The complexity of constructing empowerment measuring instruments: a delphi study*

ABSTRACT

The value of empowerment as a central concept for social work implies the potential benefits of measurement to understand how this value is materialised in social work. The way in which instruments are developed and constructed impact their quality. With regard to the complexity of measuring empowerment, this article aims to provide insights in experts’ views on the theoretical requirements to measuring empowerment in social work. A Delphi study of 3 rounds was conducted to achieve consensus into the appropriate method of instrument development for empowerment. By retrieving, organising, exchanging and combining opinions and arguments from different experts, this study provides insights into their views on requirements when developing measuring instruments for empowerment. The results show elaborations on substantive requirements and considerations that should be taken into account when developing empowerment measures in social work and that have led to the development of new guidelines. These provide social work and its researchers the opportunity to construct instruments that align with methodological standards, theoretical requirements, and ethical principles.

SAMENVATTING

De waarde van empowerment als centraal concept in het sociaal werk, impliceert de potentiële voordelen van metingen om te begrijpen hoe deze waarde vorm krijgt in het sociaal werk. De manier waarop instrumenten worden ontwikkeld en geconstrueerd, is van invloed op de kwaliteit ervan. Met betrekking tot de complexiteit van het meten van empowerment, wil dit artikel inzicht geven in de opvattingen van experts over de theoretische vereisten voor het meten van empowerment in het sociaal werk. Hiertoe is een Delphi-onderzoek van 3 rondes uitgevoerd om overeenstemming te bereiken over de juiste methode voor instrumentontwikkeling voor empowerment. Door meningen en argumenten van verschillende experts op te halen, te ordenen, uit te wisselen en te combineren, geeft dit onderzoek inzicht in hun visie op belangrijke voorwaarden bij het ontwikkelen van meetinstrumenten voor empowerment. De resultaten laten uitwerkingen zien van inhoudelijke vereisten en overwegingen waarmee rekening moet worden gehouden bij het ontwikkelen van empowermentmaatregelen in het sociaal werk en hebben geleid tot de ontwikkeling van nieuwe richtlijnen. Deze bieden het sociaal werk en haar onderzoekers de mogelijkheid om instrumenten te construeren die aansluiten bij methodologische standaarden, theoretische vereisten en ethische principes.

KEYWORDS:

• Empowerment
• social work
• measuring instruments
• guidelines
• delphi

TREFWOORDEN:

• Empowerment
• Sociaal Werk
• meetinstrumenten
• richtlijnen
• Delphi

Introduction

The value of empowerment as a central concept for social work implies the potential benefits of measurement to understand how this value comes about. Reviews conducted in social work-related contexts provide insights into the multitude of measures for empowerment (Barr et al., 2015; Herbert et al., 2009; Noordink et al., 2021). The way in which these instruments are developed and constructed impact their quality (Noordink et al., 2019; Steenssens et al., 2017).

Empowerment has different values and effects when viewed from different perspectives. It is both an important process and an important goal in social work professional practice, as demonstrated in the international definition of Social Work, in which ‘empowerment’ is stated as a core concept (IFSW, 2014); this is supported by the World Health Organization (Capone & Petrillo, 2013). Empowerment also relates to socio-political perspectives. The relevance of empowerment’s socio-political meaning is reflected in work by Rappaport and Zimmerman (Rappaport, 1987; Zimmerman, 2000) who elaborate on the word ‘power’ and its various interpretations, on theories on blaming the victim versus blaming the system, and on the concept of empowerment as a way to analyse and handle social problems (Van Regenmortel, 2002). Finally, the value of empowerment for service-users, for those who require social support, has been discussed by Boevink (2017) who repositioned empowerment and its importance for mental healthcare, together with service-users. As such, it is important to understand to what extent measurement instruments account for these differing perspectives and values.

