Sources and descriptions of constructs in model

In this study, the psychological constructs being studied are unobservable. As a result, these variables need to be inferred from measured variables. Hair et al. (1998) describes how one of the first steps in SEM is to list the constructs that the model will comprise of. This process begins with the theoretical definition of the constructs involved. These constructs are then operationalised by selecting the measurement scale items and scale type. Often a number of scales are available to researchers although in some situations due to a lack of established scales a researcher must develop a new scale or significantly modify an existing scale to a new context (Hair et al. 1998). Punch (2003) recommends that where possible, the use of existing instruments is preferable to developing one’s own, especially when working with established variables. In this study scales from previous research involving values, the TPB and NAM have been used, some without modification although the majority have been modified to reflect a different research context.

The questionnaire for this study (see Appendix 1) consisted of 43 pre-coded items measuring 14 latent variables and 4 items to record respondents’ demographic information. 12 latent variables comprised of at least 3 measurement items while two were measured with a single

measurement item. The psychological constructs in the questionnaire, their source and description are discussed below.

Measurement items and scales for the three values types - transcendence, self-enhancement and openness to change – were adopted from Hunecke, Haustien, Böhler and Grischkat (2010) and measures the importance of these values as guiding principles in a respondent’s life. Each value type was measured with three items from Schwartz’s value inventory on a nine point scale (-1 = opposed to my values, 0 = not important, 3 = important, 6 = very important, 7 = of supreme importance). The Finnish translation was obtained from the study of Puohiniemi (1995).

The ‘attitude to engaging in pro-environmental air-travel related behaviours’ construct was adapted from Fielding et al. (2008) and captures attitudes respondents have towards engaging in pro-environmental air-travel related behaviours. This construct is assessed using five semantic differentials on a seven point scale from 1 to 7. For example: I think engaging in pro-environmental air-travel related behaviours in the next 12 months is: good / bad (1 = extremely good, 7 = extremely bad).

Items for the personal norms construct were adapted from three different studies – De Groot and Steg (2009), Harland et al. (1999) and Steg and de Groot (2010) - as a result of no single scale appearing entirely appropriate. This construct captures the extent to which a respondent feels a sense of personal obligation to engage in pro-environmental air-travel related behaviours. Items were assessed using four items on a seven point scale from 1 to 7. For example: I would feel guilty if I did not engage in pro-environmental air-travel related behaviours in the next 12 months (1 = strongly agree, 7 = strongly disagree).

Injunctive norms captures a respondent’s perception of the extent to which important others approve of engaging in pro-environmental air-travel related behaviours and was adapted from Fielding et al. (2008). Injunctive norms were assessed using three items on a seven point scale from 1 to 7. For example: If I engage in pro-environmental air-travel related behaviours, people who are important to me would (1 = completely disapprove, 7 = completely approve).

Descriptive norms were adapted from Nigbur et al. (2010) and captures a respondent’s perception of the extent to which other people engage in pro-environmental air-travel related

behaviours. Descriptive norms were assessed using three items on a seven point scale from 1 to 7. For example: Most other travellers engage in pro-environmental air-travel related behaviours (1 = completely disagree, 7 = completely agree).

The ‘self-identity’ construct was adapted from Terry et al. (1999) and captures the extent to which air-travel is an important part of a respondent’s self-identity. Self-identity was assessed using three items on a seven point scale from 1 to 7. For example: To engage in air-travel is an important part of who I am (1 = no, definitely not, 7 = yes, definitely).

Perceived behavioural control was adapted from Fielding et al. (2008) and captures a respondent’s perceived capability of engaging in pro-environmental air-travel related behaviours. Perceived behavioural control was assessed using five items on a seven point scale from 1 to 7. For example: For me to engage in pro-environmental air-travel related behaviours in the next 12 months is (1 = very difficult, 7 = very easy).

Past flying behaviour was developed from Ouellette and Wood (1998) and is a proxy measure for the strength of a person’s air-travel habit. Past behaviour (flying) was assessed by a single item measuring the number of flights a respondent had taken in the previous year on a seven point scale from 1 = ‘None’ to 7 = ‘11 or more’.

Past offsetting behaviour was adopted from Whitmarsh and O’Neill (2010) and is a proxy measure for the strength of a respondent’s habit of not purchasing carbon offsets. It was assessed using a single item that asked whether a respondent had ever purchased carbon offsets in the past.

The three behaviour intentions – the intention to fly less frequently, the intention to use alternatives and the intention to pay to offset emissions – were adapted from Fielding et al.

(2008). The three intentions capture the extent to which a respondent intended to perform one of the three corresponding pro-environmental air-travel related behaviours. Each intention was measured with three items on a seven point scale. For example: I intend to fly less frequently (or not at all) in the 12 months (1 = extremely unlikely, 7 = extremely likely).