basic issues in measuring gender attitudes

Post on 23-Mar-2016

56 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

DESCRIPTION

Basic issues in measuring gender attitudes. Ko Oudhof Statistics Netherlands. What will I tell you?. Just for a start: your own contribution Subjective indicators What are attitudes Measurement issues Analytic issues Here and there: gender/ international comparability. - PowerPoint PPT Presentation

TRANSCRIPT

Basic issues in measuring gender attitudes

Ko OudhofStatistics Netherlands

workshop UNDP/UNECE gender statistics 2004

What will I tell you?

• Just for a start: your own contribution

• Subjective indicators

• What are attitudes

• Measurement issues

• Analytic issues

• Here and there: gender/ international comparability

workshop UNDP/UNECE gender statistics 2004

Before I tell you anything

• Think of one short statement on the role of women or men in decision making that according to yourself would make it possible to distinguish respondents into advocates and opponents of gender equality by looking at their (dis)agreement with your statement

workshop UNDP/UNECE gender statistics 2004

Introductory vocabulary

• Indicator

• Cognitive

• Evaluative

• Affective

• Item

• Scale

• One digit, evaluating, goal-related

• About seeing, knowing and thinking

• About good or bad

• About like or dislike

• Statement or question

• List of items

workshop UNDP/UNECE gender statistics 2004

Policy and role of indicators

Selection policy goals

Assessing policy process

Evaluate policy results

workshop UNDP/UNECE gender statistics 2004

Subjective indicators and policy-1

Selection policy goals what do people (not) want? (Worries, aspirations,

satisfactions) what do people need or get rid off? (immaterial

needs, happiness)

Assessing policy process Public support (trust, support) Assess course of policy (predictions, prognosis)

Evaluate policy results Goal attainment (health, inequality, perceived

safety, xenophobia)

workshop UNDP/UNECE gender statistics 2004

Subjective indicators and policy-2 No ‘objective’ observation?

Subjective condition real policy objective Direct measurement Both subjective and objective indicators depart

from implicit assumptions on each other in some implicit psychological model on behavior!

Vague? Limits to aggregation! Measuring all possible wrongs? Indicators with a large mandate needed

Statistical weaknesses No money and no counting Monetary value or size of subjective condition? Specific measure and methodology Experts needed

workshop UNDP/UNECE gender statistics 2004

Subjective indicators

• Policy-relevance (issues)• Need- or Behavior-related (predictability)• Variability (daily fluctuations versus almost

invariable states)– Now – indicators – In these times – indicators– Long term perspective - indicators

workshop UNDP/UNECE gender statistics 2004

Subjective conditions and the world

• Needs and wants• Emotions• Perception• Experience• Learning• Motives• Goals• Etc.

• Fysical environment

• Social environment

Now -response

Structured

stable

behavior?

Now - feedback

workshop UNDP/UNECE gender statistics 2004

Attitudes (common elements in most definitions)

• Oriented on object, person, institution or event

• Evaluative component• Cognitive component • Affective component • Stable condition or construct• Intermediary between object stimulus

and behavioural response: consistency

workshop UNDP/UNECE gender statistics 2004

Relatives with likeness Opinions (now)

stability less more cognitive and not always evaluative behavioral relation weaker

Values (long term) general and less object-oriented stability higher behavioral relation more indirect

Norms (derivative) prescription of behavior stability higher behavioral relation stronger and more direct less cognitive and less affective

workshop UNDP/UNECE gender statistics 2004

Relation subjective elements

Abstraction

Time

opinion

attitude

value

workshop UNDP/UNECE gender statistics 2004

social norm

attitude

perceived behavioural control

behavioural intention behaviour

Model Theory Planned Behaviour (Ajzen)

workshop UNDP/UNECE gender statistics 2004

attitude-object

cognitive responses/

considerations

affective responses/

considerations

attitude (evaluative response)

behavioural response

weight

weight

General Model (Van der Pligt & De Vries)

workshop UNDP/UNECE gender statistics 2004

Ajzen & Fishbein, 2004

Relation attitude – behavior ( reasoned action approach in 2004)

workshop UNDP/UNECE gender statistics 2004

Attitudes and gender policy

Hardly any NSI Why gender attitudes? Attitude change as objective? Defensive in discussion? Same question elsewhere?

Macro-economy: confidence consumers/producers Business world: marketing Politics : voting behavior Health: perceived health Crime: feeling of insecurity

workshop UNDP/UNECE gender statistics 2004

Gender attitude research and tools in practice Mainly academic or ad hoc research Few international research projects Gender role (labor market or household)

main topic Hardly any standardisation Example: attitudes on female decision

making Support preferential policies Attitudes among decision makers Acceptance of female management Effects of leadership styles

workshop UNDP/UNECE gender statistics 2004

Engendering attitudes

• Objects• Explaining behavior• Measurement tools• Analysis• Interpretation• Presentation

• Gender Issues• Engendered concepts• Gender validity• By sex or more*?• By sex or more*?• By sex or more*?