The complexity of empowerment theory

Studies show the importance and complexity of a theoretical concretisation of the concept to be measured (Christens, 2012; Cochran, 1992; Peterson, 2014; Zimmerman, 2000). Empowerment as a multi-level construct can be manifest at both individual and collective levels, and these levels are mutually connected and interdependent. They can be both cause and effect (Van Regenmortel, 2002). The multi-dimensionality of the construct shows the different dimensions of empowerment that together form the construct or that are a manifestation of the construct, depending on whether empowerment is interpreted as a formative construct or an aggregate construct (Christens, 2012; Peterson, 2014). An old lament about empowerment research that still holds is that it is mainly limited to the individual level of empowerment (Riger, 1993). Rather, empowerment research should align with a more ecological theoretical approach based on the principle that the individual and the social structures are not considered as two entities to be studied separately (Speer & Hughey, 1995). Furthermore, empowerment is an open-ended construct, meaning that it is context-specific and, as such, can have different meanings for different people and contexts (Rappaport, 1987). This also infers that empowerment is dynamic and can change over time, instead of having a fixed, static character. This open-ended character of empowerment also explains why universal measures are undesirable (Zimmerman, 1995), as instruments should ideally account for the specific characteristics of the target population and context. This was recognised by Spreitzer (1995), who emphasises that specific operationalisations of empowerment are not transferable to other organisations. This is also supported by Rogers et al. (1997), who found that pre-existing scales do not cover the interpretation of empowerment for the study’s target population. Akey et al. (2000), who developed a measure of empowerment for parents of children with a disability, also address the issue of generalizability, stating that the dimensions of empowerment may be displayed differently depending on the group of individuals and the context, as empowerment items are specific to a population.

Finally in this regard, empowerment can be both a process or a means to an end, and an outcome (Zimmerman, 2000). Narayan (2005) describes these processes as the instrumental value of empowerment and outcomes as the intrinsic value of empowerment.

The ethical aspect of measuring empowerment

In empowerment theory, it is important to emphasise the more ethical aspect of measuring empowerment and discuss the importance of target group involvement (Rappaport, 1987; Van Regenmortel, 2002). In line with the disability rights movements slogan nothing about us without us, which has since been used by many other interest groups, the position of the people of concern is fundamental to empowerment research. Rappaport (1987) emphasises this by stating that the people of concern should be treated as collaborators. Research should not just be about them; people have to be given opportunities to be involved in research activities. Those people whom programmes and interventions are about should have a primary role in developing future strategies by which they gain greater control over their lives (Cochran, 1992). The way in which research is conducted can either contribute or undermine the empowerment of those involved (Florin & Wandersman, 1990). At the very least, measuring empowerment should be in line with the experiences of the target population as they themselves would describe and recognise (Zimmerman, 1995).

Aim

Knowledge about how empowerment should be measured in social work is fragmented and insufficiently systematically linked. Apart from Steenssens et al. (2017) work, in which she and her colleagues took the initiative to develop theoretically substantiated requirements, not much integrated research has been done. The present research aims to provide insights into academic experts’ views on the theoretical requirements for measuring empowerment in social work.

Methods

The Delphi method was chosen as a suitable design for this study. A Delphi study uses a multi-round process for consensus-based group decision making. The core of this method is the systematic and repetitive consultation of experts to explore complex problems or relatively unexplored territory (Linstone & Turoff, 1975). Another key feature of the Delphi method is the principle of intersubjectivity of those involved; it is not facts that are collected, but rather expert opinions and arguments. By retrieving, organising, exchanging and combining these opinions and arguments, this study provides insights into experts’ views on theoretical requirements for the construction of empowerment measures in social work.

Participants

Respondents were recruited through a select sample. A targeted search was made for respondents who can be regarded as experts in relation to the empowerment paradigm, developing measurement instruments, and social work. Specific expertise on the combination of these three themes, summarised as ‘measuring empowerment in the context of social work’, is relatively scarce. Participants were considered to be experts if they demonstrably had expertise in at least two of the three sub-areas, as shown by having multiple publications in this regard, or by holding an academic position in relation to the central themes, such as full professors. This led to different combinations of expertise in the panel.

A total of 39 international academic experts were approached. Reviews were searched to create a valid respondent group (Barr et al., 2015; Herbert et al., 2009; Noordink et al., 2021). The scope was delineated by limiting the relevant context to OECD countries, in line with the scope of the reviews. Experts cited in these studies were approached to participate in the study, after which a snowball method was used to involve additional experts. A total of 18 experts committed and participated. Their characteristics are shown in Table 1. Reasons for not participating were mainly limited time, sick leave, or the respondent’s own assumed lack of expertise.

Table 1. Expert characteristics.

