed 350 acres of row crops (Table 2), and 58 % had completed at least some college (Table 3).
Hypotheses
Our hypotheses regarding the relationships between attitudes toward adaptive and mitigative management and beliefs and concerns regarding climate change and its potential impacts are:
-
H1:
Concern Apprehension about the potential negative impacts of climate change (CONCERN) will be positively associated with support for land protection (PROTECT), investment in agricultural drainage (DRAIN), and government efforts to reduce GHG emissions and other drivers of climate change (MITIGATE);
-
H2a:
Human ingenuity and adaptation Belief that human ingenuity will facilitate adaptation to climate change (INGENUITY) will be positively associated with support for land protection (PROTECT), and investment in agricultural drainage (DRAIN);
-
H2b:
Human ingenuity and mitigation Belief in the adaptive power of human ingenuity will be negatively related to government actions to reduce drivers of climate change (MITIGATE). The negative relationship between faith in human adaptive ingenuity and support for mitigation is proposed because the human exemptionalist paradigm is generally based on adapting to changes rather than changing behaviors that cause environmental problems;
-
H3a:
Beliefs about climate change Belief that climate change is occurring (HUM, HUMNAT, NAT) will be positively associated with both adaptation variables (PROTECT and DRAIN); and,
-
H3b:
Beliefs about causes of climate change Belief that climate change is occurring and attributable to humans (HUM, HUMNAT) will be positively related to support for public action to mitigate causes (MITIGATE), while belief that it is occurring but attributable to natural causes (NAT), is not occurring (NOCC), or there is insufficient evidence (UNCERT), will be negatively associated with support for mitigation.
Statistical approach
The goal of our statistical model is to identify important relationships among beliefs about the existence of climate change, concern about impacts, and attitudes toward different types of action in response to climate change. We analyze responses for the dependent variables using a cumulative logit model, also known as proportional odds logistic regression (Agresti 2007). The cumulative logit model is a useful option when the response variable consists of ordered categories, such as the five-level agreement scale for the climate change action items. Denote y
i
as the response of subject i to each dependent variable. The possible responses are defined as
$$ {y_i}=\left\{ \begin{array}{*{20}c} 1\ \mathrm{if}\ \mathrm{subject}\ i\ \mathrm{chooses}\ \mathrm{Strongly}\ \mathrm{Disagree} \hfill \\ 2\ \mathrm{if}\ \mathrm{subject}\ i\ \mathrm{chooses}\ \mathrm{Disagree} \hfill \\ 3\ \mathrm{if}\ \mathrm{subject}\ i\ \mathrm{chooses}\ \mathrm{Uncertain} \hfill \\ 4\ \mathrm{if}\ \mathrm{subject}\ i\ \mathrm{chooses}\ \mathrm{Agree} \hfill \\ 5\ \mathrm{if}\ \mathrm{subject}\ i\ \mathrm{chooses}\ \mathrm{Strongly}\ \mathrm{Agree} \hfill \\\end{array} \right. $$
(1)
The response follows a multinomial distribution with individual cumulative probabilities \( {p_{i,j }}=\Pr \left[ {{y_i}\leq j} \right] \). That is, p
i,j
is the probability that subject i responds in category j or below. The cumulative logit model allows the log-odds of these probabilities to be functions of covariates x
i, according to
$$ \log \left( {\frac{{{p_{i,j }}}}{{1-{p_{i,j }}}}} \right)={\alpha_j}-\mathbf{x}_{\mathrm{i}}^{\prime}\boldsymbol{\upbeta} $$
(2)
The cumulative logit model is fit for PROTECT, DRAIN, and MITIGATE. The covariates used in each model include CONCERN, INGENUITY, the five BELIEF categories (HUM, HUMNAT, NAT, UNCERT, NOCC), and the three control variables (ROWCROP, AGE, EDUC). The models are estimated via maximum likelihood, and individual covariates are tested for significance via a likelihood ratio test (Agresti 2007). A bootstrap procedure is used to estimate predicted response probabilities \( {{\widehat{\pi}}_{i,j }}=\Pr \left[ {{y_i}=j} \right] \) and confidence intervals for different combinations of the covariates.