More = differences compared to other non-gendered research as consequence of earlier steps

workshop UNDP/UNECE gender statistics 2004

Measurement of attitudes

Explicit measurement(under conscious control respondent) one item multi-item

Implicit measurement(without conscious control respondent) observation of behavior (non-obtrusive) bodily response response latency Academic research and less relevant for

statistical offices etc.

workshop UNDP/UNECE gender statistics 2004

Quality of measurement - reliability

– equal outcomes of tool when measuring the

same?

– random error

– inter-items reliability

– test-retest / split-half

– interobserver reliability

– quality measure versus external factors

workshop UNDP/UNECE gender statistics 2004

Quality of measurement - validity

– Similar results from other tools when measuring the same

– Systematic error

– Construct validity

– convergent validity – what should

– divergent validity - not what should not

– Predictive validity

– Multitrait-multimethod matrix as solid validity-testing design

workshop UNDP/UNECE gender statistics 2004

Survey?

• Insight in own attitude/opinion• Can they express the attitudes/opnions:

– personal conditions (e.g. ability)– situational conditions (e.g. individual interview?)

• Plausibility true answering– personal conditions (e.g. strategic response)– situational conditions (e.g. interviewer interaction)

• Alternative informants/ assessing documents • General considerations on survey design

workshop UNDP/UNECE gender statistics 2004

Single item or multi-item measurement?

• Quick Cheap

• all or nothing, also in time-series

• one-dimensional

• sometimes quite high and reliable

• how do you assess psychometric properties

• Response time Expensive

• Shortening scale generally possible

• Multidimensional

• Scale properties can be assessed

• International comparability and standardisation of scales (or subscales)

workshop UNDP/UNECE gender statistics 2004

Multi-item variants

Osgood scale Thurstone scale Likert scale Guttman scaling multi-object

measuring monetary methods

(WTP)

• General dimensions

• Pretested dichotomous scaled items

• Addition of multi-point (3-100) items

• Scaled statements

• Conjunct / dominance/ similarity

• Simulated markets/ hedonic price analysis/ contingent valuation (CV) or ranking (CR)

workshop UNDP/UNECE gender statistics 2004

Likert scale

• Rather simple • List of items expressing positive and negative

opinions on attitude object• Selection of relevant items by content• Choice of answering categories

– Number – meaning of scores– middle category– don’t know: yes or no

• Scale rating by summing item values (after recoding)• Self-made or standard?

workshop UNDP/UNECE gender statistics 2004

Selection of items

• Relevant for all groups (e.g. young + old)• Clear and unequivocal interpretation• No multiple question items• No double negations• No questions but statements (response set)• No confirmation bias pos + neg• Time spans: now/these days/whole life• Suggestive expression (most people…)• Biased or suggestive answering categories• Personalised or public statements (Hakim)

workshop UNDP/UNECE gender statistics 2004

More possible interferences

• Character of survey (crime or labour?)• Interviewer• Order of topics in questionnaire• Introduction of scale• Interference of different topics in one scale• Order of items• No repeats or redundancy• Social desirability overreporting or

underreporting

workshop UNDP/UNECE gender statistics 2004

So you’ve got your data

• Assessing or reassessing quality of scale? 1. Reliability aspects2. Validity aspects

• Deciding what to do considering– Objectives (employer/ supervisor)– Tools (standards?)– Methodological explanations– Explanation of results

workshop UNDP/UNECE gender statistics 2004

Item and scale analysis

• Assessing reliability of scale as given– Depending on design– Without any validity analysis of scale

• (re)assessing items + scale(s)– linearity and other assumptions?– multidimensional?– dropping items possible?– selection of techniques to assess scale

• Consistency/ homogeneity items • Analysis content via Princ.Comp./ factoran./

scaling

workshop UNDP/UNECE gender statistics 2004

Scale ratings

• Which ratings should be used?– sum– weighted sum (only part of items needed?)– factor scores

• To be used for what?– is level relevant? (breakdowns or time series)– is level confusing? (comparability)– nature of audience (general public or scientists)

workshop UNDP/UNECE gender statistics 2004

Gender & international

• Which issue or topic?• Which concept?• Which measurement tool?• Main problem for both: validity

– reduction or prevention of systematic error

• Analysis: extra = validity analysis• Interpretation = plus restraint by validity • Presentation = including reserves by limited

validity?

workshop UNDP/UNECE gender statistics 2004

More to learn

• In hand-out suggestions for further reading• Standard handbooks for students social

psychology• Look on the internet by using searching

machines: attitude, gender, survey (e.g. Ajzen)

• Search for sites on international surveys (e.g. European Social Survey) and research databases

workshop UNDP/UNECE gender statistics 2004

Evaluating both scales

• Gender dimension

• Inter-item consistency? Homogeneity?

• Valid multidimensionality?

• Quality of separate items?

• Scale quality

• Etc.

top related