	Expertise	Nationality
Expert 1	S,E,M	The Netherlands
Expert 2	S,E	The Netherlands
Expert 3	S,E,M	The USA
Expert 4	E,M	The Netherlands
Expert 5	S,E,M	Belgium
Expert 6	S,E	Belgium
Expert 7	S,E	The Netherlands
Expert 8	S,E	The Netherlands
Expert 9	S,E	The Netherlands
Expert 10	S,E	The Netherlands
Expert 11	E,M	The Netherlands
Expert 12	S,E,M	The USA
Expert 13	S,E	Belgium
Expert 14	E,M	The Netherlands
Expert 15	S,E	Ireland
Expert 16	S,E	The UK
Expert 17	S,M	The Netherlands
Expert 18	S,E,M	The USA

Note. Expertise of experts: Social Work = S, Empowerment = E, Measurements = M.

This study was approved by the Ethical Research Committee of the HAN University of Applied Sciences under reference number ECO 259.04/21.

The process of data collection and analyses

As the collection and analysis of data in a Delphi study go hand in hand, alternate, and follow each other, this process is described in an integrated manner. This Delphi study consisted of 3 rounds. The results of each round were collected, processed, analysed and formed the input for the next round. This facilitated discussions between the panel members.

Round 1

In the first round, the emphasis was on collecting different contributions, opinions and statements, explaining the open-ended nature of this round (Custer et al., 1999). Theoretical requirements for measuring empowerment, established in previous research (Steenssens et al., 2017), were presented to experts in a questionnaire, asking them what they find important, redundant and missing. Responses were analysed based on a combination of an inductive and a deductive approach, using Atlas.ti. A deductive approach was used to classify the statements per predetermined category, in line with theoretical requirements (Steenssens et al., 2017). The responses were further analysed by means of an inductive approach: by bundling, ordering, comparing and combining all opinions and arguments per category into a core labelling scheme (Baarda et al., 2018). This resulted in an overview of coherent statements per category. A summary was then written for each category. The analyses after each round and the summaries were done by the first author and jointly studied and discussed with a co-author, as such applying the four-eyes principle. Any deviations in findings were discussed in search of consensus. The other two authors then provided the results of this exercise with feedback. The first author incorporated this feedback and resubmitted the analyses and summaries to the other authors until the result was satisfactory.

Round 2

In round two, 27 statements derived from the first round were presented to the respondents. Respondents were asked to score the extent to which they agreed with the statements on a 5-point Likert scale and to substantiate their score. The Likert scale was used to give researchers a specific impression of the extent of agreement; it was not used for quantitative analysis. Respondents were instructed as such. The scores and statements of all respondents were jointly studied to determine the extent to which respondents agreed with the statement and why or why not. Hsu and Sandford (2007) describe how in Delphi studies different standards can be applied to determine the extent of agreement necessary to accept consensus, varying from 70% to 80%. For this study, we chose a norm of 75%, a commonly accepted norm (Delbecq et al., 1975)

Round 3

In the final round, the combined results from rounds 1 and 2 were presented. The statements and issues on which consensus was achieved were described, giving the respondents a final opportunity to reflect on those statements. Statements or issues on which consensus was not achieved were elaborated by describing the various conflicting points of view and arguments. Again, respondents were asked to reflect on the ambivalences. This resulted in an overview of reflections on the issues on which consensus had been achieved, and also in reflections on those issues on which respondents are ambivalent. These reflections were ordered per statement and substantively analysed in search of similarities and differences in points of view and overarching patterns per topic.

Results

The integrated results of all three rounds were used as the results of rounds 2 and 3 did not contrast with those of round 1, but reinforced, complemented and nuanced them. On those topics on which consensus was not achieved, the three consecutive rounds led to substantiations, nuances and organised arguments per statement. The integrated results are presented per subject, both on aspects on which consensus was reached and aspects on which disagreement persisted.

Purpose of empowerment measure

Without exception, experts agreed that developing a measure for empowerment should always be preceded by first establishing why empowerment is measured and in whose interest it is. Reasons can include gaining knowledge and understanding of sociological issues, but also to improve services. Measures being used for the assessment and sanctioning of clients or professionals were considered undesirable by the panel. Nevertheless, if a practice did not contribute to empowerment, even though intended, and if this was made transparent by means of measurement, practices should be adjusted accordingly. It is the co-responsibility of researchers who measure empowerment to prevent results of measurements from being abused, within the limited scope of their influence. This risk should at least be described in publications about the instrument.

