3.5 MEASURES
3.5.2 Measures: Stage B
3.5.2.1State affect
Warr et al.’s (2014) Multi Affect Indicator was utilised to measure the three affective states
pertaining to this research: High activated unpleasant affect (anxious, tense), high activated pleasant affect (enthusiastic, inspired), and low activated pleasant affect (relaxed, calm). The dimension of low activated unpleasant affect was not relevant to the hypotheses and was therefore not measured to allow for a quick response time needed of an ESM. The utilisation of just one or two items for each construct is typical in ESM designs (cf. Fisher & To, 2012), due to the repetitive nature of the surveys and the short-time frame in which respondents are reporting (To, Fisher, et al., 2015). Twice per day participants were instructed via the ESM application to “indicate the extent which [they] feel this way right now… on a five point Likert scale 1 representing very slightly or not at all, and 5 extremely. The Cronbach’s alpha coefficients indicate good reliability for these scales, with scores of 0.88, 0.76, 0.81 for high activation pleasant affect, low activation pleasant affect and high activation unpleasant affect respectively. This compares well with the alpha coefficients reported by Warr et al., (2014) which range between 0.79-0.90, 0.75-0.86, and 0.80-0.87 across six studies.
Numerous issues have been identified in the self-reporting of affective experiences (Dasborough, Sinclair, Russell-Bennett, & Tombs, 2008). Self-reports, although common place, are used at the cost of greater objectivity and comparability. The main failing of self- reports in the context of affective research is that it assumes individuals are able of appraising their emotions accurately in the moment of the experience (Dasborough et al., 2008; Frijda,
1986). This concerns concepts such as “emotional awareness” (2008, p. 201) as well as
memory bias. The second being that it relies on the individual’s unfiltered sharing of their affective experience. Issues surrounding the concealment of appraised affect may be of particular cause for concern within contexts where affect is stigmatised (Warr et al., 2014), such as in the military forces. It is not expected that an entrepreneurial context would stigmatise affect at levels greater than general workplaces, but it still possible that some responses will be the result of filtering and may demonstrate some social desirability bias. Notwithstanding, the use of self-reported measures of behaviour and affect is not without its merits. Particularly in organisational research, the mere use of self-reported ESM surveys
conducted via smartphones is hailed as “innovative” (Rodrigues, Kaiseler, & Queirós, 2015, p. 204), as it reduces memory recall issues and smartphone applications tend to be mobile- optimized, making prompt responses possible in daily-life research. Furthermore, the ability to collect multiple data across the same day, and then for multiple days in the week, exceed the typical cross-sectional single survey approaches.
3.5.2.2Innovative behaviour
Self-reported innovative work behaviour was utilised in this research. The rationale behind this, briefly discussed next, is that the collection of data via objective sources, or external reports, is controversial in an entrepreneurial, and short-time unit of analysis, context. Objective business performance measures, although useful for analysing gross outputs such as sales (Warr et al., 2014), are difficult to use within short time frames. Further, subjectively measuring increments in innovation within a timescale of half a day is difficult to assess.
“External ratings” as opposed to self-reports was not a viable for this research either, as
entrepreneurs generally do not have supervisors, and may (Harper, 2008) work alone. Additionally, the likelihood of colleagues or supervisors accurately reporting innovative work behaviour is questionable (See Janssen, 2000, p. 292 for a disccusion) thus self-reported innovative work behaviour is considered a satisfactory measure in workplace research.
Innovative work behaviour was measured with five of De Jong and Den Hartog’s (2010) ten
item innovative work behaviour measure. Only five items were used to allow for rapid responses required of a high-frequency experience sampling methodology (Fisher & To,
2012) as discussed previously. Every second item of De Jong and Den Hartog’s measure was
utilising – which results in each dimension of innovative work behaviour being included in
the research. Questions were slightly adapted to reflect short time periods. Items included: “during the past few hours, to what extent have you… Searched out new working methods, techniques or instruments…. Generated original solutions for problems... Wondered how things can be improved... Attempted to convince people to support an innovative idea.... Put effort in the development of new things” (1 = none, to 5 = a great deal). Cronbach’s alpha coefficients between the five items were high 0.88, this was slightly higher than the scores reported by De Jong and Den Hartog (2010) which ranged from 0.60 to 0.74 across studies. Interestingly although the authors sought to divide the dimensions of innovative work behaviour between activities relating to idea generation, idea exploration, idea championing
dimensions (De Jong & Den Hartog, 2010). Therefore during this research there has been no attempt to divide the items by dimensions, for reasons of measure reliability.
3.5.2.3Sleep quality
The Pittsburgh Sleep Diary (Monk et al., 1994) item “how do you evaluate this night’s
sleep?” (1 = very poor, to 5 = excellent) was utilised to measure sleep quality each morning via the ESM application. This measure of sleep quality has demonstrated validity as a single item and been utilised widely (Hahn, Binnewies, Sonnentag, & Mojza, 2011; Park & Sprung, 2015; Sonnentag et al., 2008).