Theoretical Construct

The present study set out to measure kindness in individuals, however, with this being a multifaceted construct it was crucial to identify a singular definition allowing accurate measurements to be made, thus the questionnaires construct of kindness was defined by Eisenberg’s (1986, p.63) as ‘voluntary, intentional behaviours that benefit another and are not motivated by external factors such as rewards or punishments’. Individuals who elicit kindness are suggested to perform good deeds and favours for others whilst being concerned and compassionate about their welfare, which may or may not involve monetary costs (Park, Peterson & Seligman, 2004), friendliness, soft-heartedness, good natured and helpful have also been identified to stimulate kindness in individuals (Norman, 1963). A noticeable relationship was identified between these characteristics as they all involve a lack of self-interest, as Eisenberg (1986) recognised external factors such as punishment and reward must not play an influencing role in ones effort to be kind whilst others identify the need to put others welfare before their own (Park, Peterson & Seligman, 2004). When reviewing previous literature an absence of psychometric tests measuring kindness as its own distinct construct became apparent, as previous tests often included kindness as a characteristic of other constructs such as self-compassion (Kotsou & Leys, 2016), or are designed for an exclusive population such as schools (Binfet, Gadermann, & Schonert-Reichl, 2015), however the usefulness of researching personality constructs such as kindness is viewed as crucial in developing our understanding of identity and the self (Kraus & Sears, 2008). Thus, the present psychometric test does not measure kindness in conjunction with other constructs and does not direct itself towards a specific population, but instead allows kindness to be measured on a multidimensional level, being applicable to a wider population.

Item Construction

It was important that we based our work on a single definition of kindness, thus after researching and debating on the many different perspectives of kindness we believed Eisenberg’s (1986) definition fit best with the focal points we intended on measuring; behavioural, cognitive and emotional processes in kindness that is not influenced by external reward. After further brain storming and group discussion based on personal experiences we created a preliminary item pool of 30 items, 10 for each focal point, we decided a larger item pool was best to start with as it would allow us the chance to edit and cut items further along the process (DeVellis, 2012), we also agreed by including 3 focal points it would enable the identification of any correlation between the separate processes and would amplify the need for such psychometric testing (Clark & Watson, 1995).

We then proceeded by deciding a 5 point Likert scale (1 being strongly disagree and 5 being strongly agree)(Likert, 1932) would be most beneficial when asking participants to show their level of agreement, we believed it was necessary to include of a mid-point of ‘3; not certain’ as it is a valid stance for respondents to choose as often when individuals are put on the spot they can be unsure of the answer they wish to give, it is also useful in self-assessment as individuals may feel uncomfortable answering truthfully to the items they are presented with. However 9 of the items were negatively worded which would later need recoding, we used this technique as it helps diminish the impact of social desirability through making it less clear to the respondent how to score themselves in the way they wish to be viewed, reverse scoring also induces the thought process of respondents as they must consider the meaning of each item in more depth (Anastasi, 1968).

Once we had completed the creation of our item pool we needed to check face validity, subjective knowledge that our test was accurately measuring the construct of kindness, thus we sent our item pool to an expert for review, taking on board the feedback they gave we decided to omit 3 questions as they were not viewed as relevant to our construct (see appendix….).

Validity and Reliability

In order to obtain both validity and reliability of our psychometric test, we must first test reliability as an unreliable test cannot be a valid (Hale & Astolfi, 2014). Cronbach’s Alpha (Cronbach, 1951) is the most commonly used objective measure of reliability and would show if our psychometric test is free from error, determining whether the same construct has been measured throughout (Tavakol & Dennick, 2011). In order to use it we would firstly need to recode our negatively worded items through SPSS (2013). Through running the Cronbach’s alpha on SPSS (2013), an alpha score would be produced (), for our psychometric test we would expect α to meet the criteria of +0.07 or above as this shows a high correlation of results, leading to internal validity being achieved (Maltby, Day, & Macaskill, 2013).

We would then continue by testing content validity as this depicts how trustworthy the results are in measuring all facets the measure intends to (Cook & Beckman, 2006), we would therefore send our item pool alongside our formulated plan and definition of kindness to a panel of experts in the field of kindness and psychometrics, allowing them to examine, in depth, the degree to which we have represented the facets we intend on measuring (Hale & Astolfi, 2014), as they will rate each item against our definition and advise on the areas which they believe need improvement (Maltby, Day, & Macaskill, 2013; Lawshe, 1975). This will then allow us to refine our item pool by rewording ambiguous items or simply removing those that do not represent our construct or facets of measure.

Limitations

It is important to highlight the limitations that arose from the present study, especially due to measuring personality constructs. John and Robins (1994) suggested the main concern of studying constructs such as kindness evolves from the respondent themselves; due to individuals having an internal distortion of their personal character that they can be unaware of, this then leads to the individuals attempting to create a desirable persona by responding more positively to items, due to this distortion being internal it is incredibly hard to manage and therefore leads to the questioning of credibility of results. Kagan (1988) also questions whether individuals have the capability of providing accurate information about their personality, suggesting they do not have the correct self-awareness to complete self-report measures.

Other limitations present themselves from using a likert scale due to the absence of qualitative data, this causes issues for the respondent as they are unable to give reasoning for their desired answer and could lead to selecting an incorrect response, previously this has shown to lead to contradictions of data (Ogden & Lo, 2012). There is also ambiguity in the assumption that all respondents will assume equal distance between response categories, as some may perceive the weighting of strongly disagreeing completely different to strongly agreeing, thus effecting the results (Turner, 1993). The functionality of scales has also come under scrutiny when reverse scoring of items has been included as previous research has shown the positive and negative language used in such items could affect the response given due to wording, thus we would need to take extra caution when assessing the use of our likert scale (Sliter and Zickar, 2014).

Future development of the present study would need further reliability and validity tests to ensure suggested standards. One test we would not advise using is test re-test reliability, the degree to which results are consistent over time, due to maturation effects being evident during the interval of first and second testing from exposure to natural stimuli e.g. a family death can cause alterations from initial measurement, although a shorter time interval may be used, practice effects and memory of the first test has been shown to affect the second measure thus leading to unreliable data (Rosenthal & Rosnow, 1991). However we believe conducting predictive validity, whether the measure can accurately predict future outcomes, would be beneficial as it could help determine whether individual’s kindness levels are consistent overtime or whether it correlates with seasons of the year, such as Christmas, thus contributing to the ever-growing use of researching constructs, like kindness, in developing a wider knowledge of the self and our identity (Kraus & Sears, 2008).