Process or outcome

Of the 18 respondents, eight agreed that an instrument should always specify its intention and thus whether empowerment is operationalised as a process or as an outcome. The emphasis is on the recognition that empowerment can be both a means and an end, but also on the fact that this distinction must be clear in the measuring instrument.

Eight respondents indicated that they were neutral regarding this statement; three defined empowerment as either a process or an outcome – not both. Two explicitly disagreed with the statement; they noted that empowerment can be both a process and outcome, like other concepts, and that it is therefore not a specific characteristic of empowerment. The question remains whether empowerment processes can actually be measured, or only monitored when ongoing. Outcomes can be operationalised in observable, measurable entities, which experts found more complex for the empowerment processes that lead to outcomes. Above all, researchers should be transparent in this regard and elaborate on which choices are made and why.

Experts showed an interest in more research about jointly mapping empowerment processes as well as the results of those processes, while acknowledging the complexity. Processes and outcomes do not always need to be mapped out together, as the way in which something is achieved and the final result may be independent of each other.

Defining, demarcating and operationalising empowerment

When developing a measure for empowerment, it is important to make the link to a theoretical framework with which the concept of empowerment can be defined and operationalised. This was seen as obvious by the experts, something that should apply to every construct.

Experts recognised that when developing an empowerment measure, it is important to consider the assumed direction of influence between the construct and the underlying dimensions, as it is a latent construct. Experts emphasised that researchers should determine whether they conceptualise empowerment as a superordinate construct, an aggregated construct, or as a series of different constructs.

A caveat was made concerning the assumption that empowerment is a stable and constructed concept, while empowerment explicitly means something different per person, group or context. However, experts also stated that, at a meta level, common threads could be found, for example the multi-levelness and multidimensionality that can be recognised in definitions and operationalisations of empowerment.

In discussing whether reduction of empowerment levels is permitted, concerns lay in the risk that by reducing empowerment to an individual level, problems and concerns will also be individualised and the influence of organisational or community-level factors on individuals might then be ignored. This entails the risk of blaming the victim. As a result, the need to also include the mutually influencing effect of factors at the organisational or community level on psychological empowerment was supported by the majority. The question of how these layers then relate to each other remains. In this regard special attention was paid to determining the function of researching empowerment; to gain theoretical or empirical knowledge in relation to empowerment and as such studying sociological issues, or to contribute to the empowerment of people at an individual and community level.

A compromise position proposed by an individual and recognised by the group, suggested that when developing a measure of empowerment, it seems important to define the target group by determining whose empowerment is being measured, or who the benefit group is, which does not reduce the need to identify factors and influences at a societal level, for example, when considering the empowerment of an individual. Only one respondent disagreed with this compromise position.

There was discussion amongst the experts on the statement that a reduction of the dimensions of empowerment is not recommended if the aim is to measure empowerment as a whole. Twelve respondents agreed, four disagreed and two were neutral. Experts added that reduction is always the norm. Sub-areas, dimensions or components are always measured, and are then combined into a larger whole, for example ‘empowerment’.

Furthermore, the idea of demarcating empowerment to only one of its dimensions when developing measuring instruments is based on the assumption that empowerment can be divided into levels and dimensions and that, in a methodological sense, these dimensions jointly form the whole. The latter was explicitly disputed by two experts. They stated that empowerment is more than the sum of its parts. Other experts refuted this by stating that structural models can be developed precisely for this purpose, in which empowerment is included as a latent construct and, as such, measured as a whole.

Context specificity

Thirteen of the respondents agreed that it is important to make the context explicit and they noted ways of achieving this, for instance by describing for which setting the instrument was developed, for which target group, and for which question or purpose. There was mutual recognition that by describing the context as explicitly as possible, context specificity and transferability were increased. The most important repeatedly mentioned critical note deals with the question ‘to what extent can the context be operationalised so specifically that it is still valid for a certain group'? Common circumstances or factors do not necessarily mean that an instrument is suitable for a group of individuals, collectively understood as a target group.

A compromise position put forward by a few and recognised by 11 respondents, stated that when developing a measure for empowerment, it is important as a researcher to do everything possible to make the context explicit in such a way that it is as close as possible to the specific circumstances and characteristics of the target group. This is not to say that an instrument has to be tailor-made for the entire target group, but that an attempt has been made to find a balance between transferability and context specificity.

Target group involvement

When developing an empowerment measure, possibilities must be created for and with the target group to participate meaningfully in its construction. Many experts went even further and argued that a partnership approach should be the goal or that even regardless of the goal, collaboration with the target group should always be strived for. They stated that measuring empowerment should be seen as a participatory process, as empowerment processes are about developing a say in living conditions and decisions. It should be noted that there is always a risk of unnecessarily burdening the target group.

Experts recognised that phases and ways in which the target group can make a meaningful contribution must be determined together with the target group and must be in line with its strengths and limitations. For example, it is conceivable that research technical aspects should be done by researchers, while discussions and interpretations can be done jointly. Target group involvement should always go beyond ‘item checking’ and even beyond ‘item construction’. Based on arguments about empowering effects of participation, the target group should also be allowed to influence the broader exploration and interpretation of data and the application possibilities of the results.

Additionally, with the right support, the target group can be involved in and influence everything. The question of what added value this has for the people themselves has to asked. In line with this, target group involvement in empowerment research must not be done just because of methodological values such as increasing context specificity or validity. Experts underlined that researchers should guard against the risk that target group involvement in empowerment research is abused by using the target group participation strategically, for the sake of appearance. Comments in this context mainly added to why target group involvement can be relevant, for example to bring about action and change, but also to serve both methodological goals and client-related goals. It can achieve both and therefore does not have to be ‘either or’.

There was discussion about the extent to which target group involvement should have an empowering experience on those directly involved, although 10 experts explicitly agreed. The main criticism lay in the fact that this depends primarily on the research goal. If one of the goals is to increase empowerment of those involved, then participation in that research project is needed to contribute to this.

There are a multitude of reasons for involving the target population in this process. Firstly, methodological reasons such as crossing different types of knowledge, i.e. academic and experiential knowledge, leading to a more solid and robust instrument in which potential respondents will recognise themselves more easily were mentioned. Secondly, experts noted ethical-normative reasons such as giving voice to voiceless, shared decision-making or collaborating with all involved. Respondents emphasised the importance that contributing to empowerment of the target group is only one of those reasons.

Another important criticism can be summed up in that participation in instrument development does not automatically lead to empowerment. Only when participation in these research projects also leads to being allowed to influence decision-making and opportunities, can participation contribute to empowerment of those involved. As such, it would be good to create opportunities for the target group through which their participation in instrument development could lead to empowerment, without stating that involvement automatically leads to empowerment, but that it should at least be a positive experience.

Building on existing scales

In regard to the field of tension between context specificity and building on more generic, broader applicable instruments, opinions were divided, but patterns were noted. There appeared to be two groups with opposing views.

Nine experts argued in favour of continuously developing instruments to be as context-specific as possible, whereby items and scales are tailored to the specific target group for which the measurements are taken. The main message was to aim for the greatest possible context specificity, a tailored measure. This group held an even firmer position: that it is not possible to develop an instrument in such a way that it matches the specific circumstances of the person to be measured on the one hand, and yet can be used for a collective on the other. The criticism is mainly based on the idea that context specificity in one instrument is not feasible unless it is so narrowly delineated that it results in minimal data. Four experts formed the other group; they reasoned more from the idea that in a methodological sense, scales and items can be re-used provided these scales and items are then adapted to the specific group. The wheel does not need to be reinvented, it seems, provided the correct methodological steps are taken. Respondents did recognise the principle that a balance must be sought between ‘striving for the greatest possible context specificity’ and developing more widely applicable, usable instruments. Using a clearly defined theoretical framework can help in creating opportunities to develop widely applicable instruments and enhance transferability, after which items and interpretations are constructed together with the target population and in line with their perception, creating context-specificity.

Representativeness

All experts recognised that target group representativeness is important, where the emphasis must be on substantive representativeness. This means making an effort to include those who are easily excluded, and not choosing the path of least resistance.

There was discussion about whether it is important that attention is paid to differing points of view and opinions about the theme, and whether these different perspectives would ideally be given a place in a measuring instrument, or that attempts should be made to achieve a ‘negotiated truth’. However, a large majority, 12 experts, agreed with this position. Criticism regards that the extent to which a negotiated truth should be pursued and thus a multi-perspective on empowerment is sought, again, depends on research goals.

It was stated that a negotiated truth should pay attention to facilitating an equal contribution and position for the various parties and points of view involved. Especially in cases where the stances of researchers, social workers, and services users are integrated, equality should be actively facilitated. It must not be assumed that it is the researcher or client who decides, but that everyone involved has or should have an equal vote. That being said, the researcher should eventually take responsibility for choices made and be transparent about how these choices came about and why.

Discussion

In this study, an attempt was made to determine, together with a group of international experts, what the appropriate route would be for the development of measures for empowerment. The results confirm that measuring empowerment involves more than a set of methodological conditions. Measuring empowerment is about power and counter-power and has a socio-political or ethical component that must be considered when developing measures. It is about the relationship between researcher and participant and why a measure is being developed and in whose interest this is. In order to consider all these aspects, it is precisely the empowerment construct that needs to be viewed from multiple perspectives.

The results show consensus on many themes and issues. Where there was disagreement, results did reveal patterns which gave insights into ambivalences and issues to consider when developing future measures. Combined, the statements on which consensus was achieved and the patterns that could be distilled from the statements where consensus was not achieved, can form input for guidelines for the development of empowerment measures. The findings, insights and interpretations of the results are further discussed per guideline. It is worth noting that these guidelines are an interpretation of the results by the authors, not by the expert panel.

1. Determine who is the beneficiary of this exercise and what is the purpose of measuring empowerment in relation to this

This may seem obvious, but the process of developing empowerment measures should always be preceded by first establishing why empowerment has to be measured and in whose interest it is. It is important to consider that there can be a multitude of reasons for measuring empowerment, varying from gaining knowledge, learning and understanding, to improving services, health care and social work, or all of these. In this regard, power relations should be considered when determining why empowerment is measured. Empowerment measurements carry the risk that they will be used for the assessment and sanctioning of clients, professionals or organisations, which is undesirable. Sanctioning should never be the result of measurements; making meaningful adjustments to social work and its working methods should.

• (2) Determine the theoretical foundation and definition of empowerment for the instrument

A starting point for developing an instrument should be to determine the theoretical framework with which empowerment can be defined and operationalised. This again may seem obvious and it does not only apply to the construct of empowerment, but to how the open-endedness and complexity of this construct emphasise the explicit necessity. A clear theoretical framework also creates opportunities to give meaning to the construct together with the target population and from their perspective, in a transparent manner.

• (3) Determine if this instrument aims to measure empowerment as process or empowerment as end to itself

Although the panel did not reach consensus in relation to this statement, an in-depth analysis of their opinions shows that the debate on empowerment being a process, an outcome, both or neither, is vivid and relevant. We believe that whatever your point of view, it is important to determine and be transparent about how empowerment is considered within the instrument to be developed, as this choice influences how and why a measurement is conducted. This aligns with the notion that experts consistently agreed that within a study it should always be transparent and imitable which choices have been made and why. If the aim is to measure the extent to which a person or collective perceives to be empowered, outcomes are measured. If the aim is to investigate the effect of empowering factors, events or processes or the relation between empowerment processes and other goals, empowerment is seen as a means to an end and measured as a process. It is clear that outcomes can be measured and that processes lead to outcomes, they correlate. In that sense empowerment processes can provide explanations for how outcomes arise. It is worth considering jointly mapping empowerment processes as well as the results of those processes, which gives insights into how the relation between these processes and outcomes come about. Above all, researchers should be transparent in this regard and elaborate on which choices are made and why.

• (4) Determine the substantive scope of the instrument

The empowerment literature is built on the notion that empowerment is a multilevel construct, distinguishing at least an individual level and a collective level. Reducing the level of analyses is appropriate when empowerment outcomes are measured. This then measures how empowerment is perceived and viewed by that person or that group; the object of research. Level reduction is then interpreted as ‘whose empowerment is being measured’. Level reduction should not mean that the influence of factors, for example, at the organisational or community level on psychological empowerment can be ignored, especially when looking for explanatory models or when looking at the interaction between these levels in another way. In this way, researchers can prevent the risk of empowerment being individualised and that the causes and consequences are also reduced to the individual, bringing forth a risk of blaming the victim.

When empowerment is considered as a process and an instrument aims at mapping these empowering processes, a reducing the level of analysis is not appropriate as a variety of factors at an organisational, community, or societal level effect the object of research.

The different levels of empowerment can be operationalised in dimensions or components. Reducing the dimensions of empowerment is not recommended if the goal is to measure empowerment as a whole. If the aim is to only measure that specific dimension, then in a methodological sense this is possible, but care should be taken not to interpret those results as empowerment. Again, the complexity of defining and operationalising a multi-layered construct such as empowerment requires a certain amount of modesty and as such, transparency regarding how choices came about.

• (5) Determine the relation between the construct and its dimensions for this instrument

Empowerment is seen as a latent construct, meaning one that is not directly observable, but one that can be inferred from other variables that are observed. The relationship between the construct to be measured and the dimensions in which the construct is operationalised is important in this case. In a methodological sense, empowerment can be conceptualised as a superordinate construct, meaning that empowerment is manifested by its dimensions, as an aggregated construct, i.e. that empowerment is formed by its dimensions or as a series of different constructions (Peterson, 2014). When developing an empowerment measure, it is important to consider the assumed direction of influence between the construct to be measured and the underlying dimensions.

• (6) Define and describe the context as explicitly as possible

Empowerment theory suggests that a no universal measure for empowerment is possible, since empowerment has different meanings and content per context, group and person (Van Regenmortel, 2002; Zimmerman, 1995). In this regard it is important to define and describe the context as explicitly as possible, in line with what Geertz (1973) refers to as a ‘thick description’. This opens the door to creating instruments that are highly context specific, but that also enhance the possibility of transferability. As stated by the panel, it is important to do everything possible to make the context explicit in such a way that it is as close as possible to the specific circumstances and characteristics of the target group, without claiming that an instrument is therefore tailor-made for that entire target group. In that sense, the objective is to find a balance between transferability and context specificity.

• (7) Determine the extent to which the target population will be involved in the development of the measuring instrument and why

There are many reasons for including the target population in the process of instrument development. In a methodological sense, involving the target population can contribute to developing context-specific items and scales and developing valid instruments. In a more ethical sense, involving the target population can give voice to the group; this creates opportunities to empower them, to involve them in decision making, and give them a say in living conditions. Collaborating with the target group in various phases of research can lead to an applicable, valued, and accepted instrument (Depauw & Driessens, 2017). A good aim is to create opportunities for the target group through which their participation in instrument development could lead to empowerment, to their influencing decision-making and asserting control. These opportunities should align with strengths, possibilities and expertise of the target group. Giving meaning to empowerment and being involved in the interpretation of results and decision-making seem more promising aspects for the target population to participate in than data collection or analyses, for example. But even in this sense it is important to determine with the target population if, how, when, and why they want to be involved. It is the responsibility of researchers to take into consideration that the target group should not be burdened unnecessarily, therefore involving the target population should not be done just to tick boxes or for the sake of appearance. Regarding representativeness, care must be taken to avoid the risk that only ‘the usual suspects’ are involved, while the actual ‘silent or silenced voices’ that the target group also contains are excluded. This may also mean that researchers must make efforts to make involvement accessible to people with, for example, a disability.

It is imaginable that different perspectives of groups like service users, social workers and researchers are considered when constructing a measure depending on the aim of the measure, and thus if a multi-perspective on empowerment is sought. If so, care should be given to facilitating an equal contribution and position for the various perspectives, in order to achieve a negotiated truth on the elaboration of empowerment in the instrument. It is arguable whether combining or integrating the different perspectives on reality can result in a joint, evenly shared perspective on the matter (Chapin, 1995). Nevertheless, a focus on including and involving different and diverging input from all parties, instead of striving for one solution can lead to policy and practice based on a ‘negotiated truth’, based on dialogue between all parties as equal partners (Van Regenmortel, 2002). Given this, the researcher should take responsibility for choices made and be transparent about how these choices came about and why.

• (8) Determine the extent to which existing scales will be used in the development of the instrument and how data from these scales will be translated into new contexts

Context specificity is an important principle when developing empowerment instruments. Ideally, scales are tailored to the specific target group in question. However, pre-existing scales can be of value as they can form a starting point for the construction of new items. This does not mean that items can be transferred without alteration to a new context, but items can form inspiration and a starting point or conversation tool for redesigning items that do align with the new context.

In achieving context-specific instruments that still result in a foundation for more widely applicable instruments, it may be necessary to find a balance between theory on one hand, which is a certain level of abstraction and which can describe and compare empowerment outcomes, and interpretation on the other hand, which can describe context-specific or context-transcending explanations of these outcomes. In this way, a balance can be sought between striving for the greatest possible context-specificity and developing more widely applicable instruments. Using a clearly defined theoretical framework can help create opportunities to develop widely applicable instruments and enhance transferability, after which items and interpretations are constructed together with the target population and in line with their perception, creating context-specificity.

Strengths and limitations

This study explored a range of expert opinions in relation to a complex issue; how should empowerment measures ideally be developed. The Delphi method seems an appropriate strategy to do so, though its critics emphasise the subjective nature of this approach as the results are completely opinion-based. However, this methodological approach does not pretend to create new knowledge or be hypothesis-testing; it is a strategy for exploring or exposing underlying assumptions or information leading to differing opinions and to collecting information which may generate a consensus from an expert panel (Linstone & Turoff, 1975). The Delphi method in this study included three rounds; a fourth round was considered, but given the degree of repetition and apparent saturation in round 3, this was considered unnecessary. In the final round, respondents often referred to pregiven answers or indicated that they had ‘no more additions’.

Another caveat concerns the selection of experts. As the data consists of opinions, this selection is decisive for the results. It is conceivable that other respondents would have had different opinions. However, the international and diverse character of the group suggests that the experts participating in the study form a good representation, although it is important to note that the expert panel only represents OECD nations, in an attempt to delineate the scope of this study. The snowball method led to including experts with a variety of expertise in methodology, empowerment theory, and social work.

A final limitation is formed by the possible influencing role of the researchers. Although the experts responded to the various rounds without the researchers’ guidance, it was the researchers who organised, compared and processed the data into the following Delphi rounds. The integration of the results is therefore partly influenced by interpretations from the researchers.

However, the method itself can compensate for this possible bias, as the interpretations of the researchers in between rounds formed input for the new round and in this way the experts were given opportunities to react to and correct these interpretations. Furthermore, data processing in between rounds was discussed with several researchers in order to compensate for the intersubjective character of the method.

Conclusion

This study describes a set of guidelines by which measuring instruments for empowerment in social work can be constructed and how these guidelines came about. These guidelines provide social work and its researchers the opportunity to construct instruments that align with methodological standards, theoretical requirements, as well as with ethical principles.

Acknowledgement

We would like to thank the expert panel for their extensive contributing to this study: Dr. Boomkens, Prof. dr. de Brabander, Prof. dr. Christens, Dr. Damen, Depauw, MSc., Prof. dr. Hermans, Dr. Hubers, Prof. dr. Kaulingfreks, Dr. Van der Laan, Prof. dr. Linders, Dr. Moser, Prof. dr. Peterson, Prof. dr. Roose, Prof. dr. Scholte, Dr. Shier, Dr. Spencer, Prof. dr. van Weeghel and Prof. dr. Zimmerman.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Thomas Noordink

Thomas Noordink is a doctoral researcher at the Academic Collaborative Centre Social Work, Tranzo, Tilburg School of Social and Behavioral Sciences, Tilburg University, the Netherlands and lecturer at HAN University of Applied Sciences, the Netherlands.

Lisbeth Verharen

Lisbeth Verharen is professor Strengthening Social Quality at HAN University of Applied Sciences, the Netherlands.

René Schalk

René Schalk is associated with the departments of Human Resource Studies and Tranzo at Tilburg University, the Netherlands. He is extraordinary professor at the Faculty of Economic and Management Sciences of NorthWest University, South Africa

Tine Van Regenmortel

Tine Van Regenmortel is professor at the Faculty of Social Sciences KU Leuven and Head of the Research Group Social and Economic Policy & Social Inclusion at the Research Institute for Work and Society of HIVA KU Leuven, University of Leuven, Belgium; and professor Social Work at the Academic Collaborative Centre Social Work, Tranzo, Tilburg School of Social and Behavioral Sciences, Tilburg University, the Netherlands.